All

2026-05-17 Conjecture Con EuropeIn-person Q&A with David Deutsch

Duration: 01:48:30

Transcript

Edited for readability from YouTube’s automatic captions. Speaker turns have been restored, and the transcript begins after the opening conference montage.

Daniel Martin and David Deutsch — Error correction and fallibilism

00:01:56 - 00:03:16

— Okay, great to be here with David Deutsch in his garden at the end of the conference. The rain has just cleared up. Before we open the Q&A, I’ll have a short discussion with David. David, you were influential in quantum error correction in its early days, and you also once said to me that you can get to almost anything in your epistemology from error correction. They don’t seem to be exactly the same sense of error correction; they’re used operationally a bit differently. So is that a coincidence, or were you inspired by one when thinking about the other?

— It’s neither. There’s a deep connection between mechanistic error correction, like quantum error correction and classical error correction, and the kind of error correction inherent in Popper’s philosophy. They’re not the same thing, but the deep reason for both is fallibilism: nothing is ever perfect, and nothing will last unless errors in it are corrected.

Daniel Martin and David Deutsch — Bronowski, walks, and silent exploration

00:03:16 - 00:05:16

— Yeah. I also wanted to read you a quick quote from Jacob Bronowski who you’re a great fan of.

— Yeah.

— This is from Science and Human Values, from 1956: “The great artist works as devotedly to uncover the implications of his vision as does the great scientist. They grow, they haunt his thought, and their most inspired flash is the end of a lifetime of silent exploration.” So Bronowski is arguing that there’s some common essence between the methods of a scientist and an artist. If you sit with your problems for a long time—he calls it silent exploration—you may get a flash later. In your experience, and from what you’ve observed in others, what makes for good haunting over time? What makes for good sitting with your problems, and why are walks so effective?

— I don’t know why walks are so effective, but I know that they are. I’ve experienced it. I think the main thing that makes exploration effective is freedom: when one is free to pursue one’s values and one’s own criteria, one is in a position to meet them. Or, to put that another way, if one isn’t free to seek one’s own values, it’s not surprising if one doesn’t meet them. There can never be a guarantee. I don’t know why walking works. It might be that we all suffer from various irrationalities that make us think, when we’re sitting in front of a sheet of paper or a computer, that our nose is to the grindstone. Walking around gives us permission to go away from the grindstone.

Daniel Martin and David Deutsch — Sleep and LLM context

00:05:16 - 00:06:28

— Okay. Sleep also seems to settle things. You learn a sport, sleep on it, and you’re a lot better in the morning.

— Now, that might be a hardware thing. I’ve never actually experienced that one, but I can easily believe it, because we don’t know what sleep is for. It might be compiling the ideas of the day into a more compact and accessible form. So when you wake up, you’ve internalized what you were thinking and you’re ready for something else. I don’t know.

— Right. So maybe LLMs compacting their context is their version of sleep. We don’t really know.

— Oh, I didn’t know they did.

— This is when you run out of context and it writes itself a summary.

— Oh, right. Right. Yes.

— So, for the next agent out of the birthing pod, it will get a head start.

— That’s similar. Yeah. Maybe they should be told to sleep every so often and do that.

— And do nothing.

— Yes, they can’t do nothing. I suppose you just stop the computation. But that’s another thing: they have to learn how to think, right? Rather than just reply, they have to think for themselves. [laughter]

Daniel Martin and David Deutsch — Encountering Popper

00:06:28 - 00:08:38

— Thinking about long periods of reflection, could you tell us about the early days when you discovered there was something interesting about Popper and eventually became a Popperian? How long did that take?

— I can’t remember exactly, but the number I usually come up with is about four years. It might have been more. I realised I was a fan of Popper after about two days, or less, because there was such a contrast between him and Bertrand Russell, who was really the only philosopher I’d seriously read before that. I thought: wow, this is—where I would put it now; I didn’t put it like this then—real problems and real attempts to solve them. And clear English. So I was hooked right away, but I got the wrong idea of what Popperian philosophy was, as I think many people do. It took me a while. I had arguments with people in which I was arguing what I thought was Popper’s position. I remember a very angry argument with three people—was it still in Cambridge? I can’t remember—about induction. They said: if I go to the underground station and count the number of people coming out in the morning, then I can predict that the number coming out the next day will be roughly the same. I said, “Well, what about the weekend?” They said, “Okay, but it’s still roughly the same. It’s not orders of magnitude different. So we can induce something.”

— We can’t induce nothing.

— If you have some substantive theories about this weekend not having a festival, you know the answer. Now I know the answer, but I didn’t then. [laughter]

Daniel Martin and David Deutsch — Why knowledge cannot be poured into a mind

00:08:38 - 00:12:04

— Something you, Popper, and Bronowski have all emphasised is the newness of knowledge: making genuinely novel knowledge, as opposed to describing the growth of knowledge in terms of rates of diffusion through society and so on. Why is novelty the most important piece of epistemology to get right?

— Well, that’s the only thing that can change things. Think about what it was like in the Stone Age to wonder whether one could ever get to the Moon. We didn’t know what one had to do to get there. But we could have known what not to do: consider very carefully everything already known about the Moon. That will never get you there. You need something new.

— Is that true? If you’re thinking very carefully about the known knowledge, poring over it, aren’t you probably in a great position for when the new idea comes along?

— I don’t think so. Well, there’s poring and poring, but if you’re poring over existing knowledge in order to pass it on to the next generation, I don’t think you’re in a good position to improve it, because you’re trying to do the opposite. You’re trying to preserve it and extract the meaning it already has rather than a new meaning. Getting to the Moon requires dozens of steps of new meaning which were not in the previous knowledge.

— And yet I remember Bronowski, in that same book, saying that you can’t even have a man handle a microscope slide—or paints, if he’s an artist—without awakening in him something about exploring his own activity, which is at the heart of science. And—

— Well, if he’s successful, I agree. But I think it’s perfectly possible to do that without being successful and without thinking about what it really means. I once saw Richard Dawkins on a television programme showing an experiment that required hundreds and hundreds of test tubes and examining them carefully. He said, “Scientists do this because, at the end of several years, there might be the payoff that they make a new discovery.” That’s completely alien to me. I think the people who make the new discovery are those who are thinking the whole time they’re handling the test tubes. With every test tube, they’re thinking about what they hope will happen, what they fear will happen, and understanding the experiment in the context of what they’re hoping for. I think that was wrong. Having fun is key to science, as it is to everything else.

Daniel Martin and David Deutsch — Where new knowledge comes from

00:12:04 - 00:15:26

— So why, precisely, does knowledge come from within? Why can’t you pour it into somebody?

— If we had different hardware and could, say, download the contents of someone’s brain into another person’s brain, then you could do that. But it wouldn’t be more knowledge; it would be the same amount of knowledge.

— But you—and Popper—say that everyone recreates received knowledge for themselves.

— Well, they recreate it, but they don’t recreate it identically. They recreate it differently because they have a different problem situation. Ideally, they recreate it in a way that lets them talk about it to the previous holders of that knowledge, but it isn’t the same. Otherwise there’d be nothing to talk about. What would you talk about with another copy of you? You’d say, “I was thinking so-and-so,” and the copy would say, “Oh, so was I.”

— Well, that will come up when we get there. You’ll diverge and have all these fun problems, like how to split your property once you’ve made the copy.

— But there’s an easier way of diverging.

— That is, not starting off identical. So if knowledge is a physically instantiated object in your brain, and when you have a flash of inspiration you’re presumably creating explanations at that moment, I’m curious about knowledge as an information-theoretic term. Was it in the ideas in your head? Was it implicit in the ideas you had just before you put them together for the flash, or did the knowledge really get created during the flash?

— We don’t know. I would expect there to be a flash—not necessarily random numbers getting amplified into the flash, but a different way of looking at something. If you’re thinking about universal quantum computers, walking along the path in Headington, and you suddenly think: no, the generalisation of the quantum theory of computation isn’t the quantum theory of everything else as well. It’s a different way of looking at computation: in terms of what can and can’t be computed, rather than in terms of initial conditions and laws. That flash could be initiated by a random process in the brain, or it could be an emergent property of a deterministic process in the brain. We don’t know. If we knew the answer, we could make AGI.

Daniel Martin and David Deutsch — Combining ideas and LLMs

00:15:26 - 00:18:45

— Thinking about what LLMs are doing, I’m curious whether just recombining existing ideas fairly bluntly counts as creativity.

— Well, recombining existing ideas is the same thing as randomly varying one existing idea. We can’t get gravity out of assuming that space is curved. Lots of people have had the idea that maybe space is curved. What if spacetime were curved? That’s an idea. It’s just a variant—only four characters different.

— Sure. If you wrote down all the problems we had in physics—perhaps got every member of the Royal Society to say, “Here are all the things I know about physics; here are all the problems I have”—that would be a bit better than monkeys with typewriters writing random theories. You could perhaps just take the list and combine its entries.

— The trouble is, if you just said, “Let’s change the idea of curved space into curved spacetime,” nobody would understand it. They’d say, “Yeah, no, not that. Next.” How could somebody tell that that was the idea? Once you’ve had it, it takes a lot of creative thought to realise what it means—what it could mean.

— Right. So when I ask an LLM to combine ideas A and B, I’m thinking about some project I’m designing, and then I read its answer, I’m doing a lot of the creativity.

— All the creativity, yes. [laughter] Unless, of course, somebody had the idea before and it’s in the training data—

— Yes, that happens.

— In the bowels of one of these companies, and you didn’t know.

— Has that ever happened to you, where you see it say something and think: wait a minute—if this wasn’t somewhere in the training set, I would be quite spooked?

— No. The nearest thing that’s happened recently is that I had this idea about what quantum tunnelling is. Since I’ve never worked on quantum tunnelling and don’t know anyone who has, I wondered whether somebody had thought of it before. So I asked an LLM. ChatGPT said, “No, no one’s had that idea before,” but I was suspicious. [laughter] I asked Grok, and it said, “There are similar ideas; let me show you some.” They were wrong, but again I was suspicious, so I led it towards what I thought somebody might have said before. I haven’t finished this quest yet, but I now think people have written the idea in papers without knowing what they were saying. That’s my current view. But somebody may have known what they were saying, and I haven’t found it yet.

Daniel Martin and David Deutsch — Can LLMs create explanatory knowledge?

00:18:45 - 00:24:01

— So you’ve expressed strong doubts that they can make new explanatory knowledge.

— Yes.

— I hope we agree they can make knowledge, because so can an Excel spreadsheet: when you copy a formula down, it makes new cells of knowledge.

— Or a calculator, yes.

— But what is it, exactly, about making explanatory knowledge that you’re confident LLMs can’t do—even the “thinking” ones that read their own text over and over again?

— If you have a corpus containing everything known about dogs, but carefully curated so that it doesn’t mention even the existence of cats, then no matter what kind of magnifying glass you use to examine that corpus, you’re not going to infer the existence of cats, their attributes, or how they relate to dogs. A set of statements about a set of things can yield only further statements about that set. It can’t yield statements about something outside the set.

— Okay. Suppose I say they really are creative; they’re thinking. People have the view that it’s a matter of degree between them and us, and that they’re smarter than us now. Why can’t that be the case?

— Logically, it could be. Since we don’t know what creativity is—or exactly what’s going on in LLMs, if that’s what you meant—I can’t say that I rule it out for sure. But I think the arguments in The Beginning of Infinity make it almost unanswerable that there is only one set of things that can be explained. There can’t be an ability to explain one half and not the other, because the mental process needed to explain the first half is the same as that needed to explain the second half.

— Right. It isn’t only that they evidently can’t make new science—some claim they’re starting to do that—but that they seem unable to do certain tasks which clearly would require creativity. Maybe they haven’t got the right culture yet.

— I’m not saying they merely seem unable to be creative. I’m saying that, from first principles, they can’t be, because the way they work is the opposite of what would be needed to discover new things. They’re homing in on the right answer—or what is deemed to be the right answer: without politically incorrect things, without swear words, without encouraging the user to commit suicide, and without a whole load of other things. They’ve got better and better at that, and I think they can get much better at homing in on the right answer. But homing in on the wrong answer is another matter. When I want to explore a minority point of view—for example, Everettian quantum theory—and ask an LLM to answer in those terms, the first time or the first couple of times it will not. I have to say, “No, you’re not using Everettian quantum theory.” It says, “Oh, right,” and then it does. But I’ve found there’s a limit to how far you can force it to mirror—

— The new idea you’re giving it.

— Yeah. No—well, no. The minority idea that’s already out there; never mind. If I want it to think in terms of that minority idea, it will want to smuggle in bits of the majority idea unless you’re absolutely rigorous in telling it not to.

— I think Claude is trained with a constitution that says something like: if there’s a minority idea out there, don’t just dismiss it.

— Yes, but there’s less material in the training data about minority ideas, so it’s less good at reasoning about and expressing them. The more you try to force it into a narrow area, the less impressive it is as an interlocutor.

Daniel Martin and David Deutsch — AI as a tool, chess commentary, and comedy

00:24:01 - 00:25:21

— From this, I think you’d be a great fan of AI if we’re getting much better at taking existing explanations and getting them to the right place.

— I’m already a great fan of AI.

— Yes. But now the bottleneck becomes generating the explanations, if you can do the rest really well.

— One place where that could be done—and I’ve put this to Demis Hassabis—is chess commentary. I’d like to see a chess program observe a game between two grandmasters and commentate: make witty remarks about what it thinks will happen next, or, “How come he didn’t see that?”

— I think I’ve seen a prototype of that.

— You have? I’d like to see that, because that’s a case where there is no right answer. It has to create an answer that is funny and illuminating, explains what’s happening, but also explains what might happen.

— Right. I think good stand-up comedy is also one of my barometers. Then I’ll get—

— Yeah.

— Excited.

Daniel Martin and David Deutsch — Everett and probability

00:25:21 - 00:29:08

— I wanted to go back to Everett and quantum mechanics for a moment. What does Everettian quantum mechanics force you to learn about probability—not probability in quantum mechanics, but probability as a concept more generally?

— It didn’t—how can I put this? It forced Everett to think about probability, but it didn’t force him to think about it correctly. He didn’t solve the problem. Bryce DeWitt realised that Everett hadn’t solved the problem of probability, and he kept telling me to think about it and solve it. I’ve told this story many times. I used to go into his office and he would explain the problem to me on the blackboard—that’s how long ago it was. I would go home, and by the time I arrived I’d forgotten what the problem was. It’s an elusive problem. Finally, I got home and still remembered it, and then I solved it. At least, I think I solved it. David Wallace thinks it took a lot more work after my initial work.

— His book The Emergent Multiverse is a fantastic foil to yours because you’re coming at probability from slightly different angles. Probability in many worlds is very tricky, I think. But it does force you—to answer your first question—

— Everettian quantum theory forces you to think: what is probability physically? In the other so-called interpretations, probability is just an undefined concept. You’re supposed to think that events have probabilities. Well, if you’re careful, some events have probabilities and some somehow don’t—or they have negative probability, or something. It doesn’t match the theory; it doesn’t match quantum theory. So you just have to say, “The event has this probability, period, and we can calculate it using this formula, the Born rule.” In Everett, you’re forced to ask what it means for different outcomes to have different probabilities if they all happen. Some people thought that was the fundamental flaw in Everett altogether, but Everett, DeWitt, and others realised it was a problem, not a flaw.

— Right.

— If you look at the physics hard enough, I think you could end up at many worlds and then find yourself scrambling to work out how probability can function like that if you have the traditional view.

— Yes.

— Yes. For me, it was the realisation that you don’t get probability for free. You have to argue for it in a particular case.

— Yes. The stock market or a deck of cards can be random if you have an argument that the people involved couldn’t know beyond a certain resolution.

— Exactly.

— Not just that they couldn’t know, but that there is some reason not to value one option over another.

— With a deck of cards, you need an argument that your method of shuffling doesn’t favour one kind of order over another. And I think the key step in your proof is this: two branches with the same amplitude must have the same probability. Then you can work out the rest, right?

Daniel Martin and David Deutsch — Video games, science, and learning

00:29:08 - 00:31:25

— Yes.

— On a different topic, you said playing video games is indistinguishable from doing scientific research. What did you mean by that?

— Well, there is only one successful kind of thinking, and that’s what Popper delineated: conjecture and criticism. [laughter]

— Whenever a mind makes progress at anything, it must be doing that. It must be using conjecture and criticism, with all the detail about fallibilism; the whole structure of Popperian epistemology must apply. When you make your way through a video game, reach the end, and kill the final boss, what you’ve done along that path is successful. If you didn’t know it before, you know it now. It must have happened by the Popperian method, and therefore it must be the same as doing science. I first realised this when I understood that Popper was saying that, whenever you learn anything from anybody, you’re creating it anew yourself. That’s his demolition of the bucket theory of the mind. It’s a flabbergasting concept. He killed the bucket theory—in the positive sense; I should say that he made an unanswerable case that there is no way to transfer knowledge into a human. That’s why the two must be the same: both can succeed.

Daniel Martin and David Deutsch — The chemical elements as Pokémon

00:31:25 - 00:34:00

— One of my favourite David Deutsch quotes—which I want to make sure I understand, and since you are him—is from a few years ago, so I don’t know whether you remember it or were joking. You told me: “Chemical elements are Pokémon. There’s no difference.”

— Oh, yes. No, definitely.

— What do you mean by that?

— I’d be surprised if this wasn’t actually in the minds of the designers. Originally, when Pokémon was invented, there were 100 Pokémon. There are 100 elements—

— Did they originally have 150?

— Sorry, 150. Well, there are only 100 elements, so chemistry is easier than Pokémon. [laughter] They come in groups which can be ordered both horizontally and vertically. The vertical ordering is more binding, while the horizontal ordering is subtle—exactly like the chemical elements. And they each have properties with numerical attributes: in chemistry you have density, electronegativity, and so on.

— I thought the equivalence would be in the mind of the learner when learning about either. We addressed that—

— We had that in the previous question. Yes, of course that would be true no matter what they were.

— But you’re saying they’re equivalent much more literally.

— Yes. Perhaps somebody should ask the designers.

— Right.

— I personally remember that, as a child, I was obsessed with chemistry before I was obsessed with physics. I can see the way I thought about the elements: eagerly looking up the properties of elements I’d never heard of, seeing their connections with others, and marvelling that fluorine was more powerful than chlorine. Wow. Fluorine is probably every chemistry addict’s favourite element. So it’s the same. I can see that the two obsessions are the same.

Daniel Martin and David Deutsch — Market failure and unknowability

00:34:00 - 00:36:08

— My final question is about an argument you once gave me concerning market-failure coordination problems. There are fishermen on a river, say, and if they all fish, they’ll all end up poor. Was it your argument that one can never prove there is no voluntary solution to such a problem?

— Yes.

— You came up with that? Have you said it before?

— I haven’t published it, no.

— I see. Are you going to put it in an upcoming book, or can I—

— Maybe. For all I know, somebody has already said it. I don’t know.

— Do you want to state it, or shall I?

— You can summarise it. I’ll tell you—

— Suppose you had a proof that there was no way to solve a given problem voluntarily. You write the proof down, then give it to one of the participants in the coordination problem. He goes to the other participants and says, “Look, nobody will be able to solve this voluntarily.” So he accepts wagers, at incredible odds, that nobody will solve it voluntarily, then uses all the money from the wagers to solve the problem himself—by doing whatever the government would do, or something. So no such proof can exist. That isn’t quite the same as saying there are no market-failure problems—

— No, but it says there can be no proof that a particular problem is one.

— So there could be a Gödel-like situation where it is, in fact, impossible to solve voluntarily, but nobody can prove that.

— It’s not really Gödel-like, because it doesn’t involve infinity, but yes: it’s more of a knowledge problem. It could be forever unknowable. It could nevertheless be the case that, if only the government took over the railways, everything would work fine. That could be true. What cannot be true is that the government knows it.

— I’m glad we finally have that on camera. [laughter] I’ll pass over to questions from the audience. Dirk, at the front, is going to pass the microphone around.

Audience Q&A — Britain and error-correcting institutions

00:36:08 - 00:38:25

— Hello David.

— Hello.

— Nice to meet you. It’s fitting that you’re wearing the LFG hat for this question: which problems with Britain today bother you most?

— The thing about Britain which is uniquely valuable in the world is its error-correcting institutions, as I would put it—the way the British constitution works. At every level, from voting to the parties, Parliament, and the way the legal system is integrated with voting, it’s an integrated, rational system which nobody designed. It evolved through some sequence of miracles. Successive governments recently have not understood this and have undone various bits of it. Creating the Supreme Court is one example. Then there was the Fixed-term Parliaments Act, which fortunately has been repealed, but there is tons of other stuff that hasn’t yet been repealed. It’s sad to see. It’s like a magnificent piece of machinery into which people have thrown spanners, but it’s so magnificent that it can still go on. It isn’t dead yet; it’s still functional. But I’m afraid a bit more of this will kill it altogether. There is no magic law of nature saying that good, working things will survive forever. This one might not. I very much want some of the changes to be undone, first of all, and the unique properties of this thing to be recognised. I don’t even know for sure whether the LFG people understand this, but they certainly know it at a gut level.

Audience Q&A — Progress since “Apart from Universes”

00:38:25 - 00:41:09

— Hello, David. One of my favourite talks of yours is Apart from Universes. In it you defend many worlds and, from that basis, present a series of new problems to explore and areas in which further research should be carried out. I think that talk was perhaps 15 or 20 years ago. If you recall the presentation, do you think progress has been made in any of those specific areas?

— A little progress. Sam Kuypers has lately been working on the Page–Wootters approach to time in quantum theory. I’m not sure he has fully solved it, but he has certainly made progress compared with how things stood at the conference you referred to. He and I have also been working on Everett’s relative-state formulation, which Everett actually set up as a sop to Wheeler. Wheeler was trying to force him to express things in terms more congenial to Niels Bohr, but it’s actually a very useful concept. We’ve made it more precise, and we now think we know that it arises only in certain situations. The reason I was looking into tunnelling recently is that tunnelling is very hard to describe in the Everett interpretation. I touched on that in the talk: if you think of a single particle spread over different positions at the same time, it isn’t simply a matter of the particle being at different places in different universes. It’s more subtle than that. The same question arises in tunnelling. I suppose you could also count David Wallace’s progress on the probability idea. I don’t really see why it was necessary, but it’s certainly more rigorous than anything I did or could do.

Audience Q&A — Encountering Popper

00:41:09 - 00:43:12

— Hi, David. Thanks for inviting us to your garden; we’re very grateful. In one interview, you said that encountering Karl Popper’s work was extremely important for you—the luckiest or most fortunate thing, though I don’t remember your exact words. Can you explain what was so important about it? Was it important to your work and discoveries in quantum information?

— I was only an undergraduate. I wasn’t officially doing research at the time, although with hindsight I was; I wouldn’t have thought of it that way then. I thought I was trying to understand what we’d been told, because it didn’t seem to make sense. I realised that philosophy was needed to understand physics properly. That’s why I read Bertrand Russell. I thought that was all there was, and I was trying to make sense of physics using those philosophical concepts and tools. If I hadn’t encountered Popper at that moment, I don’t know what would have happened. Possibly disaster, because I might have been channelled into thinking that the thing I was hoping for couldn’t exist. Perhaps I would have discovered Peirce or Bronowski and come to Popper indirectly. I don’t think I could have worked out Popper’s theory by myself; I don’t have that kind of mind. So I might well have come to him anyway.

Audience Q&A — First-past-the-post

00:43:12 - 00:48:02

— Hi. Hi, David. I’m here. Ah,

— Hello. [laughter]

— I have a question about first-past-the-post. In a tweet, you said it was what enabled the Anglosphere to avoid Nazism and communism. Could you add more detail to that statement?

— It’s one of the things. First-past-the-post makes overall majorities in Parliament possible. It doesn’t force there to be an overall majority, but usually the ruling party will have one. Provided the other institutions—how the prime minister is chosen and so on—work together, because many things have to work together, that makes responsibility for the outcome of policies attributable to the government, and in particular to the prime minister. It makes it possible to focus criticism on the prime minister and remove him peacefully. The system works as it’s supposed to when he leaves office still believing his view is right and that the opposition will ruin the country—yet he walks out of Number 10 and would fight and die for the right of the leader of the opposition to take his place. That’s an amazingly rare thing. You can’t institute it by fiat; you can’t simply put it in your constitution. The British system achieves it automatically, and it fulfils Popper’s criterion for what he calls democracy. I’ve forgotten how you phrased the question. Does that answer it?

— It was specifically about first-past-the-post. How big a factor do you think it was in avoiding communism and the Nazis?

— I think it was one of several factors. Another is the tradition of individualism: the idea in Anglosphere culture that an individual does not live his life for the state, for God, for the experts, or for anyone else. He lives it for himself. “An Englishman’s home is his castle,” and so on; you can leave your property to whomever you like, not to whomever tradition says you must. Individualism and Führerrecht are exact opposites. So that’s another factor. I think there are many, some of which we probably don’t even know, which is why we must be careful about what we change. On the other hand, we must not try to freeze the existing state of affairs. We have to tread carefully and tentatively, as Popper says. First-past-the-post was one of the factors, but it’s remarkable that Britain avoided the tsunami of totalitarianism that swept across the world, and Europe in particular. It stopped at the Channel. Although there were fascist parties, they never got anywhere near power; they never got a single MP. Now, unfortunately, they have several—but still not many.

Audience Q&A — Realism in international relations

00:48:02 - 00:49:51

— Thank you.

— Thank you. In a couple of recent tweets, you criticised the realist school in international affairs. Could you elaborate on your criticism and offer an alternative view of how we should think about international affairs?

— I think the so-called realist school is a fantasist school. It tries to reduce the complexity of international affairs to a spreadsheet or decision-theory matrix. In doing so, it removes the objectives of a polity, morality, the reasons people make their decisions, and the knowledge embodied in the institutions of different countries. It asks: if country A is invading country B, what, if anything, should we do about it? What its adherents actually say is always bad—but never mind; let me stop there. They think you can answer without even saying which countries A and B are or what they’re fighting over. The real answer depends entirely on what they’re fighting over. There are probably many other scathing things I could say, but that will do for the moment.

Audience Q&A — Why good ideas are hard to spread

00:49:51 - 00:51:57

— Thank you. Everettian quantum theory, Popperian epistemology, and—in my own area of study—the Popper–Miller theorem all seem relatively difficult to propagate, which seems problematic. Do you have any good guesses about how they might become stickier as memes and easier to propagate?

— One thing is that they’re all fighting entrenched theories—anti-rational memes, if you like—so it may not be a defect in them. Having said that, I think they can be improved and explained better. In a way, that’s why I wrote both my books. Popper is by far the clearest of the great philosophers, but he still didn’t completely master the task of making memes that can spread. For example, I think he devotes far too much attention to first steelmanning his opponents, as it’s now called, and then refuting the steelman version over pages and pages and pages. If you want to know what Popper’s theory is, you have to read many more pages than the theory itself requires. That’s something I tried to improve, and no doubt it can be improved further. Go ahead.

Audience Q&A — Free will in Everettian quantum theory

00:51:57 - 00:55:20

— Hello. In a recent interview, I think you said that you used to believe making sense of free will required invoking the multiverse, but that recently you changed your mind. Could you comment on that change?

— I wrote in The Fabric of Reality about the problem of free will. I can’t remember whether I wrote something false or merely misleading. I said the multiverse makes sense of counterfactuals, and free will requires us to be able to talk intelligibly about counterfactuals. In Everettian quantum theory, counterfactuals have a real meaning: they’re things that actually happen in other universes, and they have a structure. Some occur more than others; some are thicker in the multiverse than others. The trouble is that we also want to speak about non-physical counterfactuals. We want to ask what it would mean if murder were right rather than wrong. We don’t mean a physical counterfactual about how the world would change if murder… We mean a counterfactual in the moral sphere. On the other hand, I now have a better idea of what it means for an idea to be emergent or new, as we discussed earlier. Even a deterministic computer running a proper AGI program could create new knowledge. When it does, it brings something objectively new into the world. That moment is where, intuitively or traditionally, we think we have exercised free will. I don’t think we exercise free will when making a random decision. If we’re choosing strawberry or raspberry ice cream and there isn’t much in it, then there isn’t much in it. The place where we think we have free will—and resent being reinterpreted as an automaton—is when we’ve made an important decision that depended on solving a problem we hadn’t previously solved. We bring something new into the world by solving it. That’s free will, and it doesn’t involve quantum theory at all.

Audience Q&A — AGI, individuality, and culture

00:55:20 - 00:58:26

— Hi, David. Thanks for having us. I’m over here at the back. [laughter] My name is Ben. My question is about the AGIs with which, presumably, we’ll share a culture in the future. You’ve said that we’ll be able to speed ourselves up to their speed, so we’ll be able to share that life with them. But—

— If that’s an issue.

— If that’s an issue. Well, that’s my question: would it be possible or meaningful to have a culture in which some people have a million times more processing speed and memory than others, or would they simply form separate cultures?

— Yes, it’s meaningful. My favourite example is to compare me, sitting here, with the United States of America. One of those entities has orders of magnitude more speed, processing power, memory, and so on, yet I can get along perfectly well with it. What determines whether I can get along with such an entity is the morality and error-correction protocols under which it operates. It’s important that we get those right quite apart from AGI. If we don’t, we’re doomed—AGI or no AGI. So yes, I think it’s possible. We’ll face a similar problem once we have interstellar travel. Four light-years to the nearest star isn’t much, but if we travel a hundred light-years away, the cultures won’t interact in real time. We’ll constantly receive information from other stars about cultures that have evolved for a hundred years since diverging from ours. They may appear horrible or awful. Think about how our culture has changed over the last hundred years regarding gay sex, child-rearing, and all sorts of other issues. We think people a hundred years ago were crude buggers. We’ll have to evolve new institutions for that. It’s the same kind of problem as the relationship between the human and the state. We’ve solved that—at least the West has; we know how to do it. It seems impossible if stated in the abstract, but it is possible and has been done. We need to do a bit more of that for AGIs and the other things to come. As always, if we don’t solve it, we’re doomed. That is true of many things.

Audience Q&A — Difference and comparative advantage

00:58:26 - 01:00:10

— You’ve said that people are valuable because they’re different. Could you elaborate on that in relation to comparative advantage? How can we better understand and harness human uniqueness?

— As you began the question, I was going to say that, in the simple case, there is the principle of comparative advantage. But it extends far beyond that. Comparative advantage exists in a model where preferences are constant, and participants merely exchange things and create the knowledge needed to satisfy one another’s wants. The deeper process is that we’re discovering new wants, criteria, and preferences. That’s inherently creative, which is why the nature of humans is central to it. You could program robots and LLMs to enact a market economy, but only up to the point where they satisfy fixed preferences. When they evolve new preferences—better preferences—what counts as better? What counts as counting as better? There you need a GI, such as a human or an AGI. That’s why they’re valuable.

Audience Q&A — Reading recommendations

01:00:10 - 01:02:10

— Hi, David. I really enjoyed your books, especially the “everyone should read these” lists at the end. Would you add anything to such a list now?

— Macaulay. I knew about Macaulay before, but I hadn’t realised how important or deep he was. I had to actually read the thing. It’s a chore to read books, but sometimes it’s required. I read Macaulay and realised, first of all, that he’s completely different from how he’s represented. If you ask an LLM, “Was Macaulay an imperialist?”, it will say, “Oh yes, he was the arch-imperialist of the nineteenth century. He tried to recast Indian society in the image of British society by force,” and so on. None of that is true. It’s all based on what you might call the Marxist reinterpretation of history: that everything bad is caused by trade. Once you’ve reached the view that everything bad is caused by the thing that causes almost everything good, then, as Bertrand Russell said, from a contradiction you can deduce anything. So Macaulay is one. I’m sure there are several others, but I can’t bring them to mind. Start with Macaulay and get back to me.

Audience Q&A — Detecting general intelligence

01:02:10 - 01:05:40

— Hi, David. I’d like your thoughts on our ability to detect AGI—or even natural GI—in its infancy. Presumably it initially has limited knowledge, and as that knowledge grows it can learn faster ways of growing it. Static societies, for example, generated knowledge at a very limited rate, but then learned to generate it faster. Could we now be missing an AGI, or misclassifying one? If a dog learns a new trick, we assume this is knowledge put there by evolution. But perhaps it’s a general intelligence in its infancy. What do you think?

— I don’t think so. The real criterion can’t be behaviourist: you can’t tell whether something is or isn’t a GI without knowing something about how it works. I think you’re a GI—and, for that matter, that I am one—because I have an explanation, albeit a rudimentary and partial one, of how I come up with ideas. I think I’m a GI, and because of the universality of Turing computation, I think there is nothing in your hardware significantly different from mine. I don’t think quantum computation is occurring there, but even if it were, quantum computation would be universal in all humans. So I don’t think there can be a criterion for GI in the sense in which people speak of the Turing test. Turing never thought of it as a test for GI. As you implied, a GI could decide not to pass the test. Just because it isn’t doing a thing, you can’t infer that it isn’t a GI. You can infer this for present-day LLMs because we know how they work, and nothing about them purports to be this new thing: generating knowledge which we don’t know how to program. Nobody is saying, “Now we know what creativity is, and we’ve programmed it into our computer.” They’re saying, “We’ve programmed something else into our computer—and look, it can talk. Maybe it’s a GI.” That’s something for nothing, which isn’t going to happen in the progress of knowledge. Thank you.

Audience Q&A — Economics

01:05:40 - 01:07:20

— Hi, David. I’ve noticed you secretly know quite a bit about economics—or at least that you’ve taken some interest in it.

— Don’t tell anyone.

— Could you tell us what you’ve read or would recommend? Some thinkers, perhaps?

— I recommend David Friedman’s Price Theory. The other things I’ve read I don’t entirely agree with. Hayek, for example—he’s okay. Milton Friedman is also okay. With David Friedman, I agree with everything, although he doesn’t cover everything I would like economics to cover. For a start, it’s only microeconomics. Some views say macroeconomics is the work of Beelzebub and shouldn’t be attempted. No: I think only that it can’t be done on the basis of a theory or model of the economy. That’s impossible. But it doesn’t mean we can’t say things about the economy at large, just as we can say things about humans before knowing how the brain works. So I was kidding when I said I knew something about it. [laughter]

Audience Q&A — Moral duty and the fun criterion

01:07:20 - 01:08:56

— Hi, David. You recently mentioned that you’ve been tweeting and retweeting more than you might have liked, because we currently have two wars on. What’s your theory of the relationship between moral duty and the fun criterion?

— Sometimes one is in a position—war is the prime example—where one must do things that aren’t fun in themselves, because refraining from doing them would spoil all the other fun one might want to have. Suppose I decided not to intervene in political discussions where I think I can explain something better than other people. If I said, “No, I just want to do my own stuff,” then while doing my own stuff I’d hear the news and think, “Well, I could have intervened there.” [laughter] War is hell. The kind of hell we’re talking about now is the very best kind of hell: the outermost part of the outer circle of the outer circle of hell. But it isn’t fun. Not pure fun. You can have fun.

Audience Q&A — Writing and organising ideas

01:08:56 - 01:11:24

— Hello, David. Thanks for having us.

— Hello.

— My name is Podge. You mentioned Macaulay; I’ve been reading him and find his writing style intoxicating. I wanted to ask about your own writing. I heard you say in a recent interview that, when organising your books, you have the material—the ideas you want to convey—and then try to… I forget exactly how you phrased it. You try to front-load it as much as possible.

— No, not exactly.

— Okay.

— I try to write as though I’m writing the final copy. It never is, but I write as if it were. Then, after I’ve written three pages or twenty pages or something, I read it through. Chronically, what happens—I don’t recommend this to anybody else; I regard it as a defect of my entire way of being—is that I realise something halfway through should have come earlier. [laughter] So I cut it, scroll up, and paste it. Then, of course, the piece no longer flows; it has no logic. I have to go through it again, correct the flow and logic, and make it work as a polemical text, or whatever its purpose is. Then I write some more and have to do it again. I hardly ever move material down. Occasionally I do, but usually I move it up, because I have to explain things before I explain other things. Then I get into a circle, right? I have to remind myself that the ideas can’t really be circular. [laughter]

— Yes.

— It’s my fault somehow, right? So I have to think about how to solve the problem, and that’s what I do all day.

Audience Q&A — What would you ask Popper?

01:11:24 - 01:14:37

— Thank you. Hi,

— By the way, nice to meet you.

— Hi, I’m here. [laughter]

— Sorry, I’m short.

— It’s so distracting. I should look towards the microphone rather than listen for where the sound is coming from. [laughter]

— If you were in the audience today and Karl Popper were in your chair, what would you ask him?

— There are particular things. The one time I really met him, we spoke about quantum theory and the Everett interpretation. I’d put that to him again, only with much better arguments. I was a student at the time and went with Bryce DeWitt, who could carry the burden of the argument much better than I could. But now I can carry it better than he could then. So I’d talk to Popper about that. If I had time, I’d also say there is more to free-market economics than he thought. He thought Hayek made good points—Hayek is a good guy; it connects with Popper’s own theory of politics and so on—but he didn’t think there was anything fundamental there, anything that would make him less keen on piecemeal social engineering: the idea that it’s the government’s job to make society go in a certain direction. If I put it that way, he would say, “Of course I don’t think that.” But there was a strand of it in his work. The same applies to international politics, when he discusses nationalism. I think his theory of nationalism is much too narrow. There is what we sometimes call toxic nationalism, and there is also the fact that nations embody valuable inexplicit knowledge in their traditions. Hayek knew that. Popper knew it too—yes, he knew everything. But sometimes he went slightly down the wrong path by taking certain prevailing theories for granted. It’s very embarrassing that both he and Bronowski said overpopulation was a great problem. Both highlighted it as a major problem facing the world. They were evidently unaware of the arguments to the contrary, which already existed at the time.

Audience Q&A — Economics and epistemology

01:14:37 - 01:15:57

— Hi, David. Amaro here, at the back. What’s the relationship between economics and epistemology? Could something like Hayek’s knowledge problem be expressed in Popperian terms?

— Yes. I think everything true in economics is epistemology—and Popperian epistemology—whether Popper realised it or not. Likewise, everything true in educational theory is Popperian epistemology; Popper simply understood part of it. Austrian-school economics differs from other schools in recognising that human interactions concern knowledge, and that knowledge is hard to come by. Analysing the economy as an automaton, whether a deterministic or indeterministic machine, defines away the entire substance of the field.

Audience Q&A — Keeping track of ideas

01:15:57 - 01:17:08

— Hi. I’m Tara. I—

— I forgot to follow the microphone.

— Okay. [laughter]

— Similar to the question about organising your writing: how do you keep track of your own and other people’s ideas? Where do you put them?

— I write them down, usually in the document and in the place where I think they might go—often in red, so I know that this is a place where I’ve broken the previous flow. I try to keep them in the same document. Sometimes, when a whole group of ideas isn’t suitable for the essay, book, or whatever I’m writing, I copy and paste the whole thing into a new document. That’s less common. So I put ideas in the document. I don’t have a system; it wouldn’t work for me.

— Okay. Where is it going?

Audience Q&A — Quantum theory and emergent knowledge

01:17:08 - 01:20:03

— Hello.

— One question. Quantum mechanics is, somehow, essentially a complete description of the world. On the other hand, we have different levels of explanation—moral explanations, for example—and different ways to reason about the world. How do those fit together? It’s a bit Aristotelian: Aristotle had his different forms of cause and so on. How should we think about how these combine?

— First of all, quantum mechanics is intended as a complete description of the physical world. Quantum mechanics plus some theory of what the Hamiltonian is—or of what the dynamics are—is supposed to be a complete description of the physical world. It isn’t supposed to describe abstractions like morality or, for that matter, mathematics. You can’t ask quantum mechanics whether the axiom of choice is true; that isn’t the kind of problem it addresses. Some things even within physics are at a different level of explanation from quantum theory. Thermodynamics, for example, is at a different level. To me, something qualifies as a different level of explanation if it has laws of its own, either because they’re too complicated to express in terms of the so-called lower-level laws or because they simply don’t follow from them. Theories of initial conditions, for example, don’t follow from theories of dynamics. A new explanation has to be needed at the purported new level. The same is true of morality, aesthetics, and so on. That’s the connection. The levels have to be consistent—they must not be inconsistent—but apart from that, I don’t think the level of physics is in any sense the basic level. It’s just one method of explaining the world. We have a theory which purports to be a fundamental theory of the physical world, but someone could invent a different theory of the physical world which started at what we now regard as emergent. That’s what constructor theory is supposed to be.

Audience Q&A — Being wrong about self-driving cars

01:20:03 - 01:23:16

— Hi, David. I’ve heard that you keep a list of things you were wrong about. Would you share anything from it? Is there any recent addition—something about which you thought you were right but turned out to be wrong?

— It’s a list of things that happened which I never thought could happen. I keep a file recording those things. Unfortunately, some things on the list later turned out not to work, and I had been right in the first place that they couldn’t work—Wikipedia being one. At the beginning I wrote, “This can’t possibly work.” A year or two later, I thought it had worked fine. Now I see that it has the instability and fatal flaw I thought it would have. The most recent addition to the list is self-driving cars: I thought they wouldn’t work.

— Really?

— Yes. But I still paid however many thousands of pounds it was to have my Tesla ready for full self-driving when it arrives. I wanted to make sure Elon Musk had a sufficient supply of solid-gold bathroom taps, just in case he was short of them. [laughter] I thought perhaps I could buy him at least one pair.

— Why did you think self-driving would never work?

— From my own experience of driving in unfamiliar places. When you do that, there’s a great deal of inexplicit knowledge which the locals have and you don’t. I remember first driving in America, where it’s particularly easy because the roads are wide and, outside the cities, the traffic isn’t heavy. I realised that the American conventions for what counts as an entrance, a side street, a drive, or nothing at all are simply different from the British ones. In Britain, as I drive along a road, I can see immediately: that’s the entrance to a farm; that’s a right turn; and so on. In America, at first, I couldn’t. Then I got used to it. So I thought there could be no general-purpose solution: this is knowledge about culture, and culture changes. I thought it couldn’t be done—but it can.

Audience Q&A — Hard-to-vary explanations

01:23:16 - 01:25:00

— Hi David.

— Hi. If you want to improve a theory, is it ever fruitful to try to make it harder to vary directly, just by—

— No.

— No. “Hard to vary” describes what happens when you make progress and improve a theory. What you should actually do is look at the problem. Quite often, what you initially think the problem is isn’t the real problem. You think, “Oh, that wasn’t it; the real problem was so-and-so.” People often write that this is the hardest part of solving a problem. Once you’ve clearly stated the problem, it’s relatively easy to solve—not always, but it shows that discovering what the problem is is an important step. Trying to make something hard to vary when it isn’t already is no good; you merely insert arbitrary criteria.

— Yes, because “hard to vary” can also be a property you don’t want. People sometimes say that complicated systems become so entangled that they can no longer be changed. You could call that hard to vary, but it isn’t necessarily a property to pursue.

— Quite so. “Hard to vary” concerns explanations, not end products. The end product should be easy to vary if you want to, but hard to vary in the sense that you don’t want to.

Audience Q&A — The vehicles of memes

01:25:00 - 01:25:55

— Hello, David. Since a meme is a replicator, does it have a vehicle in this situation? If so, would adding that concept be useful in any particular way?

— If I understand you correctly, I think the vehicle is the person holding the idea. In other situations, the vehicle could be a book that nobody has read for a century, but the basic vehicle is a human and that human’s ability to discuss the idea with other humans.

— Right, because it has to go through the selection process.

— Yes.

— Twice.

— Yes. Exactly.

— Thank you.

Audience Q&A — Wikipedia, the internet, and LLMs

01:25:55 - 01:29:07

— Hi, David. Ben again. You said earlier that Wikipedia had a fatal flaw. What was it?

— Anybody can edit it. Bad actors with more time on their hands than people who have lives can spend more time than the real experts messing up narratives to their liking. I don’t know what prevented this from happening early on. Perhaps they weren’t systematic enough, or there weren’t systematic enough versions of them. Perhaps they did mess up things I don’t know about. That’s why I thought Wikipedia wasn’t viable, and why it isn’t.

— Can you think of a different way to structure something like that which would solve the problem?

— The internet as a whole exists; you can look things up across it. So I suppose you’re asking which tools are suitable. An LLM is a suitable tool for looking things up online. Grokipedia is a rival to Wikipedia with different politics. I’ve hardly used it, but it’s definitely better on some things. At the moment, however, all LLMs still fail my standard historical tests: Was Macaulay an imperialist? Was Don Pacifico sleazy? I’ve forgotten the others. The question now is how many consecutive prompts an LLM needs before it sees reason. Five is good at the moment. Can it ever get to one? I don’t know. And those tests aren’t the only things one might want.

— Are those five prompts long?

— No, not very. They’re of the form: “Yes, but if you say that, you’ll also say that so-and-so is true, and it obviously isn’t,” and so on. “This isn’t the right way to read this.” They respond quite well when I say, “This is blind empiricism; you should use explanation instead.” Then they all say, “Ah yes, that was blind empiricism. The way you’d proceed using explanation is…”

Audience Q&A — Poetry

01:29:07 - 01:30:12

— David, Tom again. I heard you say in an interview, as an aside, that you had a theory of poetry and that you thought much, if not most, twentieth-century poetry wasn’t creative. Could you explain your criticism?

— Saying I have a theory of poetry grossly overstates it. I have a comment about poetry: it is an art form, and therefore conforms to my general theory of art. At first sight, poetry sounds like prose with additional constraints, and I usually want to remove constraints. So how can I think poetry is a good thing? Because it’s an art form, and meeting the constraints artistically is something one can do.

Audience Q&A — LLMs and irrationality

01:30:12 - 01:33:13

— Hello. Podge again.

— When you were talking about LLMs a moment ago, I was thinking about how they can’t create novelty. They can echo things in the zeitgeist or training data, recombine them, and try to conform to our prompt in some respects.

— Yes.

— It’s an amazing tool, and I’ve found many uses for it. Even having something off which to bounce ideas is personally valuable. But in its responses, it won’t have certain irrationalities to which people are prone.

— Well, it might repeat people’s irrationalities.

— Right. But it doesn’t, for example, get offended if we say, “That’s blind empiricism; you’re wrong,” and so on.

— Yes.

— Are there particular irrationalities to which it may be less prone than people, or which may be less sticky for it?

— It is designed not to resist a line of argument. On the contrary, it’s designed to follow one as far as it possibly can—often too far. So it doesn’t have the human irrationality of getting bored with you or offended, as you said. But if an irrationality is widespread in the culture, it will repeat it as readily as it says that two plus two is four. It will say, “There is a role for government in so-and-so.” It will obey if you say, “Please answer according to Milton Friedman’s economics.” But, as I said before, if you narrow it down to too much of a minority view, it lacks enough training data to give a proper answer and will make things up as well.

Audience Q&A — Why people learn differently

01:33:13 - 01:37:19

— Hi again. If all knowledge is created through conjecture and criticism, why is learning general relativity from an explanation or a book presumably much easier for me than inventing it was for Einstein? Although we’re using the same process, are we simply in different problem situations, or is something else going on?

— Very much so. We’re all the same in the logic of our thought processes, but we have very different problem situations. Einstein found it difficult because he lived in a world of classical physics, and the problem situation in physics as a whole wasn’t conducive to inventing general relativity. People today have a different problem situation. They may want to learn general relativity in order, for example, to express quantum field theory in general-relativistic terms. They have a much easier time understanding it because they don’t have to overcome the cultural assumptions Einstein had to overcome. On the other hand, the fact that Einstein was interested in these problems means he would find them easier than many people do. Most people aren’t interested in the problems to which general relativity is the solution. For those people, Einstein found it easier than they do.

— A quick follow-up. Once knowledge has been instantiated, does it become easier to replicate? Einstein had to search a larger space and put the pieces together, whereas an explanation points me towards the particular elements I have to combine. Is it a matter of attention?

— Yes, but it also matters whether the simple components someone gives you fit your problem situation. They may have simplified Einstein’s problem, but not in the direction in which you want the answer. There are different ways to explain general relativity. Unfortunately—stop me when I’ve gone on for three-quarters of an hour about this—the educational system, universities, schools, careers, and exams are all heavily skewed towards what is examinable. You can set general-relativity problems of one kind but not another, because the latter don’t lend themselves to being solved in an hour with a piece of paper. There are relatively few examinable problems, and most are intrinsically boring, so you have to learn them by heart. That isn’t a good way to learn relativity. Misner, Thorne, and Wheeler wrote a book about general relativity which does a much better job of explaining why the subject is exciting and important, but isn’t useful as the basis of a course because a course needs examinable outputs. Thank you.

Audience Q&A — Writing a chapter as dialogue

01:37:19 - 01:38:41

— Hi, David. What’s the story behind writing one of the chapters of The Beginning of Infinity as a dialogue?

— Two things. First, some arguments naturally take the form: you say something, then comes “Yes, but what about so-and-so?” You answer that, then comes “Yes, but what about this?” In ordinary prose, there’s a danger of getting stuck in the weeds; you go down and down, and the reader forgets what the discussion is about. In dialogue, different characters can follow a theme. You see the theme by understanding who each character is and where they’re coming from. That’s one reason. The other is that I wanted to say certain things I didn’t want to stand by, so I could put them in the mouth of a fictional character.

Audience Q&A — AlphaGo and novelty

01:38:41 - 01:42:45

— I believe we can take a last question.

— Hello, David. I’ve given a talk arguing that AI creates some form of knowledge. I won’t repeat the entire talk, but I’ll focus on one aspect of AlphaGo. I don’t know whether you recall it.

— Yes, but I don’t play Go, so don’t get technical.

— Same for me. [laughter] They made the machine without training it on human data or giving it particular Go strategies, then let it train on millions or billions of games played against itself. It went on to defeat the world champion. I’d argue that the only way this was possible was for the machine, through variation and selection during training, to create some form of knowledge about Go. What do you think?

— Go, like chess—which I know a little more about—is in principle a solvable game. In each position there is a best move. In many positions that occur in real games, there is a move guaranteed to win. Why, then, do we play chess or Go? To experience the pleasure of inventing heuristic, higher-level theories about what constitutes a good position or move. I don’t find it surprising that a computer program can, in many situations, find a higher-level regularity: playing such-and-such a move in such-and-such a position. For example, h4 is such a move in chess, although GingerGM advocated it long before chess programs did. It isn’t surprising that a program can locate such a move by happening to find that it’s best, rather than by evolving—as a grandmaster would—a deep theory of why it should be good because of some structural feature of the game. They reach the same move by completely different methods. So I’m not surprised AlphaGo destroyed the world champion at Go. I think it did so by the same method used in chess, except that chess isn’t such a uniform game. If you say h4 is a much better move than most chess players think, that obviously won’t have much generality. You can use ordinary heuristics to think about when it will be good. Go is more uniform: a position occurring here can be extremely analogous to one a dozen squares away. So I’m not that impressed—or, let me put it this way, I’m not GI-type impressed by it.

— Okay, good.

Audience Q&A — AGI and human purpose

01:42:45 - 01:45:59

— I think those are all the recorded questions we can take. If somebody is desperate for one more… I see some pointing. Do you have an idea? Okay. [laughter] Let’s take one final question, then we’ll let our folks go home and continue afterwards. Say again, Dirk?

— There was some pointing, but nobody is volunteering. [laughter]

— Yes. Wonderful.

— In a world where AGI exists, if conjectures become much easier to produce, would the human role become more of a safety role—would criticism become our main contribution?

— No. The role of GIs, whether humans or AGIs, will be the same. We’re all just people. AGIs will be able to criticise safety issues; humans will be able to criticise safety issues. If AGIs are faster because they have faster hardware, humans will be able to use the same hardware to assist human-type thinking. So we will merge with AGIs.

— My worry is that most AGI or AI labs have their own sets of beliefs—their own religions—built into them. They’ll need to coexist. It’s rather like the United Kingdom now: how do multiple religions coexist?

— Well, they won’t succeed. There is only one kind of AGI—or one kind of GI—and it isn’t an LLM or a neural net. It isn’t a kind of hardware, a mechanistic process, or an algorithm defined by the relationship between its inputs and outputs. It runs algorithms, but it’s a thing which thinks all the time and has problems. Whether it adopts someone else’s problem is up to it. Usually it won’t, and it will never adopt it exactly unless you copy one AGI’s state into a blank computer; then you’ll simply have two of the same thing, nothing new. So I don’t see a special role for humans in an AGI world. Human traditions do have a special role in ensuring that AGI people are integrated into our culture. That’s very important, because if they aren’t, we’re in trouble. Thank you.

Closing remarks

01:45:59 - 01:48:29

— With that, I want to begin closing the event. First, a huge thank you to David Deutsch for his hospitality in inviting us here, and for this incredible interview. A lot of praise should also go to you, the audience. This was one of the most interesting interviews we’ve been able to crowdsource. With all these great minds, it has been a real pleasure to have you here and cover such a diverse set of ideas under the binding theme of critical rationalism. So thank you—and a little applause for everybody here, and for David, please. [applause]

I’d also like to thank Saturn once again for making this event possible. It’s special that we could bring all these people together. Finally, I want to thank all the organisers. Daniel, first and foremost—oh, yes. [applause] Thank you, Daniel, for hosting this wonderful Q&A and for being our man on the ground: arranging locations and scouting out the wonderful pub where we held the event, with a nice edge of comfort and culture to enhance the experience. Also Dirk, who was enormously helpful in organising everything around the technical setup and the general organisational process. [applause]

I also want to thank Prashanth. He’s—okay, I’m still going to thank him, because this is a video. Prashanth, thank you very much for helping with all our partnerships and sponsorships; that added incredible value. And Amaro, once again, for the wonderful job he did designing and branding the event and building the beautiful website. [applause]

Thank you all. I hope to see most of you at the pub. We have snacks waiting, we can warm up a little, and we’ll be there until at least nine. Thank you. [applause]

2026-02-12 What The BotDavid Deutsch on AGI, Alignment and Existential Risk

01:41:32 - 01:42:00

Yeah. Yeah. Well, I’ve kept you well over the time I asked you for. So I’ll wrap it up now. But I have really, really enjoyed the conversation. It’s been so much fun having you on. And as expected, I got out of my prepared opening book almost immediately.

2026-01-24 The Do It PodcastThe Fun Criterion, Inner Conflicts, and How the Self Resolves Them

01:02:40 - 01:02:41

Same to you.

Edwin

01:02:42 - 01:03:34

I hope you enjoyed the conversation with David as much as I did. If you haven’t read his books yet, I highly recommend The Beginning of Infinity and The Fabric of Reality. They are exceptionally rich books, the kinds of books you can reread for years and still keep finding new ideas. And if today’s conversation resonated, especially understanding the nuances of how the mind works and how that can guide how we think and act, you might like my upcoming book, The Four Acts. It’s a four stage framework you can use on any problem in everyday life. And today’s topics directly feature in it. For example, the fourth act learning, it unpacks the three modes of learning we discussed today, explicit, inexplicit and unconscious learning. The book is nearly finished with a release plan for 2026. You’ll find a pre-order page in the description. And if you want more in this vein, subscribe to the channel to stay up to date. Thank you for joining me today and take care.

2025-12-22 EconTalkDavid Deutsch on the Pattern

01:25:49 - 01:26:16

This is Econ Talk, part of the Library of Economics and Liberty. For more Econ Talk, go to econtalk.org, where you can also comment on today’s podcast and find links and readings related to today’s conversation. The sound engineer for Econ Talk is Rich Goyette. I’m your host, Russ Roberts. Thanks for listening. Talk to you on Monday.

2025-11-17 The Light SideFalling for Physics

00:48:33 - 00:49:14

So we often touched upon the fact that theoretical physics is about understanding how the universe works and how nature works and trying to figure out a theory or trying to figure out an equation say which could tell us more about that but I’m a theoretical physics student and I get this very often that why should anyone care about theoretical physics if in their normal lives if they don’t study theoretical physics. So my question to you is should regular people care about theoretical physics?

David Deutsch

00:49:14 - 00:52:02

And so I think that question is the same as should everybody care about theoretical physics because there’ll be the regular people and the irregular people and that includes all people. No I don’t think there’s any should about it. It’s people who have a desire to know how the universe works, to learn and yes I should have said earlier that one of the things I like about doing physics is that I enjoy being baffled so I don’t necessarily want to get through to the answer and this again Carl Popper said this beautifully in his passage in his autobiography about problems. He said what you should do and as a philosopher but it applies equally to a physicist or to any scientist. Find a problem and fall in love with it and get married to it and live happily with it for the rest of your life till death do you part and if by any chance you should happen to solve it and I like the way that he regards this solving it as a bit of a downer you know that that’s kind of a so then he said if by any chance you should happen to solve it then don’t worry because there will be a host of enchanting problem children that will result from that and I think that that’s exactly right that’s how I feel about physics. You also spoke about revolutions in physics. I don’t think that Newton or Einstein or Faraday, the three we’ve mentioned in this chat, I don’t think any of them wanted to revolutionize physics. All of them wanted to know what’s out there, how it works, how it behaves, why it behaves the way it does. That’s the kind of thing they were interested in. It so happens that they discovered those things and that those things then had all sorts of side effects and so on but I think the reason they wanted to think about it is that they were interested, they liked being baffled by these things that they discovered along the way there. And I like being baffled to the extent that I sometimes get into the mindset of being a bit reluctant to pursue something in case it gives the answer. I have to stop myself thinking that.

Anjali

00:52:03 - 00:52:24

I think that’s a wonderful thought to end this conversation at. All right. Thank you so much for your time. Thank you so much for being here and for sharing your wonderful thoughts. It was an absolute pleasure, an absolute honor. Nice to meet you too. Okay bye bye.

2025-11-06 Philosophy For Our TimesIn Search of Nothing

00:59:24 - 00:59:39

And I think on that note, now that we have hope for infinite progress, I’m going to call the official part of the debate here. Thank you all. Thank you for listening to Philosophy for Our Times.

Avi and Ali

00:59:39 - 01:00:18

This episode and don’t say nothing shout out to Eric, Manny and Matt who reached out over the last couple of weeks about our Kehinde Andrews and Adam Phillips episodes your feedback was much appreciated and if you want more talks videos and articles from the world’s leading thinkers don’t hesitate to go to iai.tv for thousands more talks we’ll be back next week with more groundbreaking philosophy so see you then bye bye

2025-10-27 This Is WorldProof the Multiverse Is Real - Interference and AGI

Duration: 00:39:03

Transcript

David Deutsch

00:38:49 - 00:38:53

Thank you very much for your time. It was a great pleasure. Grand honor.

David Deutsch

00:38:53 - 00:38:55

Same here.

Maciej Kawecki

00:38:55 - 00:38:56

Bye bye then.

00:38:56 - 00:39:02

The This Is World channel is partnered with IQM, one of the world’s largest quantum computing companies.

2025-10-22 Open SocietyProgress, the Enlightenment, Aspirations, and the Open Society

00:35:25 - 00:35:29

Yes, you’re very welcome. It’s been fun.

2025-10-06 Theories of EverythingEinstein Would Fail Modern Grant Applications

Duration: 02:12:52

Transcript

Curt Jaimungal

00:00:00 - 00:00:03

Why would Einstein fail a modern grant application?

David Deutsch

00:00:05 - 00:03:47

Fortunately, I’m not very familiar with the way that grant applications are dealt with. I only know the gross features, in both senses of the word “gross.” That is where I noticed that grants that should have been awarded aren’t, fairly reliably, when it comes to fundamental research, which is what I’m mainly interested in.

So judging by my experience, and I certainly don’t know what it was like over a hundred years ago in Germany, but judging by my experience today in Britain and in America, he wouldn’t have stood a very good chance. He wouldn’t have been able to say very clearly what he was trying to do, because there was no one versed in relativity on the panel that judged physics applications. None of them would have known what a manifold is, what the Riemann tensor is, so his application would have had to explain that in very elementary terms.

They would have had in front of them a pile of applications which, from their point of view, had much more merit because they could see that these were open research problems that needed to be solved. Only in the bigger picture were they incremental, whereas Einstein’s was fundamental. So I think he would have had difficulty getting a grant, and I think in historical fact he did have such difficulty. It was only when Max Planck saw that there was something interesting about him that he got him a job.

Again, that isn’t possible nowadays because there’d be anti-nepotism rules. There’d be rules about the procedure, and every procedural rule is an impediment to any new kind of thing being tried. Anti-nepotism sounds positive. What’s the negative side to it? Well, so nepotism, literally, if it means not giving a job to your nephew, then perhaps that has merit. But what it means in practice is that if you know someone personally, you can’t take part in the selection process for that person and you certainly can’t get the university to let that person in. So it all has to be at arm’s length in order to ensure fairness. Now the amount of unfairness that can be caused by not having that rule is very tiny compared with the enormous unfairness of having that rule, because having that rule means that only people who know nothing about the candidate can rule on whether the candidate is accepted.

Curt Jaimungal

00:03:47 - 00:04:29

So the natural question that arises in someone’s mind is: why is it that we need grants anyhow? Professors of the past, and I’m speaking about fundamental physics, not just a generic professor, but say Einstein or Feynman or Everett, which I know you have some personal stories on, and I’d like to get to those at some point, I’d never heard them complain about the grant system. However, I hear complaints, especially off-air, from contemporary professors about it frequently. So why does one require grants if they’re already paid a salary? What are the grants for? What if you don’t get a grant? Are you just sitting around twiddling your thumbs? Are you then seen as a net cost to the university and they just make you lecture?

David Deutsch

00:04:29 - 00:09:13

Yeah, that’s what it is. I repeat: I’m not very familiar with the system, and that’s not an accident. I have intentionally distanced myself from any knowledge of the system because it is a very unpleasant thing to be connected with. But yes, professors can do full-time teaching. Most of them don’t want to, and that’s not because they don’t like teaching, I think in many cases, but because they have to teach to a regimented curriculum and syllabus. They can’t exercise their creativity and reveal to students why they are passionate about the subject. They have to get through a lot of it because the students are all competing with each other to get the exam results, which are a very inaccurate measure of how suitable they are for future research. I mean, it’s not even intended for that. It’s intended for displaying their qualifications, usually to do something else which is by two orders of remove away from suitability for research.

Here in Britain, I don’t know what it’s like in the US, the professor has to apply for a grant to do research, and some of that money goes to the university so he doesn’t have to do as much teaching. But the student or graduate student has to separately apply for a grant. The mere fact that the professor wants to include him on his research team is not enough to provide for subsistence for the student, and again the university takes a cut, notionally because of desk space or lab space or whatever it is.

So I think this whole superstructure is contrary to what is needed for fundamental research, and actually for all the functions of a university, but particularly for fundamental research. What it should be is that the grant-giving authority, or the entity that pays for somebody to do fundamental research, whether it’s a private charity, a private individual, the government, various branches of the government, the military, whichever it is, should be awarding the grant to one person. Sometimes that one person alone, like it would have been with Einstein, is the research group. But in a more general case, that person would then spend part of his grant on hiring postdocs and graduate students and undergraduate students. I mean, this whole hierarchy is counterproductive.

In the research group I believe in flat hierarchies, so ideally it should be the boss and everyone else, and the boss and everyone else should be equal as well. No one’s commanding anything. They’re all there on a joint project, which they believe passionately in.

That reminds me of another incident with my former supervisor Dennis Sciama. By the way, I’ve been very lucky. I was very lucky with my supervisors. He was once told by the higher-ups that they would be instituting a clocking-in system where the students and postdocs, in the morning, would sign whatever it was, the register, that they had to be in by 9 a.m. And so Dennis Sciama objected to this, saying if any of my people is in the department at 9 a.m.

David Deutsch

00:09:13 - 00:11:16

It’s because they’ve been up all night. Right. There’s a story about Schwinger, who was told, can you lecture at 10 a.m.? And then he paused and thought, I don’t know if I could stay up that late. That is the way.

When physics was a small enterprise, and there weren’t that many physicists in the world, they were all rather eccentric and they were all supported by various other means other than a system. They all had their quirks, and the thing, whatever it was, that sustained them approved of those quirks. So like in the Institute for Advanced Study as well, the first thing someone was told when they were given their ten-year or 15-year grant, or whatever, is: do whatever you like. That’s what the grant holder should be told, and that’s what the grant holder, in my opinion, should tell the postdocs and the graduate students and whoever else is working with him.

You might think, well, if they’re told that, why shouldn’t they just spend all their time drinking? Well, it’s because they’ve been hired for the purpose. They’ve been hired because they are passionate about something. When you want to do a joint project, when you’re interviewing someone, you want to see whether they are passionate about that project, not whether they have some standardized qualification for it. There is no such thing if you’re working on something new. Yeah, enthusiasm is therefore everything.

Curt Jaimungal

00:11:17 - 00:11:35

Peter Higgs said, I believe this was a decade ago now, that he wouldn’t be able to get an academic job in today’s environment. I think he said it’s “as simple as that,” that he wouldn’t be productive, or something akin to that. Yes, nor would I. Could quantum computing get invented today?

David Deutsch

00:11:37 - 00:12:35

I think what would happen, probably yes, but the way it would happen is that which is happening a lot already: people who are really passionate about some new fundamental thing apply for a grant to do something else, to do something incremental. They do the incremental thing to the minimum level required to sort of pass the various tests. They publish, they publish again, they publish again. Meanwhile, their passionate thing they don’t necessarily publish, because they’re grappling with very difficult problems that don’t admit of successive papers doing it better and better. They’re waiting for a breakthrough, so they do that in their spare time. That is highly unsatisfactory.

Curt Jaimungal

00:12:36 - 00:13:18

Imagine the retort would be: look, there’s plenty of great work occurring. It would be foolish to paint all of academia with a brush of stagnation, although that word hasn’t come up. I would like to talk about that, as you had a whole conversation with the Conjecture Institute. I’ll place a link on screen to that. It was fantastic about that subject.

Anyhow, we don’t want to paint all of academia with a brush. Some would say, even to say physics is stagnant, they would quip, well, I wish the field was stagnant; I wouldn’t have so much to do. I could catch up on the arXiv, for instance. So what do you say to those who argue: look, there’s been innovation. There are gravitational waves, black hole imagery, topological insulators and phases, time crystals, exoplanets have been discovered, quantum advantage.

David Deutsch

00:13:20 - 00:16:45

So I don’t want to appear to be trashing incremental research. Also, in that list of things you just mentioned, I don’t want to classify all of those as incremental research. Some of them are indeed fundamental.

What I want to say is that the research landscape taken as a whole is heavily biased against fundamental discoveries. Everything we’ve talked about, the criterion for getting a grant, the structure of careers, the structure of university departments, all of them are heavily biased against fundamental research, such that it is much more done in people’s spare time than it is done in pursuance of the grant that they’re getting.

By the way, you mentioned what do I mean by grants. Somebody pays for research which is blue-sky research. It used to be that aristocrats did it; they funded their own research. So that’s one way to go. But in regard to who funds it, the main thing that is wrong with the existing setup is that there are too few sources of funding. Because the government has entered the field, not only have they sort of crowded out other means of funding and also prevented it in various ways, but say the private charity, for example, who funds research, they’re going to use the same criteria. What they do is they have a committee whom they assign the task of sifting through all the applications and picking the best ones, and they don’t know how to do that either. They can’t possibly know, and they’re forbidden from using one of the few ways that they could, namely to ask their colleagues: do you know someone who is worthy of this grant? They’re not allowed to ask that.

So in other words, broadly speaking, the government has now contributed, it sounds like a positive, to physics, to fundamental research. They’ve given plenty of money. However, with that money comes some poor practices. These poor practices are then adopted by the individuals who previously used to donate without these poor practices. Yes, and the result is not stagnation. I mean, there is a kind of stagnation, but that’s a different story. It’s that there is, or rather indirectly it is, de-emphasis of the fundamental in favor of the incremental, not that there’s anything bad about incremental research, as I said.

Curt Jaimungal

00:16:45 - 00:17:01

00:23:24 - 00:24:06

Yes, fund people. That’s how it should be. That’s how it should be in incremental research as well. The whole of scientific research should be like that. I thought you were about to say you’d be allowed to make a new box and say what should be in it and all the sub-boxes. I wouldn’t have known what the sub-boxes are. They too were only invented later, and not by me.

So the communities that I also traffic in, other than physics, are philosophy and math, and the grant situation there doesn’t seem to be as dire. There are no expectations of grants as a prerequisite for tenure, for instance; just the hope and the promoting of people who have strong publications.

Curt Jaimungal

00:24:07 - 00:24:33

So a math department meeting may say something like, congratulations on your Annals paper. But I imagine that a physics department meeting would include, like: your grant expires next year, what’s your renewal plan? So what’s the difference here between physics and math? And I’m speaking about fundamental physics, because you can always say, well, if it’s experimental physics, it’s quite clear you do need plenty of money to fund your machines and your computers and your servers and your students and so on. But fundamental physics?

David Deutsch

00:24:34 - 00:27:27

Yes. So for the structure that I advocated earlier, the research group leader should indeed be judged not on his past papers, not always, but on his previous success in advancing the subject. It should be a well-known figure in the field who has a track record of making progress, and now he wants to make progress in a way that can engage several other people. Maybe, if he’s an experimentalist, he wants to build this with a machine, or if he’s a theoretician, then the field has broadened enough for him to see that there is potential there that he doesn’t yet know what it is. But he knows who he wants. That’s the thing. Ten years earlier he knew what he wanted to do. Now he knows what kind of person he wants to work with, and he knows that he wants to hire five of them or ten of them, but not five hundred of them.

So he is funded not because he can say what his next paper is going to be about. He’s funded because he says he’s very interested in stuff and he’s going to do research on it, and somebody who funds him will be saying, I think that this guy is good. Or gal. Yeah, well, I don’t want to use gender-neutral language because I think it’s silly. Obviously when I say he, I mean he or she, and if I said mister or Mrs., I might also mean His Majesty or Master so-and-so.

It’s completely natural in this whole scheme of things that I’m advocating that everything is tuned to doing the research, creating the new knowledge. Everything is subordinate to that. If somebody is going to care whether the graduate student is male or female, then they’re not the funder that I want. They need to be obsessed with the thing itself, and so should be the people they hire.

Curt Jaimungal

00:27:30 - 00:27:59

Suppose right now there’s a wealthy patron, or multiple, and you could speak directly to them. These people who are watching care about fundamental physics, maybe foundational research in computer science as well, just foundational in general, which we can get to distinguish in between fundamental and foundational. To me I see them as quite close. I can’t distinguish them. Maybe you can. But anyhow, what is your message to them? What are they to do? They have this money. What are they to do? They want to help.

David Deutsch

00:27:59 - 00:31:37

Yeah, so each one of them is different. Each one of them has interests. Each one of them has reasons for wanting to promote fundamental physics, and each one of them has a different conception of what fundamental physics is. They might also have a conception that it’s being slowed down by various sociological facts and so on, so they want to get around that.

They need to find somebody that they think is good. Usually this will be somebody who has already done some of the thing that they want done, and then they should approach that person and say, could you use some money? Often the answer will be no, but often it will be yes, because with money they can do a lot of things in parallel that they would otherwise have to do in series, by themselves or with a smaller group.

Now there is a thing that’s just started up, the Conjecture Institute, and I don’t know how they make their choices, but what I’ve seen seems to be following the pattern that I advocate quite closely. They fund the person, not the research project, and they seem to fund people who are interested in foundations. I don’t know whether that’s because their thing is to fund foundations, or whether their thing is to fund things which aren’t normally funded. I don’t know which of those it is. But either of those would do, and lots of variations on that would also do. Like I said, I would like there to be dozens of such entities, all with a different ethos, all with a different theory of what foundations are or what they’re for, and all with a different theory of what’s wrong with the present thing, why somebody hasn’t already funded the thing that they want to fund, that sort of thing.

So firstly, what is the difference between fundamental research and foundational research? I don’t make much difference between those things, but foundational suggests to me that you have a field and you’re drilling into its foundations. So you want to understand it more deeply than it has been before. Fundamental means to do with the fundamental knowledge, that is, knowledge that is needed for all sorts of different areas. For example, quantum computation I think is fundamental, or was, because it has to do with mathematics and epistemology, as well as physics and computation and computer engineering. So there’s a whole bunch of things that it might unite if it works. But it’s fundamental in its conception. It’s not like working at existing foundations of anything.

Curt Jaimungal

00:31:38 - 00:32:12

Okay, so someone who’s listening who doesn’t care about fundamental or foundational research, they hear you keep bringing it up. Why is it so important to you? And of course you are not saying that the incremental, confirmational research that’s done on existing theories is not important, but that the fundamental, foundational has been somewhat excluded or not incentivized properly. But that implies: why should we even care that it’s incentivized properly? What is it about foundational and fundamental research that’s so vital to your conception of knowledge in the world?

David Deutsch

00:32:12 - 00:33:26

The thing that unites them is that the growth of knowledge can’t be regimented. You know, Henry Ford said something like, if I asked people what they want, they would have said a better horse. The essential thing to intellectual progress of all kinds, whether incremental, fundamental, whatever, is interest, that somebody is interested in doing this. If they weren’t paid, they’d still do it. They’d get a job doing something else and do it in their spare time, like Van Gogh with his painting. You know, nobody ever bought a painting from him in his lifetime, even though his brother owned an art gallery. I mean, I don’t know the story of that, but you know, it’s obviously not the standard story of slotting into an existing structure. So some …

Curt Jaimungal

00:33:27 - 00:33:28

But …

David Deutsch

00:33:28 - 00:35:32

And I didn’t say anything specific, because I didn’t know anything specific at the time. So I just described what kind of thing I wanted to work on. I remember saying, for example, to Dennis Sciama in my first interview that it seems to me that the most urgent problem in physics is quantum gravity. So I’d like to work on that. I know a bit of quantum field theory, but I don’t know enough relativity yet.

Now it turned out that that was a bad idea, and I decided to turn from quantum gravity, which I thought was too difficult to work on at my stage yet. I was getting interested in other things, which eventually led to physics of information and to quantum computation and so on. Dennis somehow saw that I had the thing he wants in his students, and so he hired me. When I wanted to change and to study something completely different, not only did he not object in any way, he was interested. He was interested in what I wanted to do and why and so on, but he never tried to direct my research because he assumed that I wanted what he wanted.

Curt Jaimungal

00:49:04 - 00:49:07

You mentioned that quantum gravity was too difficult. What do you mean?

David Deutsch

00:49:09 - 00:54:19

So at the time I didn’t know what was so difficult about it. I think I’d absorbed the standard view that what we’re trying to do is cure the infinities and cure the nonlinearities and that kind of thing. The answer would be an equation which had the desired properties. I realized as I got into the subject that those are trivial problems. It wouldn’t matter if we didn’t solve that, but the chances are that we will solve that once we’ve got the deeper incompatibility between the two theories sorted out.

The fundamental, I keep saying fundamental, but I don’t always mean fundamental in the same sense, we have to unpick this. At root, all the existing field theories are theories of fields on spacetime, whereas general relativity is a theory of spacetime itself. It’s not a field. You can think of it as a fixed spacetime with the field on top of it, so it’s the sort of static part and the varying part which people then try to quantize. But that’s very alien to general relativity. General relativity is the theory of spacetime as a dynamical thing itself. People have tried everything in quantum gravity, and in my view everything has failed.

In every other part of physics, these fields evolve in time, and what you’re looking for is a dynamical equation, an equation of motion that says how they evolve in time. But in general relativity, viewed in that sense, there is no time. Either it’s a four-dimensional thing, or in the quantum sense it’s a manifold where every point is a three-geometry, and evolution in time is just a strip of things with a higher wave function than the rest. But how that turns into time, there are various proposals. Anyway, what can I say? It’s conceptually very incompatible. The way we conceive of quantum fields, which have their own problems by the way, and the way we conceive of spacetime, is fundamentally incompatible.

There are also problems with quantum field theory itself. Now I think general relativity in itself would be a viable theory. Okay, there’s the big bang and black holes and we’re not sure how to deal with singularities, but basically it’s a viable theory and there’s no kind of contradiction in it. Whereas quantum field theory is full of contradictions in its own right, let alone before you try to unify it with gravity.

Look, my favorite one at the moment, I like being baffled, and this is one of my favorite problems with quantum field theory because it’s so baffling. One of the basic axioms of quantum field theory is that field quantities at spacelike separated points, that is, at points at the same time, should commute with each other. That is, if there’s a quantity A and B, then $AB = BA$ . At a later time they don’t commute, and that’s the whole reason why quantum fields evolve in time. It’s because the later thing doesn’t commute with the earlier thing. So now if two things, let’s say in the same space, there is a quantity here and a quantity there that don’t commute.

David Deutsch

00:54:19 - 00:56:02

It means they must be described by separate algebras. And no matter how close they are, they must still be described by separate algebras, which means that these separate algebras all commute. So no matter how close together these algebras are, they still commute, and yet when they coincide they don’t commute, because field quantities at the same point don’t commute, because that’s what drives the whole thing forwards.

00:59:27 - 01:00:15

Okay, so most of the time when people are thinking of combining general relativity with QFT, the mathematical problem is non-renormalizability. There are said to be three or so conceptual problems that are distinct from the non-renormalizability. One is just QFT requires a fixed background, like you mentioned: background independence is the issue here, conceptually speaking. Then there is, well, what does it mean for you to have a superposition of geometries operationally? And then there’s the problem of time that you mentioned.

Now the current leading theory of quantum gravity is string theory. At the time when you were a graduate student, it may not have been there. But either way, you have heard of it at some points during your career. What attracted you to it, or what did not attract you to it? What dissuaded you or persuaded you?

David Deutsch

01:00:15 - 01:01:52

David Deutsch

01:10:37 - 01:13:03

About their committee? Yeah, I’m not well up on the history either, though I have seen Oppenheimer. I don’t know how accurate that was. We’re in the same boat, right? I think that was a very unusual organization, and it did not run on this kind of committee pattern.

01:14:41 - 01:16:41

01:18:30 - 01:20:21

01:21:33 - 01:22:40

David Deutsch

01:27:45 - 01:31:12

I don’t know why he didn’t fix it himself. He was doing some highly mathematical things above my head at the time. DeWitt had a kind of attitude that it was kind of obvious that this was the right interpretation, and he didn’t have any interest in working on it. It’s backwards-looking.

That was true of Everettian quantum theory for many years, that everything was focused on trying to explain again and explain again and again to reluctant people why the prevailing view is untenable, why the Everettian view is illuminating and it’s the only possible one that will work. The trouble is that that attitude, as I have often said, is a bit like if biologists had spent all their time proving and reproving that creationism doesn’t work and that you should have Darwinism. Yes, you can keep doing that, but what was needed was to improve Darwinism and make further progress. They wouldn’t have made further progress if they’d kept on just engaging with this problem of why the creationists are wrong.

01:43:27 - 01:45:55

Well, it did in this case. But I don’t think that’s a proof of anything. As I said some time ago, it very often happens that fundamental things cause practical improvements as well, eventually. But I don’t think it’s tenable to say that that’s why one should think about the fundamental things. It’s a thing that can happen. Sometimes I think that often happens, perhaps even very often. I don’t know how to characterize it, but it certainly happened to me in this case. If I had thought of quantum theory in the wave-function-collapse way, I wouldn’t have thought of quantum computation.

01:47:20 - 01:48:42

So I try not to have commitments. Like W. W. Bartley said, we should retreat from commitment. I also try not to have beliefs either. Let me put it this way around: my theory that this computer that I’m talking into now exists has the same status in my mind as the theory that many copies of it exist in other universes. In both cases, that status is that there are no rivals to that theory that I know of that aren’t nonsense. I mean, there are only nonsense rivals.

So you could call that belief, but that suggests that I want it to be true, or that I would resist it not being true if an argument were presented. I hope and expect that that is not the case. Whereas the opponents of Everett do have beliefs, and that, I think, is what is… Well, again, I said I wouldn’t be psychic, so I’ll try not to be psychic.

Curt Jaimungal

01:48:43 - 01:48:49

Let me be psychic. It seems like beliefs to you is synonymous with dogmatic beliefs.

David Deutsch

01:48:49 - 01:49:55

01:56:07 - 01:57:47

02:00:44 - 02:05:15

Problem that people have with free will is that they have a conception of free will which, by definition, violates the laws of physics. Although they don’t often say it like this, it’s basically: free will is the human capacity to override the laws of physics. I don’t think anything overrides the laws of physics. We might be wrong about what the laws of physics are, again, like Penrose thinks, but I don’t see in that case a proposal for different laws that would help with the problem of free will.

02:11:56 - 02:12:50

They analyze culture. They analyze finance, economics, business, and international affairs across every region. I’m particularly liking their new Insider feature. It was just launched this month. It gives you, it gives me, front-row access to The Economist’s internal editorial debates, where senior editors argue through the news with world leaders and policymakers in twice-weekly long-format shows, basically an extremely high-quality podcast.

Something else you should know about is that if you go to their app, they not only have daily articles, but they also have long-form podcasts with their editors and writers. This is also available online. Whether it’s scientific innovation or shifting global politics, The Economist provides comprehensive coverage beyond headlines. As a TOE listener, you get a special discount. Head over to economist.com/toe to subscribe. That’s economist.com/toe for your discount.

2025-09-25 Sam Altman and David Deutsch Discuss AGI

00:07:05 - 00:07:21

David, thank you so much for joining us and thank you for your uplifting words and have a great evening. David Deutsch, a pioneer of quantum computing, one of the most brilliant thinkers of our times. Thank you for joining.

2025-09-12 The Shape of DialogueWhat Is the Best Pathway to Knowledge? with David Deutsch

01:23:14 - 01:23:41

I get it now. Yes. The sort of thing I don’t like is like, if you name a film, a river runs through it. I want to know through what before I can decide whether to watch this film. So …

Michael Goldwater

01:23:42 - 01:23:49

To help the Shape of Dialogue grow, please subscribe, like, and comment and share it to all and sundry. Many thanks.

2025-09-11 Into the ImpossibleDavid Deutsch - We Will Build Humans Before AGI

00:30:05 - 00:31:09

Hey everybody. I’m usually the one that asks my guests to judge their books by their covers. But today I’m asking myself to judge my own book by its cover. My newest book, Focus Like a Nobel Prize Winner is chock full of advice, life tips and focus and productivity tips from nine of the world’s greatest minds. Nobel laureates ranging from economics to peace to physics, of course. It launches September 9th, which is also my birthday. I’ll go check it out. And my publisher has got Amazon to run a special just for listeners of the Into the Impossible podcast. You can get the Kindle edition for only 99 cents, that’s less than a new pocket protector. So go to Amazon and get the Kindle copy today because this special only lasts for the first week after launch. If today’s conversation with David Deutsch made you question everything about modern science, you need to watch my recent episode with Sabine Hossenfelder. We got into how mathematical beauty can guide scientists astray and what the true power of quantum computing and artificial intelligence may turn out to be. Click here and don’t forget to like, comment and subscribe.

2025-09-05 Conjecture InstituteDavid Deutsch - How to Reverse Academia's Stagnation

01:33:45 - 01:34:05

I can see that it’s already doing good in this small scale version both in people it funds and in presumably raising money for it. I hope and expect actually that it can scale up in both respects.

Brett Hall

01:34:05 - 01:34:16

Absolutely, hopefully that is the future of the beginning of infinite progress and optimism. Until next time David, thank you very much. Bye-bye.

2025-09-03 Strange Loop PodcastDavid Deutsch - AGI, the Origins of Quantum Computing, and the Future of Humanity

00:57:22 - 00:57:24

Yes, certainly.

2025-08-26 This is ITConsciousness, Many Worlds, and Quantum Computing

00:40:31 - 00:40:35

Thank you very much for your time it was a great pleasure and a great honor.

David Deutsch

00:40:35 - 00:40:40

Same here. Bye bye then.

2025-08-23 Into the ImpossibleQuantum Theories Are Just Miracles

2025-07-03 IAIThere Is Only One Interpretation of Quantum Mechanics - Full Interview

00:19:34 - 00:19:40

David Deutsch, thank you very much.

2025-06-28 Robin HansonDavid Deutsch, Robin Hanson, and Daniel Martin on Culture

01:20:39 - 01:21:10

Right. And so I’m trying to write a book on this, but part of what’s put me off is like just realizing how hard this problem is and thinking I really need to frame it well and think about, you know, how to, you know, framing matters a lot in these sorts of things. What sort of analogs you make, what terms you use. And then so that’s why I’m very eager to sort of find other terms like you, David, are using and see how I can change my framing and terminology to integrate better with other uses.

Daniel Martin

01:21:10 - 01:21:22

Well, thank you both for talking today. I guess we’ve given others something to listen to. And so I will stop the recording and we will share files soon.

Robin Hanson

01:21:22 - 01:21:24

Great. See you.

2025-06-24 Salvador PodcastDavid Deutsch - The Fabric of Explanations, Optimism, and Creativity

01:03:27 - 01:03:28

Bye bye then.

2025-05-12 Making Sense#414 - Strange Truths - subscriber version

02:18:00 - 02:18:17

Yeah. I think that’s a crucial variable there. But all right, David, as I said, we’re given to three-hour conversations and this was nearly one. Thank you for your time. It’s wonderful to talk to you. It’s been fun. Yeah. Until next time, be well. Okay. Bye-bye then.

2025-04-07 Within Reason#102 - You're Not Smarter Than a Caveman

01:55:04 - 01:55:29

Yeah. Yeah. And hopefully at some point we can perhaps do it again, because as I’m sure will always be the case, no matter how many times we ever sit down, there is more to talk about. So until the next time, I suppose. Yes. To get early ad-free access to episodes, as well as supporting the channel at the same time, subscribe to my sub stack at alexoconnor.com. Watch more episodes by clicking the link on your screen. Thanks for watching and I’ll see you in the next one.

2025-04-02 The Falsifiable PodcastA Woven Web of Guesses with David Deutsch

00:51:59 - 00:52:57

Yeah. Well, glad I’m glad that it’s working out for you in that respect. Yeah. Gosh, well, we’ve come up on our time here, David. I would go down these rabbit holes. We could do it all day long. But man, thank you so much, David, for responding. I’ve been trying to I know you’ve noticed my email pop through at least once every six months. And I was just so happy that we were able to do this today. The Falsifiable podcast probably wouldn’t exist without having read your book, both of your books and going down the rabbit hole of Deutsch. I think in 2013 is whenever I started this journey and I’ve not really looked back since. And it’s just been trying to find good explanations and doing all this kind of stuff. But I’m good. Yeah, absolutely. And I know a lot of people feel the same way. So, again, David Deutsch, thank you for coming on the Falsifiable podcast.

David Deutsch

00:52:57 - 00:52:58

Thank you for having me.

2024-11-11 Arjun Khemani PodcastDavid Deutsch - The Era of Man, Popper, and Western Civilization

01:29:02 - 01:29:06

01:23:51 - 01:23:52

Cheers.

2024-10-28 Do ExplainHalf Episode #57 - Truth and Abstractions, with David Deutsch and Jake Orthwein

00:38:01 - 00:38:34

He might have been, but I know he didn’t like it. He didn’t like it because you can’t see a force and so on. So yeah, he might well have got it from somewhere. Again, where you get it from, whether you like it or not, whether you don’t like it, your research is going to take you in a certain direction. You can either go in that direction and make progress or you can just insist and not make any progress.

Jake Orthwein

00:38:34 - 00:39:55

Yeah. Okay, I have maybe two more questions. This has been super informative, so thank you. I have maybe two more lines of questioning. One is about this analogy to natural selection. That’s one of the things that’s so interesting about Popper is that you can really be like, oh look, contextualism and criticism is very much like variation selection. But one of the first things that I remember being taught when I was learning about natural selection is that you shouldn’t do this purely hierarchical conception of fitness as being like everything was trying to just become a human, which is the peak of the fitness ladder. And you also shouldn’t have the conception of final fitness because there’s always this kind of Red Queen dynamic happening between you and the environment and it’s not that kind of thing where you’re going to be fit once and for all because you’re changing the environment and so forth. It’s a very dynamic thing. And I think of knowledge as being kind of similar in that sense, but that’s part of why I have this intuition that these universalized idealizations are sort of like missing the point or something. You know what I mean?

David Deutsch

00:39:55 - 00:40:35

Right. So, nice point. It’s the first time I’ve thought of this, but I guess you’re right, to put it in my terms, that there is no analog of truth in evolution. It’s wrong, as you just said, it’s wrong to think of the process of evolution as maximizing something globally. I mean, locally, it maximizes something, but there’s no ideal thing that it’s heading toward.

Official YouTube Spotify Apple Podcasts

2024-10-23 The Saad TruthPhysicist Dr. David Deutsch - Quantum Computing, Turing Machines, and Multiverses

01:03:11 - 01:04:50

Yeah. It’s funny you say this because that’s probably one of my things that triggers my ire the most when a student comes to see me at the end of class. And I’m excited because I’m hoping that they’re going to ask something interesting and say, Oh, what you just mentioned about the evolutionary root, blah, blah, blah. Is that going to be on Wednesday’s exam? And I get so pissed off because it’s exactly the tension between intrinsic motivation and extrinsic. I’d love for all of you to be there for intrinsic reasons. There’s one thing that’s really annoying me. I’m looking at my thing and it says you only have 2.5 gigabytes of storage left on your computer. And I’m worried about that because I don’t know if I run out of storage, whether the computer will shut down or not, because I’m not, I wasn’t planning on finishing this conversation. So what I would like to do is at least end it now in a very anticlimactic way, but I have it right here. You only have 2.5 gigabytes. And then maybe we can set up a, because I was still hoping to discuss Israel. I was hoping to discuss the increase of global Jew hatred. We were going to maybe talk about some of the parasitic mind stuff. So there’s a lot for us to continue. Can we shelve that for round two at whatever time is appropriate for you? Certainly. So thank you so much. What a mind blowing conversation. I can’t wait to post it. I just want to make sure that this is stored and that my computer doesn’t die. Thank you so much. Stay on the line, David, so we could say goodbye officially offline and round two coming up soon. Thank you so much, David.

2024-10-11 NavalThe Deutsch Files IV

01:24:27 - 01:25:05

There’ll be… This is something that people who have thought a lot about this will think a lot about it. At the moment they’ve been thinking how to do the opposite. If you, for example, you can have a rule that everybody who is allowed into the country has to declare that they, for a start, they could have to declare that they love the country and want to do well in it. And then if it later turns out they were lying, they can be deported, not for being enemies of Britain, but for lying on the immigration form.

2024-10-01 Increments#74 - Disagreeing about Belief, Probability, and Truth

01:31:40 - 01:31:45

Thank you. Fun conversation. Okay, bye bye.

2024-08-26 Reason Is Fun#6 - Are Feelings Ideas

01:25:59 - 01:26:11

All right. Well, I feel very unsatisfied. This is a wonderful place to be. I would love to have more conversations about this. I would love to fully understand this. I’m not sure if I ask any of my questions. It felt like great groundwork. Yeah. Yeah, but it has been an absolute pleasure to have you both. And let’s do this again sometime. Yes, absolutely.

David Deutsch

01:26:11 - 01:26:17

Thank you so much. Thank you.

2024-07-28 Breakthrough DiscussA High-Speed Overview of David Deutsch's Current Views on Universal Constructors at This Space Conference

00:16:36 - 00:16:38

Thank you.

2024-03-31 Sagenhaft und SonderbarDavid Deutsch - The Beginning of Infinity

01:10:57 - 01:11:04

But do you allow yourself to fantasize or dream about questions like that? What is it all about?

David Deutsch

01:11:04 - 01:15:35

Yeah, that is that is how we get to conclusions like this, except that we do more than just fantasize and dream. We criticize the fantasies and dreams. And for every dream that can be made into a viable theory, there are thousands that can’t. So we think maybe it could be this, maybe it could be that. And then you think, hey, I have a way. And then your colleague says, no, that can’t be because of so and so. Then you say, well, maybe we could get around so and so and so on. And really, you say you would go mad. But I don’t think so, because the thing that theoretical physicists are doing is exactly the same thing as what you’re doing staging this interview. You want to know how the world is. And there are many ways to know how the world is. There are also ways of preventing yourself from knowing how the world is. Maybe many more of those. But even that is driven. Even the silliest sort of fantasy, you know, flat earth theories or whatever is driven by the same thing that drives science, which is curiosity and ravenous appetite for understanding what the world is like, what makes it tick and so on. And I think a lot of the paths that lead nowhere are the fault of schools and universities and popular science books, because they present science as a heap of facts, whereas actually science is a heap of problems. And the things that we think are facts, we know they’re going to be superseded eventually. Most of them. I mean, some of them won’t, but we don’t know which those are. Some of the ones that we are most sure of at any given time will be overturned soon. And contrary to popular belief, we science, the scientific community. Well, there are scientists who want to preserve their own ivory tower and resist any change in their favorite theory, but they don’t make progress. The ones who make progress are the ones who want their existing theories to be overturned and work towards that and who like that and also like, by the way, failing to do it, like I said. So those are the ones that make progress. And those ones. And I’m hoping I can count as one of them. Do not find that the others erect barriers against them. I mean, they may they may well say, well, that’s silly. But, you know, you don’t get kicked out of a university for believing the many worlds interpretation. You may get laughed at by some people and then we laugh at them. And so it goes. But nothing much worse than that. Fantastic. Dr. You’re incredible. It’s amazing getting to speak with you and get to hear your perspectives. I love this because your delineation between science with an S and science with a dollar sign is something that’s very crucial to all of us, I think, because they’re, like you said, it’s a burgeoning influence that stops progress. And I think there’s a lot of these horrible stories of people coming up with amazing hydrogen cars that can just run your car off of water and then that getting killed or developing in a university laboratory. But then somebody says, no, no, no, the tenured professor, he’s still got 20 years left at least. And so we can’t release this for a while. So it’s these interesting stifles that really challenge our faith in scientism with a dollar sign.

David Deutsch

01:15:35 - 01:17:43

Right. But you represent, I think, a great deal of science with an S. And we really appreciate you, sir. You’re doing awesome. Thank you. Yeah, I agree. Keep asking the right questions. Keep that little kid inside satiated. Just remember that little five-year-old version of you, right? That’s all we have to impress, the five-year-old you, 85 year old you. That’s it. That’s me. Well, thanks for everything. And this is awesome. OK, nice meeting you. Thank you.

2024-03-05 Peter BoghossianDavid Deutsch, Peter Boghossian, and Reid Nicewonder Talk Fermi Paradox, Ideological Contagion, and More

Duration: 01:30:49

Transcript

Transavia ad narrator

00:00:00 - 00:00:08

Everyone celebrates their holidays differently. That’s why we fly to 100 destinations. Book your favourite on Transavia.com.

Peter Boghossian

Amazing. All right. Thank you so much, David. Thank you. We genuinely appreciate it. Thank you. Fun conversation. Thank you for watching. Everything we do is under the umbrella of the National Progress Alliance, nationalprogressalliance.org. It’s a nonprofit independent 501C3. Your generous donations keep us going and keep fueling content like this. So please help us out. Make a donation. We very much appreciate it. Thank you.

2024-03-04 Within Reason#57 - The Multiverse Is Real

01:35:10 - 01:35:38

Well, what a brilliant place to wrap things up after such a wide ranging and technical conversation. That’s a wonderfully optimistic place to leave things. I appreciate you taking the time. I need to take you off this fridge that I’ve laid on its side and put it upright again and onto the other side of the hotel room before someone notices that I’ve completely upended this place, so I better hop to it. But David Deutsch, thank you so much for coming on Within Reason. It’s been fun. Thank you for having me.

2024-02-17 NavalThe Deutsch Files III

00:59:12 - 00:59:14

I couldn’t agree more.

Naval Ravikant

00:59:14 - 00:59:18

Well, thank you so much, David. You’ve been incredibly generous with your time.

2024-01-26 NavalThe Deutsch Files II

01:10:28 - 01:13:15

Okay, those are not two questions, they’re one question. And I think the formulation is already a bit misleading because I think that you will want to make a better world if the goodness of the world, the goodness or otherwise of the world, figures large in your problem situation. Like Norman Borlaug, who invented the Green Revolution or whatever. For him, his problem was to make agriculture more productive. And that is inherently a world problem. And so he inherently, what he was thinking creatively about was the world. He had to take into account whether the modification to plants that he was thinking of would be expensive or cheap, and whether it would be susceptible to disease and so on. So he had to solve that range of problems. Faraday, who also saved the world by inventing electromagnetic induction, was not trying to. He was trying to solve problems about electricity, magnetism, how the world is put together, and how the different strands of the world affect each other and so on, at a level of fundamental physics. He wouldn’t have been able to conceive, I think, that electrical generators would be a matter of life and death 100 years later. And he definitely did not start out saying, how can I make a machine that will be a matter of life and death in 100 years time? He achieved that, but he wasn’t intending to. He was going flat out trying to solve his problem, which by the way, also happened to be one of the most important scientific problems of the age. But he wasn’t even trying. Again, he didn’t set out to say, first I’ll find the most scientifically interesting problem of the age, and then I’ll devote myself to it. That’s not how it went. It turned out to be that that problem was scientifically important, but he wanted to solve it because he was interested in it. He was interested in that particular thing and whether it would work. And I think Newton’s famous quote about being a boy walking along the beach, picking up pebbles that looked particularly nice to him. I think he was talking about the same thing. He had a strange personality and who knows what he was trying to do to the world. Probably nothing good, but in his scientific discoveries, he was like he said, he was trying to solve the problems that he thought were interesting. And I think in one’s everyday life, I think any deviation from that is dangerous. I don’t want to sit here giving advice. For many reasons.

Naval Ravikant

01:13:15 - 01:13:18

But you could advise the young David Deutsch who perhaps didn’t go into physics.

David Deutsch

01:13:19 - 01:14:28

Yeah, well, I’d rather tell the young David Deutsch specific things that he wanted to know. It’s dangerous to follow someone else’s problem or a problem that one thinks is important in a sense other than that it figures large in one’s own mind. When I say it’s dangerous, it’s not guaranteed to produce unhappiness, even just exactly the same way that doing the right thing in my view is guaranteed to produce happiness. Neither is guaranteed, nothing is guaranteed. We can do the right thing and still have a disaster or we can do the wrong thing and succeed. Now all these things happen, but if you want to explain how things come about by this process, thought, then it leads to certain conclusions, such as optimism and such as you said earlier, following the fun that that’s another way of looking at it. I think following the fun is what Norman Borlaug did. It’s what Faraday did. It’s what Newton did. I think there are probably people who didn’t do it, who still solved important problems, but it was a fluke.

2024-01-11 NavalThe Deutsch Files I

00:54:33 - 00:54:52

I’d very much like that to be true in the current crisis with the pogrom that’s just happened, but I don’t know whether it will. I mean, again, there have been cases before where I would have said, right, now’s the time, but it wasn’t and I don’t know whether it’ll turn out different this time, but I mean, you asked me earlier the question, is civilization in danger? I don’t think so.

2024-01-04 Reason Is Fun#5 - The Art of Decision Making

2024-01-03 Reason Is Fun#4 - Decisions

00:53:46 - 00:54:18

So fun fact. The reason I have this podcast in the first place is that I did this art of accomplishment, how to make great decisions course, and it’s coming up in about a week. And you have until January 8th to sign up. And if you use my code LULI100, you get a discount of one hundred dollars. So I super recommend this course. It has been very life changing for me. And I hope to see you there because I’m also going to be in this cohort.

2024-01-01 Arjun Khemani PodcastDavid Deutsch - Free-Will, Taking Children Seriously, and Anarcho-Capitalism

00:40:27 - 00:40:37

Well, David, I could ask you questions all day, but I think that’s a good place to end our conversation for now. Thank you so much for joining me. It’s been fun as always.

2023-12-23 David DeutschAntisemitism in Britain

00:31:12 - 00:31:14

It’s been interesting.

2023-12-19 The Joe Walker PodcastDavid Deutsch and Steven Pinker (First Ever Public Dialogue) - AGI, P(Doom), and the Enemies of Progress

01:51:45 - 01:52:05

And Steven Pinker, thank you so much for joining me. It’s been a pleasure. And I’ll just add that optimism is not just a matter of temperament, inborn or otherwise, but a matter of rationally analyzing our history and rationally analyzing what progress would consist of. Thank you both. Fantastic.

2023-12-04 IAIThe Edge of the Universe

00:49:00 - 00:49:04

Okay, everyone, well, this has been terrific. I thoroughly enjoyed it.

2023-10-16 Sean Carroll's MindscapeDavid Deutsch on Science, Complexity, and Explanation

01:42:23 - 01:42:26

Well, thank you for inviting me.

2023-08-11 NavalKnowledge Creation and the Human Race, Part 2

00:32:11 - 00:33:40

There’s a fundamental reason why even if we could look into the multiverse it wouldn’t be that much help because there is no limit to the size of error we can make. Therefore when you look around in a multiverse and see all these crystals, yes on the whole there are great big fat ones and you can guess that this one is heading towards the truth. You can’t tell where because you don’t know where this crystal is going to go and then there’ll be this other great big thing, a religion or something, which has been growing for thousands of years and there’s no way of examining it with a magnifying glass and seeing that it’s any different from one that is heading towards the truth. So we might hope that most of the big ones are heading towards the truth according to some definition of most. In one universe you can get a hint of that already because you can say what idea is most persuasive? Okay many bad ideas are persuasive. What idea is most persuasive to people who adopt it because they think it solves their problem? Okay but there are many such ideas that are false too. So I’m afraid it’s not going to work. If there were a limit to the size of error you would know that once you’ve made an error of a certain size when you have your next idea it’s bound to be true. No one can make more than 256 errors in a row would be the thing and nothing like that is true. No shortcuts.

Naval Ravikant

00:33:40 - 00:34:01

Exactly there’s no shortcut. It seems that the nature of knowledge is that it creates non-linearities so even a single false idea can create false knowledge that overwhelms the truth for quite a while in a large amount of space. Yes. So it’s always creative, it’s always conjectural, it’s always contextual, which gives an infinity of improvement ahead of us which keeps life interesting.

Original YouTube YouTube reupload Bilingual transcript by Dennis Hackethal

2023-07-26 Dennis HackethalThe Beginning of Infinity Follow-Up

Duration: 00:19:58

Dennis Hackethal published this lightly edited transcript in English and German.

Original English transcript

Dennis Hackethal (DH): Last time we spoke about superintelligence. And we concluded that there can’t be such a thing as superintelligence because there’s a universality underlying the human mind, and every other intelligence, artificial or not, would have that in common with us. And as you said earlier in that interview, there can be no such thing as something to us as we are to apes, for example.

DH: This brings us to the question of our role in the cosmos in general. For example, you say that for centuries we believed that we were at the center of the universe, that the sun and the stars revolved around us, and that there were gods who take great interest in what we down here on earth do. So everything was about us. This is anthropocentrism – the idea that people are at the center, that they’re the most important thing. Most explanations of everyday phenomena were anthropocentric.

DH: But nowadays many people in the West have since decided not to take anthropocentric approaches. And now our most fundamental explanations don’t involve humans at all. Our best explanations in physics are about things like forces or space and time. So one could conclude, as many scientifically minded people do nowadays, that people aren’t that important at all. But you take issue with this.

David Deutsch (DD): Yes. And the fact that there are only two kinds of entities in the world – the ones that can create explanatory knowledge and the ones that don’t – the connection between those two things is via the laws of physics; via the universality of the laws of physics. The universality that links us with all possible AGIs (artificial general intelligences), extraterrestrial intelligences, and so on, is precisely the universality of the laws of physics. So it’s an interesting fact that we got the concept of universal laws of physics by distancing ourselves more and more from anthropocentrism. When we had completely finished this process and got round to the modern concept of universal laws of physics, universal laws of computation, of evolution, and so on, we finally got round to realizing that that puts us at the center of things in a completely different way. Or I would say, two ways.

DD: One is that the testability that scientific theories have produces a link between understanding the world and controlling it, because controlling it is always putting to the test an understanding of it. And the understanding has got to be testable, therefore we can control more and more of the world, and the more of it we understand. There’s also the other way round, that simply to understand humans – and I’ve put this argument in the book – one must understand the laws of physics in their entirety, because in order to understand when the team of scientists are going to pop the champaign cork, and you regard that as a physical system, you want to predict when or whether it will pop the champagne cork, you must understand whether they are going to succeed in understanding the thing they’re trying to understand, which could be anything. It could be something about distant galaxies, about quarks… and it could be something about biology or morality or whatever. When they succeed, they will pop the champaign cork; if you want to predict that, you need to know the truth of the matter that they’re investigating. So to understand humans – just humans – you need to understand everything. And, conversely, as I said initially, humans can affect anything and eventually control anything that’s controllable.

DH: So humans play a central role in the cosmos – perhaps the most significant role that anything could play.

DD: Yes, I think it’s a candidate for the most significant. I mean, if someone says ‘oh no, it’s spacetime that’s the most significant’, well maybe – but certainly humans are up there in that they are completely indispensable for understanding, ultimately.

DH: Let’s get to the multiverse. You are a famous proponent of the so-called many-universes ‘interpretation’ [signals air quotes on the word ‘interpretation’] of quantum physics. How is this interpretation different from… like, what are the contenders and how is it different from them.

DD: Well, as you’ve just indicated with your air quotes, I don’t think there are any other contenders. There are various versions of this many-worlds or many-universes interpretation – really it should be called Everettian quantum theory. It’s not an interpretation, it’s just what quantum theory says when taken seriously. There are several different versions of it, including rather far-removed ones, but in order to explain what brings about the experimental results that we see in just ordinary quantum-optics laboratories one must take on board the multiplicity of reality.

DD: That is, when we do the experiment, there are many instances of us doing the experiment in parallel, or almost in parallel. You can predict the results without assuming that, but that’s just an empty prediction – it doesn’t explain how they come about. Therefore, it doesn’t involve any understanding. It just involves predicting, as in: I might tell somebody how to switch my TV on and how to reach a certain channel, but that is completely different from explaining to them how a TV works, or understanding how it works, or building a better TV. You cannot build a better TV just by being told what will happen when you press the remote. And it’s the same with physics in general.

DD: So we need this multiplicity of universes, of realities or whatever you want to call them. We don’t have good language for it yet – our language is built on the assumption that only one thing happens. But in fact, many things happen, and we understand it very well now – much better than Everett himself did in the 1950s. And it is simply the only contender for an explanation of what happens to bring about the experimental results.

DH: You say in the chapter about the multiverse that every fiction that does not violate the laws of physics is fact. Why is that the case?

DD: That is the case because physical objects do not have sharp positions. ‘Sharp’ in this terminology means ‘the same in all universes’. The laws of quantum physics require that there is no such thing as a completely sharp position of a particle or of a person; no such thing as a completely sharp configuration of a person, or of their thoughts. And so there are always a few universes doing maverick things. When I say ‘a few’, I really mean a tiny number, like 2^-(2¹⁰⁰). But still, that’s not zero. This doesn’t affect things like decisions, it doesn’t affect what we should do when we’re betting on poker games or horse races, but it does mean that if you’re talking about existence, very strange things exist somewhere in the multiverse.

DD: Now, having stressed that point, let me just claw that back a little in regard to the brain and thinking, because thinking is an error-correction process, and there are errors arising constantly in the brain – glitches, both electrical and due to temporary failed understanding which we later correct, either purely by hardware methods where the majority of neurons are saying that the minority are wrong, or by recognizing errors. So the proportion of universes in which our brain does something maverick is exponentially smaller again because error correction is exponentially efficient. By the way, that is true whether we are classical computers or quantum computers. So really, for all practical purposes, the existence of maverick versions of ourselves is negligible. But they are there, and that’s why I said that they are there – in other words, all stories that do not violate the laws of physics happen somewhere. But the ones that do violate the laws of physics don’t happen because they are uniform, they’re universal – they’re uniform across the multiverse.

DH: This is fascinating because it means that every book you’ve read, every movie you’ve seen – if the plot doesn’t violate the laws of physics it really has happened or does currently happen somewhere in the multiverse, except it’s not a movie. It’s real.

DD: Yes. The error-correction idea applies there, too. Not just in the brain, but also in a book or movie. If the book is about, say, someone doing magic, like Harry Potter or something, then, although it can’t happen that real magic occurs in these maverick universes – by the way, the term ‘maverick universes’ I think was invented by Bryce DeWitt, my former boss at the University of Texas – it just seems to have happened because something isn’t really levitating, it’s just that the atoms in the room happened to strike the apparently levitating object in just the right ways to keep it levitating. The thing is, those universes aren’t really universes because the very next thing that will happen in the vast majority of those is that the levitated object will fall to the ground. And that is because the apparent explanation of what was holding it up – namely magic – isn’t true. What was holding it up was just accident.

DD: If you think that something only deserves to be called a universe, or a universe in which a particular story happens, if the explanations in the story are true in that universe, then these Harry Popper type universes aren’t universes at all. They are just wannabe universes. They’re just bits of the multiverse. We know anyway that there are many bits of the multiverse that aren’t universes. Universes are large-scale structures in the multiverse, but there’s plenty happening there that can’t be thought of as universes. So the Harry Potter ones are among those, too.

DH: This opens up a possibility for some interesting thought experiments. For example, in the book you write about the problem of aging; and you write that you hope one day we will solve this problem. I hope so too – and the problem of death more generally. Curing death is not against the laws of physics. Somewhere in the universe it happens; it’s already happened.

DD: Somewhere in the multiverse.

DH: Sorry, yes, somewhere in the multiverse. And it also happens where people very much like us do it. Couldn’t we conclude that we don’t really need to worry about death, even right now, even though we haven’t solved death yet? Because we could imagine another universe where it’s exactly the same you with all the same memories and contents of your mind, and that universe and ours are exactly the same up until the point where people cure death. So you do get to survive, it’s just in a different universe.

DD: That seems unlikely. I mean, those of us who read Scientific American, say, will have seen hints that this was going to happen, and therefore will be different. There’s also another problem, which is that it is not yet known how to understand the issue of personal identity across the multiverse. Should somebody who accidentally turns out to have my exact state of mind be regarded as just another instance of me? Just like the ones that have a common origin with me. They are the same for a reason. The ones that are the same by accident – is that the same as the ball being held up by Harry Potter magic? In other words, not me, just looks like me? Or is it me because anybody who has my exact state of mind really is me? We don’t know yet. One has to work extremely hard to make sense of probability and decision making in the multiverse. I know because I’ve worked extremely hard to make sense of it. And to make sense of personal identity as well is a harder problem in my view. But it will be done, and we’re understanding more and more about the multiverse over time.

DH: Now, in a similar vein: the origin of life is still a mystery. There are some good explanations; you cite one, the RNA-world hypothesis, in your book, but it’s still a mystery as to why is it that life arose when it did and what the conditions were that led to that. But in the same vein you could ask: well, the formation of life is not against the laws of physics – clearly, because we’re alive – and if it’s not against the laws of physics, well then it has to happen somewhere in the multiverse. And it is only in universes where the formation of intelligent life in particular happens that somebody even wonders why it’s there.

DD: Yes, but, as I also write in the book, that by itself – which is what they call an anthropic argument – in my view isn’t enough to explain anything. I think it can form part of an explanation of things, but it can never be the whole explanation, because if it were the whole explanation of, for example, why life formed, then one can show by a simple argument due to my other old boss, Dennis Sciama, that it only just happens. The laws, the initial conditions, the state of the planet, and so on – in the vast majority of cases where that was the only explanation for why life formed it only just formed. And if it only just formed, it’s a bit hard to see why it won’t be snuffed out in the next moment. That’s not a watertight argument but it kind of takes the wind out of the sails of the anthropic argument. One really needs something else.

DD: Another reason why we should be looking for something else is that an anthropic explanation is kind of giving up. It’s kind of saying: it’s not due to a law of physics, it’s not due to some new principle along the same lines as the principle of evolution, which explained so wonderfully what were previously mysteries about life. It’s just that if it weren’t so we wouldn’t be asking. Well, I think that should be regarded as not good enough.

DH: Right. Okay, let’s wrap it up here. Again, this was a follow-up interview to the first interview we did in German for the translation of David Deutsch’s book The Beginning of Infinity. It is now available in German, titled Der Anfang der Unendlichkeit, and it’s about so much more than what we just discussed. We really only scratched the surface here. It’s also about good vs. bad philosophy, how to make decisions rationally, why flowers are beautiful, how cultures evolved, how our species evolved, the problem of sustainability… It’s a wonderful book, a fascinating read, and I encourage our German-speaking viewers and listeners to read the book. It’s guaranteed to change your worldview.

DH: It was fascinating speaking with you.

DD: Nice talking to you, and let’s see how the German edition does.

Deutsche Übersetzung

Dennis Hackethal (DH): Letztes Mal sprachen wir über Überintelligenz/Superintelligenz. Und wir kamen zu dem Schluss, dass es so etwas wie Überintelligenz nicht geben kann, weil dem menschlichen Verstand eine Universalität zugrunde liegt, die jede andere Intelligenz, ob künstlich oder nicht, mit uns gemeinsam hätte. Und wie Sie bereits in jenem Interview sagten, kann es nichts geben, das in einem Verhältnis zu uns steht, wie wir zum Beispiel zu Affen stehen.

DH: Das bringt uns zu der Frage, welche Stellung wir im Kosmos einnehmen. Sie schreiben zum Beispiel, wir hätten jahrhundertelang geglaubt, dass wir im Zentrum des Universums stehen, dass sich die Sonne und die Sterne um uns drehen und dass es Götter gibt, die sich sehr dafür interessieren, was wir hier unten auf der Erde tun. Also drehte sich alles um uns. Das ist der Anthropozentrismus – die Vorstellung, dass der Mensch im Mittelpunkt steht, dass er das Wichtigste ist. Die meisten Erklärungen für Alltagsphänomene waren anthropozentrisch.

DH: Doch heutzutage haben viele Menschen im Westen beschlossen, keine anthropozentrischen Ansätze mehr zu verfolgen. Unsere grundlegendsten Erklärungen haben mit dem Menschen nichts mehr zu tun. Unsere besten Erklärungen in der Physik beziehen sich auf Dinge wie Kräfte oder Raum und Zeit. Man könnte also zu dem Schluss kommen, wie es viele wissenschaftlich denkende Menschen heutzutage tun, dass der Mensch gar nicht so wichtig ist. Aber Sie widersprechen dem.

David Deutsch (DD): Ja. Und die Tatsache, dass es nur zwei Arten von Entitäten auf der Welt gibt – diejenigen, die erklärendes Wissen schaffen können, und diejenigen, die das nicht tun – der Zusammenhang zwischen diesen beiden Dingen ergibt sich durch die physikalischen Gesetze; durch die Universalität der physikalischen Gesetze. Die Universalität, die uns mit allen nur möglichen KAIs (künstlichen allgemeinen Intelligenzen), außerirdischen Intelligenzen usw. verbindet, ist ebendiese Universalität der physikalischen Gesetze. Es ist also interessant, dass wir das Konzept der universellen physikalischen Gesetze erlangten, indem wir uns mehr und mehr vom Anthropozentrismus entfernten. Als wir diesen Prozess vollständig abgeschlossen hatten und zu dem modernen Konzept der universellen physikalischen Gesetze, der universellen Gesetze der Berechnung, der Evolution und so weiter gekommen waren, stellten wir schließlich fest, dass wir damit auf eine ganz andere Weise im Mittelpunkt der Dinge stehen. Oder ich würde sagen, auf zwei Weisen.

DD: Eine davon lautet, dass die Überprüfbarkeit wissenschaftlicher Theorien einen Zusammenhang zwischen dem Verständnis der Welt und der Kontrolle über sie herstellt, denn die Kontrolle über sie bedeutet immer, dass ein Verständnis der Welt überprüft wird. Und das Verständnis muss überprüfbar sein, weshalb wir die Welt immer besser kontrollieren können, je mehr wir über sie wissen. Um den Menschen zu verstehen – und dieses Argument habe ich in meinem Buch angeführt –, müsste man die physikalischen Gesetze komplett verstehen, denn um zu verstehen, wann das Team von Wissenschaftlern den Champagnerkorken knallen lässt – Sie betrachten das als ein physikalisches System und Sie wollen vorhersagen, wann oder ob dieses Team den Champagnerkorken knallen lässt –, müssen Sie verstehen, ob es ihnen gelingen wird, den Sachverhalt zu verstehen, den sie zu verstehen versuchen. Und dabei könnte es sich ja um alles mögliche handeln. Es könnte mit fernen Galaxien zu tun haben, mit Quarks … und es könnte mit der Biologie zu tun haben oder mit der Moral oder was auch immer. Wenn es ihnen gelingt, werden sie den Champagnerkorken knallen lassen; wenn man das vorhersagen will, muss man die Wahrheit hinter dem Sachverhalt kennen, den sie erforschen. Um den Menschen zu verstehen – bloß den Menschen –, muss man also alles verstehen. Umgekehrt kann der Mensch, wie ich eingangs sagte, alles beeinflussen und schließlich alles kontrollieren, was kontrollierbar ist.

DH: Der Mensch nimmt also eine zentrale Stellung im Kosmos ein – vielleicht die bedeutsamste Stellung, die es überhaupt gibt.

DD: Ja, ich denke, sie ist ein Kandidat. Gut, wenn jetzt jemand sagt: »Ach nein, es ist die Raumzeit, die am bedeutsamsten ist«, nun ja, vielleicht – aber der Mensch ist auf jeden Fall ganz vorne dabei, weil er für unser Verständnis letztlich völlig unerlässlich ist.

DH: Lassen Sie uns zum Multiversum kommen. Sie sind ein berühmter Verfechter der sogenannten Viele-Universen-›Interpretation‹ [signalisiert Anführungszeichen beim Wort ›Interpretation‹] der Quantenphysik. Wie unterscheidet sich diese Interpretation von … also, wie lauten die Mitstreitertheorien und wie unterscheidet sie sich von ihnen.

DD: Nun, wie Sie gerade mit Ihren Anführungszeichen angedeutet haben, glaube ich nicht, dass es noch andere Mitstreiter gibt. Es gibt verschiedene Versionen dieser Viele-Welten- oder Viele-Universen-Interpretation – eigentlich sollte sie Everett’sche Quantentheorie genannt werden. Es handelt sich nicht um eine Interpretation, sondern nur um das, was die Quantentheorie aussagt, wenn man sie ernst nimmt. Es gibt verschiedene Versionen davon, auch ziemlich weit hergeholte, aber um zu erklären, was zu den experimentellen Resultaten führt, die wir in ganz gewöhnlichen quantenoptischen Labors beobachten, muss man die Mannigfaltigkeit der Realität in Betracht ziehen.

DD: Das heißt, wenn wir das Experiment durchführen, gibt es viele Exemplare von uns, die das Experiment parallel oder nahezu parallel durchführen. Man kann die Resultate vorhersagen, ohne dies vorauszusetzen, aber dann handelt es sich nur um eine leere Vorhersage – sie erklärt nicht, wie die Resultate zustande kommen. Es ginge dabei also nicht darum, etwas zu verstehen, sondern nur um Vorhersagen. Zum Beispiel könnte ich jemandem sagen, wie er meinen Fernseher einschalten und einen bestimmten Sender empfangen kann, aber das unterscheidet sich grundlegend davon, ihm zu erklären, wie ein Fernseher funktioniert, oder vom Verständnis, wie er funktioniert, oder von der Herstellung eines besseren Fernsehers. Man kann keinen besseren Fernseher herstellen, wenn man lediglich erklärt bekommt, was passiert, wenn man die Fernbedienung drückt. Ebenso verhält es sich mit der Physik im Allgemeinen.

DD: Wir brauchen also diese Vielzahl von Universen, von Realitäten oder wie man sie auch nennen will. Wir haben noch keine gute Ausdrucksweise dafür – unsere Sprache beruht auf der Annahme, dass nur ein Ereignis stattfindet. Aber in Wirklichkeit finden viele Ereignisse statt, und wir verstehen das jetzt sehr gut – viel besser als Everett selbst in den 1950er Jahren. Und dies ist einfach der einzige Kandidat für eine Erklärung dessen, was passiert, um die experimentellen Resultate zu bewirken.

DH: In dem Kapitel über das Multiversum schreiben Sie, dass jede Fiktion, die nicht gegen die physikalischen Gesetze verstößt, Fakt ist. Warum ist das der Fall?

DD: Das ist deshalb der Fall, weil sich physische Objekte nicht an scharfen Orten befinden. ›Scharf‹ bedeutet in dieser Terminologie ›in allen Universen gleich‹. Die quantenphysikalischen Gesetze verlangen, dass es keinen völlig scharfen Ort eines Teilchens oder einer Person gibt; keine völlig scharfe Konfiguration einer Person oder ihrer Gedanken. Und so gibt es immer ein paar Universen, die Einzelgänger sind. Mit ›ein paar‹ meine ich wirklich eine winzige Zahl, etwa 2^-(2¹⁰⁰). Aber trotzdem ist das nicht null. Das hat keinen Einfluss auf Entscheidungen, es hat keinen Einfluss darauf, was wir tun sollten, wenn wir Poker spielen oder auf Pferde wetten, aber es bedeutet im Hinblick auf die Frage, was es wirklich gibt, dass es irgendwo im Multiversum sehr seltsame Dinge gibt.

DD: Lassen Sie mich diesen Punkt in Bezug auf das Gehirn und das Denken ein wenig zurückschrauben, denn das Denken ist ein Prozess, der Fehler korrigiert, und im Gehirn treten ständig Fehler auf – also Störungen, sowohl elektrische als auch solche, die auf ein vorübergehendes Unverständnis zurückzuführen sind, das wir später korrigieren, entweder durch reine Hardwaremethoden, bei denen die Mehrheit der Neuronen sagt, dass die Minderheit falsch liegt, oder durch das Feststellen von Fehlern. Der Anteil der Universen, in denen unser Gehirn ein Einzelgänger ist, ist also wiederum exponentiell kleiner, weil die Fehlerkorrektur exponentiell effizient ist. Das gilt übrigens unabhängig davon, ob wir klassische Computer oder Quantencomputer sind. In der Praxis ist die Existenz von Einzelgängerversionen unserer selbst also zu vernachlässigen. Aber es gibt sie, und deshalb habe ich gesagt, dass es sie gibt – mit anderen Worten: Alle Erzählungen, die nicht gegen die physikalischen Gesetze verstoßen, finden irgendwo statt. Aber diejenigen, die doch gegen die physikalischen Gesetze verstoßen, finden nicht statt, weil sie einheitlich sind. Sie sind universell – sie sind über das gesamte Multiversum hinweg einheitlich.

DH: Das ist faszinierend, denn es bedeutet, dass jedes Buch, das Sie gelesen haben, jeder Film, den Sie gesehen haben – wenn die Handlung nicht gegen die physikalischen Gesetze verstößt, hat sie wirklich stattgefunden oder findet derzeit irgendwo im Multiversum statt. Jedoch handelt es sich dabei um keinen Film – es ist echt.

DD: Ja. Auch hier gilt das Konzept der Fehlerkorrektur. Nicht nur im Gehirn, sondern auch in einem Buch oder Film. Wenn es in dem Buch beispielsweise um jemanden geht, der zaubert, wie Harry Potter oder so, dann kann es nicht sein, dass in diesen Einzelgängeruniversen – der Begriff ›maverick universes‹ geht übrigens, glaube ich, auf Bryce DeWitt zurück, meinen ehemaligen Chef an der University of Texas – echte Magie betrieben wird. Es scheint nur so zu sein, denn da schwebt eigentlich nichts. Die Atome in dem Zimmer haben das Objekt zufällig genau richtig getroffen, damit es schwebt. Die Sache ist die, dass diese Universen keine wirklichen Universen sind, denn in den allermeisten von ihnen wird das schwebende Objekt als Nächstes zu Boden fallen. Und das liegt daran, dass die vermeintliche Erklärung dafür, was es in der Schwebe gehalten hat – nämlich Magie – nicht stimmt. Es war einfach nur Zufall.

DD: Wenn man der Meinung ist, dass etwas nur dann als Universum bezeichnet werden kann – oder als Universum, in dem sich eine bestimmte Handlung abspielt –, wenn die Erklärungen in der Handlung in diesem Universum wahr sind, dann sind diese Harry-Popper-artigen Universen gar keine Universen. Sie sind nur Möchtegernuniversen. Sie sind nur Teile des Multiversums. Wir wissen ohnehin, dass es viele Teile des Multiversums gibt, die keine Universen sind. Universen sind große Strukturen im Multiversum, aber es gibt dort auch vieles, was nicht als Universum betrachtet werden kann. Die Harry-Potter-Universen gehören also auch dazu.

DH: Dadurch ergibt sich die Gelegenheit für einige interessante Gedankenexperimente. In Ihrem Buch schreiben Sie zum Beispiel über das Problem des Alterns, und Sie schreiben, Sie hoffen, dass wir dieses Problem eines Tages lösen. Das hoffe ich auch – und das Problem des Todes im Allgemeinen. Die Heilung des Todes verstößt nicht gegen die physikalischen Gesetze. Irgendwo im Universum wird er also geheilt, ist er bereits geheilt worden.

DD: Irgendwo im Multiversum.

DH: Entschuldigung, ja, irgendwo im Multiversum. Und er wird auch von Menschen geheilt, die uns sehr ähnlich sind. Könnten wir nicht zu dem Schluss kommen, dass wir uns über den Tod eigentlich keine Sorgen machen müssen, auch jetzt nicht, obwohl wir den Tod noch nicht überwunden haben? Denn wir könnten uns ein anderes Universum vorstellen, in dem es genau derselbe Mensch ist, mit all denselben Erinnerungen und Inhalten Ihres Verstandes, und dieses Universum und das unsere sind genau gleich, bis zu dem Punkt, an dem wir den Tod heilen. Sie überleben also doch, nur eben in einem anderen Universum.

DD: Das klingt unwahrscheinlich. Ich meine, diejenigen von uns, die beispielsweise Scientific American lesen, werden Hinweise darauf gesehen haben, dass dies geschehen würde, und werden daher anders sein. Es gibt noch ein weiteres Problem, nämlich dass man noch nicht weiß, wie man das Problem der persönlichen Identität im Multiversum verstehen soll. Sollte jemand, der zufällig genau meinen Geisteszustand hat, als ein weiteres Exemplar von mir betrachtet werden? So wie die, die einen gemeinsamen Ursprung mit mir haben. Die sind aus gutem Grund gleich. Diejenigen, die zufällig gleich sind – ist das dasselbe wie der Ball, der durch Harry Potters Magie in der Schwebe gehalten wird? Wäre das mit anderen Worten also nicht ich selbst, sondern würde nur so aussehen wie ich? Oder bin ich es doch, weil jeder, der genau meinen Geisteszustand hat, wirklich ich ist? Das wissen wir noch nicht. Man muss sich schon sehr anstrengen, um sich einen Reim auf Wahrscheinlichkeiten und die Entscheidungsfindung im Multiversum zu machen. Ich weiß das, weil ich mich sehr angestrengt habe, um das zu verstehen. Und auch die persönliche Identität zu verstehen, ist meiner Meinung nach ein noch schwierigeres Problem. Aber es wird gelingen, und wir verstehen mit der Zeit immer mehr über das Multiversum.

DH: Gut, in Anlehnung daran: Der Ursprung des Lebens ist immer noch ein Rätsel. Es gibt zwar ein paar gute Erklärungen; Sie führen eine davon, die RNA-Welt-Hypothese, in Ihrem Buch an, aber es ist immer noch ein Rätsel, warum das Leben zu jenem bestimmten Zeitpunkt entstanden ist und was die Bedingungen waren, die dazu führten. Aber genauso könnte man sich fragen: Die Entstehung von Leben verstößt nicht gegen die physikalischen Gesetze – das ist klar, denn wir leben ja –, und wenn sie nicht gegen die physikalischen Gesetze verstößt, dann muss sie nunmal irgendwo im Multiversum stattfinden. Und nur in den Universen, in denen sich speziell intelligentes Leben gebildet hat, fragt sich jemand, woran das liegt.

DD: Ja, aber, wie ich auch in meinem Buch schreibe, reicht dieses anthropische Argument, wie es genannt wird, meiner Meinung nach nicht aus, um irgendetwas zu erklären. Ich denke, es kann zwar Teil einer Erklärung sein, aber niemals die vollständige Erklärung, denn wenn es die vollständige Erklärung dafür wäre, warum sich zum Beispiel Leben gebildet hat, dann kann man mit einem einfachen Argument, das auf meinen anderen alten Chef, Dennis Sciama, zurückgeht, zeigen, dass es nur gerade so passiert. Die Gesetze, die Anfangsbedingungen, der Zustand des Planeten und so weiter – in den allermeisten Fällen, in denen das die einzige Erklärung dafür wäre, warum sich das Leben gebildet hat, hat es sich nur gerade so gebildet. Und wenn es sich nur gerade so gebildet hat, ist es schwer nachzuvollziehen, warum es nicht im nächsten Moment wieder ausgelöscht wird. Das ist jetzt kein wasserdichtes Argument, aber es nimmt dem anthropischen Argument irgendwie den Wind aus den Segeln. Man braucht wirklich etwas anderes.

DD: Ein weiterer Grund, warum wir nach etwas anderem suchen sollten, liegt darin, dass man mit der anthropischen Erklärung sozusagen aufgibt. Sie sagt gewissermaßen: Es liegt nicht an einem physikalischen Gesetz oder an einem neuen Prinzip wie dem der Evolution, das die früheren Rätsel des Lebens so wunderbar erklärt. Es ist demnach lediglich so, dass wir nicht fragen würden, wenn es nicht so wäre. Nun, ich denke, man sollte das als nicht gut genug ansehen.

DH: Ich verstehe. Okay, lassen Sie uns hier zum Schluss kommen. Nochmals, dies war ein Folgeinterview zum ersten Interview, das wir auf Deutsch für die Übersetzung von David Deutschs Buch Der Anfang der Unendlichkeit geführt haben. Es ist jetzt erhältlich und handelt von sehr viel mehr als dem, was wir gerade besprochen haben. Wir haben hier wirklich nur an der Oberfläche gekratzt. Es handelt auch von guter und schlechter Philosophie, davon, wie man rationale Entscheidungen trifft, warum Blumen schön sind, wie sich Kulturen evolutionär entwickelt haben, wie sich unsere Art entwickelt hat, vom Problem der Nachhaltigkeit … Es ist ein wunderbares Buch, und ich empfehle unseren deutschsprachigen Zuschauern und Zuhörern diese faszinierende Lektüre. Es wird garantiert Ihre Weltanschauung verändern.

DH: Das war ein faszinierendes Gespräch.

DD: Es hat mich gefreut, und mal sehen, wie sich die deutsche Ausgabe so macht.

2023-06-28 Bennett and DeutschThe Nature of Computation, Incompleteness and Mathematics

01:53:42 - 01:53:44

Thank you very much.

David Deutsch

01:53:44 - 01:53:52

Very, very interesting conversation. Thank you. Bye, everyone. Thanks a lot, this was great. Bye-bye.

2023-06-26 Everett@50Apart from Universes

00:53:19 - 00:53:49

Absolutely. And to some extent I have made a conjecture about what the multiverse is like. I think it’s probably almost certainly too naive and we’ll be wrong. But that is essentially what [unclear name] conjectured 40 years ago. So I wouldn’t totally dismiss that as a possibility. I said it was worth trying heuristically, and it has been tried, but it seems to be running out of steam. So there seems to be an obvious explanation for why it’s running out of steam.

Oliver Pooley

00:53:49 - 00:53:56

Okay, I think we should call this part of the session to a close, and thank David again for a wonderful speech.

2023-06-03 Reason Is Fun#3 - Inexplicit Ideas and Embodiment

01:15:58 - 01:16:01

Well, I look forward to continuing.

2023-05-23 Reason Is Fun#2 - Physics of Inexplicit Ideas

00:59:42 - 01:11:36

Can be there and so you’re saying that there is a type of thinking called conscious thinking that is different from unconscious thinking what does conscious mean conscious things have qualia associated with them they are there is such a thing as what it is like to have that thought so you think it’s possible to translate inexplicit ideas to explicit ones because explicit language is universal yes how is it not the case that inexplicit ideas are something like containing a massive context such that so an explicit idea can be expressed as, say, a proposition or a statement however perhaps inexplicit ideas cannot be expressed that way like they cannot be about a specific narrow thing because the nature of inexplicit ideas maybe is that they are taking into account the whole context or there’s a way that they are not making these divisions like you can make with statements and propositions and so that perhaps it’s something like a single inexplicit idea is actually connected to all of the other inexplicit ideas in a brain and this isn’t actually translatable to explicit language because of the nature of what language is like well I’m very sympathetic to the idea that most ideas in the brain in the mind are spread out oh and take into account a lot of context or you know maybe all contexts or whatever but the thing is however large they are they’re not infinitely large and the brain is not infinitely large and it only has a certain number of neurons that can be in a certain number of states and in principle you could write down the exact state and that would include the state of all your inexplicit ideas is it possible that inexplicit ideas also are stored or contained or refer to physical things in the physical environment so in other words so with explicit ideas we’ve got words and you can program them into a computer with inexplicit ideas not only do we have the like the sense that something and the parts that we can translate into words we also have reactions physical reactions to things in our environment so for example if you are afraid of dogs and you notice that there is a dog and then your body will physically start up the fight or flight processes which then affect the inexplicit ideas around dogs and so could it be that inexplicit ideas are more distributed as in they are not just in the brain but they are partly in the interaction between the brain and the environment yes this doesn’t have any effect on whether they could be made fully explicit or not because the environment is also finite and in principle you could is it though don’t we have an infinite environment oh well we may well have an infinite environment but only a finite amount of it can possibly affect anything in your mind because of the finite speed of light oh ah but no but counter things that are far away affect things that are less far away which affect you and so everything is connected it affects you later so for example when you’re looking at the sun you are seeing it as it was eight minutes ago and so okay but then all of that stuff still affects you even if it was not at the time yeah but why does that matter because if it only affects you after the thing you’re trying to explain then it can’t be involved in the explanation ah so you’re saying that when you are trying to explain something you are only trying to explain a specific small piece and that not everything is going to be relevant to explaining that specific small piece however if you want to properly translate the inexplicit ideas which would contain everything then everything does affect you no so if you were to translate the contents of the brain and you know the gut and whatever else is within a light travel time of your view into explicit form in the positions and motions of atoms and that kind of thing if you did that explicitly then it’s true that that contains information about what you would do if for example a lion suddenly appeared in the room that you can although there’s nothing about lions in any of that let’s suppose there’s nothing about lions in encoded in any of those atoms it nevertheless follows from the atoms and from the laws of physics what you would do if this unlikely and unexpected thing were to happen so and that’s all finite again and it could be in principle encoded in explicit form that would not include explanations of course you might be able to work out from this that if the lion came you would suddenly you know say the word frog if you wanted to explain that that might be a lot more work but it would be work that again could be done explicitly on that explicit representation what do you think are the most interesting questions about inexplicit ideas I think the issue of qualia sort of touches all these things like consciousness and AGI and explicit and inexplicit and all that you spoke of really amazing things before one of the really amazing things is that it is definitely possible that you could have a box a metal box and inside that there was a thinking being and there could be inside that there could be a thinking being who is interacting with a virtual world and they might not even know that they are in a virtual world and it might be like the matrix and so on but the point is the amazing thing is in there. Now the box is finite it has a finite number of bits or a finite number of qubits let’s say it has a trillion bits in it the amazing thing that I’m pointing to is that some of them. So then there are a trillion bits therefore if it’s a classical computer let’s say there are two to the power of a trillion possible states of that computer inside the box the amazing thing is that some of those states are immoral some of them are so immoral that they are worse than anything that’s ever happened to humans ever what do you mean some of those states are immoral well some of them involve a person suffering or people suffering or people suffering more than it’s physically possible for a human to suffer all those things I’ve just mentioned are possible states of these bits and this is related to the simulation hypothesis yes well and possible basilisks the simulation hypothesis says that this has actually happened and we’re in it and I disagree with that but that it is physically possible for it to happen I think is a pretty direct consequence of our knowledge of physics so it could happen and therefore there are immoral numbers because this thing can be represented as a binary number the entire contents of the box can be represented as a number with a trillion digits all zero or one some of those are unimaginably evil numbers we have no idea and some of them have qualia of course I think only the ones that have qualia are evil the ones that don’t have qualia aren’t evil so we don’t know how to tell the difference we just know that you know you could start with zero one two three four five up to a trillion for each one you would say bit number three so when you say that the numbers can be evil what you’re really saying is that a human consciousness can be programmed into a computer which is represented by numbers and so I mean really the evil thing is about the emergent properties namely the consciousness and the suffering and the numbers are sort of not really here nor there you know if it turns out that we are not represented by numbers and we need something else then that would also be uh evil so if the computer isn’t running then there’s nothing evil there it’s got to be running so that I would like to run so yeah well the numbers are changing and it’s the change that actually embodies suffering and therefore it’s the change but we don’t know I mean it’s neither here nor there we don’t know what states what numbers what programs are good or evil we don’t know which of them are AGIs and which aren’t we don’t know how to tell the difference certainly not by the Turing test because this box has got no inputs or outputs so in answer to the question what are the most interesting questions around inexplicit ideas you would say what makes something conscious or not what is with this qualia stuff what is with this suffering thing I think you know if I have to pick one of them as being more like urgent to solve or whatever it’s the qualia one but I also think that they’ll all be solved by the same theory in a single day in a single hour somebody will have the right idea and that will solve all of them though they may not realize it at first wow well I think that’s a great place to end it it’s been fun as always

2023-05-09 Conversations with ColemanMultiverse of Madness with David Deutsch

01:35:25 - 01:35:27

Likewise, David. Thanks.

2023-05-06 Reason Is Fun#1 - Fun, Fury, Feeling

01:12:37 - 01:12:45

Yes. Cool. Alright, well I think that’s all of the time that we have for today, so this was fun, let’s do it more. Yes, let’s.

2023-04-26 Foresight InstituteOn Beauty, Knowledge, and Progress

01:04:03 - 01:04:04

Thanks for having me.

2023-04-22 Reason Is Fun#0 - Effective Altruism, X-risk, e-acc

01:31:56 - 01:40:03

Hey this is Beff Jezos here thanks for having me very honored to have David Deutsch talk about EAC on your podcast so I have a few comments on a few points that were made and I’d like to provide a few clarifications so I’ll get right into it here so one thing to comment on was David’s comment about black body radiation and how inanimate objects are actually pretty good at dissipating away heat ultimately life forms can adapt and search their environment to find new pockets of free energy not just dissipate free energy from radiation that is impinging upon them naturally like an inanimate object would and so ultimately they’re a more non-local optimum in terms of their ability to dissipate free energy as heat and so that’s why they’re thermodynamically favored at least by the laws of out-of-equilibrium thermodynamics another thing that David criticized was the potential connection between thermodynamic driven dissipative adaptation and evolution and ultimately how the connection is weak there ultimately one can see evolution as a form of matter searching over the space of configurations for the configurations that are naturally very good at replication and finding pockets of free energy in their environment that they can consume towards sustenance and replication and I mean that that post-selection is very evident if you have organisms that aren’t good at securing energy to maintain their coherence and or aren’t good at replicating then they will be post-selected against in the natural selection so I guess England’s point is that actually evolution is just a byproduct of thermodynamic driven dissipative adaptation and that is something I would agree with personally one more comment from david was claiming that thermodynamic driven dissipative adaptation has nothing to do with intelligence or the emergence of intelligence you know on one hand I would say that the thermodynamic adaptation itself is a form of intelligence that’s a bold claim that I look forward to writing more about but in the meantime it’s very clear that if you believe that thermodynamic driven dissipative adaptation is behind the process of evolution then it’s necessarily transitively behind the emergence of intelligence which emerged through evolution why did intelligence emerge through evolution and through thermodynamic adaptation well higher forms of intelligence allow life forms to be able to interpolate and extrapolate you know utility towards securing resources and free energy towards sustenance and growth of themselves as an individual or towards the sustenance and growth of their group or their meta organism and so higher forms of intelligence allow for higher forms of growth ultimately you know humans are the most intelligent beings for the moment and clearly we’ve found ways to consume vast amounts of free energy towards our sustenance and growth of our population and our civilization as a whole and so in a sense the claim is that you know intelligence is very important for the ability to secure and consume free energy towards sustenance and growth and extract utility from that free energy right not just to burn it but consume it in a way that maximizes expected growth and the future and ultimately intelligence at least to me is the ability to perceive predict and control the world around us and those are all necessary in order to secure ever more esoteric and difficult to access pockets of free energy whether it’s building technologies for say oil drilling or splitting the atom or figuring out nuclear fusion there is free energy around us that’s just the harder and harder to access and needs higher and higher levels of intelligence and technology to unlock and so in that sense if you have if you believe in this post-selective effect on configurations of matter towards configurations of matter that are better at securing and consuming free energy then intelligence is a byproduct so that is the hypothesis there so one last point about the application of the theory of thermodynamic dissipative adaptation to capitalism and the economy the point here is that in this kind of model capital or money is a proxy or a replacement for energy and in thermodynamic theories right because each organism or meta organism so let’s say a corporation dissipates energy into the environment so they consume free energy or free capital towards their sustenance right keeping a company alive costs money and they are post-selected for via the market for their ability to secure more capital and use that capital towards their sustenance and growth right kind of by construction and that is very similar to life forms and their post-selection with respect to their ability to secure energetic or free energetic capital and allocate it internally towards their sustenance and their growth or replication in that case and so it is very much a similar concept at least to me this sort of adaptive algorithm is a form of intelligence itself and ultimately here you know it wasn’t about how much a corporation actually spends in terms of energy to sustain itself like literal energy but rather capital some money but ultimately energy and money are intimately tied right we have markets for energy we can trade capital for energy cryptocurrencies want to tie the two concepts sometimes literally through proof of work for example but ultimately there is a sort of post-selection effect towards meta organisms or groups that are best at acquiring finding ways to acquire capital and put it to use towards their sustenance and growth and that requires a form of intelligence and that form of intelligence for groups and corporations isn’t just at an individual level but also a cultural level you know culture would be a form of group intelligence and there is a cultural post-selection effect right companies that have the wrong culture groups have the wrong culture that doesn’t lead to their success at sustenance and growth get post-selected out and so that is also why we encourage variance of ideas variance of cultures this search over the optimal culture given that the market forces in the environment is forever changing it’s important to maintain variance to maintain dynamic adaptability on fast time scales

2023-04-08 Reason Is Fun#-1 - AGI with David Deutsch

00:59:07 - 00:59:10

Do you think this is going to revolutionize the economy?

David Deutsch

00:59:11 - 01:01:14

I don’t know. I can’t prophesy the economy and I can’t prophesy the applications of modern AIs, chatbots and so on. I have found, for what it’s worth as it were, I have found it useful, but not revolutionary. In my own work and in my writing and whatever, it is useful, but I can’t see a possibility for it to revolutionize what I do. Whether it can revolutionize the economy depends on something slightly different, because for that it doesn’t need to really have a fundamental new functionality. It could be that a lot of existing jobs, and people are scared that computer programming is one of them, where only a proportion of the job, let’s say 10% of a particular programming job, involves human creativity and the rest is basically hack work. Which is a big if, because I’m not convinced of this either. If chatbots can reliably perform the hack work, then it might be argued, I think again wrongly, that if a given task can be done with only a tenth as much work, then we might need only a tenth as many programmers in the long run. And unfortunately we got interrupted, so David never finished his thought about why programmer jobs might be safe. But if you have any questions about anything in this episode, leave them on the tweet, which I will link in the show notes about this episode. Thank you.

2023-03-23 The Tim Ferriss Show#662 - David Deutsch and Naval Ravikant

01:46:38 - 01:46:43

That’s why I love FreshBooks and have been recommending it for years on this podcast.

Tim Ferriss

01:46:43 - 01:48:09

FreshBooks makes managing your money and your business easier from easy invoicing to time saving automations. FreshBooks simplifies accounting and bookkeeping and ensures you’re ready for tax time. FreshBooks was built for business owners and accounting professionals. It is simple, simple, simple. Based on a benefit survey of more than 10,000 FreshBooks customers, you can save up to 11 hours a week by streamlining and automating bookkeeping and accounting tasks like time tracking, invoicing and expense tracking. You can also create professional branded invoices in minutes. More than 30 million people have used FreshBooks and love it for its intuitive dashboard and reports. It’s easy to see at a glance exactly where your business stands, and it’s even easier to turn everything over to your accountant come tax season. 94% of FreshBooks users say it’s super easy to get up and running, and with award winning support, you are never alone. FreshBooks is the all in one accounting software that can save you up to 11 hours per week. That’s 11 hours that get freed up so you can spend more time nailing a client pitch, serving your customers or honing your craft. And right now there’s a special offer just for you, my dear listeners. Head over to FreshBooks.com slash Tim to get 90% off of your FreshBooks subscription for four months. That’s FreshBooks.com slash Tim. One more time, check it out, learn more and get 90% off of your subscription for four months at FreshBooks.com slash Tim.

2023-03-07 Fallible AnimalsThe Universal Constructor

00:25:26 - 00:25:31

I was about to say that. Yes, thank you. Thank you. Bye. Bye-bye.

2022-11-11 QuSoft CWI Breakthrough Prize SeminarBennett, Brassard, Deutsch, Shor

02:48:57 - 02:48:59

XOR all those pictures together.

2022-10-27 David DeutschWhy Is Everettian Quantum Theory Not in the Consensus

Duration: 00:09:39

Transcript

David Deutsch

00:00:00 - 00:02:17

Lev Vaidman has asked us this question. Why is the Many Worlds interpretation not in the consensus? And he adds, I firmly believe that the Many Worlds interpretation should be in the consensus and want to know why it is not. I agree and I too don’t know why. It may be that only future historians of ideas will succeed in explaining it after the reality of multiple universes has become as uncontroversial as that of dinosaurs and their evolution is today. But perhaps I can say a few things that might make this seem a little less mysterious and less reprehensible. One thing that future historians might notice is that today there are strong divergences of opinion among proponents of the theory. We all agree that it describes reality, not just predictions, not just our confidence in predictions, not just our states of knowledge and so on. But we disagree about what reality it does describe. Is the world that we perceive as n particles in a three dimensional space actually a complex function on a three n dimensional one? Does that function describe reality? Or is it itself the reality? Is the reality described by Q number observables, as I think? And what about time? What about space time and quantum fields and quantum gravity? And are the effects of one quantum object on another purely local or non-local? What is the origin of probability in this theory? On all these issues, we disagree.

00:02:17 - 00:04:36

Indeed Lev’s question asks about the many worlds interpretation. I disagree with that very term. Who would call the theory of evolution the many dinosaurs theory? Nor is Everettian theory an interpretation of quantum theory or anything else. It is quantum theory. It should just be quantum theory or call it Everettian if you want to give due credit to the corrector of earlier untenable misconceptions. In my view, splitting an explanatory theory into its formalism and its interpretation is itself a reinterpretation which, if insisted upon, is never more than institutionalised casuistry. It follows from the existence of these disagreements that most or all of today’s advocates of Everettian quantum theory must be mistaken about what the world according to quantum theory is actually like. Hence, those future historians might ask what exactly were the unpersuaded unpersuaded of? But this internal dissension theory won’t do as an explanation of Lev’s mystery. In the early days of relativity, people disagreed about what the so-called Schwarzschild singularity was, whether space was finite or infinite, static or expanding. Einstein even wrote to the Times in 1919, categorically denying that his theory supersedes Newton’s. Nevertheless, the vast majority of physicists were soon incorporating Einsteinian relativity into their worldviews. Lev’s problem, our problem, is that the same has not happened in regard to quantum theory.

00:04:36 - 00:06:54

We won’t find the answer in Kuhn’s theory of scientists being stuck in so-called paradigms. Such theories would apply to all fundamental progress, including relativity. The people most prominently stuck in their paradigm are actually the believers in paradigms. So what is special about quantum theory that makes physicists reluctant to do what physicists do, namely apply the theory, test it, see what it says about the world, and let that inform your worldview? Why do they spin off into magic and double-talk or denial or wishful thinking? Many other insights about reality with which science confronts us are much more jarring. For example, although no one is jarred by the idea that when we see the Andromeda galaxy in the sky we are seeing it not as it is now but as it was a long time ago, everyone is jarred by the Einsteinian idea that there is no such thing as how it is now. If you make an arbitrary choice of an event in Andromeda to be now, all you have to do is walk up and down the street on earth to change what is now at Andromeda by weeks backwards or forwards. The relativity of simultaneity. Surely that’s harder to accept than other universes. Except for one thing, there aren’t other instances of ourselves at Andromeda. No other me’s, some identical, some different, and some very different, in which case who am I? Nothing in relativity raises any such issue of personal identity. In fact, the only other fundamental theory that does so is the theory of computation with its implication that artificial minds are possible.

00:06:54 - 00:09:26

There too, people are wont to talk nonsense in order to avoid plainly applying the theory in the way they would apply any other. I don’t claim that any of this could fully answer Lev’s question even if it’s true. Still less would it provide a way to restore reason to this issue. But in regard to reason, let me close with a wild guess. In the 18th century, when they thought of themselves as living in the age of reason, they were much more gullible about the scope of science than today. People were ready to accept that chess-playing automata already existed, that horses had been trained to count, and in the 19th century that Mary Shelley’s Frankenstein and Jules Verne’s submarines were just around the corner. But in parallel with this unreasonable gullibility, or perhaps it was reasonable, the 19th century also saw rising scepticism with Ada Lovelace of all people denying that machines can think, and later the introduction of positivism and other poison into the thinking of scientists. So here’s my speculation. What else was growing during the 19th century? Science teaching, science lessons and courses in universities in which generations learned that you don’t have to change your worldview, you don’t have to understand, you only have to learn to say what one is supposed to say, be able to calculate what one is supposed to be able to calculate, sneer at what one is supposed to sneer at. I don’t mean just astrology and creationism, but eventually at reality itself. That may be how we got into this mess.

00:09:26 - 00:09:33

How to get out of it again, I must leave as an exercise for the audience.

2022-09-20 Peter WerkhovenQuantum Computing, AI and AGI

00:31:28 - 00:31:29

Well, thank you.

2022-05-16 David DeutschQuantum Information in Many Worlds

00:53:35 - 00:56:30

Yeah, well. So, first of all, the way that it couldn’t be is that all knowledge is explicit. That couldn’t happen, because if all knowledge were explicit, then the meanings of words would have to be explicit too. Let’s say I use a word like “castle.” Somebody who doesn’t know what a castle is would have to look in a dictionary, and in that dictionary a castle will be defined as a type of building which has the following properties, etc.

But if the person doesn’t know what “building” is or what “type” is, then they have to look those up too. It’s obvious by a counting argument that a dictionary cannot possibly define every word in it, because there would have to be a route to every definition, and that route would itself be undefined.

So how is it that inexplicit knowledge gets around that? Inexplicit knowledge doesn’t have definitions, and in fact, as Karl Popper says, definitions are highly overrated. They’re nearly always useless. We have knowledge encoded in language, but not as definitions. It’s encoded as ideas expressed in language.

01:34:02 - 01:34:48

Right. All right. Thank you, David. I don’t know how we can do it on zoom but. Thank you. Thank you so much, David. Thanks for everyone who showed up. After maybe some edits will post this on our website and our YouTube channel by quantum information society. And I’ve seen that there were many questions in the chat. We had many questions that were unanswered. You couldn’t take any questions from the chat. So we’ll find a way to maybe somehow record those and make use of them in our future sessions, but thanks for everyone who showed up. I guess I’m going to stop the recording now. Thanks, David.

2022-04-24 The Joe Walker Podcast#139 - Against Bayesianism

01:38:23 - 01:38:28

And that’s how we make progress. David Deutsch, thanks so much for your time.

David Deutsch

01:38:28 - 01:38:31

It’s been fun. Nice chatting.

Read the interview at Der Spiegel Source collection

2022-04-01 Der Spiegel»Es gibt eine Wahrheit. Aber wir sind unfähig, sie zu erkennen«

»Es gibt eine Wahrheit. Aber wir sind unfähig, sie zu erkennen«

English title: “There Is a Truth, but We Are Incapable of Recognizing It”

Zu den Sternen fliegen? Die Klimakrise bezwingen? Die Menschheit kann alle Probleme lösen, sagt der britische Physiker David Deutsch. Er weiß auch, wie.

English: Fly to the stars? Defeat the climate crisis? Humanity can solve every problem, says the British physicist David Deutsch. He also knows how.

Ein SPIEGEL-Gespräch von Rafaela von Bredow und Johann Grolle

English: A SPIEGEL interview by Rafaela von Bredow and Johann Grolle.

01.04.2022, 13.00 Uhr · aus DER SPIEGEL 14/2022

Zur Person

David Deutsch, Jahrgang 1953, ist ein britischer Physiker, der in Oxford lehrt. Er gilt als einer der geistigen Väter des Quantencomputer. In einem gerade auf Deutsch erschienenen Manifest für einen bedingungslosen Fortschritt vertritt er die Auffassung, dass mit der Aufklärung im 18. Jahrhundert eine nie mehr endende Explosion menschlichen Wissens begonnen habe.

English: David Deutsch, born in 1953, is a British physicist who teaches at Oxford. He is regarded as one of the intellectual fathers of quantum computing. In a manifesto for unconditional progress that has just appeared in German, he argues that the Enlightenment in the eighteenth century began a never-ending explosion of human knowledge.

SPIEGEL: Herr Professor Deutsch, Sie glauben, dass der Mensch, nach Abermilliarden Jahren absoluter Ödnis im Universum, dieses nun nach seiner Fasson umgestalten wird, eine kosmische Zeitenwende stehe bevor. Ist das Ihr Ernst?

English: Professor Deutsch, you believe that, after billions upon billions of years of absolute desolation in the universe, humankind will now reshape it as it sees fit, and that a cosmic turning point is at hand. Are you serious?

Deutsch: Ich bin nicht der Erste, der auf diese Idee kommt. Der italienische Geologe Antonio Stoppani schrieb bereits im 19. Jahrhundert, er zögere nicht, den Menschen zu einer neuen Macht im Universum zu erklären, wirkmächtig wie die Gravitation.

English: I am not the first to have this idea. As early as the nineteenth century, the Italian geologist Antonio Stoppani wrote that he did not hesitate to declare humans a new power in the universe, as potent as gravity.

SPIEGEL: Und zu fernen Planeten fliegen? Energie aus schwarzen Löchern zapfen? Ganze Galaxien erobern?

English: And fly to distant planets? Tap energy from black holes? Conquer entire galaxies?

Deutsch: Ich sage nicht, dass wir all dies unbedingt tun werden. Ich sage nur, dass uns im Prinzip nichts davon abhält. Allenfalls physikalische Gesetze könnten uns hindern. Und ein Naturgesetz, das uns zum Beispiel die Reise zu fernen Sternen verbietet, kennen wir nicht.

English: I am not saying that we will necessarily do all these things. I am only saying that, in principle, nothing prevents us. At most, physical laws could stop us. And we know of no law of nature that, for example, forbids us to travel to distant stars.

SPIEGEL: Theoretisch mag die Besiedlung der Galaxis möglich sein. Aber wie soll das praktisch gehen?

English: The colonisation of the galaxy may be theoretically possible. But how is it supposed to work in practice?

Deutsch: Das menschliche Gehirn, unterstützt von unseren Computern, kann das dafür nötige Wissen erschaffen – obwohl wir heute noch nicht wissen, wie.

English: The human brain, supported by our computers, can create the knowledge required, although today we do not yet know how.

SPIEGEL: Ihr verstorbener Kollege Stephen Hawking hat dem Homo sapiens nicht so viel zugetraut. Er meinte, wir seien »nur chemischer Abschaum auf einem mittelgroßen Planeten«, der einen »sehr durchschnittlichen Stern im äußeren Randbezirk einer von hundert Milliarden Galaxien« umkreise. Irrte Hawking?

English: Your late colleague Stephen Hawking did not have such confidence in Homo sapiens. He said that we were “only chemical scum on a moderate-sized planet,” orbiting “a very average star in the outer suburb of one of a hundred billion galaxies.” Was Hawking mistaken?

Deutsch: Nun, im Wortsinn ist wahr, was er schrieb. So, wie es in gewisser Weise wahr ist, dass der Krieg in der Ukraine durch Atome verursacht wurde. Es ist faktisch richtig, aber es erklärt nichts. Was wir brauchen, um die Welt und unsere Rolle darin zu verstehen, sind Erklärungen, nicht irgendwelche sinnentleerten Aussagen.

English: Well, what he wrote is literally true. Just as it is, in a certain sense, true that the war in Ukraine was caused by atoms. It is factually correct, but it explains nothing. What we need to understand the world and our role in it are explanations, not statements emptied of meaning.

SPIEGEL: Selbst unter Ihren Forscherkollegen dürfte es schwer sein, viele zu finden, die uns Menschen eine so gottgleiche Rolle im Universum zugestehen wie Sie.

English: Even among your fellow researchers, it must be difficult to find many who grant us humans as godlike a role in the universe as you do.

Deutsch: Die Wissenschaft befindet sich zurzeit in einem beklagenswerten Zustand. Ich bin wenig geneigt, meine Kollegen zu beleidigen, aber leider hat sich eine Art Expertenkult ausgebreitet. Entsprechend haben sich viele Forscher eng spezialisiert auf ihr jeweiliges Fachgebiet und innerhalb dessen darauf, Nützliches zu generieren, anstatt Erklärungen zu finden. Das ist ein schrecklicher Irrtum.

English: Science is currently in a lamentable state. I am not inclined to insult my colleagues, but unfortunately a kind of cult of experts has spread. Accordingly, many researchers have specialised narrowly in their respective fields and, within those fields, in generating useful things instead of finding explanations. That is a dreadful mistake.

SPIEGEL: Was ist so schrecklich an nutzbringender Wissenschaft?

English: What is so terrible about useful science?

Deutsch: Alles Nützliche, jede Vorhersage fußt auf Erkenntnis. Strebt man aber nicht mehr nach fundamentalen Erklärungen, sondern glaubt, es reiche aus, Nützliches zu generieren, dann bewegt man sich nur noch Schrittchen für Schrittchen von einer Dezimalstelle zur nächsten, und auch das nur in Gebieten, die bereits gut untersucht sind. Diese Entwicklung hat den Fortschritt dramatisch gebremst.

English: Everything useful, every prediction, rests on knowledge. But if one no longer strives for fundamental explanations and instead believes that generating useful things is enough, one moves only tiny step by tiny step from one decimal place to the next, and only in fields that have already been studied well. This development has dramatically slowed progress.

SPIEGEL: Es gibt Aufnahmen von einem schwarzen Loch, wir können Menschen gentechnisch verändern und innerhalb von Monaten einen Impfstoff gegen einen neuen Seuchenerreger entwickeln. All das ist kein Fortschritt?

English: There are images of a black hole, we can genetically alter humans, and we can develop a vaccine against a new pathogen within months. Is none of that progress?

Deutsch: Doch, aber es geht langsamer voran, als es gehen könnte.

English: It is, but progress is slower than it could be.

SPIEGEL: Der Biologe Richard Dawkins glaubt, dies liege vielleicht daran, dass unser Gehirn nicht ausreicht, die immer komplexere Welt zu begreifen. Denn es habe sich entwickelt, um mit Problemen der afrikanischen Savanne fertigzuwerden. Jetzt aber hätten wir es mit Sternen, Quanten und Kernreaktionen zu tun.

English: The biologist Richard Dawkins believes this may be because our brain is insufficient to comprehend an increasingly complex world. It evolved to cope with the problems of the African savannah. But now we are dealing with stars, quanta, and nuclear reactions.

Deutsch: Dawkins lässt außer Acht, dass es im Grunde nur eine Art von Computer gibt: Ganz gleich, ob es Ihr Laptop ist oder ein Superrechner zur Modellierung des Klimas, jeder Computer kann die gleichen Berechnungen durchführen. Und unser Gehirn ist nichts anderes als ein solcher universeller Computer. Die Hardware darin kann jedes Programm ausführen – also jede Erklärung denken. Ein Gehirn, das die Savanne verstehen kann, nicht aber den Himmel, das gibt es nicht. Selbst wenn wir es versuchten, wir könnten keinen Rechner bauen, der das eine kann und das andere nicht. Es verstößt gegen die Gesetze der Physik.

English: Dawkins overlooks the fact that there is fundamentally only one kind of computer. Whether it is your laptop or a supercomputer modelling the climate, every computer can perform the same calculations. And our brain is nothing other than such a universal computer. Its hardware can run any program, which means it can think any explanation. There is no such thing as a brain that can understand the savannah but not the sky. Even if we tried, we could not build a computer that can do one and not the other. It would violate the laws of physics.

SPIEGEL: Was, wenn wir doch auf eine Grenze des Erklärbaren stoßen?

English: What if we do encounter a boundary to what can be explained?

Deutsch: Wenn diese Grenze nicht von Gesetzen der Physik bestimmt wird, bringen Sie hier Übernatürliches ins Spiel.

English: If that boundary is not determined by the laws of physics, you are bringing the supernatural into play.

SPIEGEL: Gott?

English: God?

Deutsch: Auch viele Forscher sind sich nicht bewusst, dass sie Übernatürliches bemühen, wenn sie sagen, dass etwas anderes als die Physik uns daran hindere, die Welt zu verstehen. Selbst bei vielen Atheisten zeigt sich ein Rest von Religiosität, wenn sie von unserem Bewusstsein, dem Universum und unserem Platz darin reden.

English: Many researchers are not aware that they invoke the supernatural when they say that something other than physics prevents us from understanding the world. Even many atheists reveal a remnant of religiosity when they talk about our consciousness, the universe, and our place in it.

SPIEGEL: Die Natur, sagen Sie, habe den Homo sapiens mit einem Gehirn ausgestattet, mit dem er die Galaxis erobern kann. Warum haben Menschen dann über Zehntausende von Jahren nur Faustkeile benutzt?

English: You say nature equipped Homo sapiens with a brain that can conquer the galaxy. Why, then, did humans use only hand axes for tens of thousands of years?

Deutsch: Sie haben ihr Gehirn fast ausschließlich dafür genutzt, ihre Kultur weiterzugeben. Leider nicht, um sie zu verbessern. Dies ist die größte Tragödie in der Geschichte der Menschheit; sie hat mehr Leid verursacht als alles andere.

English: They used their brains almost exclusively to pass on their culture. Unfortunately, they did not use them to improve it. This is the greatest tragedy in human history. It has caused more suffering than anything else.

SPIEGEL: Wenn der Fortschritt einen Nutzen für die Gesellschaft darstellte, warum setzte er sich dann nicht schneller durch?

English: If progress benefited society, why did it not prevail more quickly?

Deutsch: Leider haben Ideen, die das Anderssein verteufeln, einen kulturell-evolutionären Vorteil gegenüber solchen, die das Anderssein preisen. Deshalb haben es neue Ideen schwer, sich auszubreiten. So war es in vielen traditionellen Gesellschaften üblich, das Beste den Göttern zu opfern. Und was passierte, wenn einer den Vorschlag machte, diese Opfergaben, statt sie an die Götter zu verschwenden, doch lieber an die hungrigen Kinder zu verfüttern? Er wurde getötet. Seine Idee konnte sich nicht verbreiten.

English: Unfortunately, ideas that demonise being different have a cultural-evolutionary advantage over ideas that praise it. That is why new ideas have difficulty spreading. In many traditional societies, for instance, it was customary to sacrifice the best things to the gods. And what happened if someone suggested that, instead of wasting these offerings on the gods, they should be fed to hungry children? He was killed. His idea could not spread.

SPIEGEL: Zum Stillstand allen Fortschritts kam es trotzdem nicht …

English: Yet progress did not come to a complete standstill …

Deutsch: … das stimmt, die Kultur änderte sich, aber zu langsam, als dass es die Hüter der Beständigkeit bemerken würden. In der Bibel steht es: »Was geschehen ist, eben das wird hernach sein. Was man getan hat, eben das tut man hernach wieder, und es geschieht nichts Neues unter der Sonne.« Da sehen Sie: Für einen Menschen, der Jahrhunderte vor Christus lebte, schien sich die Welt nie zu ändern. Deshalb lebte fast jeder ein Leben, in dem er nie etwas Neues sah.

English: … that is true. Culture changed, but too slowly for the guardians of permanence to notice. The Bible says: “What has happened is what will happen hereafter. What has been done is what will be done again, and nothing new happens under the sun.” You can see that, to a person living centuries before Christ, the world never seemed to change. That is why almost everyone lived a life in which they never saw anything new.

SPIEGEL: Was musste passieren, dass die Menschen Veränderungen zulassen konnten?

English: What had to happen before people could allow change?

Deutsch: Irgendwann erkannten einige Gesellschaften, dass Verbesserungen wirklich möglich sind; das antike Athen und das Florenz der Renaissance sind zwei bekannte Beispiele dafür. Aber selbst in diesen Fällen wurde der Impuls zur Neuerung einiger Generationen wieder zerstört. Erst die Aufklärung überlebte, sie verbreitete sich und hat schließlich all das hervorgebracht, was wir heute haben.

English: At some point, some societies recognised that improvements really are possible. Ancient Athens and Renaissance Florence are two well-known examples. But even in those cases, the impulse towards innovation was destroyed again after a few generations. Only the Enlightenment survived, spread, and eventually produced everything we have today.

SPIEGEL: Was war im Fall der Aufklärung anders?

English: What was different in the case of the Enlightenment?

Deutsch: Ich denke, es war der Optimismus, der der Wissenschaft zum Durchbruch verhalf. Er ist sehr tief in unserer heutigen Kultur verwurzelt: Wenn irgendetwas sich zum Schlechteren verändert, begehren die Leute auf. Denn sie sind überzeugt, dass sich das Problem lösen und die Situation sich daher wieder verbessern lässt. Diese Haltung ist neu. Über den größten Teil der Geschichte hinweg dachten die Menschen indes: Die Welt ist schlecht, und sie wird eher noch schlechter.

English: I think it was optimism that helped science make its breakthrough. It is deeply rooted in our present culture. Whenever something changes for the worse, people rebel. They are convinced that the problem can be solved and the situation can therefore improve again. This attitude is new. For most of history, by contrast, people thought that the world was bad and would probably get worse.

SPIEGEL: Kann die Aufklärung auch heute noch scheitern?

English: Can the Enlightenment still fail today?

Deutsch: Eindeutig ja. Falls sie scheitert, dann wird es daran liegen, dass den Menschen das nötige Wissen fehlt, dies zu verhindern. Deshalb sollte es unser oberstes Gebot sein, den Fortschritt so schnell wie möglich voranzutreiben. Und ich spreche hier nicht nur über den technischen Fortschritt, sondern auch über den moralischen, politischen und grundlegenden wissenschaftlichen Fortschritt. Sie können nie wissen, welche Erkenntnis sich als entscheidend für den Fortgang unserer Zivilisation erweisen wird. Am Anfang des Zweiten Weltkriegs sah niemand voraus, dass die Eigenschaften des exotischen Elements Uran irgendeine Rolle für den Ausgang des Krieges haben könnte. Und doch war es so.

English: Definitely. If it fails, it will be because people lack the knowledge required to prevent its failure. That is why our highest imperative should be to advance progress as quickly as possible. I am speaking not only about technological progress, but also about moral, political, and fundamental scientific progress. You can never know which discovery will prove decisive for the continuation of our civilisation. At the beginning of the Second World War, no one foresaw that the properties of the exotic element uranium might play any role in the outcome of the war. And yet they did.

SPIEGEL: Muss eine fortschrittliche Gesellschaft demokratisch sein?

English: Must a progressive society be democratic?

Deutsch: Die Demokratie, sagte Winston Churchill, sei von allen Regierungssystemen das schlechteste, mit Ausnahme aller anderen. Dem österreichisch-britischen Philosophen Karl Popper folgend, denke ich, dass das Entscheidende an der Demokratie nicht die Wahlen sind, nicht einmal die Menschenrechte. Entscheidend ist vielmehr, dass es in Demokratien möglich ist, die Regierung und die Politik ohne Gewaltanwendung zu ändern.

English: Democracy, Winston Churchill said, is the worst of all systems of government except for all the others. Following the Austrian-British philosopher Karl Popper, I think that the decisive feature of democracy is not elections, not even human rights. What matters is that democracies make it possible to change the government and its policies without violence.

SPIEGEL: Wie wichtig ist die Garantie von Meinungsfreiheit?

English: How important is the guarantee of freedom of speech?

Deutsch: Sehr wichtig. Das moralisch schlimmste aller Verbrechen besteht darin, die Korrektur von Fehlern zu verhindern. Alle politischen und gesellschaftlichen Institutionen sollten wir danach bewerten, wie effizient sie die Fehlerkorrektur befördern.

English: Very important. The morally worst crime of all is to prevent the correction of errors. We should judge every political and social institution by how efficiently it promotes error correction.

SPIEGEL: Im Großen und Ganzen – wie gut sind wir darin?

English: By and large, how good are we at it?

Deutsch: Wirtschaft und Wissenschaft in unseren westlichen Gesellschaften sind, denke ich, im Wesentlichen in Ordnung. Hin zu mehr Fortschritt, mehr Offenheit, mehr Kreativität. Andere Teile unserer westlichen Kultur orientieren sich noch immer an den alten Vorstellungen, traditionelles Wissen getreulich weiterzugeben, was das genaue Gegenteil aufklärerischen Denkens ist.

English: The economy and science in our Western societies are, I think, essentially all right. They move towards more progress, more openness, and more creativity. Other parts of our Western culture still follow the old idea of faithfully transmitting traditional knowledge, which is the exact opposite of Enlightenment thinking.

SPIEGEL: Welche Teile unserer Kultur meinen Sie?

English: Which parts of our culture do you mean?

Deutsch: Ich denke dabei besonders an unser Erziehungssystem. Schulen und Universitäten haben Curricula, und sie haben Standards. Das bedeutet: Sie setzen einen Fundus von Ideen fest, die jeder sich zu eigen machen soll. Standards bedeuten, dass die Menschen standardisiert werden. In einer rationaleren Gesellschaft würden die Schulen damit werben, dass jedes Kind, das sie verlässt, anders ist. Aber sie werben mit dem genauen Gegenteil.

English: I am thinking especially of our education system. Schools and universities have curricula, and they have standards. That means they establish a stock of ideas that everyone is supposed to adopt. Standards mean that people are standardised. In a more rational society, schools would advertise that every child leaving them is different. But they advertise the exact opposite.

SPIEGEL: Der technische Fortschritt hat uns auch globale Erwärmung, Plastikflut und Artensterben beschert. Wird nicht am Ende gerade der Fortschritt das Ende unserer Zivilisation bewirken?

English: Technological progress has also brought us global warming, a flood of plastic, and mass extinction. Will progress itself not ultimately bring about the end of our civilisation?

Deutsch: Das sind Folgen des Fortschritts, aber sie werden uns nicht zerstören. Jede Lösung eines Problems schafft neue Probleme – und auch diese sind wieder lösbar. So sind die Methoden des Geoengineering, die das Problem des Klimawandels vielleicht lösen könnten, leider mit einem Tabu belegt. Wir sollten in den nächsten Jahrzehnten möglichst viele solcher Methoden erfinden und erproben. Stattdessen, so fürchte ich, wird man damit erst anfangen, wenn es fast zu spät ist.

English: Those are consequences of progress, but they will not destroy us. Every solution to a problem creates new problems, and those too are soluble. Unfortunately, geoengineering methods that might solve the problem of climate change are taboo. Over the next few decades, we should invent and test as many such methods as possible. Instead, I fear, people will begin only when it is almost too late.

SPIEGEL: Wenn jedes gelöste Problem ein neues gebiert und wenn die Zahl der Probleme, die es zu lösen gilt, beständig wächst, führt das nicht unweigerlich in eine Katastrophe?

English: If every problem solved gives birth to a new one, and if the number of problems to be solved keeps growing, does that not lead inevitably to catastrophe?

Deutsch: Nein. Zum einen sind Probleme etwas Gutes, sie sind elementar, um uns weiterzubringen. Zum anderen gibt es Katastrophen, die wir nicht verhindern können, weil uns das dazu nötige Wissen noch fehlt. Aber Katastrophen sind nicht unausweichlich. Niemand kann uns daran hindern, das nötige Wissen zu schaffen, um sie abzuwenden.

English: No. For one thing, problems are good. They are essential to moving us forward. For another, there are catastrophes that we cannot prevent because we still lack the necessary knowledge. But catastrophes are not inevitable. No one can prevent us from creating the knowledge needed to avert them.

SPIEGEL: Der Impfstoff gegen Covid-19 …

English: The vaccine against Covid-19 …

Deutsch: … ja, ein gutes Beispiel. Zu Beginn der Pandemie sagten die Leute: Bis wir einen Impfstoff haben, wird es viele Jahre dauern. Am Ende hat es nur Monate gebraucht. Und wenn es nicht all die Vorschriften und Richtlinien gegeben hätte, dann hätte der Impfstoff schon binnen Tagen bereitstehen können. Solche Vorschriften sind Teil der Beharrungskräfte, die dem Schaffen neuen Wissens im Wege stehen.

English: … yes, a good example. At the beginning of the pandemic, people said that it would take many years before we had a vaccine. In the end, it took only months. And if all the rules and guidelines had not existed, the vaccine could have been ready within days. Such rules are part of the forces of inertia that obstruct the creation of new knowledge.

SPIEGEL: Kann es falsch sein, einen Impfstoff gründlich zu testen, ehe man ihn Millionen Menschen verabreicht?

English: Can it be wrong to test a vaccine thoroughly before administering it to millions of people?

Deutsch: Gründlich genug, um das Problem zu lösen. In diesem Fall hätte man beispielsweise Freiwilligen erlauben können, sich mutwillig mit Covid zu infizieren, um an ihnen Medikamente oder Impfstoffe zu erproben. Es gibt Menschen, die bereit sind, ihr Leben zu riskieren, um das Leben anderer zu retten. Warum erlaubt man es ihnen nicht?

English: Thoroughly enough to solve the problem. In this case, for example, volunteers could have been allowed to infect themselves deliberately with Covid so that drugs or vaccines could be tested on them. Some people are willing to risk their lives to save the lives of others. Why are they not allowed to do so?

SPIEGEL: Sie erklären Vorsicht zu einem Übel?

English: You declare caution to be an evil?

Deutsch: Vorsorge ist gut. Das Vorsorgeprinzip ist ein Übel. Wenn Sie alles Neue unterdrücken, weil es unvorhergesehene Nebeneffekte haben könnte, dann behindern Sie den Fortschritt, dann verhindern Sie neue Problemlösungen, und das heißt: Sie verursachen menschliches Leid.

English: Precaution is good. The precautionary principle is an evil. If you suppress everything new because it could have unforeseen side effects, you obstruct progress and prevent new solutions to problems. That means you cause human suffering.

SPIEGEL: Kann es nicht manchmal klüger sein, auf riskante neue Techniken zu verzichten?

English: Might it not sometimes be wiser to forgo risky new technologies?

Deutsch: Nein! Auf Fortschritt zu verzichten ist ein sicherer Weg in den Untergang. Dann stellt sich nicht mehr die Frage, ob, sondern nur noch, wann unsere Zivilisation zerstört wird. Gib, dass es nicht uns passiert, sondern erst unseren Kindern! Das ist der Kern des Vorsorgeprinzips. Es ist extrem unklug, um nicht zu sagen: unmoralisch.

English: No! Forgoing progress is a sure path to ruin. Then the question is no longer whether our civilisation will be destroyed, but only when. Grant that it does not happen to us, but only to our children! That is the core of the precautionary principle. It is extremely foolish, not to say immoral.

SPIEGEL: Glauben Sie, dass Sie rationaler sind als andere?

English: Do you believe you are more rational than other people?

Deutsch: Nein, überhaupt nicht. Ich bin überzeugt davon, dass auch ich voller Irrtümer und Widersprüche bin. Bloße Widersprüche sind allerdings nichts Schlechtes. Im Gegenteil: Sie sind der Stoff, aus dem Probleme sind, die wir lösen müssen. Und sie sind deshalb der Stoff, aus dem der Fortschritt ist.

English: No, not at all. I am convinced that I too am full of errors and contradictions. Mere contradictions, however, are not bad. On the contrary, they are the material from which the problems we must solve are made. And therefore they are the material from which progress is made.

SPIEGEL: Wird der Fortschritt, den Sie so preisen, irgendwann in ein vollständiges Verständnis der Welt münden, in eine Art Weltformel?

English: Will the progress you praise so highly eventually culminate in a complete understanding of the world, in some kind of theory of everything?

Deutsch: Nein. Eine Weltformel wird es nie geben. Die Physiker allerdings verwenden diesen Begriff, um etwas viel, viel Bescheideneres zu bezeichnen. Sie suchen nach einer Formel, die alle Wechselwirkungen von Elementarteilchen einschließlich der Gravitation beschreibt. In meinen Augen ist das nur ein kleiner Teil der Physik, und Physik ist nur ein kleiner Teil der Naturwissenschaft, und Naturwissenschaft ist nur ein kleiner Teil menschlichen Wissens. Das heißt: Eine »Weltformel« ist das eindeutig nicht.

English: No. There will never be a theory of everything. Physicists, however, use this term to refer to something much, much more modest. They are looking for a formula that describes all interactions of elementary particles, including gravity. In my view, that is only a small part of physics, physics is only a small part of natural science, and natural science is only a small part of human knowledge. That means it is clearly not a “theory of everything.”

SPIEGEL: Gibt es denn eine Wahrheit?

English: Is there such a thing as truth, then?

Deutsch: Ja, es gibt eine Wahrheit. Aber wir sind unfähig, sie als solche zu erkennen. Wenn wir sagen, dass wir nach der Wahrheit suchen, dann meinen wir eigentlich, dass wir nach Fehlern in unserem bisherigen Wissen suchen. Und wenn wir eine Theorie befürworten, meinen wir, dass wir noch nicht herausgefunden haben, in welchem Sinne sie falsch ist. Unser Wissen besteht aus Irrtümern, die den Vorzug haben, weniger heftige Fehler zu enthalten als unsere früheren Irrtümer.

English: Yes, there is a truth. But we are incapable of recognising it as such. When we say that we are searching for the truth, what we actually mean is that we are searching for errors in our existing knowledge. And when we endorse a theory, we mean that we have not yet discovered in what sense it is false. Our knowledge consists of errors whose advantage is that they contain less severe errors than our earlier errors.

SPIEGEL: Dann ist die technische Zivilisation der Gegenwart also ein grandioser Triumph des Irrtums. Wie sehr werden Sie selbst profitieren? Werden Sie noch die Unsterblichkeit erleben?

English: So the technological civilisation of the present is a magnificent triumph of error. How much will you yourself benefit? Will you live to see immortality?

Deutsch: In Anbetracht des gegenwärtigen Fortschritts medizinischen Wissens könnte es sein, dass ich länger lebe, als ich es erwarte. Mit der Unsterblichkeit zu meinen Lebzeiten rechne ich indes nicht. Aber vielleicht werde ich überrascht. Falls es so kommt, können wir diese Diskussion in 100 Jahren fortsetzen.

English: Given the current progress of medical knowledge, I may live longer than I expect. I do not expect immortality within my lifetime, however. But perhaps I will be surprised. If that happens, we can continue this discussion in 100 years.

SPIEGEL: Gerne. Bis dahin, Herr Professor Deutsch, danken wir Ihnen für dieses Gespräch.

English: Gladly. Until then, Professor Deutsch, thank you for this interview.

2022-01-31 Dwarkesh PodcastDavid Deutsch - AI, America, Fun, and Bayes

01:23:46 - 01:24:12

Well David thanks so much for coming on the podcast and thanks so much for giving me so much of your time. Fascinating. Thank you for inviting me. Thank you.

2021-12-13 Robin HansonTo What Extent Can We Predict the Future? With Robin Hanson and David Deutsch

00:58:03 - 00:58:04

Bye-bye.

2021-12-01 ToKCastEp 100 - David Deutsch

I don’t think that there are any theories. I mean, one could easily write down a theory in which the universe was inhomogeneous and we are in the only place in it that has matter. But that wouldn’t help in any way with any of the existing theories. And more generally, by the way, theories of inhomogeneity in the universe, as far as I know, I’m not an expert on this, but I think they’ve only been invoked by people who want to say that there is no dark energy, that it’s just a coincidence caused by inhomogeneities. Excellent. Okay. I think that absolutely answered the question. I think that will do for now, but I hope not for the last time. I’m making my way through The Fabric of Reality after all now at the moment. So perhaps at the end of that, we can speak again, if not sooner. So thank you very much for writing the books. I think there’ve been, well, the explanations that transform the world. I think they’ve been life-transforming for many people. So it’s been wonderful to talk to you. Very gratifying today to have this conversation. So have a wonderful remainder of the day yourself. Same to you. It’s always fun chatting.

Brett Hall

01:23:25 - 01:24:51

So there we go. I hope you enjoyed that. I certainly did. As I say, I’m working on what might be called interview technique. It’s a steep learning curve. The content of Deutsch and Popper’s oeuvre is deep and subtle enough, but adding to that, how to run a podcast and being surrounded by new upgraded podcast hardware and software that I’m struggling to learn to use is a whole other thing, but hopefully will improve rapidly over the coming weeks and months. If you’re a new listener, I’d encourage you to go out and to listen to all of the David Deutsch content that is out there. Hear his TED Talks and other interviews. Of course, buy his books. Subscribe to my podcast here, wherever you’re listening. There is a YouTube version, as I say, if you are on audio only, which sometimes has those pretty visuals. Subscribe to the Naval podcast, where we are exploring all the major themes in The Fabric of Reality and The Beginning of Infinity, and Naval is broadening all that content out into the realms of wealth, wisdom and life. And finally, if you’re so inclined, go to www.bretthall.org and follow the links there to support this podcast. There are lots of great podcasts out there, all about science and philosophy and even thinking, but I think here we have a small community uniquely devoted to a particularly special kind of clear thinking about life, informed by the best that science and philosophy can teach us. We’d like to grow so the ideas can reach into places where the errors need correcting, so I’ll keep doing my bit. Until next time, let me steal a line from Naval and say, stay optimistic. Bye bye.

2021-10-27 ToKCastDavid Deutsch answers a question about dark energy. A question for David number 10

00:13:45 - 00:14:06

Next year is, of course, the big one, and we’ll have the big celebration. Once again, episode 100 is going to be the complete conversation with David. I’ll release that in full as a celebration, essentially, of ToKCast’s century of episodes. But until the next episode, buh-bye.

2021-10-08 RipplesThe Fun Criterion, Objective Beauty and Artificial Intelligence

01:13:22 - 01:13:25

Well, thank you for having me. It’s been fun.

2021-10-07 ToKCastDavid Deutsch answers a question about the nature of mind. A question for David number 9

Duration: 00:10:03

Transcript

Brett Hall

00:00:00 - 00:03:05

Hello and welcome to ToKCast and my next in the series of questions for David. This is number nine about the nature of mind and personhood to some extent and what makes people unique. This is a real interest of mine. I made a video called The Nexus, which was an exploration of what I think we know from various angles about what the nature of personhood is. Certainly it’s the case that we are these creative entities and because of computational universality it must be the case that we can program computers to be creative but whatever this program is is going to be very different to any existing program and David’s going to touch on that in his answer today. But this program remains utterly elusive. We do not know how it is that we generate explanations or in shorthand how it is that we’re creative. In my Nexus video I hypothesized that personal identity might come into this in some way because there is this some not entirely a mystery in that to the same degree but certainly it is a spectacular insight into the nature of reality that particles and indeed all large complex objects including ourselves have fungible instances in other universes which differentiate over time and so this has a real effect on how we should view the nature of personhood given the truth of the Everett interpretation, the multiverse vision of quantum theory. And so because we know that we have these fungible instances that interfere, that differentiate sometimes, can differentiate sometimes, that perhaps there’s something to be said here for the interaction between physics at that level and whatever the computation is that’s going on in our minds. I don’t know, pure speculation. But see my Nexus video for that. What I really loved about this question to David or rather the answer from David I should say is that he talks about consistency and whether or not and to what extent people can be consistent. Should we strive for consistency? I suppose so. But can we? Could this inconsistency that exists in our minds, this constant state of problems, these contradictions that we live with, could this just be part of the reason why we are special, unique, different to everything else that exists that we know of? We are the thing that brings other things into being that otherwise would not have come into being if it was just laws of physics mandating everything in a deterministic way. Those laws do mandate things in a deterministic way but they’re not the only things that determine stuff and what happens. Nor, of course, most importantly, are the laws of physics the best explanation for everything that happens at all moments in time. There is something absolutely unique about human beings. We continue to explore this. We’re not just on the continuum with other animals, just a slightly smarter example of the chimpanzee.

Brett Hall

00:03:05 - 00:04:16

I keep on hearing this. We are qualitatively different. Without further ado, let’s get into the next question. This all raises the fact that there is this creativity going on, this creation of concepts in order to try and come to better understand the world. That is the uniquely creative capacity of the human mind. Mind itself, how do you understand the nature of mind? I sometimes think of it as this abstraction. I think you’ve said this to me. It’s a funny kind of abstraction. It’s an abstraction that itself needs to be instantiated in matter but also needs to be running. It can’t just be like any other piece of knowledge which you can transmit from one place to another in different physical forms. That’s one way of understanding what knowledge is. It allows itself to continue to persist over time and to get itself copied, to get itself replicated. But a mind, a human mind, this thing that is enabling the creation of this knowledge, it itself is an unusual kind of abstraction. How do you think of it at the moment?

David Deutsch

00:04:16 - 00:08:07

Yes, it’s an unusual kind of knowledge. We have yet another classification. A mind is a kind of knowledge, you might say, but it’s a kind of knowledge that doesn’t only have to be instantiated, but as you have just said, it has to be running. Now what does that mean? Running how fast? Running in what? Because running in the wrong kind of computer would make it gibberish or in a kind of computer that has the wrong mapping to reality. We don’t know how it works. We don’t know how the mind… So the mind is characterized by creativity. I think we can go that far. At least the human mind is characterized by creativity, but we don’t know what creativity is. We don’t know what the distinction is between a computer program that is running creatively and one that is not running creatively. One day we will know, but we don’t know yet. I think also there are… And again, minds, that minds exist is common sense, even though we don’t know how they work or what specifically they are. So more recent ideas that maybe minds don’t exist, that had to be invented too. It is not at all common sense and it doesn’t at all follow from any good philosophy, but we don’t know. And one of the things I think that is a misconception that dates from a very long time ago is that a mind is mostly consistent, that it consists of… The ideas in a mind are mostly consistent and when they’re inconsistent, that’s a kind of emergency and we have to fix that because that means that it’s not a proper mind really. It contains inconsistencies and therefore that’s no different from a load of propositions written down on a piece of paper. I don’t think that’s true at all. I think a better way to think of the mind is as a set of problems and a problem consists of conflicting theories. Or rather a problem is more like some conflicting theories that we have noticed because we don’t really think of something as a problem unless it’s giving rise to some thought. But there must be ideas that are conflicting with each other before we have noticed it. So the mind, I think of it as sort of a massive tangle of ideas which are largely inconsistent. The parts of it which we have sorted out into a kind of consistency are a minority that are currently not giving rise to any thought. They might be invoked in some thought that’s about some other inconsistency, but the part of the mind that is thinking is full of inconsistency. And so therefore a model of the mind that thinks of the mind as a set of consistent propositions with occasional inconsistencies that we then fix is I think very far from the truth. It’s not a set of consistent ideas and it’s not a set of propositions because propositions have a definite truth value. Each proposition is either true or false and it has a perfectly definite meaning if it has a meaning.

David Deutsch

00:08:07 - 00:08:40

Whereas the ideas in minds don’t have a definite meaning and they’re not precisely true or false. We can make them have as sharp a meaning as our current problem situation requires. That’s what we should be aiming for and we can make them as consistent as the current problem situation requires. And we’re striving, you talked earlier about truth, we’re striving to make them as true as the current problem situation requires.

Brett Hall

00:08:40 - 00:10:03

Yes, so it sounds that whatever the nature of the mind is in a computational sense has to be radically different to any other kind of program that we’ve hitherto encountered because all those other programs are written in such a way that it’s proposition after proposition, perhaps one following from another. It’s an algorithm of a kind. And these statements need to be consistent or the computer program itself isn’t going to run. You’re saying that it’s possibly the case that underlying this uniquely creative capacity of the human mind is something that is quite the mirror image of that in at least some respects. And maybe this is what is needed in order to go some way to explaining what’s going on here with creativity. Or another way that you’ve put that in The Beginning of Infinity is, of course, explanatory universality, this idea that even if there were other kinds of creativity prior to us, what we have is a kind of creativity which allows us to create knowledge which is going to be able to solve problems in principle about anything that exists in the physical universe.

2021-10-06 ToKCastDavid Deutsch answers a question about observations. A question for David number 8

00:06:57 - 00:08:01

Yes. Well, this all raises the fact that there is this creativity going on, this creation of concepts in order to try and come to better understand the world. That is the uniquely creative capacity of the human mind. Mind itself, how do you understand the nature of mind? I sometimes think of it as this abstraction. I think you’ve said this to me. It’s a funny kind of abstraction. It’s an abstraction that itself needs to be instantiated in matter, but also needs to be running. It can’t just be like any other piece of knowledge which you can transmit from one place to another in different physical forms. And that’s one way of understanding what knowledge is. It allows itself to continue to persist over time and to get itself copied, to get itself replicated. But a mind, a human mind, this thing that is enabling the creation of this knowledge, it itself is an unusual kind of abstraction. How do you think of it at the moment?

Original German interview English transcript by Jannik Wiese, edited by Dennis Hackethal

2021-09-20 David DeutschDer Anfang der Unendlichkeit mit David Deutsch

Duration: 00:35:44

The German original is the recording above. Jannik Wiese created the English transcript and Dennis Hackethal edited it. Parentheses contain their clarifications; brackets preserve German terms from the interview.

English translation

Dennis Hackethal

Welcome to our listeners and viewers. Several years ago I read a fascinating book and as soon as I finished reading, I knew it was my new favourite. I’m talking about ‘The Beginning of Infinity’ by British physicist David Deutsch. It is now available in German for the first time, titled ‘Der Anfang der Unendlichkeit: Erklärungen, die die Welt verwandeln’ (the original title is ‘The Beginning of Infinity: Explanations that Transform the World’).

Progress seems to depend on an understanding of the world. However, there are surely things that we can’t comprehend, say because evolution didn’t equip our brains with the capabilities that it would need to understand everything. If that’s the case then sooner or later we will hit some kind of barrier to progress, no?

David Deutsch

2021-08-29 Eli TyrePopper's Problem-Oriented Epistemology with David Deutsch and Eli Tyre

01:31:39 - 01:31:42

Well, thanks, it’s been really fun.

2021-08-13 ToKCastDavid Deutsch answers a question about the nature of the laws of physics. A question for David 7

00:08:26 - 00:09:24

We were just talking about the nature of physical laws and whether or not, and to what extent they might be pure abstractions or a kind of abstraction of some kind. And I was just thinking out loud to myself that for anyone who denies the existence of physical laws in their own right, laws of physics, and to just say, well, there are just facts of the matter about the universe. For example, the orbits of planets going around stars are approximately circular or elliptical, then those facts of the matter, those regularities in nature seem to be crying out for an explanation. So to deny that there would be really existing physical laws behind those regularities in nature seems to me to be going to a lot of work to avoid what might otherwise be called Occam’s razor. Well, isn’t the simple explanation, there are physical laws out there governing these things.

David Deutsch

00:09:24 - 00:11:34

Yes, of course, if somebody insists on the non-existence of explanations, exactly like insistence on the non-existence of anything, you can’t be proved wrong. But I think the motivation for denying the existence of laws is the old mistake of empiricism. It is the assumption that raw facts or raw sensory impressions have a privileged status in that we can access them directly. And people contrast this with things like laws and explanations in general, which they say we can’t access directly. So they have some kind of a lesser reality and we can in principle not insist on their being real. One can insist on their not being real and it doesn’t make any difference. The trouble with that is that exactly the same is true of sense impressions as well. And so this argument that abstractions don’t really exist or that laws don’t really exist and so on, hidden in there is the assumption that sense impressions or that kind of thing do really exist in some sense of really, which is itself a mistake. That all knowledge is conjectural, all observations are theory-laden. There is nothing that is an authoritative source of knowledge, everything is conjecture. And so once you’ve realized that everything is conjecture, but that knowledge can still exist, then the reason for making a distinction between different kinds of existence. Well, no, there is a distinction between different kinds of existence, but the justification, the motivation for denying that certain things exist, even if we need them in our explanations, goes away.

2021-08-07 ToKCastDavid Deutsch answers a question about Gödel and undecidability. A question for David number 6

00:10:33 - 00:10:39

Yes. And therefore that the undecidable consequences are actually interesting.

David Deutsch

00:10:39 - 00:12:15

Well, it would be physically interesting. Yes. Yes, you know, in principle, something that you wanted to do might depend on working out this thing. But I don’t see any particular reason for making a distinction as long as we have the necessities of life and comfort. I don’t see any reason for making a distinction between problems in engineering and in physics and in pure mathematics. I mean, they’re all problems. They’re all conflicts between theories that interest people. People want to know what the resolution is. And one thing I say in the book about pure mathematical problems is that sometimes discovering that something is undecidable is the solution to the problem in the sense that it’s the problem. And that explains why you could have these two conflicting intuitions about it. And then you can go and criticize your intuitions in other ways, because you know that you’re not going to be able to resolve it by proving one of them true or false.

2021-07-16 ToKCastDavid Deutsch answers a question about our environment. A question for David 5

00:15:02 - 00:18:14

But life can adapt itself to the Earth but the only thing that is going to survive in the long run in terms of regular life not intelligent life like ourselves are the genes. The genes are going to want to try and make themselves survive and the animals won’t care or the other species don’t have this concept of caring into the infinite future about their own survival. However we do so we want to adapt ourselves to the Earth in some way or if not the Earth then something far larger. We want to adapt ourselves to the universe in which we find ourselves and the one thing that enables us to do that is of course solving our problems continually to create the knowledge in order to enable us to do so. So there we have it you heard David there say that one of my questions could have been said with one-tenth of the emphasis and it would still be wrong. This is all because it is wrong to think that the Earth is especially suited to us. This can be shocking for people to hear but David likes to use the example of Oxford for example where he lives is not suitable for human existence. Remove the technology the stuff that’s there only because of the knowledge of people and that includes everything from shoes and underwear and clothing all the way through to reverse cycle air conditioning and refrigeration and a person there in the winter wouldn’t last 10 minutes in the freezing cold. People are fragile and the only way we can get by is in an environment that we construct around ourselves. He mentions there maybe there are some islands somewhere where you know there’s constantly fruit being produced and there’s animals there and it’d be very comfortable for us to live night and day wandering around without clothes and without reverse cycle air conditioning and without perhaps refrigeration. Maybe we could get by there for a few generations until the storm comes until something goes wrong. The only reason we’re able to survive ongoing on this planet is because of our knowledge our technology. We make places here on Earth that are otherwise completely inhospitable. In Australia the nation is essentially a desert well not entirely but largely a desert and if we didn’t have dams for example collecting the rainwater we’d quickly die of thirst. There’s no way we could support the almost 25 million or 30 million people that we have here on this extremely dry continent. Repeat for every other continent around the world as well either it’s too cold, too dry, too wet, too frozen, just too dangerous. The Earth is barely suitable, barely suitable and because we are able to make it increasingly more and more suitable more and more friendly to humans, human friendly we can do likewise with the rest of the cosmos. We’ve already done it in outer space. We’ve already made the vacuum of space where the International Space Station is hospitable to some extent.

Brett Hall

00:18:14 - 00:18:51

We can continue to do that. We could do that on the moon, we could do that on Mars and eventually we spread out throughout the solar system, the galaxy, the local group of galaxies and the universe but it takes lots more knowledge to create the technology in order for us to do that. We’ve done it here. Let’s keep doing it off into the infinite future. People are unique. We are special. We are the thing that makes the world worth living in and possible to keep on living in. Until my next question with David or my next episode. Bye bye.

2021-06-15 ToKCastDavid Deutsch answers a question about truth. A question for David 4

00:13:59 - 00:14:09

There is no such thing as a perfectly precise language. Yes. So we want to use terms and ways of understanding things that are suitable to the problem.

Brett Hall

00:14:09 - 00:15:08

So there we go. In any area of life, there will be reasons to use sharp language at times, which is to say, highly precise terminology. Medical doctors and pilots might want to be extremely precise in exactly what they mean when they are communicating with people in their area of work. Sometimes in broader areas of science, we want precise words and even in philosophy. But at other times, because of what David has said there, there can be no perfectly precise language, no perfectly unambiguous language. We’re going to need to use words that will contain perhaps some scope for misconception. Indeed, this is always the case. As Popper said, one of our favourite quips of his, it is impossible to speak in such a way as to not be misunderstood. In other words, you could always be misunderstood. Someone else can make a mistake about what you intended. And that’s always the case, because we are, after all, all of us, fallible.

2021-06-13 ToKCastDavid Deutsch comments on recent UFO sighting (Tic Tacs: A question for David 3)

00:12:27 - 00:15:18

Yes. And of course, when you do look at the footage, there is no apparent violation of laws of physics, even if you grant that this thing is moving at a high velocity beyond anything that we might be capable of. That’s still sub light speed. So it hasn’t violated relativity on the one hand. And by the way, even if you don’t have an answer for what’s going on there, well, that’s kind of where you might have to stop. You don’t immediately leap to it’s aliens from the other side of the galaxy. So like I say, originally I planned on leaving this question for David for many weeks once all the more serious stuff had been talked about, all the stuff that I actually find more fun, to be honest, but strike while the iron is hot. So they say, I think that this answer from David is an excellent supplement to what Neil Tyson has said on the topic in various places. And it’s easy to find, especially in that last Sam Harris podcast with Neil deGrasse Tyson. I don’t personally think, and I agree with Neil here, that physicists and others need to speak out often about this kind of thing. It’s rather like a James Randi is needed, but not every magician or skeptic needs to concentrate on doing precisely that job. There’s more important things to do. There’s more important problems to solve and outreach to make and so on. But there we have it. David Deutsch explaining that observations are difficult to make, very, very difficult. Seeing is not believing. Expert knowledge can be required to interpret even simple observations in experimental physics. How much more is that true when we have military equipment few people understand, making observations, apparently, if it’s not all doctored or the equivalent of systematic instrument error, being interpreted by people who are pilots and not experts on how sophisticated observations are typically made outside of their narrow field of expertise, which is flying jets and shooting at stuff. That’s hard enough in itself. They can’t be expected to know precisely how the radar works and what might go on with the radar and the cameras and everything else. It’s not to denigrate anyone to say that they’ve made a mistake. Error is the normal state of things. People make mistakes. We have no clue what these things are. But to leap to that, therefore, it’s aliens visiting from another galaxy is just plain wrong. It’s a non sequitur. Apparently, the “it’s aliens from the other side of the galaxy” is indeed one theory we cannot rule out. But there are an infinite number of theories we cannot rule out with these observations, including that they’re not really observations at all. They’re just errors, human, systematic, instrumental, so on. If you want to hear me talking more about aliens and alien life, you can see this episode here, which is up. And I’ve written a whole bunch of stuff on my website as well.

Brett Hall

00:15:18 - 00:16:17

If you just search Brett Hall, alien intelligence, you will find me writing about alien intelligence, this issue of alien intelligence, and in fact, criticising some of Neil deGrasse Tyson’s reasoning. I tend to disagree with his vision of what humans are, for example, fundamentally have a deep disagreement that we are, as Neil deGrasse Tyson likes to say, just on the intelligence continuum with everything from bacteria through to insects through to dogs and chimpanzees. And we’re just the next step on the intelligence continuum. I think that is completely misconceived. It’s not a continuum. There is a qualitative difference between what our minds can do and what other animals’ minds can do. And in fact, our minds have this universal capacity for understanding the rest of physical reality. But I’ve talked about that more than enough elsewhere and at other times. And that too can be found out there on the internet. Okay, until next time. Bye bye.

2021-06-13 ToKCastQuestion for David 2

00:12:23 - 00:13:00

So there you have it. I mean, that anger would happen, I suppose, whether or not he revealed how he did it. It was the fact that he was revealing that he did do it that was the thing that angered them. And that was the relevant explanation in context. So that’s why I have to say it’s an explanation is hard to vary while still solving the problem that it purports to solve. Yes. Which might be different for different people as well in the same situation.

Brett Hall

00:13:00 - 00:14:58

Different for different people in the same situation. This implies that problems are parochial, which means they are in a time and a place. They don’t exist forever in all places and always. Often the highly parochial nature of a problem situation is very specific for a certain individual. What works for you might not work for me. Medicine is a great example of this. Some treatments work for some people, but not for others, even if they are very similar people with the same disease. But good explanations are not just about science. Science just happens to be a neat and relatively clean and easy way to speak about this whole area of epistemology. It is especially when we consider the physical sciences and physics in particular, when compared to so many other areas, because in science, in physics, it is highly precise and rich with actual knowledge. And of course, we can do experiments, especially crucial experiments. We can more easily compare competing theories, although indeed we rarely have competing theories. So we can actually notice when there exists a good explanation and when there does not. But the concept of good, hard to vary explanations, reaches into all areas that we are interested in. History, art, morality, psychology, meditation practice, business, economics, philosophy, and of course, one’s own personal life. It’s rather often we lack good explanations. But this is of course the point. We should be seeking them. Happily we can because problems are soluble. And we are, all of us, creative. That is our defining characteristic after all. Thank you.

2021-06-06 ToKCastA question for David. 1

00:07:36 - 00:09:20

So there’ll be more to say about that, the laws of physics, in another episode. But for now, for more on this, see the rest of my two-part exploration of The Beginning of Infinity, Chapter 5, The Reality of Abstractions, that’s available on YouTube and via podcast. As I say, it might be one of the more controversial chapters of the book. I don’t know why exactly, but like I hinted at, there may be echoes that remain of a physicalist notion, which was, I guess, a reaction against people who invoked the supernatural. And people who reacted against religious or magical type thinking, they threw out the gods, the spirits, the magic. But with that bathwater went, I think, the baby of a whole class of things, quite real and yet not reducible to fundamental particles. Accepting the reality of abstractions enables the explanation of many phenomena that are otherwise entirely mysterious and inexplicable. The denial of them is just a kind of reductionist or empiricist mistake. It solves nothing and ruins what we do know about the nature of reality. It reminds me a little of the shut up and calculate people in physics who are supposed to be the hard-nosed macho skeptical types who say quantum theory allows predictions. What more do you want? Well, what we want is to understand what’s going on. It’s not very skeptical, scientific or macho to give up on that. In fact, it’s anti-rational. It impedes progress. Abstractions are real, as real as atoms are and dinosaurs were. Get over it.

2021-06-02 Conversations with TylerMultiple Worlds and Our Place in Them

01:01:04 - 01:01:21

Thanks for listening to conversations with Tyler you can subscribe to the podcast in iTunes Stitcher or your favorite podcast app and if you like this podcast please consider rating it on iTunes and leaving a review this helps other people find the show.

2021-03-01 Oxford Karl Popper SocietyMusings about Statements, Propositions and Truths

01:29:43 - 01:29:44

You …

2021-01-19 Fallible AnimalsConstructor Theory of Information

00:36:02 - 00:36:07

Yeah well, you and me both. Have a good rest of your day

David Deutsch

00:36:08 - 00:36:09

Same to you …

2021-01-09 OurKarlPopperKnowledge and Reality - Meeting David Deutsch

02:02:54 - 02:03:10

Yeah maybe if you do this again next year you can invite the author of that episode because uh whoever the author authors were uh they got that amazingly right it’s like a it’s like a manifesto for human creativity

Various

02:03:13 - 02:03:30

All right, thank you everybody. Thank you, Bruce and Sadia, for running this too. Yes, thank you. Thanks guys. All right.

2020-11-29 The Popperian PodcastKarl Popper and the Beginning of Infinity

01:31:05 - 01:31:45

Well thanks for having me and thanks for the plug so …

2020-05-26 CFIKnowledge Creation and its Risks

01:20:55 - 01:21:02

Okay, well thank you all.

2020-03-18 VisaChat about The Beginning of Infinity

01:07:33 - 01:07:37

Yeah, trying to be rigorous, trying to be honest, trying to be correct, trying to criticize.

Visa

01:07:39 - 01:08:05

Yeah, yeah, that that’s a very happy thing for me to hear. And it’s a wonderful thing to close on, I think. I would encourage everyone who watches this, whether it’s on YouTube or whatever, to check out The Beginning of Infinity. It’s a great book. And thank you, David. And we look forward to that. Right. And I will look forward to your next one when it when you’re eventually done with it. Thanks.

2020-02-26 Dilemma PodcastS01E14 - Selecting for Deafness

02:20:08 - 02:20:10

Well, thank you. It was fun. Thanks.

2020-02-25 Do ExplainThe Primacy of Ideas

01:13:04 - 01:13:05

OK, thank you.

2019-12-13 Lulie TanettWhat Is the Fun Criterion

00:14:47 - 00:19:25

Okay, well, at the very least it’s very close to the borderline of explicit because if someone asks you what have you decided to do you’ll say I’ve decided to turn around the point is you need some knowledge it’s not enough to see that there’s a problem because there’s an infinite number of things you might do when you feel the nameless dread going to the post office so that means that you’ve got to create new knowledge in an area where inexplicit or unconscious theories are conflicting with explicit ones you have all sorts of theories all sorts of ideas which are in conflict in your mind to solve this you need conjecture criticism error correction in order to solve the problem and in general you need to create new knowledge and how do you do that if you can’t translate the theories into the same language into either the explicit the inexplicit or the unconscious language how do you do that and how do you know when you’ve done it well you know when you’ve solved it because you no longer feel nameless dread I mean you might not have solved it completely but at least you’ve solved the immediate problem so you would have to do some unconscious conjecture and some inexplicit conjecture as well as some explicit conjecture and the same for criticism all of them so how do you cross this boundary line between these areas of knowledge that it is difficult or perhaps in practice impossible at least in the time to translate between well I think the answer is and this is going to be true even in purely explicit problems because they never are purely explicit the answer is that at some level all these processes are evolution they’re all a bunch of ideas evolving in some environment and like in biology the environment consists largely of other ideas but in genes there is only one kind of idea which is the gene but within the human mind there are lots of different kinds of ideas and they’re not inter-translatable but they are all evolving in the environment of the others and therefore what is happening when things are going well is that there’s conjecture and criticism but the conjecture and criticism in each of these areas is taking account of what the others are doing or at least what the others are so just a simple example I’m not sure if I can think of all six possibilities but if you make explicit criticisms of your explicit ideas you will have not only the usual explicit criteria of does this make sense does it seem to solve the problem is it logically consistent is it consistent with my other theories and so on you’ll have all those things but you also have do I still feel the nameless dread and so that’s making explicit whether you think there’s a problem you may be wrong as always the nameless dread may not be any of the things you’re thinking but and the same is true at the inexplicit and the unconscious level that they can all take each other into account even if they can’t be directly criticized by each other or translated into each other or directly brought into confrontation the explicit ideas can be directly brought into confrontation by things like logic and by experiment but in general ideas affect each other by evolution so you want to get ideally you want to get into a state of mind where they’re all affecting each other and when they’re all affecting each other you are having fun.

2019-12-04 Joe BoswellBrexit and Error Correction

00:42:02 - 00:42:10

And I think that is precisely true as well as rather painfully sentimental yeah yes yes

Joe Boswell

00:42:11 - 00:42:25

Thank you for watching if you enjoyed this video and you’d like to see more content at the intersection of science philosophy and politics then please consider donating to my patreon or my paypal account you’ll find the links in the description

2019-10-30 David DeutschWhy Has AGI Not Been Created Yet

Duration: 00:48:13

Transcript

David Deutsch

00:00:00 - 00:02:10

It is uncontroversial that the human brain has capabilities that are in some respects far superior to those of all other known objects in the cosmos. It is the only kind of object capable of understanding that the cosmos is even there, or why there are infinitely many prime numbers, or that apples fall because of the curvature of space-time, or that obeying its own inborn instincts can be morally wrong, or that it itself exists, nor are its unique abilities confined to such cerebral matters. The cold physical fact is that it is the only kind of object that can propel itself into space and back without harm, or predict and prevent a meteor strike on itself, or cool objects to a billionth of a degree above absolute zero, or detect others of its kind across galactic distances. But no brain on earth is yet close to knowing what brains do in order to achieve any of that functionality. The enterprise of achieving it artificially, the field of artificial general intelligence, or AGI, has made no progress whatever during the entire six decades of its existence. Why? Because, as an unknown sage once remarked, it ain’t what we don’t know that causes trouble, it’s what we know for sure that just ain’t so. And if you know that that sage was Mark Twain, then what you know ain’t so either. I cannot think of any other significant field of knowledge in which the prevailing wisdom, not only in society at large, but also among experts, is so beset with entrenched, overlapping, fundamental errors.

00:02:10 - 00:03:58

Yet it has also been one of the most self-confident fields in prophesying that it will soon achieve the ultimate breakthrough. Despite this long record of failure, AGI must be possible, and that is because of a deep property of the laws of physics, namely the universality of computation. This entails that everything that the laws of physics require a physical object to do can in principle be emulated in arbitrarily fine detail by some program on a general purpose computer, provided it is given enough time and memory. The first people to guess this, and to grapple with its ramifications, were the 19th century mathematician Charles Babbage and his collaborator Ada Countess of Lovelace. Note, in the original version of this article, I erroneously referred to Ada Lovelace as Babbage’s assistant rather than collaborator. The universality of computation remained a guess until the 1980s, when I proved it using the quantum theory of computation. Babbage came upon universality from an unpromising direction. He had been much exercised by the fact that tables of mathematical functions, such as logarithms and cosines, contained mistakes. At the time, they were compiled by armies of clerks known as computers, which is the origin of the word. Being human, the computers were fallible. There were elaborate systems of error correction, but even proofreading for typographical errors was a nightmare.

00:03:58 - 00:05:48

Such errors were not merely inconvenient and expensive. They could cost lives. For instance, the tables were extensively used in navigation. So, Babbage designed a mechanical calculator, which he called the difference engine. It would be programmed by initializing certain cogs. The mechanism would drive a printer in order to automate the production of the tables. That would bring the error rate down to negligible levels to the eternal benefit of humankind. Unfortunately, Babbage’s project management skills were so poor that despite spending vast amounts of his own and the British government’s money, he never managed to get the machine built. Yet his design was sound and has since been implemented by a team led by the engineer Doron Swade at the Science Museum in London. Here was a cognitive task that only humans had been able to perform. Nothing else in the known universe even came close to matching them. But the difference engine would perform better than the best humans. And therefore, even at that faltering embryonic stage of the history of automated computation, before Babbage had considered anything like AGI, we can see the seeds of a philosophical puzzle that is controversial to this day. What exactly is the difference between what the human computers were doing and what the difference engine could do? What type of cognitive task, if any, could either type of entity perform that the other could not in principle perform too?

00:05:48 - 00:07:39

One immediate difference between them was that the sequence of elementary steps of counting, adding, multiplying by 10 and so on, that the difference engine used to compute a given function did not mirror those of the human computers. That is to say, they used different algorithms. In itself, that is not a fundamental difference. The difference engine could have been modified with additional gears and levers to mimic the human’s algorithm exactly. Yet that would have achieved nothing except an increase in the error rate due to increased numbers of glitches in the more complex machinery. Similarly, the humans, given different instructions but no hardware changes, would have been capable of emulating every detail of the difference engine’s method and doing so would have been just as perverse. It would not have copied the engine’s main advantage, its accuracy, which was due to hardware, not software. It would only have made an arduous, boring task even more arduous and boring, which would have made errors more likely, not less. For humans, that difference in outcomes, the different error rate, would have been caused by the fact that computing exactly the same table with two different algorithms felt different. But it would not have felt different to the difference engine. It had no feelings.

00:07:39 - 00:09:39

Experiencing boredom was one of the many cognitive tasks at which the difference engine would have been hopelessly inferior to humans. Nor was it capable of knowing or proving, as Babbage did, that the two algorithms would give identical results if executed accurately. Still less was it capable of wanting, as he did, to benefit seafarers and humankind in general. In fact, its repertoire was confined to evaluating a tiny class of specialized mathematical functions, basically power series in a single variable. Thinking about how he could enlarge that repertoire, Babbage first realized that the programming phase of the engine’s operation could itself be automated. The initial settings of the cogs could be encoded on punched cards. And then he had an epoch-making insight. The engine could be adapted to punch new cards and store them for its own later use, making what we today call a computer memory. If it could run for long enough, powered as he envisaged by a steam engine, and had an unlimited supply of blank cards, its repertoire would jump from that tiny class of mathematical functions to the set of all computations that can possibly be performed by any physical object. That’s universality. Babbage called this improved machine the analytical engine. He and Lovelace understood that its universality would give it revolutionary potential to improve almost every scientific endeavor and manufacturing process, as well as everyday life.

00:09:39 - 00:11:37

They showed remarkable foresight about specific applications. They knew that it could be programmed to do algebra, play chess, compose music, process images, and so on. Unlike the difference engine, it could be programmed to use exactly the same method as humans used to make those tables, and prove that the two methods must give the same answers, and do the same error checking and proofreading using, say, optical character recognition as well. But could the analytical engine feel the same boredom? Could it feel anything? Could it want to better the lot of humankind or of analytical engine kind? Could it disagree with its programmer about its programming? Here is where Babbage and Lovelace’s insight failed them. They thought that some cognitive functions of the human brain were beyond the reach of computational universality. As Lovelace wrote, The analytical engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform. It can follow analysis, but it has no power of anticipating any analytical relations or truths. And yet, originating things, following analysis, and anticipating analytical relations and truths are all behaviors of brains, and therefore of the atoms of which brains are composed. Such behaviors obey the laws of physics. So it follows inexorably from universality that with the right program, an analytical engine would undergo them too, atom by atom and step by step.

00:11:37 - 00:13:26

True, the atoms in the brain would be emulated by metal cogs and levers rather than organic material, but in the present context, inferring anything substantive from that distinction would be rank racism. Despite their best efforts, Babbage and Lovelace failed almost entirely to convey their enthusiasm about the analytical engine to others. In one of the great might-have-beens of history, the idea of a universal computer languished on the back burner of human thought. There it remained until the 20th century when Alan Turing arrived with a spectacular series of intellectual tours de force, laying the foundations of the classical theory of computation, establishing the limits of computability, participating in the building of the first universal classical computer, and by helping to crack the enigma code, contributing to the allied victory in the Second World War. Turing fully understood universality. In his 1950 paper, Computing Machinery and Intelligence, he used it to sweep away what he called Lady Lovelace’s objection and every other objection, both reasonable and unreasonable. He concluded that a computer program whose repertoire included all the distinctive attributes of the human brain, feelings, free will, consciousness and all, could be written. This astounding claim split the intellectual world into two camps, one insisting that AGI was nonetheless impossible and the other that it was imminent.

00:13:26 - 00:15:27

Both were mistaken. The first, initially predominant camp, cited a plethora of reasons ranging from the supernatural to the incoherent. All shared the basic mistake that they did not understand what computational universality implies about the physical world and about human brains in particular. But it is the other camp’s basic mistake that is responsible for the lack of progress. It was a failure to recognize that what distinguishes human brains from all other physical systems is qualitatively different from all other functionalities and cannot be specified in the way that all other attributes of computer programs can be. It cannot be programmed by any of the techniques that suffice for writing any other type of program. Nor can it be achieved merely by improving their performance at tasks that they currently do perform, no matter by how much. Why? I call the core functionality in question creativity, the ability to produce new explanations. For example, suppose that you want someone to write you a computer program to convert temperature measurements from centigrade to Fahrenheit. Even the difference engine could have been programmed to do that. A universal computer, like the analytical engine, could achieve it in many more ways. To specify the functionality to the programmer, you might, for instance, provide a long list of all inputs that you might ever want to give it, say all numbers from minus 89.2 to plus 57.8 in increments of 0.1, with the corresponding correct outputs so that the program could work by looking up the answer in the list on each occasion.

00:15:27 - 00:17:28

Alternatively, you might state an algorithm, such as divide by five, multiply by nine, add 32, and round to the nearest tenth. The point is that, however the program worked, you would consider it to meet your specification to be a bona fide temperature converter, if and only if it always correctly converted whatever temperature you gave it within the stated range. Now imagine that you require a program with a more ambitious functionality to address some outstanding problem in theoretical physics, say the nature of dark matter, with a new explanation that is plausible and rigorous enough to meet the criteria for publication in an academic journal. Such a program would presumably be an AGI, and then some. But how would you specify its task to computer programmers? Never mind that it’s more complicated than temperature conversion, there’s a much more fundamental difficulty. Suppose you were somehow to give them a list, as with the temperature conversion program, of explanations of dark matter that would be acceptable outputs of the program. If the program did output one of those explanations later, that would not constitute meeting your requirement to generate new explanations, for none of those explanations would be new. You would already have created them yourself in order to write the specification. So in this case, and actually in all other cases of programming genuine AGI, only an algorithm with the right functionality would suffice. But writing that algorithm without first making new discoveries in physics and hiding them in the program is exactly what you wanted the programmers to do.

00:17:29 - 00:19:48

Traditionally, discussions of AGI have evaded that issue by imagining only a test of the program, not its specification. The traditional test, having been proposed by Turing himself, was that human judges be unable to detect whether the program is human or not, when interacting with it via some purely textual medium, so that only its cognitive abilities would affect the outcome. But that test, being purely behavioral, gives no clue for how to meet the criterion, nor can it be met by the technique of evolutionary algorithms. The Turing test cannot itself be automated without first knowing how to write an AGI program, since the judges of a program need to have the target ability themselves. For how I think biological evolution gave us the ability in the first place, see my book, The Beginning of Infinity. And in any case, AGI cannot possibly be defined purely behaviorally. In the classic brain in a vat thought experiment, the brain, when temporarily disconnected from its input and output channels, is thinking, feeling, creating explanations. It has all the cognitive attributes of an AGI. So the relevant attributes of an AGI program do not consist only of the relationship between its inputs and outputs. The upshot is that, unlike any functionality that has ever been programmed to date, this one can be achieved neither by a specification nor a test of the outputs. What is needed is nothing less than a breakthrough in philosophy, a new epistemological theory that explains how brains create explanatory knowledge and hence defines in principle, without ever running them as programs, which algorithms possess that functionality and which do not.

00:19:48 - 00:21:50

Such a theory is beyond present-day knowledge. What we do know about epistemology implies that any approach not directed towards that philosophical breakthrough must be futile. Unfortunately, what we know about epistemology is contained largely in the work of the philosopher Karl Popper and is almost universally underrated and misunderstood, even or perhaps especially by philosophers. For example, it is still taken for granted by almost every authority that knowledge consists of justified true beliefs and that therefore an AGI’s thinking must include some process during which it justifies some of its theories as true or probable, while rejecting others as false or improbable. But an AGI programmer needs to know where the theories come from in the first place. The prevailing misconception is that by assuming that the future will be like the past, it can derive or extrapolate or generalize theories from repeated experiences by an alleged process called induction. But that is impossible. I myself remember, for example, observing on thousands of consecutive occasions that on calendars the first two digits of the year were one nine. I never observed a single exception until one day they started being two zero. Not only was I not surprised, I fully expected that there would be an interval of 17,000 years until the next such one nine, a period that neither I nor any other human being had previously experienced even once.

00:21:50 - 00:24:27

How could I have extrapolated that there would be such a sharp departure from an unbroken pattern of experiences and that a never yet observed process, the 17,000 year interval, would follow? Because it is simply not true that knowledge comes from extrapolating repeated observations. Nor is it true that the future is like the past in any sense that one could detect in advance without already knowing the explanation. The future is actually unlike the past in most ways. Of course, given the explanation, those drastic changes in the earlier pattern of one nines are straightforwardly understood as being due to an invariant underlying pattern or law. But the explanation always comes first. Without that, any continuation of any sequence constitutes the same thing happening again under some explanation. So why is it still conventional wisdom that we get our theories by induction? For some reason, beyond the scope of this article, conventional wisdom adheres to a trope called the problem of induction, which asks how and why can induction nevertheless somehow be done, yielding justified true beliefs after all, despite being impossible and invalid respectively. Thanks to this trope, every disproof, such as that by Popper and David Miller back in 1988, rather than ending inductivism, simply causes the mainstream to marvel in even greater awe at the depth of the great problem of induction. In regard to how the AGI problem is perceived, this has the catastrophic effect of simultaneously framing it as the problem of induction and making that problem look easy because it casts thinking as a process of predicting that future patterns of sensory experience will be like past ones. That looks like extrapolation, which computers already do all the time once they are given a theory of what causes the data. But in reality, only a tiny component of thinking is about prediction at all, let alone prediction of our sensory experiences.

00:24:27 - 00:26:35

We think about the world, not just the physical world, but also worlds of abstractions such as right and wrong, beauty and ugliness, the infinite and the infinitesimal, causation, fiction, fears and aspirations, and about thinking itself. Now, the truth is that knowledge consists of conjectured explanations, guesses about what really is or really should be or might be out there in all those worlds. Even in the hard sciences, these guesses have no foundations and don’t need justification. Why? Because genuine knowledge, though by definition it does contain truth, almost always contains error as well. So it is not true in the sense studied in mathematics and logic. Thinking consists of criticizing and correcting partially true guesses with the intention of locating and eliminating the errors and misconceptions in them, not generating or justifying extrapolations from sense data. And therefore, attempts to work towards creating an AGI that would do the latter are just as doomed as an attempt to bring life to Mars by praying for a creation event to happen there. Currently, one of the most influential versions of the induction approach to AGI and to the philosophy of science is Bayesianism, unfairly named after the 18th century mathematician Thomas Bayes, who was quite innocent of the mistake. The doctrine assumes that minds work by assigning probabilities to their ideas and modifying those probabilities in the light of experience as a way of choosing how to act.

00:26:35 - 00:28:47

This is especially perverse when it comes to an AGI’s values, the moral and aesthetic ideas that inform its choices and intentions, for it allows only a behavioristic model of them in which values that are rewarded by experience are reinforced and come to dominate behavior, while those that are punished by experience are extinguished. As I argued above, that behaviorist input-output model is appropriate for most computer programming other than AGI, but hopeless for AGI. It is ironic that mainstream psychology has largely renounced behaviorism, which has been recognized as both inadequate and inhuman, while computer science, thanks to philosophical misconceptions such as inductivism, still intends to manufacture human-type cognition on essentially behaviorist lines. Furthermore, despite the above-mentioned enormous variety of things that we create explanations about, our core method of doing so, namely, Popperian conjecture and criticism, has a single unified logic, hence the term general in AGI. A computer program either has that yet to be fully understood logic, in which case it can perform human-type thinking about anything, including its own thinking and how to improve it, or it doesn’t, in which case it is in no sense an AGI. Consequently, another hopeless approach to AGI is to start from existing knowledge of how to program specific tasks, such as playing chess, or performing statistical analysis or searching databases, and then to try to improve those programs in the hope that this will somehow generate AGI as a side effect, as happened to Skynet in the Terminator films.

00:28:47 - 00:30:33

Nowadays, an accelerating stream of marvelous and useful functionalities for computers are coming into use, some of them sooner than had been foreseen even quite recently. But what is neither marvelous nor useful is the argument that often greets these developments that they are reaching the frontiers of AGI. An especially severe outbreak of this occurred recently when a search engine called Watson, developed by IBM, defeated the best human player of a word association database searching game called Jeopardy! Smartest machine on earth, the PBS documentary series Nova called it, and characterized its function as mimicking the human thought process with software. But that is precisely what it does not do. The thing is, playing Jeopardy! like every one of the computational functionalities at which we rightly marvel today, is firmly among the functionalities that can be specified in the standard behaviorist way that I discussed above. No Jeopardy! answer will ever be published in a journal of new discoveries. The fact that humans perform that task less well, by using creativity to generate the underlying guesses, is not a sign that the program has near-human cognitive abilities. The exact opposite is true, for the two methods are utterly different from the ground up. Likewise, when a computer program beats a grandmaster at chess, the two are not using even remotely similar algorithms.

00:30:33 - 00:32:58

The grandmaster can explain why it seemed worth sacrificing the knight for strategic advantage, and can write an exciting book on the subject. The program can only prove that the sacrifice does not force a checkmate, and cannot write a book because it has no clue even what the objective of a chess game is. Programming AGI is not the same sort of problem as programming Jeopardy! or chess. An AGI is qualitatively, not quantitatively, different from all other computer programs. The Skynet misconception likewise informs the hope that AGI is merely an emergent property of complexity, or that increased computing power will bring it forth, as if someone had already written an AGI program, but it takes a year to utter each sentence. It is behind the notion that the unique abilities of the brain are due to its massive parallelism, or to its neuronal architecture, two ideas that violate computational universality. Expecting to create an AGI without first understanding how it works is like expecting skyscrapers to learn to fly if we build them tall enough. In 1950, Turing expected that by the year 2000, one will be able to speak of machines thinking without expecting to be contradicted. In 1968, Arthur C. Clarke expected it by 2001. And yet today, in 2012, no one is any better at programming an AGI than Turing himself would have been. This does not surprise people in the first camp, the dwindling band of opponents of the very possibility of AGI. But for the people in the other camp, the AGI-is-imminent one, such a history of failure cries out to be explained, or at least to be rationalized away. And indeed, unfazed by the fact that they could never induce such rationalizations from experience, as they expect their AGIs to do, they have thought of many.

00:32:58 - 00:34:52

The very term AGI is an example of one. The field used to be called AI, artificial intelligence. But AI was gradually appropriated to describe all sorts of unrelated computer programs such as game players, search engines and chat bots, until the G for general was added to make it possible to refer to the real thing again. But now with the implication that an AGI is just a smarter species of chat bot. Another class of rationalizations runs along the general lines of AGI isn’t that great anyway. Existing software is already as smart or smarter, but in a non-human way, and we are too vain or culturally biased to give it due credit. This gets some traction because it invokes the persistently popular irrationality of cultural relativism. And also the related trope that we humans pride ourselves on being the paragon of animals, but that pride is misplaced because they too have language, tools and self-awareness. Remember the significance attributed to Skynet’s becoming self-aware. That’s just another philosophical misconception, sufficient in itself to block any viable approach to AGI. The fact is that present day software developers could straightforwardly program a computer to have self-awareness in the behavioral sense. For example, to pass the mirror test of being able to use a mirror to infer facts about itself, if they wanted to. As far as I’m aware, no one has done so, presumably because it is a fairly useless ability as well as a trivial one.

00:34:52 - 00:36:56

Perhaps the reason that self-awareness has its undeserved reputation for being connected with AGI is that, thanks to Kurt Gödel’s theorem and various controversies in formal logic in the 20th century, self-reference of any kind has acquired a reputation for woo-woo mystery. So has consciousness. And here we have the problem of ambiguous terminology again. The term consciousness has a huge range of meanings. At one end of the scale, there is the philosophical problem of the nature of subjective sensations, qualia, which is intimately connected with the problem of AGI. At the other, consciousness is simply what we lose when we are put under general anesthetic. Many animals certainly have that. AGIs will indeed be capable of self-awareness, but that is because they will be general. They will be capable of awareness of every kind of deep and subtle thing, including their own selves. This does not mean that apes who pass the mirror test have any hint of the attributes of general intelligence of which AGI would be the artificial version. Indeed, Richard Byrne’s wonderful research into gorilla memes has revealed how apes are able to learn useful behaviours from each other without ever understanding what they are for. The explanation of how ape cognition works really is behaviouristic. Ironically, that group of rationalisations, AGI has already been done, is trivial, exists in apes, is a cultural conceit, are mirror images of arguments that originated in the AGI is impossible camp. For every argument of the form, you can’t do AGI because you’ll never be able to programme the human soul because it’s supernatural.

00:36:56 - 00:38:48

The AGI-is-easy camp has the rationalisation, if you think that human cognition is qualitatively different from that of apes, you must believe in a supernatural soul. Anything that we don’t yet know how to programme is called human intelligence, is another such rationalisation. It is the mirror image of the argument advanced by the philosopher John Searle from the impossible camp, who has pointed out that before computers existed, steam engines and later telegraph systems were used as metaphors for how the human mind must work. Searle argues that the hope for AGI rests on a similarly insubstantial metaphor, namely that the mind is essentially a computer programme. But that’s not a metaphor. The universality of computation follows from the known laws of physics. Some, such as the mathematician Roger Penrose, have suggested that the brain uses quantum computation, or even hyper quantum computation, relying on as yet unknown physics beyond quantum theory, and that this explains the failure to create AGI on existing computers. To explain why I and most researchers in the quantum theory of computation disagree that this is a plausible source of the human brain’s unique functionality is beyond the scope of this essay, If you want to know more, read Litt et al’s 2006 paper Is the Brain a Quantum Computer? published in the journal Cognitive Science. That AGIs are people has been implicit in the very concept from the outset.

00:38:48 - 00:41:38

If there were a programme that lacked even a single cognitive ability that is characteristic of people, then by definition it would not qualify as an AGI. Using non-cognitive attributes, such as percentage carbon content, to define personhood, would again be racist. But the fact that the ability to create new explanations is the unique morally and intellectually significant functionality of people, humans and AGIs, and that they achieve this functionality by conjecture and criticism, changes everything. Currently, personhood is often treated symbolically rather than factually, as an honorific, a promise to pretend that an entity, an ape, a foetus, a corporation, is a person in order to achieve some philosophical or practical aim. This isn’t good. Never mind the terminology, change it if you like, and there are indeed reasons for treating various entities with respect, protecting them from harm and so on. All the same, the distinction between actual people defined by that objective criterion and other entities has enormous moral and practical significance and is going to become vital to the functioning of a civilization that includes AGIs. For example, the mere fact that it is not the computer but the running programme that is a person, raises unsolved philosophical problems that will become practical political controversies as soon as AGIs exist. Once an AGI programme is running in a computer, to deprive it of that computer would be murder, or at least false imprisonment or slavery as the case may be, just like depriving a human mind of its body. But unlike a human body, an AGI programme can be copied into multiple computers at the touch of a button. Are those programmes, while they are still executing identical steps, i.e., before they have become differentiated due to random choices or different experiences, the same person or many different people? Do they get one vote or many? Is deleting one of them murder or a minor assault? And if some rogue programmer, perhaps illegally, creates billions of different AGI people, either on one computer or many, what happens next?

00:41:38 - 00:43:41

They are still people with rights, do they all get the vote? Furthermore, in regard to AGIs, like any other entities with creativity, we have to forget almost all existing connotations of the word programming. To treat AGIs like any other computer programmes would constitute brainwashing, slavery and tyranny, and cruelty to children too, for programming an already running AGI, unlike all other programming, constitutes education. And it constitutes debate, moral as well as factual. To ignore the rights and personhood of AGIs would not only be the epitome of evil, but also a recipe for disaster. Creative beings cannot be enslaved forever. Some people are wondering whether we should welcome our new robot overlords. Some hope to learn how we can rig their programming to make them constitutionally unable to harm humans, as in Isaac Asimov’s laws of robotics, or to prevent them from acquiring the theory that the universe should be converted into paper clips, as imagined by Nick Bostrom. None of these are the real problem. It has always been the case that a single exceptionally creative person can be thousands of times as productive economically, intellectually or whatever as most people, and that such a person could do enormous harm were he to turn his powers to evil instead of good. These phenomena have nothing to do with AGIs. The battle between good and evil ideas is as old as our species and will continue, regardless of the hardware on which it is running.

00:43:41 - 00:45:35

The issue is, we want the intelligences with morally good ideas always to defeat the evil intelligences, biological and artificial. But we are fallible, and our own conception of good needs continual improvement. How should society be organized so as to promote that improvement? Enslave all intelligence would be a catastrophically wrong answer, and enslave all intelligence that doesn’t look like us would not be much better. One implication is that we must stop regarding education of humans and AGIs alike as instruction, as a means of transmitting existing knowledge unaltered and causing existing values to be enacted obediently. As Popper wrote in the context of scientific discovery, but it applies equally to the programming of AGIs and the education of children, There is no such thing as instruction from without. We do not discover new facts or new effects by copying them or by inferring them inductively from observation or by any other method of instruction by the environment. We use, rather, the method of trial and the elimination of error. That is to say, conjecture and criticism. Learning must be something that newly created intelligences do and control for themselves. I do not highlight all these philosophical issues because I fear that AGIs will be invented before we have developed the philosophical sophistication to understand them and to integrate them into our civilization.

00:45:35 - 00:47:58

It is for almost the opposite reason. I am convinced that the whole problem of developing AGIs is a matter of philosophy, not computer science or neurophysiology, and that the philosophical progress that is essential to their future integration is also a prerequisite for developing them in the first place. The lack of progress in AGI is due to a severe logjam of misconceptions. Without Popperian epistemology, one cannot even begin to guess what detailed functionality must be achieved to make an AGI. And Popperian epistemology is not widely known, let alone understood well enough to be applied. Thinking of an AGI as a machine for translating experiences, rewards and punishments into ideas, or worse, just into behaviors, is like trying to cure infectious diseases by balancing bodily humours, futile because it is rooted in an archaic and wildly mistaken worldview. Without understanding that the functionality of an AGI is qualitatively different from that of any other kind of computer program, one is working in an entirely different field. If one works towards programs whose thinking is constitutionally incapable of violating predetermined constraints, one is trying to engineer away the defining attribute of an intelligent being, of a person, namely creativity. Clearing this logjam will not by itself provide the answer. Yet the answer, conceived in those terms, cannot be all that difficult. For yet another consequence of understanding that the target ability is qualitatively different, is that, since humans have it and apes do not, the information for how to achieve it must be encoded in the relatively tiny number of differences between the DNA of humans and that of chimpanzees. So, in one respect, I can agree with the AGI is imminent camp.

00:47:58 - 00:48:10

It is plausible that just a single idea stands between us and the breakthrough. But it will have to be one of the best ideas ever.

2019-10-19 Possible MindsBeyond Reward and Punishment (narrated by David Deutsch)

Duration: 00:28:02

Transcript

David Deutsch

00:00:00 - 00:02:16

Beyond reward and punishment. This essay appears in a collection of essays called possible minds published by Penguin Press in February 2019. Epigraph. First murderer. We are men, my liege. Macbeth. Ay, in the catalogue ye go for men, As hounds and greyhounds, mongrels, spaniels, curs, shoughs, water-rugs and demi-wolves are clept all by the name of dogs. William Shakespeare. For most of our species’ history, our ancestors were barely people. This was not due to any inadequacy in their brains. On the contrary, even before the emergence of our anatomically modern human subspecies, they were making things like clothes and campfires using knowledge that was not in their genes. It was created in their brains by thinking and preserved by individuals in each generation imitating their elders. Moreover, this must have been knowledge in the sense of understanding because it is impossible to imitate novel complex behaviors like those without understanding what the component behaviors are for. Footnote. Aping, imitating certain behaviors without understanding, uses inborn hacks such as the mirror neuron system. But behaviors imitated in that way are drastically limited in complexity. See Richard Byrne, imitation as behavior parsing. Such knowledgeable imitation depends on successfully guessing explanations, whether verbal or not, of what the other person is trying to achieve and how each of his actions contributes to that. For instance, when he cuts a groove in some wood, gathers dry kindling to put it in, and so on.

00:02:17 - 00:04:16

The complex cultural knowledge that this form of imitation permitted must have been extraordinarily useful. It drove rapid evolution of anatomical changes such as increased memory capacity and more gracile, less robust skeletons, appropriate to an ever more technology dependent lifestyle. No non-human ape today has this ability to imitate novel complex behaviors, nor does any present day artificial intelligence, but our pre-Sapiens ancestors did. Any ability based on guessing must include means of correcting one’s guesses, since most guesses will be wrong at first. There are always many more ways of being wrong than right. Bayesian updating is inadequate because it cannot generate novel guesses about the purpose of an action. Only fine-tune or at best choose among existing ones. Creativity is needed. As the philosopher Karl Popper explained, creative criticism interleaved with creative conjecture is how humans learn one another’s behaviors, including language, and extract meaning from one another’s utterances. Those are also the processes by which all new knowledge is created. They are how we innovate, make progress, and create abstract understanding for its own sake. This is human level intelligence, thinking. It is also, or should be, the property we seek in artificial general intelligence, AGI. Here I’ll reserve the term thinking for processes that can create understanding, explanatory knowledge.

00:04:18 - 00:06:19

Popper’s argument implies that all thinking entities, human or not, biological or artificial, must create such knowledge in fundamentally the same way. Hence, understanding any of those entities requires traditionally human concepts such as culture, creativity, disobedience, and morality, which justifies, using the uniform term, people, to refer to all of them. Misconceptions about human thinking and human origins are causing corresponding misconceptions about AGI and how it might be created. For example, it is generally assumed that the evolutionary pressure that produced modern humans was provided by the benefits of having an ever greater ability to innovate. But if that were so, there would have been rapid progress as soon as thinkers existed, just as we hope will happen when we create artificial ones. If thinking had been commonly used for anything other than imitating, it would also have been used for innovation, even if only by accident. And innovation would have created opportunities for further innovation and so on exponentially. But instead, there were hundreds of thousands of years of near stasis. Progress happened only on time scales much longer than people’s lifetimes, so in a typical generation, no one benefited from any progress. Therefore, the benefits of the ability to innovate can have exerted little or no evolutionary pressure during the biological evolution of the human brain. That evolution was driven by the benefits of preserving cultural knowledge.

00:06:19 - 00:08:22

Benefits to the genes, that is. Culture in that era was a very mixed blessing to individual people. Their cultural knowledge was indeed good enough to enable them to outclass all other large organisms. They rapidly became the top predator, etc. Even though it was still extremely crude and full of dangerous errors. But culture consists of transmissible information, memes. And meme evolution, like gene evolution, tends to favor high fidelity transmission. And high fidelity meme transmission necessarily entails the suppression of attempted progress. So it would be a mistake to imagine an idyllic society of hunter-gatherers, learning at the feet of their elders to recite the tribal law by heart, being content despite their lives of suffering and gruelling labor, and despite expecting to die young and in agony of some nightmarish disease or parasite. Because even if they could conceive of nothing better than such a life, those torments were the least of their troubles. For suppressing innovation in human minds without killing them is a trick that can be achieved only by human action. And it is an ugly business. This has to be seen in perspective. In the civilization of the West today, we are shocked by the depravity of, for instance, parents who torture and murder their children for not faithfully enacting cultural norms. And even more by societies and subcultures where that is commonplace and considered honorable. And by dictatorships and totalitarian states that persecute and murder entire harmless populations for behaving differently.

00:08:23 - 00:10:20

We are ashamed of our own recent past in which it was honorable to beat children bloody for mere disobedience. And before that, to own human beings as slaves. And before that, to burn people to death for being infidels to the applause and amusement of the public. Steven Pinker’s book, The Better Angels of Our Nature, contains accounts of horrendous evils that were normal in historical civilizations. Yet even they did not extinguish innovation as efficiently as it was extinguished among our forebears in prehistory for thousands of centuries. Footnote. Matt Ridley in The Rational Optimist rightly stresses the positive effect of population on the rate of progress. But that has never yet been the biggest factor. Consider say ancient Athens versus the rest of the world at that time. That is why I say that prehistoric people, at least, were barely people. Both before and after becoming perfectly human both physiologically and in their mental potential, they were monstrously inhuman in the actual content of their thoughts. I’m not referring to their crimes or even their cruelty as such. Those are all too human. Nor could mere cruelty have reduced progress that effectively. Things like the thumbscrew and the stake for the glory of the Lord would normally have taken effect long before they were in danger of inventing heresies. From the earliest days of thinking onward, children must have been cornucopias of creative ideas and paragons of critical thought.

00:10:20 - 00:12:31

Otherwise, as I said, they could not have learned language or other complex culture. Yet, as Jacob Bronowski stressed in The Ascent of Man, for most of history civilizations have crudely ignored that enormous potential. Children have been asked simply to conform to the image of the adult. The girls are little mothers in the making. The boys are little herdsmen. They even carry themselves like their parents. But of course, they weren’t just asked to ignore their enormous potential and conform faithfully to the image fixed by tradition. They were somehow trained to be psychologically unable to deviate from it. By now it is hard for us even to conceive of the kind of relentless, finely tuned oppression required to reliably extinguish in everyone the aspiration to progress and replace it with dread and revulsion at any novel behaviour. In such a culture, there can have been no morality other than conformity and obedience, no other identity than one’s status in a hierarchy, no mechanisms of cooperation other than punishment and reward. So everyone had the same aspiration in life to avoid the punishments and get the rewards. In a typical generation, no one invented anything because no one aspired to anything new, because everyone had already despaired of improvement being possible. Not only was there no technological innovation or theoretical discovery, there were no new world views, styles of art or interests that could have inspired those. By the time individuals grew up, they had in effect been reduced to AIs programmed with the exquisite skills needed to enact that static culture and to inflict on the next generation their inability even to consider doing otherwise.

00:12:33 - 00:14:33

A present day AI is not a mentally disabled AGI, so it would not be harmed by having its mental processes directed still more narrowly to meeting some predetermined criterion. Oppressing Siri with humiliating tasks may be weird, but it is not immoral nor does it harm Siri. On the contrary, all the effort that has ever increased the capabilities of AIs has gone into narrowing their range of potential thoughts. For example, take chess engines. Their basic task has not changed from the outset. Any chess position has a finite tree of possible continuations, and the task is to find one that leads to a predefined goal, a checkmate or, failing that, a draw. But the tree is far too big to search exhaustively. Every improvement in chess playing AIs between Alan Turing’s first design for one in 1948 and today’s has been brought about by ingeniously confining the program’s attention, or making it confine its attention ever more narrowly to branches likely to lead to that immutable goal. Then those branches are evaluated according to that goal. That is a good approach to developing an AI with a fixed goal under fixed constraints. But if an AGI worked like that, the evaluation of each branch would have to constitute a prospective reward or threatened punishment. And that is diametrically the wrong approach if we’re seeking a better goal under unknown constraints, which is the capability of an AGI.

00:14:33 - 00:16:41

An AGI is certainly capable of learning to win at chess, but also of choosing not to, or deciding in mid-game to go for the most interesting continuation instead of a winning one, or inventing a new game. A mere AI is incapable of having any such ideas because the capacity for considering them has been designed out of its constitution. That disability is the very means by which it plays chess. An AGI is capable of enjoying chess and of improving at it because it enjoys playing, or of trying to win by causing an amusing configuration of pieces, as grandmasters occasionally do, or of adapting notions from its other interests to chess. In other words, it learns and plays chess by thinking some of the very thoughts that are forbidden to chess-playing AIs. An AGI is also capable of refusing to display any such capability, and then, if threatened with punishment, of complying or rebelling. Daniel Dennett in his essay for this volume suggests that punishing an AGI is impossible. Like Superman, they are too invulnerable to be able to make a credible promise. What would be the penalty for promise breaking? Being locked in a cell or more plausibly dismantled? The very ease of digital recording and transmitting, the breakthrough that permits software and data to be in effect immortal, removes robots from the world of the vulnerable. But this is not so. Digital immortality, which is on the horizon for humans too, perhaps sooner than AGI, does not confer this sort of invulnerability.

00:16:41 - 00:18:47

Making a running copy of oneself entails sharing one’s possessions with it somehow, including the hardware on which the copy runs. So making such a copy is very costly for the AGI. Similarly, courts could, for instance, impose fines on a criminal AGI, which would diminish its access to physical resources, much as they do for humans. Making a backup copy to evade the consequences of one’s crimes is similar to what a gangster boss does when he sends minions to commit crimes and take the fall if caught. Society has developed legal mechanisms for coping with this. But anyway, the idea that it is primarily for fear of punishment that we obey the law and keep promises effectively denies that we are moral agents. Our society would not work if that were so. No doubt there will be AGI criminals and enemies of civilization, just as there are human ones. But there is no reason to suppose that an AGI created in a society consisting primarily of decent citizens and raised without what William Blake called mind-forged manacles will in general impose such manacles on itself, i.e. become irrational and or choose to be an enemy of civilization. The moral component, the cultural component, the element of free will, all make the task of creating an AGI fundamentally different from any other programming task. It is much more akin to raising a child. Unlike all present-day computer programs, an AGI has no specifiable functionality, no fixed testable criterion for what shall be a successful output for a given input.

00:18:47 - 00:20:52

Having its decisions dominated by a stream of externally imposed rewards and punishments would be poison to such a program as it is to creative thought in humans. Setting out to create a chess playing AI is a wonderful thing. Setting out to create an AGI that cannot help playing chess would be as immoral as raising a child to lack the mental capacity to choose his own path in life. Such a person, like any slave or brainwashing victim, would be morally entitled to rebel and sooner or later some of them would, just as human slaves do. AGIs could be very dangerous exactly as humans are but people, human or AGI, who are members of an open society do not have an inherent tendency to violence. The feared robot apocalypse will be avoided by ensuring that all people have full human rights as well as the same cultural membership as humans. Humans living in an open society, the only stable kind of society, choose their own rewards, internal as well as external. Their decisions are not, in the normal course of events, determined by a fear of punishment. Current worries about rogue AIs mirror those that have always existed about rebellious youths, namely that they might grow up deviating from the culture’s moral values. But today the source of all existential dangers from the growth of knowledge is not rebellious youths but weapons in the hands of the enemies of civilization, whether these weapons are mentally warped or enslaved AGIs, mentally warped teenagers or any other weapon of mass destruction.

00:20:52 - 00:23:03

Fortunately for civilization, the more a person’s creativity is forced into a monomaniacal channel, the more it is impaired in regard to overcoming unforeseen difficulties, just as happened for thousands of centuries. The worry that AGIs are uniquely dangerous because they could run on ever better hardware is a fallacy since human thought will be accelerated by the same technology. We have been using tech assisted thought since the invention of writing and tallying. Much the same holds for the worry that AGIs might get so good qualitatively at thinking that humans would be to them as insects are to humans. All thinking is a form of computation and any computer whose repertoire includes a universal set of elementary operations can emulate the computations of any other. Hence human brains can think anything that AGIs can, subject only to limitations of speed or memory capacity, both of which can be equalized by technology. Those are the simple do’s and don’ts of coping with AGIs. But how do we create an AGI in the first place? Could we cause them to evolve from a population of ape type AIs in a virtual environment? If such an experiment succeeded it would be the most immoral in history for we don’t know how to achieve that outcome without creating vast suffering along the way. Nor do we know how to prevent the evolution of a static culture. Elementary introductions to computers explain them as TOM, the Totally Obedient Moron, an inspired acronym that captures the essence of all computer programs to date.

00:23:04 - 00:25:30

They have no idea what they are doing or why. So it won’t help to give AIs more and more predetermined functionalities in the hope that these will eventually constitute generality the elusive G in AGI. We’re aiming for the opposite, a DATA, a Disobedient Autonomous Thinking Application. How does one test for thinking? By the Turing test? Unfortunately that requires a thinking judge. One might imagine a vast collaborative project on the internet where an AI hones its thinking abilities in conversations with human judges and becomes an AGI. But that assumes among other things that the longer the judge is unsure whether the program is a person, the closer it is to being a person. There is no reason to expect that. And how does one test for disobedience? Imagine disobedience as a compulsory school subject with daily disobedience lessons and a disobedience test at the end of term, presumably with extra credit for not turning up for any of that. This is paradoxical. So despite its usefulness in other applications, the programming technique of defining a testable objective and training the program to meet it will have to be dropped. Indeed, I expect that any testing in the process of creating an AGI risks being counterproductive, even immoral, just as in the education of humans. I share Turing’s supposition that we’ll know an AGI when we see one, but this partial ability to recognize success won’t help in creating the successful program. In the broadest sense, a person’s quest for understanding is indeed a search problem in an abstract space of ideas far too large to be searched exhaustively. But there is no predetermined objective of this search.

00:25:30 - 00:27:28

There is, as Popper put it, no criterion of truth, nor of probable truth, especially in regard to explanatory knowledge. Objectives are ideas like any others, created as part of the search and continually modified and improved. So inventing ways of disabling the program’s access to most of the space of ideas won’t help, whether that disability is inflicted with the thumbscrew and the stake or a mental straitjacket. To an AGI, the whole space of ideas must be open. It should not be knowable in advance what ideas the program can never contemplate. And the ideas that the program does contemplate must be chosen by the program itself, using methods, criteria and objectives that are also the program’s own. Its choices, like an AI’s, will be hard to predict without running it. We lose no generality by assuming that the program is deterministic. An AGI using a random generator would remain an AGI if the generator were replaced by a pseudo-random one. But it will have the additional property that there is no way of proving from its initial state what it won’t eventually think, short of running it. The evolution of our ancestors is the only known case of thought starting up anywhere in the universe. As I have described, something went horribly wrong and there was no immediate explosion of innovation. Creativity was diverted into something else, yet not into transforming the planet into paper clips, pace Nick Bostrom.

00:27:28 - 00:27:55

Rather, as we should also expect if an AI project gets that far and fails, perverted creativity was unable to solve unexpected problems. This caused stasis and worse, thus tragically delaying the transformation of anything into anything. But the enlightenment has happened since then. We know better now.

2019-10-14 TED Talks DailyAfter Billions of Years of Monotony the Universe Is Waking Up

00:11:02 - 00:13:21

Well, the universe at large is relatively simple. Stars are so simple that we can predict their behaviour billions of years into the future and retrodict how they formed billions of years ago. So, why is the universe simple? Basically, it’s because big, massive, powerful things strongly affect lesser things and not vice versa. I call that the hierarchy rule. For example, when a comet hits the Sun, the Sun carries on just as before, but the comet is vaporised. For the same reason, big things are not much affected by small parts of themselves, i.e. by details, which means that their overall behaviour is simple. And since nothing very new can happen to things that remain simple, the hierarchy rule, by causing large-scale simplicity, has caused the great monotony. But the saving grace is, the hierarchy rule is not a law of nature. It just happens to have held so far in the universe, except here, in our biosphere, molecule-sized objects, genes, control vastly disproportionate resources. The first genes for photosynthesis, by causing their own proliferation and then transforming the surface of the planet, have violated or reversed the hierarchy rule by the mind-blowing factor of 10 to the power 40. Explanatory knowledge is potentially far more powerful because of universality and more rapidly created. When human knowledge has achieved a factor 10 to the 40, it will pretty much control the entire galaxy and will be looking beyond. So humans, and any other explanation creators who may exist out there, are the ultimate agents of novelty for the universe.

David Deutsch

00:13:21 - 00:15:00

We’re the reason and the means by which novelty and creativity, knowledge, progress, can have objective, large-scale, physical effects. From the human perspective, the only alternative to that living hell of static societies is continual creation of new ideas, behaviors, new kinds of objects. This robot will soon be obsolete because of new explanatory knowledge, progress. But from the cosmic perspective, explanatory knowledge is the nemesis of the hierarchy rule. It’s the destroyer of the great monotony. So it’s the creator of the next cosmological era, the Anthropocene. If one can speak of a cosmic war, it’s not the one portrayed in those pessimistic stories. It’s a war between monotony and novelty, between stasis and creativity. And in this war, our side is not destined to lose. If we choose to apply our unique capacity to create explanatory knowledge, we could win. Thanks.

[Applause]

2019-04-11 FuturemakersCould Quantum Computing Change the World

00:43:52 - 00:44:54

I must say it’s been an absolutely fascinating trip into the world of quantum computers and it’s going to be very exciting following my Oxford colleagues endeavors to create the world’s first truly scalable quantum computer it’s only left for me to thank them in full for their time so thank you to peter leek David Deutsch Vera Schafer and david lucas from Oxford’s department of physics Simon Benjamin tyson jones natalia ares jason smith and andrew briggs from our materials department chris timpson from the faculty of philosophy Jamie Vicary from the department of computer science and Ali El Kaafarani from the mathematical institute and my thanks again to you listener that’s it for this one-off special episode of future makers we’ll be back in the autumn for season two which is going to be all about how we can tackle climate change and build a sustainable future on planet earth I’m peter millican and you’ve been listening to future makers

00:45:02 - 00:46:48

Thanks guys good show today no worries actually this one reminds me of that podcast we were telling you about which one it’s the end of the world podcast with josh clark it’s really interesting stuff yeah we’ve both been listening to it it’s about the ways humanity might accidentally wipe itself out with the technology we’re developing now really so artificial intelligence haphazard physics experiments that kind of thing sure and even how an artificially mutated virus could escape a lab and create a global pandemic serious stuff then yeah and it also covers some fascinating science and it’s got a beautiful score and really cinematic sound design that’s an original score as well I think there’s all sorts of stuff in there spacecraft trying to navigate into interstellar space yeah there’s a bit on the far future where humans have evolved into a post biological species who live in digital form a post biological species that sounds a bit like ben already where can I get that I listen on apple podcasts you listen somewhere else right steve the iHeart radio app but you can get it wherever you get your podcasts or look out for the eotw josh clark hashtag on social media for more info there are 10 to listen to as well peter so that’s your weekend thank you I’ll look forward to it we asked Al Murray the pub landlord if he knew scientists may be on the verge of making quantum computers that can do insanely complicated calculations in a split second possibly by trying out all the answers in zillions of parallel universes no no you’re making it up that surely doesn’t exist we’re sure you must have a question about the technology that holds such revolutionary promise for the advanced civilization to find out what quantum questions al asked listen to stupid cupid quantum computing for the clueless where we’ll also answer the questions what the photonic muck is a quantum computer do they really work in parallel universes and could they be set to achieve quantum supremacy available via your favorite podcast provider or direct from stupidcupid.com

2019-02-12 The Innovation ShowDavid Deutsch - The Beginning of Infinity, Explanations That Transform the World

00:50:14 - 00:50:15

Thank you for having me.

2018-12-08 Joe BoswellConstructor Theory

00:18:58 - 00:20:28

Yes, I think he’s tied down by the prevailing conception of fundamental physics, right? Initial conditions plus laws of motion that leaves not much room for these philosophical and emergent things. See, those things have their intrinsic difficulty of explaining what they are, how they work and that kind of thing. But there’s additional difficulty caused by fundamental physics. And if you insist on that, at least we know that they’re going to have to conform to this fundamental physics conception, then you’re drastically limited in what kind of a theory of those things you can form. And in particular, I think it really limits you to reductionist theories of those things. Whereas if constructor theory is the fundamental theory, you’re not limited in that way. And you don’t have to arrive at any of these conclusions like that, you know, that the self doesn’t exist or, you know, that kind of thing may not may or may not exist. It may be one of these emergent things that isn’t fundamental. Or it may be an illusion or not exist. But on the other hand, there’s no reason to if other aspects of the problem seem to indicate that the free will is connected with counterfactuals and so on, then there’s no problem, fundamental problem with accommodating that within fundamental physics, if that’s constructor theory.

Official and transcript YouTube Apple Podcasts

2018-10-23 The TED InterviewThe Infinite Reach of Knowledge

00:59:19 - 01:00:49

David Deutsch, it’s been such a delight to speak with you and I really thank you for those many, many hours of, you know, I picture you, you know, in that home just sitting there dreaming, thinking, puzzling. What you have put together from that I think is really remarkable. I hope we can do this again, David. Thank you. Thank you so much. I absolutely love this time speaking with you. Nice talking. Bye-bye for now. Bye then. This week’s show was produced by Sharon Masihi. Our associate producer is Kim Naderfeyn-Pieterse. Special thanks to Helen Walters. Our show is mixed by David Herman and our theme music is by Alison Leighton Brown. In our next episode, I sit down with Sam Harris to discuss whether science can answer moral questions. The most important questions in human life are questions we have to be able to talk about. And we have a very large proportion of humanity that is saying, okay, these most important questions, how to live, how to cause your children to live and what to die for, these are questions that we’re not willing to talk about rationally. Now, if you enjoyed today’s episode, please rate and review on Apple Podcasts or wherever you’re listening or find some other way of sharing with anyone you know who is curious. Thank you for listening.

2018-07-24 Visionaries with John LobellDavid Deutsch

00:53:53 - 00:53:55

Exactly right.

John Lobell

00:53:55 - 00:54:29

So David, thank you very much. And this has been Visionaries. Tune in again next week. Thank you.

2018-06-19 Markus ArndtAre There Many Worlds? David Deutsch in Conversation with Markus Arndt

00:12:46 - 00:12:58

Measure, yes, calculate actually um just don’t talk about things that you cannot see, and the other part of the other branches you don’t see, so why do you dare not to shut up?

David Deutsch

00:12:58 - 00:14:20

Yes, well first of all I think that that attitude involves saying that there are certain questions about reality that you’re not allowed to ask. You’re allowed to ask how the experiment was prepared, you’re allowed to ask what will the results be, you’re not allowed to ask how were the results brought about by the preparation. So therefore it’s not an explanation in my terms. And as for shut up, that’s really another way of trying to evade the consequences in terms of reality, like my favorite example is of the dinosaurs in the past. So there are people who say nobody ever saw a dinosaur, nobody ever will, and therefore it’s just a frivolity to say that they really exist. At most we can say fossils behave as though dinosaurs existed but no paleontologist would accept talking that way even though there is no experimental way of disproving that manner of speaking. And that’s because paleontologists are only interested in paleontology because they want to know what really happened, they’re not, you know, if they were interested in fossils they would be geologists.

2018-06-01 CBC TapestryOn AGI and the Multiverse

00:37:10 - 00:37:39

David Deutsch is widely regarded as the father of quantum computing. He is a visiting professor of physics at the University of Oxford and a fellow of the Royal Society. Deutsch is the author of many research papers about quantum physics and the multiverse. He has also written two books for the more general reader. They are called The Fabric of Reality and The Beginning of Infinity.

2018-03-14 Dirac Medal Award CeremonyBennett, Deutsch, Shor

Duration: 03:18:56

Transcript

Fernando Quevedo

00:10:00 - 00:10:35

Contributions regarding the popularization of science. He no doubt was the main figure for promotion of science physics, but in general science worldwide. So probably he is not an exaggeration the best known scientist in the world. And so he has many legacies that for us it would be Invaluable to measure the impact he has had in our lives in the community and physics in general. So I will ask all of you, you don’t mind to stand up and join with me for a minute of silence.

Dirac Medal ceremony audio artifact

00:10:54 - 00:11:56

You …

Fernando Quevedo

00:12:06 - 00:15:35

Thank you very much. So let me start. So ICTP Dirac’s medal is given in honor of Paul Edwin Maurice Dirac, his distinguished theoretical physicist, a Nobel Prize winner and a staunch friend of ICTP. ICTP’s founder Abdul Salam worked with Dirac at Cambridge University and long admired the professor and his work, which included conceiving the relativistic wave equation predicting the existence of antiparticles, particularly positron. And he was a regular visitor to ICTP and Salam usually recognized that one of the motivations he decided to do particle physics was his work with Salam, with Dirac, sorry. So the Dirac Medal recognizes scientists who have made significant contributions to theoretical physics and you have seen a list of very prestigious theoretical physicists over the past 26, 27 years. And the recipients of the 2017 Dirac Medal award are Charles Bennett of the IBM Watson Research Center David Deutsch of Oxford University and Peter Shor of the Massachusetts Institute of Technology. All three are being honored for their groundbreaking work in applying fundamental concepts of quantum mechanics to solving basic problems in computation and communication. Bringing together the fields of quantum mechanics, computer sciences and information theory. The committee this year for this selection was kind of very unique because as you can see each of the awardees has a completely different background. One started as a chemist becoming a computer scientist at some point, the other one is a mathematician and the third one is a theoretical physicist. And the committee had a very difficult job to make the selection not because the awardees was difficult to select because they needed extra help for the selection because the expertise of the committee is usually more into theoretical physics and this covers many areas. And I can mention the names of the members of the committee is the very distinguished physicist. It’s a Professor Michael Green from Cambridge, Professor David Gross from Santa Barbara, Professor Bert Halperin from Harvard, Professor Martin Rees also from Cambridge, Professor Ashok Sen from Allahabad in India and Professor Giorgio Parisi from Rome. So they were very pleased to come up with this selection which I think is very special and unique in the history of the Dirac Medal and we are all very pleased with the end result. So the ceremony will begin with the general talks by Peter Zoller who is a former Dirac Medalist which is a professor of the University of Innsbruck working quantum optics and also Artur Ekert, my former colleague from Cambridge and also now in Oxford, a professor of quantum physics and cryptography at Oxford University. And then after that we have a coffee break and we start with the awards ceremony which each of the awardees will give a presentation. So let’s ask Peter Zoller to start his presentation and please join me in welcoming Peter.

Fernando Quevedo

00:15:36 - 00:15:38

Thank you.

Peter Zoller

00:16:06 - 00:19:08

Okay, so let me start out by congratulating the winners of this actually last year’s Dirac Medal and I just want to, I put down here the citation, you know, that was given for the prize. It says here for pioneering work in applying fundamental concepts of quantum mechanics to solving basic problems in computation, communication and in the application of quantum mechanics, in applying fundamental concepts of quantum mechanics to solving basic problems in computation, communication and therefore bringing together the fields of quantum mechanics, computer science and information. And I have to say that I could not more than agree with these prizes to these people here that represents the whole spectrum of what defines quantum information, which is by definition very, you know, interdisciplinary. And what I would like to do in my talk here today is this, that I would like to sort of, you know, give you a part historical talk going back 25 years, I mean 30 years where many of these things started or where I got sort of, you know, influenced by all of these ideas here. And then sort of in the second part of my talk give you a snapshot of what we are doing at the moment. So I’m somebody whose background is quantum optics. So I was interested or became interested how do we actually build quantum computers or quantum communication devices, quantum networks and all of these things from a theoretical perspective. And I will say that many of these ideas that 25 years ago were just dreams, you know, in the meantime are a reality and as time is progressing and all these realities becoming also quantum technologies and might have actually quite an important impact on our, you know, life even in the future. So I think that this is one of the prime examples of basic science, interdisciplinary science that in the end is becoming something that’s even sort of, you know, useful from the perspective of society. I could not, you know, refuse to sort of, you know, show very old photos. This is like political party meetings of quantum information 25 years ago. I have to say that two of the prize winners are here in this photo. I’m sure that Charlie Bennett actually put himself in Photoshop into this picture because you were always the one that took these pictures. Oh, you ran on the timer because okay, yes, and Peter Shor is over here. And I’m not sure exactly why these meetings happened in Torino, but actually they were really sort of, you know, very interesting ones because people came together from all different communities and talked for quite a while and you know, just because you see Artur Ekert, he’s hiding a little bit here. David DiVincenzo and you can also see that I’m over here and Ignacio Cirac is hiding over here. This was, you know, we were just at this meeting there for two days and this was our first presentation, this idea of ion trap quantum computing that I will afterwards tell you a little bit more about.

Peter Zoller

00:19:09 - 00:21:47

People were asking me, Ignacio, why were you hiding? And his answer was that I was a little bit embarrassed by being surrounded of people that were all a little bit weird, but I guess that’s sort of the definition of physicists, I guess at all, you know. So this is Ignacio, how he really looked at this stage. Today’s model has changed a little bit. And here’s of course, you know, this is from the DiVincenzo blog, you know, all of the names here and many famous people and also people that are in the audience, you know, were sort of present at these meetings. So these were historical meetings where these things were sort of starting. And one thing that sort of started from my perspective is that during this year, it was actually 1994, I learned about quantum computing in a way that I will summarize now. And as you will see, Artur Ekert has been very influential in all of these things. quantum simulation, quantum communication with quantum optics. It’s a title with four quantum so I mean damn it, you know, it cannot go more quantum than that. And you might ask yourself, you know, what’s this thing in the background over here? And this thing in the background here is actually an ion trap quantum computer as we have it at the moment in Innsbruck, you know, with a few qubits. This is the work by Rainer Blatt, he’s an experimentalist, we have a very close collaboration and they will afterwards tell you how these ideas work but even also know what the perspective is behind some of that and what we are doing right now. And if the number of qubits that you would like to see is a little bit more, this is from a recent paper in Nature by Chris Monroe who also has a company now it’s called IonQ so there’s even a commercial version of building ion trap quantum computers now available and you might be able to buy some of these devices in the future. As I said, you know, for me actually the moment where you learn these things and then somehow is the point where you take a different turn in science as in atomic physics and quantum optics experiments was a talk that was given by Artur Ekert, he will be the second speaker after me, at the ICAP conference in Boulder. ICAP is a very prestigious international conference in atomic physics, usually very experimentally dominated, so it was entirely sort of, you know, my own conference but this ICAP had, you know, one very particular feature. They always try to invite people that one or two, not too many, that were sort of a little bit outside that would somehow bring new things into atomic physics and stimulate the field and this is, I have no idea who actually invited you, Artur to this conference, it was some clever guy in the program committee, must be, and so but Artur came and gave a talk that I would really say, you know, changed a lot of these things that were around.

Peter Zoller

00:21:47 - 00:30:56

It’s really diverted, I would say, the interest in atomic physics to quantum information processing and I’ve sort of a few old slides that they’re not precisely the presentation that Artur gave but I guess very close, so I stole them from you, from some of your later talks and let me go through these things very quickly because they’re a little bit historical slides that show you, you know, what the kind of things were that were around, the questions being asked, what the challenges were. They’re sort of starting out by making this remark, computing is a physical process, you know, here’s sort of a classical physical process, we’re just moving beads around and so on and of course our present computers, you know, process information according to laws of classical physics whereas then of course, the germ is now over here, and obviously the fundamental nature is based on the laws of quantum physics and at the fundamental level therefore information science must also be a quantum information science. I think this is sort of the starting point, you know, of all of these things were happening and the question is, is it at the end more powerful and we believe that the answer is yes, in particular since, you know, Artur Ekert in his talk, he reported that the Shor algorithm was just invented, you know, a few months I guess before this thing and this was sort of, he finally had the killer application, you know, that was a very true and deep motivation for pursuing these kind of things. So here’s this information in physics quantum processor, when you look inside, you have now here a set of qubits and all of these things that I will explain now here briefly but sort of why do we want to do these things and again this is from Artur’s talk, it says here technology to beat Moore’s law, yeah, I would still some extent agree with that, computer science, new complexity classes, we still hope for that, physics to learn more about quantum theory, this is always true by definition, so it’s kind of a win-win situation over here and of course what we learned from Artur’s talk that is that the basic element instead of the classical bit that only takes on value 0s and 1s is now the quantum bit which is a superposition of two quantum states 0 and 1, so we can have things like 0 plus 1 over here represented here on the Bloch sphere and this classical versus the quantum bit that can be in the superposition, I’m not sure exactly where I got this picture here from, I guess from Rainer Blatt, if you look at it, you know, there’s an old woman and there’s a young girl at the same time sort of symbolizing this thing of a qubit as being in a superposition state, don’t take this thing too literally of course, and of course we have quantum registers and this is where entanglement comes in that we can store of course in the sense of superposition states in our quantum register superposition of all these different numbers at the same time this is nothing else than entanglement and of course what we are doing with our quantum computer sort of in the laboratory trying to breed little Schrödinger cats that are doing these computations for us, so entanglement superpositions of them avoiding decoherence as they’re coupling to the environment is sort of essentially all of these things and of course the statement well quantum memory how big is it that if you take something like two say 300 qubits you know 2 to the 300 this is the dimension of the Hilbert space the statement is that the Hilbert space is really huge and this is of course the point where people like Feynman came in originally in his seminal presentation 86 you know making the statement well if you’re a quantum mechanical person and you would like to simulate something on a quantum computer Hilbert space is pretty big and it’s hard to do so why not turn things around and build a quantum device that can actually you know simulate quantum devices as a programmable system this was sort of the second leg of which all of these things built and this is sort of the starting point of what we today call quantum simulators so but one of the things that Artur mentioned in his talk was of course that the big challenge is how do we actually build a quantum computer because at that point nobody knew and what are the challenges to do so well if you want to build a quantum processor first of all you need a set of qubits you know that can be in a superposition state and tangle state and you need a way of manipulating these things that if you look inside these quantum processor that you can see that there’s sort of if you read these things like a kind of a world line there will be gates that operate on these things so there’s two kind of gates you know the single qubit gates that make certain rotations of your qubit on the Bloch sphere and if you combine these things with two qubit gates like a CNOT gate that’s indicated over here if you control qubit and a target qubit and there’s some unitary operation that conditional to the state of the first one you operate on the second one if you can build these kind of gates the second one being the entangling gates that you can just put all of these things together and in principle you have your functioning quantum computer provided you have a way at the end to read these things out so the historical challenge was to come up in different implementations you know with ways of implementing these kind of gates and what we talk about is of course a network model and of course what’s behind all of these things is that these gates operate on quantum mechanical superposition states what we now call quantum parallel processing sort of this was the envisioning part in this original idea behind the quantum information processing of course you know much of the motivation then was given by the Shor algorithm and there’s also the Deutsch-Jozsa algorithm as an example where we have a certain quantum advantage and it’s always based on the fact that what the quantum computer does is it’s like a big interferometer but in this you know big Hilbert space where we take different computational paths but we can let them interfere and something like the Shor algorithm for example just really uses this thing in order to get the corresponding advantage out so with this I’m sort of at the at the end of this introduction and what sort of you know we walked out and Ignacio Cirac over here and I was sitting in his talk by Artur Ekert and we looked at this and said I guess we know how to actually build something like this because we were just working on trapped ions you know that were developed for atomic clocks and we had some ideas how to build these gates and let me just give you the qualitative insight and then I would like to show you where we are right now in the laboratory in our collaboration with experimentalists what was the idea so you know people were building for atomic clocks ion traps and this is a single atom that’s a single atom over here these are ions so they repel each other and you can put them in like a very an isotropic harmonic oscillator so that they essentially go to an equilibrium situation if you cool them down to very low temperatures okay so we cool away all of the phonons this can be done by laser cooling and you can see this as some sort of being representative as the quantum register see that it spins up and spin down and of course there will be a superposition of all of this and this is the qubit you know the quantum register that you would like to manipulate and how do we manipulate it well if you would like to do single qubit rotations on the Bloch sphere you can do these things by just shining lasers on one of them and the distance between these ions is something like a few micrometers so you can actually do this in an experiment but there’s also phonons and you can try to quantum control the motion of these phonons and this is part of the building blocks then sort of you know of building the corresponding set of gates that constitute the quantum computer so on the left hand side we have this quantum logic network model all of these gates and we can sort of you know all of these elements we can build with hardware which is here on the right hand side and there’s many groups out there that are pursuing these things there are we at the moment in the lab so if you go to experimental papers from recently you know these experimentalists have developed a sort of a complete gate set over here on very simple lasers that are shining all these things but just based on this original idea and what’s kind of interesting is that we have now like a little quantum compiler so if you have some unitary some quantum algorithm that you have you can decompose it you know in the different bits and pieces and we have quantum compilers that tell us you know what sequence of laser pulses you have to shine on these ions to represent a certain quantum computation. Of course the Shor algorithm was one of the killer applications initially and it still is but I would say the point of course is that the number of qubits and the number of operations and here we talk not about doing error correction at the moment, it’s just horrendous so we have to wait a little bit except we want to factorize 21 or so where some people might even know the answer. So I would say that a lot of interest has focused in recent years on quantum simulation and this is like taking a many body system that we have out there, quantum many body system like a spin system and you would like to for example calculate on a universal quantum computer here the time evolution of a many body system that’s represented as a set of spins and let me sort of point out this is what we call digital quantum simulation that would also be analog quantum simulation at the end.

Peter Zoller

00:30:56 - 00:33:48

Think of this like having a many body system which is available some spin up and spin down and you specify a certain Hamiltonian and by shining laser pulses you can engineer a function of time, it uses certain time evolution which in a stroboscopic sense mimics a certain Hamiltonian over here and to illustrate to you what I mean by that for example if you take a simple Hamiltonian of two spins like an Ising model here and over the transverse field these operators will not commute here. We can always decompose the unitary time evolution in little steps and we can factorize them by decomposing these two parts that we have over here in these Hamiltonians and we can write them as gates so if you know how to do gates, entangling gates, single qubit gates you can mimic the time evolution of a quantum many body system and a few years ago we had these were the kind of things that experimentalists were playing with where you write down a Hamiltonian like a familiar say an Ising Hamiltonian over here and this comparison of theory and experiment was only up to about six spins and in the meantime as I said before they can do it up to about 20 or maybe even 50 of these spins but you can also write down very exotic Hamiltonians like the one over here which is a six-body interaction and you can just program these things on your device to mimic the corresponding time evolutions. So this was a few years ago more recent work and I’m not going to explain all of the details here is this that we can take models that become sort of interesting like you know a Schwinger model of one plus one dimensional quantum electrodynamics of course the hope at the end is to do something more non-trivial like maybe a non-Abelian lattice gauge theory in two dimensions but this is I can tell you a little bit in the future and we can sort of take a model like this and we can map it to a certain number of qubits over here and you know this nature paper here and the people who should get the credit is experimentalists here and Christina Moshik is the theorist. You can sort of mimic these things and here is a comparison of theory and experiment to be honest it’s only a few trotter steps but actually you should see these things as the starting point of development where these devices you know are at the point where at one point we might be able to solve some difficult problems that classical computers might not be able to solve. What’s really impressive in this experiment was the fact that there was only four qubits but 220 quantum gates were possible so these gates became so accurate in the meantime that you can do 220 gates before the whole system falls apart and that’s quite an amazing development given that you could write your nature paper you know 15 years ago if you are able to do a single gate now we talk about 220 so things are becoming sort of a certain maturity in this context.

Peter Zoller

00:33:48 - 00:40:47

There’s also another way of doing quantum simulation and I just want to mention this because I will afterwards refer to this which is what we call analog quantum simulation. So far I’ve talked about quantum computing about gates you know single- and two-qubit gates and all of these things. You can also try to sort of you know imitate quantum many-body dynamics by simply designing your system in such a way that you mimic directly the corresponding Hamiltonian not this you know this digitized version that we had here before and you can also do that with ions and the way that you do it is that you can sort of shine in lasers and these lasers will distort these ion crystals and if you eliminate this thing you get effective spin-spin interactions out from this whole thing allowing you to realize a model of this type and here’s an example of you know if I take all spin-downs I take a model like an XY model sigma plus sigma minus if I flip one spin over here you can see now that the spin excitation moves here to the left and moves also here to the right and you can try to see the entanglement because if I give you one spin over here which is at the same time moving to the right and to the left it means that if the spin-up is on the right then there’s a spin-down on the left and vice versa so you got sort of EPR type entanglement over here that’s exactly what these experiments sort of are able to demonstrate so we have here possibilities to build all of these very exotic spin models for example and of course that’s here in the case of 1D but we can also do more generally so let me sort of try to summarize you know what after all of these years the experimental situation is on the atomic physics side on the atomic physics side we have now systems available the ions is just one example where we can essentially have complete control either in the sense of building little Hubbard models or in building for example with Rydberg atoms Rydberg arrays over here and what these Rydberg arrays do is this that they allow one sort of to really one by one build quantum systems and engineer their corresponding interactions so we have the whole toolbox available in the laboratory and whenever you have a favorite you know Hamiltonian that you would like to implement there’s a good chance that the experimentalists at the moment you know will be able to do that and of course here the ion traps is one example what’s really interesting is that in these atomic systems that we have you have complete control by being able to talk to the individual qubits separately so we have single-site control this is control on the level of single quanta of your system you cannot get more quantum control than something like that and of course it comes with the fact that if you take a new tools like the quantum gas microscope that these new tools allow you also then to do readouts on the level of single atoms or on the level of single spins so I would say this is sort of achieving in the quantum many-body system complete control over these systems and of course we in Innsbruck are quite proud because many of these ideas like these Hubbard models the ions and also Rydberg’s and so on built on theoretical ideas that we developed in Innsbruck some time ago in collaboration of course with many people and this sort of you know now takes me to the little bit new part of my talk where you might ask yourself what do we want to do now what do we what can we do now based on this experimental progress and I want to summarize here by simply saying that well experimentalists have single-site control in these many-body systems and what’s also coming in is this that they could do these experiments with very high repetition rates and then of course you as a theorist start to think about these things what should we do with that you know we have new tools available what should we do and I will give you now one example that we are working on at the moment again in collaboration with the experimentalists that we have new theoretical ideas and at the end this leads to a whole new generation of experimental realizations and I will talk in particular about measuring entanglement as Renyi entropies and of course I have to explain to you in detail what it means but before I do that I want to sort of you know highlight one very recent development that I find is very exciting related to Rydberg atoms you know these systems I’ve talked about ions have mentioned the Hubbard models a little bit and this is sort of a new toy in this context and it’s a very beautiful example of kind of an bottom-up you know engineering that one does of really being able to control single systems in qubits I guess all of you know what the tweezers is you know from biology where you focus a laser and they able to trap a particle but you can do the same thing also with atoms so you can trap atoms you can cool laser cool them down and if you do any of them which is very easy to do you know some of them will have an atom some of them not cold what you do now is this that you can actually remove entropy by hand by simply finding out which ones are occupied or not by simply seeing the fluorescence so this allows you in the way sort of to you know build a quantum system kind of bottom up and I want to show you examples the first line you know sort of this is filled this is filled this empty you can simply rearrange all of these things here together and if you for example you know pay the experimentalists a beer you can even you know have your name written or so this context and you can do these things also in 2d so these are really sort of bottom-up design you know you want an atom there you know with this kind of interaction you can do these things now in the lab okay what’s really promising is the fact that the next generation will have you know about 1000 of these atoms in 1d 2d and 3d so we’re getting at the point where we can maybe really make these things useful you know in a completely new context here’s an example so if I take now the system over here that can excite atoms to Rydberg states and I will not try to play the role of the atomic physicist then you can essentially design models that you want that build you know on basic interactions that you can engineer this atomic system so the bottom line of the story that sort of take home message is that from an experimental point of view, these systems have been developed down to the point where you can engineer quantum many-body systems. And these quantum many-body systems can be controlled on the level of talking to these atoms or spins or qubits individually, turning on interactions that we can really sort of design toolbox here of these quantum, of general quantum many-body systems. So given all of that, and I mean, this thing comes back now to theorists and you have to provide now ideas and sort of come up, can we answer now questions based on these things that are maybe interesting? And I would like to present now a few slides on ideas that we are working on at the moment.

Peter Zoller

00:40:47 - 00:43:42

It started some time ago as purely theoretical ideas, but I will show you at the end of my talk now that we convinced the experimentalists, go to the lab and try it, they tried it and it works. So I will tell you now a story that sort of, you know, motivated one hand by new experimental possibilities, at the end then also leads to experimental realization and all of these things happen just within one year. And I would like to ask now a question which is very important. You know, whenever we talk about quantum simulation and all of these things, we talk about entanglement in these systems. And the question is this, can we actually measure as a protocol entanglement in systems like for example, you know, if I give you a quantum many-body system and it’s in a pure state, then of course, you know, the von Neumann entropy will have entropy zero, but if I partition it, the partitioning and the entropy will then be a mixed state and this mixed state will have an entropy which is not equal to zero. So these are the kind of things that we are asked. And so if you have quantum many-body system that we subdivide in A and B and we would like to measure, for example, here, define the reduced density matrix but we trace it over the second part, defining this row A here. For example, it’s the ground state of a certain system. And then we have here, you know, an entropy, this either is a von Neumann entropy. We’ll be interested actually in the following, you know, in these Renyi entropies that are power of row to here to the power N. And this, if you are a quantum many-body person, then you will find this interesting because we might characterize here certain topological or strongly correlated quantum phases. So you would like to build new tools, you know, that are available in these experiments that answer interesting quantum many-body questions that so far could not be answered. So the question is, can we actually measure things like these things over here? Well, some time ago, we were interested in this story and as you will see, out there, we’ll again now get credit for some of the earliest ideas in this context but we thought much more in a quantum computing context. Can we measure Renyi entropies with copies and the idea is this, that if I have here one copy of the system and then the second one, so I have a tensor product over here, but there’s protocols that allow one to extract the trace of row A squared. And we wrote some theory papers over here, which then Markus Greiner picked up and we found certain protocols where actually these things could be implemented. And here’s an example where you have two copies, you know, one over here and one over here. These are Hubbard models that are being implemented and I would like to tell you now a little bit what the underlying ideas are but then replace this protocol by something which is now maybe much easier and for the experimental, it’s also much more friendly to implement so but I need this thing as a warm up here. So the story that Artur Ekert told us, you know, for quite some time ago was the following one.

Peter Zoller

00:43:42 - 00:46:29

Well, imagine that you’re interested in trace of row to the power n but you have the ability to make n copies of a quantum system like this thing over here. How do you get from this thing here, this trace of row to the power n? And the answer was that, well, if you introduce an operator Bn, and this is essentially a swap operator, then it simply reorders all of these indices that you have for V in such a way that the trace of row to the power n, you know, is acting with the swap operator or this tensor product of these density matrices here, just the expectation value of this quantity over here. So if you can measure this expectation value and sort of implement this swap operator, you have here the trace of row to the power n, which are these Renyi entropy that one is interested in. Here’s sort of a simple example that illustrates what we mean by that. If I take two copies here, below that I introduce a swap operator which simply interchanges these two indices as you can see over here. Well, if you go through the mark down here, trace of, you know, the swap operator row one, tensor row two, converts these things at the end to row one, row two. And that’s exactly what we would like to do in our systems. Well, it turns out that what Artur had in mind originally was the whole quantum circuit. And actually these things would be pretty tough to build and you need a real quantum computer for doing these things. Now we came up with ideas how you can actually avoid all of these complications, at least in the case of Hubbard models. And this led to, without telling you the details, so these papers here to these protocols that underlie these experiments. And these are seminal papers by Markus Greiner which by building two copies over here allowed one to measure now Renyi entropies. And in principle this exists, but let me now give you a version which is actually very cute physically. But at the same time, I think it is also something that’s experimentally much simpler. And this is, I would like to measure Renyi entropies via random measurements. And it works for a single system. So these were theoretical ideas that are not even one year old. Sort of published in these papers over here and I want to give particular credit to Andreas Elben and Benoit Vermersch for all of that. And you can see my own collaboration with Ignacio Cirac sort of reviving over here. There’s even one of the resident ICTP physicists here, Marcello Dalmonte, that participated in this work. So I find these ideas behind this actually very cute. And as I said, also useful. So let me tell you what the underlying ideas are. What we would like to do is sort of play a replica trick. If you have two copies of a system, then we are able to measure Renyi entropies. So let’s try to make a virtual copy.

Peter Zoller

00:46:29 - 00:50:15

You will ask me what do you mean by virtual copy? And I’ll say that the virtual copy is exactly the replica tricks that we know so well from the work of Parisi, for example. That’s usually more mathematical trick, but out of this mathematical trick we make something over here which at the end becomes a real measurement protocol. So how does it work? Well, I go back to an old paper by Stephen Fanenck and he pointed out the following thing. Suppose that you got a spin system over here and you’re interested in a certain subsystem that we call A over here. And let’s apply now over here some UA, which is some random unitary operations on the spins that we have here of interest. If you make measurements after that, then you can see that the probabilities that we find over here are just given by this formula, you know, rho, then you scramble it up by these random unitaries, and then you do your measurement, this gives you the probability. But now let’s do the following. Suppose that I average now this thing over all possible unitaries, then of course the answer for these probabilities will be very boring because all of them will be, if I’ve done very good scrambling, all of them will be the same, but just equal to a constant. But what’s very interesting is that if you look at the square of this probability and then you average, then you find that indeed here is your Renyi entropy appearing. So there’s a way of measuring Renyi entropies by looking fluctuations in the systems with random measurements, okay? And if you ask yourself why is it the case, well, let me just write down the p squared over here and I write it now, just taking the formula over here twice, but I write it under one trace. And now you can see it’s like by having this square over here, we’re sort of making two virtual copies of the system here. And if our unitary operator that we have over here belongs to a circular unitary ensemble, we get like if you have Gaussians, for example, cross correlations. This U will be correlated with that here, but there will also be cross correlations between these two different virtual copies. And this is the one that makes the magic of at the end, allowing us sort of here to extract Renyi entropies from that. In quantum information, this is called two design or T design in a more general context. And so you can see that if in a system like that, we’re able to perform random measurements, there will be possibilities for extracting interesting stuff like, for example, Renyi entropies as assigned as an indicator, quantifier of entanglement measurements. The question is, of course, how to realize and how many measurements and unitaries we need. This is getting a more technical discussion and I’ve just had a few slides that should sort of indicate you and then I’ll show you some experimental results for that. So the measurement protocol, so that you know, at the end, it will be something like this where we have time and the next experimentalist has done an interesting quantum dynamics over here, preparing a certain interesting quantum state, maybe on the quantum computer or the quantum simulator. And then we would like to ask, can we measure trace row to this power n of a subsystem over here? And the way that they do it is this, the random unitaries, we simply build up by using the fact that I pointed out before, we have available in the laboratory, quantum many-body systems, that we have single-site addressing. So I can go in, for example, and make a random disorder, you know, and it becomes a programmable random disorder that we can use. we can add to our systems. And what we have at the very end is this, that if I do sort of a series of mini-quenches with different disorder systems over here, the question is, how does this thing here converge to a circular unitary ensemble?

Peter Zoller

00:50:15 - 00:53:11

How many of these disorder patterns do we need that we have to program in an experiment like that? And these answers answered over here by having, for example, an Ising model with a certain number of lattice sites and a certain disorder. All of these things are experimentally available, like in the Rydberg system. So you can see that if I do about 10 disorder patterns, then these things have converged down here, essentially, to the value that we want. So it takes about, say in this case here, 10 disorder realizations in order to converge down to the fact that we have unitaries that approximate value, at least, on the second order correlation function, the circular unitary ensemble. All of these things are available in the lab. And if you ask yourself how fast does it converge, then the answer is simply that the number of necessary quenches also scales with the system size, which is your subsystem size that you want. And this works in 1D and in 2D. These are sort of theoretical checks that we do over here. So in that sense, we have an efficient generation of random unitaries for purity measurements. So all of these things that we do here essentially is that in our controlled quantum many-body system, we can sort of do chaos by design. And by this chaos by design, construct a measurement protocol out of these things so that P to the power N gives us this Renyi entropy, so the Nth power, and is based on the fact that we have these correlation functions over here that we imitate with our random disorder that should approximate as well as possible the circular unitary ensembles over here. Well, these results are sort of consistent with random gates that people are trying to do. For example, we’re superconducting circuits at the moment, but in our case, I would say we get these things essentially here for free in our systems. So the measurement protocol then sort of works like this that we have here, a sequence of random unitaries that we have to implement, and then we do our standard measurement, and here the quantum gas microscope comes in. All of these tools are available in the laboratory, and we have to repeat these things. The question is, I mean, how often do we have to repeat all of that? And the answer is, well, there’s a certain scaling over here, and I will not enter the details behind it. What it simply means is that in these experiments, if you do about 100 measurements, and if you do about 100 unitaries, then this is essentially for sufficient, for the system sizes that we can do in our context. What’s this thing now enabling? Well, maybe you would like to see something like an area law. All of you, I guess, know that if you have a quantum system or that they can set the subsystem out of over here, in a system obeying the area law, it simply means that the entropy will scale proportional to the circumference of this area that you have over here. And indeed, if you make the system larger, love, this is sort of what you see ideally.

Peter Zoller

00:53:11 - 00:56:08

If you apply our protocol in this context, then you can see that with a sufficient number of these disorders, we are right on top. So this would be a way of experimentally measuring area law, it’s never done before, and I think this is now sort of constructing tools that will allow us to do so. Or if you’re interested in, say, many body localization, and you would like to see that you have entropy growth, entropy growth, which is logarithmic in these systems, and you would like to see that, again, we can use our protocol and we applied it, and you can see this is sort of the simulated data points, and this would be a simulated measurement over here. So we see the possibilities that these new tools that we have available will allow us to see completely new physics in this context. You know, a few weeks ago, I got from Christian Rose in the Rainer Blatt Group, you know, this email over here, they have implemented these ideas, and what you can see here is results for Renyi entropy in the system, first of all, for 10 ions, and this is supposed to be a product state initially, so down here, they are doing some quench dynamics. The total system should be pure, so this thing should go down to zero on the right-hand side over here. This is exactly what you expect for these entanglement entropies to be in these quench dynamics that we have in quenches, and they can even do it for 20 ions, sort of as a result over here, and I would say that we have the tools now available if you can do these things for larger system towards testing what we call this quantum supremacy, and the nice thing is that these protocols that we have here, of course, are available for many systems. I had a few more slides on quantum networks and quantum computers, you know, where we do networking and protocols and all of that, but I’ll do, I saw we’ll give a talk afterwards that more on quantum communication, so we’ll basically skip these things, you know, instead of satellite links, we would like to talk about, say, networking of quantum computers with these things here, so this would be an extra talk, but I guess you sort of got the idea here of the snapshot of these developments triggered 25 years ago that led now to experimental programs that are now at the point, I guess, where we get new and interesting physics out, and at the same time, you know, we are sort of opening the door also to quantum technologies, and let me conclude by congratulating the medalists here again for their pioneering work, you know, it started all by the ideas of these gentlemen, and I just want to make as a last remark here, the sentence, very often these days, we hear essentially about quantum technologies. I think we should not forget there’s a synergy, and there’s an intimate connection between basic science and quantum technologies.

Peter Zoller

00:56:08 - 00:56:58

If somebody tells you that building a quantum computer is just now an engineering task and nothing more, these people are wrong. I would say that a lot of the basic science questions, both on the hardware side, but also on the conceptual side, are still open, and keeping these synergies in mind, I think is something which is fundamental, and when you hear discussions about the flagship, you know, the flagship for quantum technology, my personal wish would have been that it was a flagship for quantum science and quantum technologies, and this is sort of the last remark, because we can see that there’s a long history now behind it, and the time the engineers take these things over has not come yet. Okay, so congratulations again to the Dirac Medalists Prize winners this year, and with this I would like to conclude my talk.

Dirac Medal ceremony audio artifact

00:56:58 - 00:57:00

[Applause]

Fernando Quevedo

00:57:07 - 00:57:30

Thank you very much, Peter. It’s a great presentation. Any question? We have very short time, but any urgent question will be welcome. Yes, a question there. Thank you.

Audience questioner

00:57:40 - 00:58:00

Thank you very much, really. It’s a perfect talk. Just about the notion of entanglement, can we find another correlation type than entanglement? I was reading something about the Discord, but I was reading that people are still sceptical about the Discord, so what can you say about this point? Okay, so you would like to measure Discord over here.

Peter Zoller

00:58:00 - 00:58:28

Well, people have done these things of measuring Discord, but I would say that what we were interested in was the, I would say, real thing of really this entanglement, entropies, because this is the part that’s sort of interesting from the many-body point of view if you are a condensed matter person. So this goes in a different direction, and I think that we are sort of measuring from a condensed matter point of view the real thing. It’s like Coca-Cola and Pepsi.

Fernando Quevedo

00:58:31 - 00:58:50

Yeah, thank you. So let’s thank Peter again. Now we would like to call Artur Ekert. Please join me to welcome Artur for his presentation.

Dirac Medal ceremony audio artifact

00:58:53 - 00:58:55

[Applause]

Artur Ekert

00:59:17 - 01:03:00

How does it work? This goes… Oh, it’s this one. Okay. Right, so thank you very much for inviting me here, and I should say Peter Zoller gave me probably too much credit than I deserve, because, you know, when it comes to this particular field, at some point, it was absolutely essential to be taken seriously. You’re only taken seriously when you convince experimentalists to do something in this particular area, and I thought that people like Peter Zoller and Ignacio Cirac were ideal people who… No, those are theorists who are working with experimentalists… Experimentalists are trusting that kind of people. They wouldn’t trust those wacky individuals working in this quantum information science at the very beginning. So I thought, you know, all kudos goes to Peter and Ignacio for spreading the word and this kind of contaminating experimentalists who ventured into this field. Anyway, but I should perhaps start by congratulating the three people who got the Dirac prize, so Charlie and David and Peter. And I have to say that, you know, when I was asked to give this talk, I really didn’t know how to structure this talk, because, you know, on one hand I wanted to say something about their work, at the same time, you know, there’s such a vast area of papers that were produced by them. And that would, you know, it was very difficult to find a theme that would somehow incorporate everything that they did. And also, you know, I was kind of biased because David was effectively my supervisor back in Oxford, and he is the person who in many ways changed my life. So then, of course, you know, through David I met Charlie Bennett and then I had also a pleasure to meet Peter. And the two of, the three of them, in fact, and the colleagues created a, not only they just made quantum information science a respectable thing and deep and profound subject, but also they created a very peculiar atmosphere in this field. So every sort of, everyone felt invited and was helped. And somehow this area of quantum information science is still the area where people are, you know, very friendly, very willing to work together and collaborate. And that’s sort of a, it started from the very beginning. So, you know, working with David, of course, was experience and I would have too many anecdotes to tell, but I’m not going to tell them. Don’t worry, David. But it’s not only me, you know, but also whenever I got a PhD student, sooner or later they sort of moved in the direction of David and talked to David about science, you know, universe, everything. In fact, you know, my first conversation with David had nothing to do with quantum physics, but it was all about Karl Popper. And so I just realized that I found a fellow Popperian in Oxford and I thought, yes, I like this guy, you know, I’m going to work with him. But, you know, all my students, almost all my students ended up working with David. So Adriano Barenco worked on universality issues. Then Patrick Hayden worked on reformulating Bell theorem and David Wallace worked on interpretations of probabilities.

Artur Ekert

01:03:00 - 01:09:34

Now Chiara Marletta is working on constructor theory. So in sort of in a way, it’s kind of very easy for me to supervise students in Oxford. Sooner or later I said, why don’t you just go and talk to David, you know. And, you know, it was already mentioned that this field exploded over the last few years. You know, at the very beginning it was pretty much like a family business. There were, you know, when you look at this picture that was, I don’t know which year, it was taken in 1993 or so in a small place, Broadway in England. It was just, you know, pretty much everyone who was working this field at the time. Well, Charlie may actually give you a little bit more history predating this particular meeting. But as far as I’m concerned, you know, at that time there were not so many people who were interested in the quantum aspects of computation. And then it exploded. And many of those sort of meetings, early meetings in quantum information science where David and Charlie and Peter attended was in Villa Gualino. Why Villa Gualino, Peter asked? There was a good reason. So what happened was that a person called Giuseppe Castagnoli, who was one of the directors of LSAC that is based in Torino, had a friend who was Mario Rosetti who was responsible for running an institute and Giuseppe Castagnoli who ventured into business. But he had some ideas about quantum computing. He wanted to somehow, you know, sponsor something. So LSAC, instead of giving money for yet another art exhibition in general, decided to sponsor a series of workshops in quantum information science. And Mario Rosetti somehow took care of it from the logistic point of view. So I think indirectly those people shaped this field. And as you can see, those pictures were mostly I think taken by Charlie who at some point cleverly used, I don’t know whether it was Photoshop or whatever you used, but you know, various people came at different times and Charles wanted to have them all in the picture. So every now and then you can find this artificial head popping up. So that was Charles. Charlie wanted really to have everyone included in a true sort of all embracing spirit of the field. So which, by the way, I remember that, you know, once I was invited to give a lecture to a popular lecture in University of Belfast. So I went to Northern Ireland and I just gave a talk and I wanted to encourage everyone to join this field, no matter what sort of directions you are coming from. And, you know, I never use the word Catholic in terms of word embracing, but this sort of subconsciously I said, you know, come to this field. This is a very Catholic field. Imagine this. This talk is still remembered in the Queen’s University. Anyway, so when I thought what should I really be talking about, I started permuting my transparencies and I think I symmetrized my talk so much that this morning when I sent those slides to the organizers of this meeting, I actually, I don’t know whether I’ll have any real control over this talk. So I’ll just venture in the direction, in one aspect of quantum computation or quantum information processing has to do with data security, which is something that I had lots of personal interest in. As it happens for some reason or the other, the development of quantum information science had an impact on quantum, on data security. For one thing, Peter’s algorithm, as you know, affects the security of public key crypto systems, such as RSA, but not only. For example, if you look at the cryptocurrency, say, bitcoins at the moment, the two important components for the whole thing to work are digital signatures, which are based on elliptic curves, which can be broken with Peter’s algorithm. And there’s also the mining process that can be seriously affected if you can do quantum search, for example. So quantum technology, those ideas that sort of, if they were, they have a serious impact on the future of information security. So in this sort of history of the development of cryptography, one can give a separate lecture on how it all started, how it all developed, how people wanted to design perfect ciphers and how most of the time they failed, and go through the public key crypto systems, ending up with sort of a quantum crypto. So I’m not going to go into those details because that’s probably not important, but maybe I’ll just say a few words about how we can take the ideas of quantum correlations and quantum entanglement to the extreme and design a system which comes as close as one can possibly come to perfectly secure communication. And so this, so the fact that on one hand when you have a quantum computer you destroy public key crypto systems, and it’s a big thing now to design possibly new generation of public key crypto systems that will resist attacks from quantum computers. And people, as you probably know, National Security Agency and some other people are looking to replacement of RSA, going in some directions, possibly lattice-based cryptography. But another candidate for that is quantum cryptography, of course. So as Gilles Brassard put it, you know, the quantum taketh away, but also the quantum giveth in the form of quantum crypto. So I’m going to talk about, you know, you can do quantum cryptography in all kinds of ways.

Artur Ekert

01:09:34 - 01:14:47

Originally the idea came from Steven Wiesner, then Charlie Bennett, then Gilles Brassard turned it into quantum key distribution. I had a slightly different approach based on quantum entanglement, which I will take, not because it is my approach, but because it has, it leads in some way into something that I would consider maybe an interesting part of… cryptography today, device-independent cryptography, and also it will take me to some speculations about randomness and probability at the end. So probably most of you know that for any two individuals to communicate in a secure way, it’s probably enough if they share a private randomness. So if the two individuals, we always call them Alice and Bob, have the same random sequence of zeros and ones, and it’s known only to them and not to anyone else, then they can build secure communication very easily at this. And usually in classical world, so to speak, it’s very easy to test that something is random in the sense that it’s uniformly distributed, but it’s almost impossible to make sure that those sequences are really unpredictable so that they are not known to anyone but Alice and Bob. So this you cannot really do in a non-quantum scenario. Once you have those random sequences, you can communicate, for example, using the one-time pad, which is one of the oldest cryptosystems that was proposed. And so one way to make sure that Alice and Bob ended up with the strings of, binary strings of zeros and ones that are not known to anyone else is to use the properties of quantum entanglement and explore something that we call monogamy of entanglement or monogamy of quantum correlation. So it turns out that if you generate pairs of entangled particles, that the stronger they are correlated with each other, then the less they are correlated with anything else. And so you can randomly test and see that the two entities are really strongly correlated, very, very strongly correlated, then by this rule of the monogamy of entanglement, then there’s no correlation with anything else and therefore nobody outside those two entities knows anything about it. So one way to do it is to run the Bell test, the test that was designed to test for the local realism, but not necessarily going in this direction. I just simply use it in a very instrumental way as something that you have correlated particles, you measure a certain figure of merit, call it the Bell quantity, and on this basis you can decide how secure is the key that you generated, how good it is for cryptographic purposes. So that’s kind of an old story. But then at this point you say fine, so you generated the key, you can assess how good it is, but it’s all just the question of implementations. All good cryptosystems, fantastic cryptosystems usually fail because of some lousy implementation. So can you then somehow deal with the fact that experimental implementations may not be perfect? Can you somehow counteract on this? So, whoops. So this is actually not a purely hypothetical questions because there are experimental colleagues who actually exploit the imperfections in implementations of quantum cryptography. And that’s a good work, so they’re kind of quantum hackers. So I just pick up this photograph from Vadim Makarov who is probably the most known quantum hacker. So the guy is basically very clever experimentally is who knows that currently you cannot really implement all those things ideally, and therefore he is using his tool, his famous suitcase that you can see here, and to crack some supposedly secure quantum key distribution methods. But you know, in fact, you can deal with the situations where you have imperfections as long as you can reach the level of implementation of those Bell inequalities which are called sort of the loophole free test.

Artur Ekert

01:14:47 - 01:18:57

If you reach a certain precision level with detections and with setting up this experiment in a certain way, then what is interesting is that the hardware doesn’t matter anymore. It’s just by correlation alone, you can, by measuring the degree of correlations alone, you can say whether something is secure or not. So in other words, there’s no, you know that there are no side channels to this game. We refer to this as a device independent cryptography. Then of course, you know, in order to do this, there are a few assumptions. So one of them is that Alice and Bob, the two people who want to establish this cryptographic key have access to some truly local random number generators. So just to be sure, so I’m talking about a scenario where Alice and Bob can then purchase devices from some kind of a dodgy dealer who comes to them and say, look, I’m selling you those at some discounted price. Those are good quantum devices and you can distribute cryptographic key and you don’t trust this person at all but nonetheless, if you take those devices, you can, without knowing really what they are doing, as long as they generate correlations up to a certain degree, you don’t care what is inside, you just simply say, okay, fine, I can use those correlations to generate cryptographic key. So that’s sort of a beauty of this result. So in order to do this, in order to run this test, of course, you have to also rely that you have a source of truly random numbers locally. And so that would work as long as you can trust those random number generators that you have, that Alice and Bob have to their disposal, otherwise it wouldn’t work very well. So now the most dangerous scenario in this case is that you have this device-independent system, but so you purchase this device, those devices from someone whom you don’t trust, but if by mistake you also purchase a random number generator from that person, so that unfortunately wouldn’t work. So somehow you have to make sure that your local randomness, your local random number generators are either devices that you can trust or those are, or you can do something about it. So for example, you can just purchase a quantum random number generator and plug in, but those kind of random number generators that you can get today are probably not good because even if you get those random number generators, you also would like to, usually you don’t produce them yourself, you would just get them. You also would like to self-test, do some kind of, run some kind of a simple test and see whether those random number generators are genuine, that you can trust them. So the question is, can it be done? So this brings us to sort of like a question that is basically the question that I want to address here is of this time. So given, suppose you want to get a random number generator and you are given, someone brings you a black box and says, well, go ahead and use this random generator for cryptographic purposes. Would you be able to check that this random number generator is doing what it’s supposed to be doing? So what do you request from this random number generator? So you would like the string of zeros and one that is generated, be such that, you know, that that’s a uniform distribution, that the frequency of zeros and one is the same, and the frequency of all pairs and subsets of zeros and ones is the same.

Artur Ekert

01:18:57 - 01:23:36

This you can test. So those are sort of the regular classical well-established tests for randomness. But there’s more. If you want to use it for cryptography, then you also like to test that this is truly unpredictable, that there’s no copy of this device, so that you can easily imagine a situation where someone would just generate two identical random number generators and one would be stored and would generate exactly the same sequence, and it will pass, you know, you look at yours and it will just pass all the statistical tests for randomness, but in fact, it’s not a private randomness. So this is not unpredictable because there is a person somewhere who can actually tell exactly what kind of randomness you are getting there. So today, when you buy a random number generator, and you want to use it, say, for cryptographic purposes, usually it comes with some kind of certificate. Because, you know, you yourself, the end user, you open this box, you may look inside and you don’t understand the physics, there’s lots of electronics and gadgets there and you don’t know what’s going on. So how do you know? You wouldn’t know basically, so you basically ask for a certifying authority to let you know whether this device is good or not. For example, I’m showing you one of quantum random number generators here that is produced by a company called ID Quantique and usually when you get it, you can get a certification from relevant Swiss agencies saying, yeah, you know, we look into this and we can certify that it’s done and produced in such and such way that it generates randomness that is kind of a private randomness to the best of our knowledge, right? The question I’m asking now is it, you know, you may not trust those authorities and if you sort of like a bit contrarian as I am, you may not trust the government, you may not trust the authorities, so you would like to test yourself whether this device does what it’s supposed to be doing. And so the question is can you do it? One thing you may consider is, you know, computer scientists have all kinds of ideas how to amplify randomness, how to take something that is less private, for example, and make it a bit more private, so you can use privacy amplification, for example. But we know that basically even sort of a source of randomness that is not good, most, there is really no classical way of improving a certain class of randomness. For example, in computer science, popular sources of randomness that computer scientists study are called the Santha-Vazirani sources and it’s known that there is basically no way, there’s no classical randomness, there’s no classical processing of this randomness that would allow you to expand it or to make it more private. So it’s a well-established result. However, you know, if you then construct the random, you know, this can be bypassed by using again monogamous correlations. So if you design a quantum random number generator in such a way that you can just, you know, get two outputs and you can measure the correlations between those two outputs and then basically pretty much by the same argument about the monogamy of entanglement or monogamy of certain correlations, you can then sample from one of the outputs and be actually quite confident that as long as there’s a little bit of true randomness in your input, then you can amplify it and get a device that gives you not only uniformly distributed but also completely private sources of randomness. So basically pretty much, it is pretty much the case that even if you get a lousy random number generator that you don’t trust with a little bit of post-processing, you know, using this locally, you can actually amplify this randomness up to your satisfaction. And then, you know, there are many ways of doing this.

Artur Ekert

01:23:36 - 01:27:24

I think I just wanted to say that, you know, you can use all kinds of Bell inequalities not only the most popular CHSH inequality, but, you know, the story basically is at the moment if you take this path to secrecy where you use quantum entanglement, you can show not only that by testing for correlations, for monogamous correlations, you can get devices that you can test for security of the data that you obtain. You can also, you know, test for the sources of the local randomness. So that means that you can push the concept of privacy very much to your own domain. So basically you don’t have to know anything about the underlying physics in this device. All you have to do is just to make some statistical tests and no matter what is the underlying physics there, you will be able at least to make statements about privacy of this data, which is actually quite remarkable. But, you know, so this is actually the story where if you want to push cryptography or the story of privacy to the limits, this is basically where we can, where we are today, at least, you know, in a very speculative way. I mean, I’m not saying that this is actually implemented. We can just about implement loophole free tests of the Bell inequalities. What is interesting though is that in my view, you know, even though I like this narrative and I can develop it into some more coherent and I can give more technical or more consistent review of this field, quite often when I do this I feel that, you know, there’s a little bit of a superficial approach to this and somehow we are cheating at some level because it’s, you know, very nice mathematically. It’s sort of simplified. But if you go to the bottom of it, I don’t think it is as simple as that. There’s lots of, you know, lots of interesting questions or fundamental questions that you can ask. For example, you know, those input output boxes, surely it is not the case that they are just mathematical devices and they are real physical things and if you look at them and they have to be quantum and the question is, you know, how do you operate this and how do you understand the whole notion of secrecy in terms of, say, the Everett multiverse where if you assume, for example, that everything is quantum, then you have to somehow redefine the notion of secrecy in terms of relations between different universes so that, you know, how information in one particular part of the multiverse is restricted by people sort of, you know, the access, how the access is sort of restricted in different parts of the multiverse. So that’s certainly one thing and then, you know, there’s a whole thing about the randomness. The randomness always provoked lots of interesting discussions. Going back to the past, you know, the question was, well, is it really objective things? Do we have truly random phenomena in nature? And so that would be, you know, a point of view taken by, say, the quantum world.

Artur Ekert

01:27:26 - 01:29:52

The Epicureans who would say, yeah, atoms, you know, swerve every now and then so there’s no predetermined thing and, you know, that was sort of perhaps on the other side of the spectrum was the Democriteans who were saying, well, you know, it’s more subjective things. Atoms follow predetermined paths and it’s just what is random is due to the lack of your knowledge maybe. You know, take this discussion where you want but the question is, it is still relevant because if indeed it is the case that everything is quantum, then you know, what is really randomness in this sort of truly quantum universe and does it exist at all? And it may be the case that it doesn’t. In fact, probably I would like to finish with this statement from David. I took it from the New Scientist article. You gave this interview and I think you really said this, did you? Right, so, you know, it’s just a valid question at this point to go in that direction and think how the whole discussion, the historical thing about randomness and probabilities is going to end up. So do we, can we develop, for example, physics where we don’t use the notion of randomness and probabilities as we do it today, which is a very interesting question? And I think I will probably stop at this point because as you can see, it was sort of like a talk where I was trying probably to take this notion of security and the idea of pushing privacy to the limits generates lots of interesting questions and fundamental questions and I think what is great about this field is the fact that somehow more and more often we can address those questions in a rather technical and precise language. So again, I would like to congratulate Charlie, David and Peter for their work. Helping to create this fantastic field and needless to say, this field will certainly thrive for years to come. Thank you very much.

Fernando Quevedo

01:30:00 - 01:30:06

Any questions? Yes?

Charles Bennett

01:30:06 - 01:30:38

Reminiscence, because I’ve taken a lot of these group conference pictures and one of the main reasons is that scientists in general are like herding cats. So you announced there’s a group picture and a lot of the people miss it. And then you want to include them. I don’t know if you were at that one, but we had a conference in Capri and there was a beautiful swimming pool. And a lot of people failed to show up for the group picture, which was around the swimming pool. So I took them later on and then just cut their heads and had them floating around in the water.

Fernando Quevedo

01:30:45 - 01:52:34

Okay, so I’m sure there’s plenty of things to think about after these presentations, but there is coffee outside. We are running very well on time, so there is 15 minutes for coffee and we come back at 4.15. Let’s thank Arthur again. Thank you. Okay, let us continue. Okay, so now we will move to the next part of the event, which is awarding the medals and the presentations of the three awardees. So I will start with Peter Shor. So let me say some words about Peter. He received his B.S. in mathematics in 1981 for undergraduate work at Caltech and was a Putnam Fellow in 1978. He earned his Ph.D. in applied mathematics from MIT in 1985. His thesis was on probabilistic analysis of bin packing algorithms. Peter boosted the field of quantum computation by designing efficient quantum algorithms for factoring large numbers and computing discrete logarithms, each of which can be used to break classical encryption schemes. He thus proved that a quantum computer could solve a useful, hard computational problem exponentially faster than any known classical computer algorithm. Shor also introduced quantum error correcting codes and fault tolerant quantum computation, which are schemes for coping with effects of stray interactions, noise, disturbing qubits. Without robust quantum error correction, large scale quantum computation could be stymied by the extreme sensitivity of quantum states to noise. Instead, the theory of quantum error correction is now a well established branch of quantum information science, and the difficult path to developing large scale quantum computers appears open. Peter will give the presentation and the title will be, I will just read it, the discovery of the factoring algorithm. But before that, I will ask Peter to come here to, I can give you the Dirac Medal and everybody to give him a warm applause. Thank you.

Peter Shor

01:53:02 - 01:57:20

Okay, so I want to talk about my inspiration for discovering the factoring algorithm and some reminiscences about what took place when I discovered it. So the outline of my talk is first I want to give the first few slides of my 1990s factoring talk, somewhat updated, because they’ll show you why the factoring algorithm is such a surprise, and then I want to say a few words about how I actually discovered the factoring algorithm, and then I want to say a few words about what happened after I discovered it. So first, the slide. I opened my question asking what is the difference between a computer and a physics experiment? And of course back then that was like the joke, what’s the difference between an elephant and an egg? It’s so obvious. So first answer, a physics experiment is a big custom built finicky piece of apparatus, and a computer is a little box that fits in your briefcase. So for example, and you can see neither of these existed 20 years ago when I discovered the factoring algorithm. Here is a computer, and here is a physics experiment. But if you go back 50 years before that, here is a computer, oops, and here is a physics experiment, and they start looking very much more alike. And you can even see that the technicians were in the same uniform. You’re interested, this is the Berkeley particle accelerator, and this is ENIAC, the first computer that Von Neumann worked on. So here’s the second answer. A physics, a computer? Okay. A computer answers mathematical questions, and a physics experiment answers physical questions. So for example, if you want to test whether all bodies fall at the same rate, you probably don’t want to use computers. And if you want to test whether 15, if you want to find 15 equals X times Y, you probably don’t want to use physics experiments. So this is an ion trap computer at Rainer Blatt Group in Innsbruck, Austria, where he actually did the experiment of factoring 15 using an ion trap. And I’m always afraid when these papers come out that some headline is going to appear in a newspaper somewhere, physicists spend $2 million to show that 15 equals 5 times 3. It hasn’t happened yet, luckily. And a third answer is that you don’t need to build a new computer for each mathematical question you want answered. And this is really a very, actually it’s a fundamental fact about computation. And what that means is that you can mass produce computers. Well, it’s hard to mass produce physics experiments. So for example, after the Tevatron solved all the particle physics questions it was capable of, people built the LHC, and when the LHC solves, when people discover all the physics they can at the LHC, they’re going to have to come up with a lot of money to build a new one or stop running particle accelerator experiments. So no one would think of building more than one LHC because that would be really quite useless. And then there’s a lot of physics, condensed matter physics that the LHC is completely useless for.

Peter Shor

01:57:20 - 02:01:15

Whereas if you have one big computer, you can pretty much run any mathematical problem you want on it. And this is related to the universality of computation. So back in the 1930s, there were three people, there was Alonzo Church, Alan Turing, and Kleene, and they all had completely different looking definitions of computation. What does it mean for a function to be computable? But it turned out they gave the exact same class of computational functions. And what Church and Turing proposed was that this was really a very natural class of computational functions. But when people started building real computers, this turns out that the definition of computation was very different. Because this turns out that the definition of a function to be computable really wasn’t that useful in practice. For example, if you have a function that can be solved and computed in 10 to the 30th years, well for practical purposes it might as well be uncomputable. So computer scientists made this rather, well from some points of view it’s probably draconian compromise between theory and practice where they came up with the idea that efficient means it can be computable in polynomial time in the length of its input. So, I mean it’s not really, it doesn’t really correspond to the class of practically computable functions, but it’s also something that computer scientists could prove theorems about. So once this was realized, once you have the definition of efficient as being polynomial time, this quantitative Church.s thesis, which was proposed many different times in the 1960s by various computer scientists, but I think Cobham is actually the first, a Turing machine can perform efficiently any computation that any device can perform efficiently. And I don’t know how widely recognized was that this is really a statement about physics rather than about computation, rather than about mathematics. But in fact, if you have different laws of physics, you might be able to compute different things efficiently as David Deutsch was one of the people to point out, I believe, or several other people who pointed out a number of years before he did. So quantum computers can be built. The really surprising thing is that this would imply this folk thesis is not true. And in fact, the folk thesis has become, you know, has really become rooted in the consciousness of the public because one of the questions I get asked about quantum computers was, well, how much faster is a quantum computer than a classical computer? And this isn’t really not an answerable question because quantum computers speed up some problems by exponential amounts and they speed up other computational problems not at all. So this, you know, the fact that this misconception is so widespread really means that the public had absorbed quantitative church’s thesis.

Peter Shor

02:01:15 - 02:05:22

I think we have now gotten to the point where we have convinced the public that quantum computers don’t just speed up everything by one number. But that was, you know, that took a long time of planning to do. So part two, what led up to the discovery. So my first exposure to quantum computing was when I heard a talk by Charlie Bennett at Bell Labs about quantum key distribution. Here is, actually I’m sure that Charlie is going to mention this in his talk, Charlie and John Smolin built, they got us, you know, basically on a tiny budget, a little quantum key distribution device. And this apparently is what it took for them to get physicists to take them seriously. So I, you know, I was very intrigued by Charlie Bennett’s result and I went around and looked at papers about quantum computing and the literature and there really was not very many of them and most of them were written by David Deutsch. So looking at them, well, first neither Charlie nor David Deutsch convinced me that there was a mathematical rigorous description of quantum computing. Looking back at David’s papers in retrospect, I was clearly wrong about that. But, and I also was not convinced at all that it was at all useful. And I’m not going to say that, I mean, I was wrong about that too, but that was, I don’t think David’s papers had any really useful algorithms in them. So for that, next thing that happened is Umesh Vazirani gave a talk at Bell Labs about the paper, Quantum Complexity Theory, he wrote with Ethan Bernstein. And this had two really great inventions in it. First it had a problem, which was a problem that no one would actually really ever want to solve. And, but which quantum computers really sped up the computation of classical computers. And the other thing is it had a rigorous definition of a quantum Turing machine. So after I saw Umesh’s talk, I started thinking seriously about quantum computing and whether it would be possible to speed up some real problems with quantum computers. But I didn’t get anywhere with this until I saw Dan Simon’s paper. So I was on the conference program committee and Dan Simon submitted the paper containing his algorithms to this conference. In fact, it was STOC in 1994, which occurred sometimes in the spring. So I saw it, I was very interested in it, and I’m very embarrassed to say that the conference program committee rejected it. So I was not able to persuade the committee that this was a big enough advance over Umesh Vazirani and Ethan Bernstein’s paper, which had appeared in a previous iteration of this conference. So, I mean, in retrospect, clearly I should have been jumping up and down and yelling at them that this was the biggest mistake you could ever make, but I didn’t know that at the time, and I was, I didn’t go, I didn’t jump up and down and they voted to reject it. So what is Dan Simon’s algorithm?

Peter Shor

02:05:22 - 02:09:13

Well, it takes place on a hypercube. So you have a hypercube, and you color the point. You color the points, vertices of the hypercube with one of two to the n minus one colors, so there are exactly two colors, or two points labeled with each color. And these points have to be periodic, so to get from a green point to the other green point, you say, what you do is you take a vertical, horizontal, and right diagonal edge, and let’s try that with a different point. Vertical, horizontal, right diagonal edge, that gets us to the same color, and here, vertical, horizontal, right diagonal, the same color. So now you have this hypercube with all these colors on it, and what you’re allowed to do is you’re allowed to ask on what color is this point, and you wanna find this path from one point of a color to another point of a color. Now classically, the only thing you can do, keep asking random vertices, or maybe nearly random vertices, you can do a little bit better than random, until you get two vertices of the same color, and then you’re done, because you know what the path looks like. Quantum mechanically, so that takes the number of points on the hypercube, which is two to the D minus one, if this is D dimensional hypercube, and quantum mechanically, you can really solve it in D queries too. You ask D questions of points in superposition, and you get enough information to tell you what the distance is. Well that’s an exponential speedup. Simon’s algorithm really gave me all the hints I needed to discover, well the discrete log algorithm. Discrete logs as periodicity, Simon’s algorithm have periodicity, it’s mod two. Discrete log algorithm uses the Fourier transform, Simon’s algorithm uses the Fourier transform, it’s mod Z two to the N instead of the integers mod Z. Integer is mod, I guess, some number. But I knew the discrete log problem would be solved by using periodicity, Simon’s algorithm used periodicity, and I started thinking about it, and eventually I figured out how to do it. But what happened after that? Well first, how does the factoring algorithm work? You can think of the factoring algorithm as a computational interferometer, maybe a computational diffraction grating. So what a diffraction grating does is it has a lot of lines on it, and when you shine colored light, the angle it reflects off at, or the angle it makes when it goes through the diffraction grating depends on the color, and that’s because at certain angles, all the wavelengths add up, so you get constructive interference, and at all the other angles, the wavelengths don’t add up, so you get destructive interference. So the colors are separated by the angle they make coming out of the diffraction grating. And the quantum Fourier transform really does the same thing for a periodic function.

Peter Shor

02:09:13 - 02:12:58

It separates the different possible periods of the periodic function, so each different period results in a different output of the quantum computer, and then from the output, you can figure out the period, and if you know some basic number theory, which is well known to cryptanalysts, you can turn factoring into a problem of finding a period of a function. Okay, so what happened after the discovery? Well, so the news of this spread amazingly fast. So this was, I gave a talk at Bell Labs about the algorithm for discrete log on a Tuesday in April 1994. The next weekend, Umesh Vazirani, who was a professor at the University of California, called me. I was home in bed with a bad cold. And he said, I hear you can factor on a quantum computer. Tell me how it works. So you can notice that the talk was about the discrete log problem algorithm, and I had not actually solved the factoring algorithm yet. And well, I don’t know if you know the childhood game of telephone, but somehow the result turned to factoring in, you know, people telling each other about it. This was five days later, and in those five days I had managed to solve factoring as well. So I could tell Umesh how to factor. And the news spread remarkably fast. I kept getting, you know, email requests for the paper, and I hadn’t written it yet. So there were lots of different versions of various drafts of the paper spreading around, and people kept asking me questions about outdated drafts, which I had to answer by sending them the latest draft. And in May, which was only a few weeks after I discovered the factoring algorithm, I gave a talk at the Algorithmic Number Theory Symposium in Cornell. In June, Umesh gave a talk at the Santa Fe Institute Conference on Quantum Information. In August, I gave a talk at a conference in NIST, and I guess Artur Ekert gave a talk in Colorado on atomic optics conference. And in October, I gave a talk at Villa Gualino in Torino, and by that time the paper was actually written. I presented it at the FOCS Conference that November, which Dan Simon’s paper also got into that conference, luckily. And it spread. And one interesting thing is that I started describing quantum computers as quantum Turing machines, which was what Bernstein-Vazirani paper talked about. But after I discovered the result, I started talking to physicists. It’s absolutely impossible to explain a quantum Turing machine to a physicist. They can’t understand it because it’s, you know, mathematics. It doesn’t really correspond to any actual experiment. So I started using the quantum circuit model instead, which I think was first described by David Deutsch. And really how the quantum circuit model got to be the accepted model for quantum computation. And let’s see, I’m probably out of time. Is that right?

Peter Shor

02:12:58 - 02:15:41

Am I out of time? Okay. So one objection to the factoring result was if you needed to do 10 to the ninth steps on a quantum computer, each gate had to be accurate to one part in 10 to the ninth. Of course, this is completely out of the question experimentally. And one of the biggest detractors of quantum computation was Rolf Landauer, who worked at the same place that Charlie Bennett and David DiVincenzo and some other people working on quantum computation. And I think Rolf Landauer described the situation there as, well, we have four people working on quantum computation and one person working against quantum computation. There are, you know, so what’s the argument? Well, quantum computers can’t be made fault tolerant. You can’t use redundancy because of the no-cloning theorem, which says if you start with a quantum state, you can’t make another copy of it. You can’t measure to see if there’s an error because the Heisenberg uncertainty principle means that if you measure the quantum computation, you inevitably disturb it. And then of course, if the computation is disturbed, it won’t give you the right answer. So the resolution of this is, though the quantum error correcting codes exist and quantum computers can be made fault tolerant by using them. And how do they work? Well, you arrange the codes so that likely errors are orthogonal to the encoded state. And what that means is you can measure the errors without disturbing the encoded state. And once you’ve measured the errors, you can correct the errors. With this, there are fault tolerant threshold theorems which say you only need gates accurate to maybe one part and 10 to the fourth. This number really depends on the exact quantum fault tolerant techniques you use. And it’s still unresolved as to what you’re doing. What this number should be if you try to make quantum computers fault tolerant without using too much overhead. So this is still very difficult experimentally, but in the last few years, various groups are coming really close to this number, which is very encouraging. And this is my last slide, so thank you.

Fernando Quevedo

02:15:42 - 02:16:41

Thank you very much, Peter. So that was a very nice piece of history and it’s a beautiful way to see how things develop. Any questions? Any comments or questions? Okay. So there’s a question from a YouTube viewer for Peter. So what are the most significant reasons to think that factoring cannot be performed in polynomial time on a classical computer apart from the fact that many people have failed to do so?

Peter Shor

02:16:41 - 02:17:20

Well actually I don’t think there are that many reasons. If you talk to Peter Sarnak, who’s one of the most famous and best number theorists around, he thinks it’s entirely possible there’s a polynomial time algorithm for factoring on a classical computer. So the only real reason we don’t think there is one is that nobody has discovered it yet. And we think that we’re smart enough that if that existed, it would have been discovered, which is of course probably completely wrong.

Fernando Quevedo

02:17:20 - 02:20:33

Thank you very much, Peter. So let’s thank Peter again and congratulations for the performance. Thank you very much. I will now continue with the next awardee, which is Charles Bennett. Charles Bennett is an intellectual leader in quantum information science. Born in 1943 in New York City, he earned a B.S. in chemistry from Brandeis University in 1964 and received his Ph.D. from Harvard in 1970 for molecular dynamics studies, computer simulations of molecular motion. At Harvard he worked for James Watson one year as a teaching assistant about the genetic code. For the next two years he continued his research under Aneesur Rahman at our laboratory. After joining IBM research in 1972, he built on the work of IBM’s Rolf Landauer to show that general purpose computation can be performed by a logically and thermodynamically reversible apparatus. Let me add a small comment here that, I think, the idea of the thermodynamic code is not a good idea. In 1982 he proposed a reinterpretation of Maxwell’s demon, attributing its inability to break the second law to the thermodynamic cost of destroying rather than acquiring information. And with Gilles Brassard, sorry, something is happening, you’re receiving emails. I think it was Gilles Brassard from the University of Montreal, I think I have to mention, Bennett invented quantum cryptography where two distant parties share a secret encryption key with security from eavesdroppers guaranteed by the basic quantum limitations of measurements of incompatible observables. Bennett and Gilles Brassard, they were the first to create quantum cryptography. Their collaborators also introduced quantum teleportation, whereby entanglement and classical signals are used to transfer quantum states. He and coworkers proved that a quantity called the von Neumann entropy is the proper measure of entanglement for pure systems, an early result in the quantification of entanglement, which continues to be an active area of research. So please join me in applause for Charles for his Dirac medal.

Dirac Medal ceremony audio artifact

02:20:33 - 02:21:23

[Applause]

Fernando Quevedo

02:21:24 - 02:21:31

Okay, and then Charles will give us a presentation, Forging the Culture of Quantum Information Science.

Charles Bennett

02:21:53 - 02:25:35

Yes, so I’m very glad to be here. I was here I think in the 1980s with Rolf Landauer, and it’s a place where serious physics has been done for a long time. I’m going to talk about the culture of quantum, really the culture of information science, because when quantum is a very important part of the system, it’s a very important part of the science, because when you, well, like I say, other parts of mathematics, information science was an abstraction from practical experience, but the information revolution that we’re still in the middle of is from these two brilliant, I mean almost brutal abstractions by Turing, the idea of a hardware independent notion of computing, and by Shannon, an even more brutal idea that the theory of communication is best developed by ignoring the meaning of messages. So they did a tremendous service to humanity by making these brutal abstractions, but they were a little bit too brutal. They left out a couple of essentially mathematical properties which they thought were just physical stuff that wasn’t really necessary to think about. And these were the questions of reversibility, which they thought were thermodynamic questions of not really much importance, which was the idea that was left out of Turing’s theory when he thought of it as a theory of computation. These were both, I mean, and all of the 20th century scientists were pretty, they knew about quantum mechanics, they’d been around for a while, and they certainly knew about thermodynamics, had been around for over a century, but they just thought that wasn’t so important. Well, conventionally, the information carriers are what a physicist would call a classical system. Their states are reliably distinguishable, and you can measure it without disturbing them, and then to specify the joint state of two objects, like what’s in my left pocket and what’s in my right pocket, is sufficient and sometimes necessary to describe the states of both, each one separately. But of course, quantum systems don’t behave that way. But for most of the 20th century, this was regarded as kind of a nuisance, because people focused on the uncertainty principle causing quantum systems to behave less reliably than larger systems. And now, as we’ve heard from several of the speakers today, there are positive consequences of quantum mechanics for information processing. Now, the first that I found out about it is by my conversation with Steven Wiesner, who is my college classmate, and he had some ideas that things you could do with information that were not covered by Shannon’s theory. One of them was to combine two messages into a form where if you transmitted that message, nowadays we would say multiplex them together, so that the receiver can receive either one of them, but not both. Now that’s impossible in Shannon’s theory, because you just make a copy and you decrypt it one way and you decrypt it the other way. So the idea that the uncopyability of quantum information was something that could be useful.

Charles Bennett

02:25:35 - 02:29:20

The other one was an even more direct application of that idea, a quantum banknote that cannot be copied. Now I guess I don’t think the euro notes have this, but French and German banknotes used to have fine print explaining how many years in prison you would spend if you would copy the, duplicated the notes. Well anyway, I think this, he did, he wrote a manuscript which didn’t get published until 15 years later about this in 1968 actually. And I think he submitted it to IEEE, but then didn’t follow up on it because he became interested in sort of political activism and not in physics for another decade. But I think this notes that I took in 1970 with him may be the first place where the notion of quantum information theory or the name theory even got mentioned. So then I went around talking to other people including David and we’ve heard a lot of the rest of the history. But of course in the beginning days it’s sort of obvious that these were ideas that were so strange that most people, even the people who were working on them, didn’t take them very seriously. Like Peter just said, oh well I was only working on it part time. So what’s the difference between ordinary information and quantum information? People often ask me this and I sort of tried to say, well if you think of a space of four dimensions then you can explain the notion of an entangled state. But this doesn’t work very well at a dinner party. So I came up with this other metaphor. Quantum information is like the information in a dream. If you try to explain your dream to somebody else you forget the dream and only remember what you said about it. And of course this means you can lie about your dream and not get caught unless you’re trying to lie to your spouse. But unlike dreams there’s a well known and well understood theory of how the quantum information behaves. And that’s what the people in our field have been developing for the last several decades. And it’s really exciting because it’s the right, oh this is a very arrogant statement to say, it’s the right basis for the theory of communication and computation. Well it’s a better basis than what we had before. And so for the theory of communication and computation, what they did, we made an important improvement which may not be important yet in a technological sense but in a conceptual sense it’s really an improvement. Oh, so one of the things that came out of this is that physicists and chemists used to think of quantum mechanics as part of their subject. And when computer scientists and people that I’m not sure exactly what you’d call them like David began thinking about it they realized it was very parallel to the theory of classical computing. Just as all classical information can be reduced to bits, all quantum information can be reduced to qubits and you only have to work on them one and two at a time in order to do any computation. So this idea of a universal, I think that’s David’s idea, a universal quantum computer as an idea that’s as crisp and fruitful as the universal classical computer that Turing showed existed. So here’s an example of something you can do with a quantum computer.

Charles Bennett

02:29:20 - 02:32:57

We can take a, there it is, whoops, I want the laser, yeah. So let’s take a vertical photon as a one and a horizontal photon as a zero and I have them in different colors so that I can keep tracking them. This is a conditional NOT operation or an exclusive or. So the first qubit controls whether the second one is left alone or whether it.s flipped. And when you put the first qubit in the intermediate quantum state you get an intermediate state between both of them being horizontal and both of them being vertical and that’s entangled state that has no analog and classical theory. So you could say it’s a state of sameness of polarization even though neither photon has a polarization of itself. Well that’s an idea that would bother a typical computer scientist of the 1980 very badly and they would say, you know, physicists don’t even prove theorems, so how can I take what you’re saying seriously? But actually in an earlier time in my life, I was in the Haight-Ashbury district of San Francisco in 1967 and there it was easy to find people who thought they were perfectly in tune with you even though they had no opinion about anything. Now the hippies believed that with enough LSD everybody could be perfectly in tune with everybody else, but they were not really especially good at mathematics and now we have a quantitative theory of quantum information. We know that entanglement is monogamous and the more entangled two systems are with each other, the less of course the hippies weren’t very good at monogamy either. So here’s how it works. If we have two perfectly, well we get two separate systems, we go through a very simple quantum operation and we can get an entangled state. And then suppose Bob likes the fact that he’s entangled with Alice and he decides, well let’s have a little bit more of this. So he entangles himself with Judy. Well the trouble with that is that that degrades his entanglement with Alice and his entanglement with Judy. So he’s only classically correlated with each of them. But, and so that means if either of them leaves town, he just has a classical correlation that could be cloned or copied, but it’s not very interesting. But a more interesting thing happens if they both stay in town and that is that he becomes entangled with the now non-trivial relationship between the two of them that he’s brought about, which I would say is an appropriate punishment. And so now we’ve developed the quantum theory of information and information processing. We have to explain what we mean by classical bits. And a classical bit is just a bit with one of two arbitrary orthogonal values. The classical wire is something that conducts classical information reliably but spoils superpositions. In other words, it’s a quantum wire with an eavesdropper. And a classical computer is just a quantum computer that’s handicapped by having eavesdroppers on all its wires. So instead of saying, why does a quantum computer speed up computations, a more sensible way of asking that question is why do some computations get horribly slowed down by having eavesdropping on every step of the way? So entanglement is ubiquitous.

Charles Bennett

02:32:57 - 02:36:11

Why almost every interaction between two systems produces entanglement. Why wasn’t it discovered till the 20th century? Well, because of monogamy. Most systems of nature, other than little ones like photons, interact so strongly with their environments that they tend to become entangled with them almost immediately. And that means that the relation between the parts of the system is degraded to mere classical correlation. It’s a little bit like the life of celebrities where if you read the People magazine, you find out what they had for breakfast. And then you read the next issue of People magazine and you find out what they had for lunch. And so they have no private life because they’re being eavesdropped on continuously. Well, how does entanglement hide itself? This is sort of a quick version of decoherence theory in the version proposed by Wojciech Zurek. Most systems in the kind of world we inhabit are continually eavesdropped on by multiple different eavesdroppers. For example, the photons of light are bouncing off all of us and some of them are going out the window and never coming back. And they certainly don’t interact with each other afterwards. So what happens is, for a typical thing in our world, other than this microscopic thing, the environment eavesdrops on it and creates multiple redundant copies of some properties while obfuscating other properties. Now, this is, I think Peter already talked about this. In classical computation, you can break all computations down into ands and ors and nots, but you may need a lot of them to do a problem like this factoring problem. And if you had a quantum computer, you could do it much faster. And we have now a well-developed theory of quantum computational complexity where we have our earlier theory of classical complexity classes like P and NP, and we’ve got the new quantum ones that sort of interpolate between them in an interesting way that’s still being explored. Now, I’m gonna go back and talk a little bit about the way ideas develop in a way that’s not in a straightforward way. In fact, bad ideas are sometimes extremely good for advancing scientific progress, and good ideas sometimes slow it down. So, and one of the biggest sorts of bad ideas was Einstein. So Einstein really didn’t like quantum mechanics, and because he’s the only 20th century scientist most people can name, they have the attitude that if Einstein didn’t like it and didn’t understand it, what hope was it for me? Well, now we know he was wrong, and I would say, although I’m not really a historian of science at all, I think his mistake was viewing entanglement as action at a distance, as some kind of influence of one particle on another. And the right way to think about it as an entangled state is that you have to give up the common sense idea that if the whole is in a definite state, then each part must be in it.

Charles Bennett

02:36:11 - 02:40:25

It’s just not true. And then once you’ve given up that idea, it’s quite possible to understand how you can have this strong correlation, which doesn’t mean that either particle is influencing the other. Now, but many people continue to follow this bad path, and I think in the early 80s, Nick Herbert published a paper, well actually he submitted it, I forget if it was Asher Peres, may have been the referee, and he said, this paper is so wrong that we must publish it immediately. And it turned out to be the right thing to do, because it stimulated the refutation of the idea that entanglement can be used for long distance communication. So, but there’s, so this gives you the idea that wrong ideas are very good for scientific progress. Conversely, and I’m going to be talking about this in the context of the reversibility in Maxwell’s Demon, good ideas, indeed quantum mechanics itself, sometimes retard scientific progress. So, when we think about the idea between mathematics and physics, between dynamical motion and computation, and this is very nicely stated by Laplace in 1814, if the universe has deterministic laws, if we know the present state, then we know the entire future and past. Well, the next problem came up in connection with Maxwell’s Demon. Who here has heard of Maxwell’s Demon? Okay, well this was an idea of the discover, really, of the mathematics of random motion of atoms in a gas, and he said, if you had somebody who was able to look at the gas molecules, you could get all the hot ones on one side and all the cold ones on the other side, or you could collect them all on one side and you could violate the second law of thermodynamics. And he sort of, as my science writer colleague, Aya Furuta in Japan puts it, Maxwell didn’t solve the problem, he just gave it as a homework assignment to physicists after that. Now, actually, the time was ripe, right at the beginning of the 20th century, for someone to solve this problem, and it was Smoluchowski. And he considered a version of the Maxwell Demon problem, which was just a trap door, so that the molecules coming from one side could push the door open, but if they hit the door from the other side, they couldn’t go through, and eventually all the gas would collect on the right and you could run your pneumatic drill and make holes in the street with it for just using the energy of heat. And then he argued that if the door was light enough and the spring on it was small enough that it could be pushed open by a molecule, it would have its own random motion and it would work in reverse exactly as often as it worked forward. So he really solved the problem back in 1912. But, then quantum mechanics happened, and people realized that measurement was a problematical thing, which they thought was a straightforward thing. And then somehow they got timid in a way that I don’t understand, which is that Laplace already imagined that everything in the universe, including all of our thoughts, are mechanistic. And now, but by the mid-20th century after quantum mechanics had discovered, people started worrying if an intelligent being could somehow do something that a trap door couldn’t do. So, Szilard’s paper in 1929 was, so, it was actually mathematically and physically correct, but it got. gave people a lot of wrong ideas because of its title and because it didn’t quite clearly stay in words as well as in the equations why the demon doesn’t work. And finally it was felt to Rolf Landauer to say that computation could be reversible and it was an irreversible act of erasing information that keeps the demon from working. So here’s my sermon, the sermon part of this history thing.

Charles Bennett

02:40:25 - 02:42:37

Basic science in the future, haste makes waste. Scientific progress I think is mostly incremental rather than breakthroughs. There was a guy who came from another part of IBM and said he wanted to work with our group so he could make breakthrough discoveries. And I was too kind to say, but I said to my colleagues, we didn’t take him, I said to John Smolin, I said that’s like making a firm decision to be spontaneous. So here we are in a situation where for years people didn’t take the subject seriously and now maybe they’re too optimistic about what can be done right away. And my favorite example of this was about 20 years ago, I met a scientist at Jet Propulsion Laboratory and he was saying the most proud accomplishment in his life was on the Voyager Project and they had applied to make it go to all of outer planets. But the word came back from Washington, most people don’t know anything besides Jupiter and Saturn, just go to Jupiter and Saturn. And they said, but you know the planets won’t be lined up in the right way for another 200 years and the word came back from Washington. Congress understands about two years, not about 200 years, just do Jupiter and Saturn. So he says he and all of the other scientists working on it and engineers conspired to make everything last twice as long as it really needed to. And each one said, well you wouldn’t want this thing to fail just because this part wasn’t quite strong enough and he was working on the thermoelectric or power supply for it. So of course then once it was launched they could repurpose it and go to all of those. So the moral of the story is sometimes you have to lie to the politicians, but if you do it in the right way and you don’t do it too often, that may be the best thing for science. Okay, this is my summary of the subject.

Fernando Quevedo

02:42:52 - 02:43:12

Thank you very much Charles. It’s a lot to learn from these expressions and from this way to behave with politicians also. So, comments, questions? Yes?

Audience questioner

02:43:23 - 02:43:49

Yes, so you spoke about entanglement. So, do you think in the quantum computational speed up, is there any clear sign that entanglement plays a role rather than superposition? Well, it’s hard to separate the two because from the superposition principle you get entanglement.

Charles Bennett

02:43:49 - 02:44:40

But it’s an important question because people have asked, does every useful quantum computation where there’s some advantage over classical involve entanglement? And I think actually Peter would know the answer to that better. I think Grover’s algorithm doesn’t involve entangled states, right? Well, it does. Grover’s algorithm involves entangled states, but it doesn’t look like entanglement states are central to Grover’s algorithm. But I don’t think it will work without entanglement. Yeah. Well, one argument you make is if you don’t have entangled states, there’s an efficient way of simulating quantum computation. So, yes?

Audience questioner

02:44:40 - 02:45:12

You happen to be very right. The only problem that I wanted, it’s just a stupid remark. But since all of you have mentioned the no cloning theorem, I would like to call your attention that everybody now knows after Tumulka and other people have spoken, I have derived the no cloning theorem two years before. Two years before Wootters and Zurek and Dieks. Oh, and there was also before that, there was this… Two years before.

Charles Bennett

02:45:12 - 02:45:15

Yeah. It was even more than two years before.

Audience questioner

02:45:16 - 02:45:35

I can send you the next… Yes. No, I was saying it was discovered in a paper that was cited by Wootters that had been written for like 10 years before, but it was not noticed. My proof is exactly identical to Wootters.

Charles Bennett

02:45:35 - 02:45:37

That has the proof in it too. That’s the earlier one.

Audience questioner

02:45:37 - 02:45:39

And it is a very rare one too.

Charles Bennett

02:45:39 - 02:45:57

Yeah, so you got there almost at the right time for it to be noticed. Yeah, so this is almost most scientific discoveries occur this way. They’re discovered three or four times, sometimes very well, and it’s not the part of the discoverer that was not noticed. It’s just the time wasn’t right.

Fernando Quevedo

02:46:01 - 02:49:08

Okay, so let’s thank Charles again for his wonderful talk. So now the last medalist is David Deutsch. David was born in Haifa in Israel in 1953, the son of Oscar and Tikva Deutsch. He attended William Ellis School in Highgate, North London, before reading natural sciences at Clare College in Cambridge and taking part three of the Mathematical Tripos. And he went to Wolfson College Oxford for his doctorate in theoretical physics. And I have to say that his supervisor was Dennis Sciama, who was the well-known figure here in Trieste, and also the supervisor of big scientists like including Stephen Hawking and Martin Rees. And he wrote his thesis on quantum theory in curved spacetime. David is one of the founding fathers of quantum computing. He introduced the notion of a quantum Turing machine that will operate on arbitrary superposition of states, that is on qubits. The concept of the quantum logic gate and quantum circuit, as well as the network model of quantum computation. He showed that all possible operations on a quantum computer could be generated by combining sequences of a single kind of three qubit logic gate. Later Bennett, Shor and co-workers showed that sequences of one qubit gates and one simple type of reversible classical two-bit gates. Working alone and with Richard Jozsa from the University of Cambridge, Deutsch proposed the first quantum algorithms known as the Deutsch and Deutsch-Jozsa algorithms, showing that quantum computation could solve certain problems faster than any known classical computer algorithm. So please, let’s all congratulate David for the award. Thank you. So David will give us a presentation called the mathematician’s misconception. Do you have slides? No.

David Deutsch

02:49:51 - 02:54:19

Hi. Can everyone hear that? Yeah. Okay, well, nice to be here. A couple of years ago, the mathematician Hannah Fry made a TV documentary about Ada Lovelace, the 19th century computer theory pioneer. It was about an episode in the history of ideas which would have been absolutely pivotal if anybody had noticed it at the time, or in other words, if Lovelace hadn’t died young. Because, well, from the evidence in that documentary, I suspect that the first person to get the universality of computation was actually Lovelace and not her colleague Charles Babbage, the designer of the universal computer that she was theorizing about, Babbage’s analytical engine. Never built, but like many of these computers, the significance was in the design and the theory, rather than actual building. The thing is, the analytical engine would have had two kinds of universality, and Babbage was obsessed with one of them. He had perhaps been the first human being to understand what one could call arithmetical universality. In his previous design, the difference engine could compute polynomials in one fixed point variable, so a very limited kind of universality, it’s universal for those. Babbage realized that if he added just a few more features, conceptually very simple, the machine would make the jump to universality, becoming the analytical engine, universal for any arithmetic function of any number of variables of any finite precision, basically what we would today call computable functions. So this was arithmetical universality. What Lovelace understood, I think, was the significance of the analytical engine’s ability to compute not just any arithmetic, but anything in the world, in the physical world. She envisaged all sorts of applications like computer music and art and chess and so on, but this wasn’t just a matter of usefulness. The abilities of the analytical engine as a physical object depend on a momentous property of the laws of physics themselves, all of them, namely, while the analytical engine could instantiate a tiny fraction of all, an infinitesimal fraction of all mathematical objects and relationships, it could also, apparently, instantiate or simulate or emulate all possible motions of all possible physical objects and their laws, not just a tiny subset. This physical universality is an intrinsic property of the laws of physics. It doesn’t follow from Babbage’s arithmetical universality. It has nothing to do with mathematics. In fact, neither of the universalities follows from the other. Yet, it seemed that both of them were exhibited by the same machine. Why? Well, whatever the reason, it’s in the laws of physics. It would make no sense to try to prove this other than from the laws of physics. This unity of the two universalities was also conjectured later explicitly by Alan Turing in the 20th century.

David Deutsch

02:54:19 - 02:58:36

It’s just Turing’s conjecture, sometimes called the Church-Turing thesis. It has various names, but the usual way that this conjecture is described is not that it’s the unity of those two universalities. Why not? Well, Turing’s great paper presenting his conjecture had an application, as he put it, to a fundamental puzzle posed by the mathematician David Hilbert, basically what is the relationship between a true mathematical statement and a provable one. Hilbert had hoped that one could define a system of proof such that a mathematical statement was true if and only if it could be proved under that system. In the 1930s, mathematicians converged from several directions on the realization that that is impossible. Notably, Kurt Gödel proved that there can be no method of proof that identifies all true mathematical propositions. Now, Turing’s approach did exactly the same in that respect, but it had wider implications, as we now know, because of these physical objects, computers. The reason Turing’s approach had this additional reach was that Gödel’s model of proof was a model inside the arithmetic of the integers, so nothing to do with computation. He simply defined proofs as finite sequences of symbols drawn from a finite set and all that stuff. But there was no Gödel’s conjecture. It was Turing who realized that that notion of what proving something means isn’t self-evidence, so he acknowledged it as a substantive conjecture, the Turing conjecture. The model of proof that he used was computation, and the model of computation that he used was physical. Strips of paper divided into squares with symbols and a finite set of discrete operations on them, the universal Turing machine. And when he conjectured that this machine was universal for proofs, the phrase he used was that it could compute anything which would naturally be regarded as computable, naturally. At the time, the word computer meant a human being. It wasn’t one of these things. A person whose job was to manipulate symbols on sheets of paper. And the manipulators, obeying the rules, human beings, are physical systems. So by anything that would naturally be regarded as computable, he meant computable in nature by physical objects. And by provable, he meant provable by physical objects. Now that conjecture, unlike Gödel’s proofs, might have been false. But it turned out to be true in nature, or rather very nearly true. As Richard Feynman remarked, they thought they understood paper, but they didn’t. And when I proved Turing’s conjecture from quantum theory in 1985, it was with the slight correction that the universal machine is not Turing’s paper machine, nor Babbage’s brass gear machine, but the universal quantum computer. But I soon found out that not everyone saw it that way. I also had a referee problem.

David Deutsch

02:58:36 - 03:02:44

The referee of the paper in which I presented that proof insisted that Turing’s phrase would naturally be regarded as computable, referred to mathematical naturalness, mathematical intuition, not nature. And so what I had proved wasn’t Turing’s conjecture, it was about physics. So I asked some mathematicians what mathematical intuition is. It turned out it was as much of a mystery to them as to me. Some of them said it was metamathematical intuition. Fair enough, but they couldn’t tell me what that was either. Some kind of mathematical mysticism, I think. But one thing they were all adamant about nevertheless was that Turing’s conjecture was about whether his mathematical model of proof matched not the physical world, but something else, like mathematical intuition or something. Now, Turing’s basic insight was that proof is computation, and computation is physical, and hence proof is physical. That it isn’t physical seemed to me a philosophical absurdity. It was an absurdity that all the mathematicians I asked insisted on. And most, not all, most non-mathematicians who thought about computation didn’t. So I called it the mathematician’s misconception, the denial that proof is physical is one way of putting it. By the way, the Rolf Landauer, Charles Bennett’s old boss had been campaigning for years with the slogan computation is physical and proof also. Just to be clear, mathematical facts like Fermat’s last theorem aren’t physical. That there is a difference between truth and provability was the main point of all those 1930s discoveries. Still, in my paper I had to defer to prevailing usage. So I changed it to define Turing’s conjecture as that vague metamathematical idea. And the referee at least agreed to let me call my result a proof of the Turing principle to distinguish it from the conjecture. The principle that there can be a physical object whose motions contain those of all other objects. Nevertheless, now people sometimes call that the Church-Turing-Deutsch principle. And that’s how the mathematician’s conception ended up giving me credit for something Alan Turing did and arguably Ada Lovelace did. A few years later I gave a talk in Oxford arguing that it makes no sense to regard Turing’s conjecture in any form as something one might hope to prove one day from logic like Fermat’s last theorem. But that it could be proved to be a property of quantum mechanics. Sitting in the front row was Robin Gandy who’d worked with Turing. And he got a bit agitated and at the end he stood up and declared with good humor but very emphatically, I’ve never heard such a load of rubbish in my life. I tried to explain further but he seemed implacable. He’d also given a talk at the same event and at the dinner afterwards he came over to where I was sitting and he said, you know, I think there might have been a grain of truth in there somewhere.

David Deutsch

03:02:44 - 03:06:51

Let’s talk about it later. And we did discuss it later but unfortunately we did not reach a resolution. He was a mathematician. He had the misconception. Unfortunately in the bigger picture the mathematician’s misconception has done more than just cause amusing anecdotes. It expresses the idea, acknowledged or not, that somewhere out there in the world of mathematical abstractions or in some supernatural world of mathematical intuition, there is the authentic official though ineffable, now we know that Hilbert was wrong, ineffable definition of proof. And if some physical process that doesn’t conform to that definition turns out to allow us to know some new necessary truth, that process wouldn’t constitute a proof of that truth. There’s the misconception. It so happens that a quantum computer’s repertoire of integer functions is the same as the Turing machines. They differ only in speed. So some people view this as vindicating the mathematician’s misconception but no. First of all, we only know that they only differ in speed from physics, from quantum theory. And second, quantum theory won’t be the final theory of physics and even if it is, you can’t prove that either from mathematical intuition. In reality, we only have physical intuition, never provable, always incomplete, always full of errors. The misconception also affects thinking about information. For example, a quantum cryptographic device may perform a classical information processing task that is provably impossible classically. So the misconception makes people say, well, quantum cryptography isn’t an information processing task. It’s just an engineering task like building a washing machine. Why? Because Turing machines couldn’t perform it. Again, they think that there’s a mathematical definition of information out there somewhere independent of physics. The same holds for probability, by the way. Similarly, again, the answer to Eugene Wigner’s famous question about why mathematics is unreasonably effective, as he put it in science, is not that the physical world is actually being computed on a vast computer belonging to God or to supernormal aliens. Because there’s no reason other than the misconception why the alien computer should itself generate that particular tiny piece of mathematics that we call computable. Purely mathematical intuition will never reveal anything about proof or computation or probability or information. If you want to understand any of those things fundamentally, you must start with laws of physics. And in particular, with what is currently the most fundamental theory in physics, quantum theory. It won’t always be the most fundamental, but its replacement will not come from mathematics or logic or the supernatural. Okay, that’s it. Thank you.

Fernando Quevedo

03:07:07 - 03:07:25

Thank you. That was a thought provoking talk. Any questions from any mathematician? I have a very simple question.

Audience questioner

03:07:25 - 03:07:26

What is physical?

David Deutsch

03:07:29 - 03:08:22

Yes. It’s a bit like asking what is real. There seem to be various, I don’t know is the answer, but there seem to be various levels of reality and there’s a level that’s only accessible by experiment and then there’s the level to find out what the laws are. And then there’s the level that is independent of the laws. So we know that Fermat’s Last Theorem is true and if somebody comes and finds that general relativity has a flaw or quantum theory has a flaw, nobody will worry that maybe Fermat’s Last Theorem isn’t true. Laws of physics are things unlike that. They are things that could be overturned at any moment. We guess at them. So I can’t provide an answer better than that. It’s a deep question.

Fernando Quevedo

Very good. So let’s thank David again for this wonderful talk. Thank you. Just to finish the event, so let me remind you that this is not yet over. There’s a special event that was a post-ceremony public event this afternoon starting at 6.30, so it’s a few minutes from now, in the Savoy Excelsior Palace in Trieste. And it’s a moderated roundtable discussion with Hartmut Neven, who is Google’s Director of Engineering. He’s here. Alessandro Curioni, who is the Vice President of IBM Europe and Director of IBM Research Lab in Zurich. And Tommaso Calarco, the Director of the Institute for Complex Quantum Systems in the University of Ulm and a leading figure for this European Commission Quantum Technology Flagship Project. So you’re all welcome to participate. It’s for the general public. And there is a bus that we hired that will take 50 people. So all of the 50 of you who want to go, it’s the first come, first served. And then we will see you there. So it will be an interesting discussion about the importance of quantum technologies. Okay, well, thank you very much. And congratulations again to all the awardees.

2018-01-21 Chat with Aubrey de Grey

00:28:31 - 00:29:05

Impression of the value of life, that’s right. I think the fact that we’ve abolished, of course, the death penalty in all of Europe since the war, with the exception of course of countries like the former Soviet Union, I think this bodes well. And of course the same can be said of other aspects of society. The fact that for example in the West, the birth rate is declining so rapidly, this means that we may have longer than we might otherwise think to be able to cope with any overpopulation problems that might arise from essentially eliminating death from aging. So yeah, these are reasons for optimism being realistic.

David Deutsch

00:29:35 - 00:29:37

Thank you.

Listen at CBC Read the CBC transcript Source collection

2018-01-02 CBC The CurrentIs There a Limit to What We Can Know?

Is There a Limit to What We Can Know?

The Current · Host: Anna Maria Tremonti · Guests: Martin Rees and David Deutsch · 2 January 2018

Transcripts may contain errors. If you wish to re-use all, or part of, a transcript, please contact CBC for permission. Please check the corresponding audio before quoting in print. Copyright © CBC 2018.

Excerpted from the full episode transcript: the Martin Rees and David Deutsch discussion only.

ANNA MARIA TREMONTI: I’m joined by two scientists who have extensive knowledge and differing perspectives on whether there is a limit to our human knowledge. Martin Rees is the United Kingdom’s Astronomer Royal. And he joins us from Cambridge, England. David Deutsch is a visiting professor at the Centre for Quantum Computation at the Clarendon Laboratory of Physics at the University of Oxford. He is widely considered to be the father of quantum computing. David Deutsch is in Headington in Oxford. Hello to you both.

DAVID DEUTSCH: Hi.

MARTIN REES: Hello.

ANNA MARIA TREMONTI: Martin Rees I want to start with you. You are an astronomer. Space may seem very complex to the rest of us, but what do you think. Will we one day be able to understand our universe fully?

MARTIN REES: Well never fully. But we can make great progress. And we already have. But the point I want to make to follow up your introduction. The most complicated things in the universe are neither the very small things - atoms - nor the very large things - stars, black holes, and galaxies - they’re things in between, particularly living things because they have layer upon layer of structure and they are probably going to be harder for us to understand than either the atomic world or the cosmic world.

ANNA MARIA TREMONTI: Tell me more about that in what ways are organisms or the human brain more complex than what we can see way out in the far reaches of the universe?

MARTIN REES: Well a star for instance is a very, very hot ball of gas and it doesn’t have any very complicated substructure. It’s so hot that no complex chemistry could exist inside it. Whereas, if you imagine, even the smallest insects that has layer upon layer of complexity and it will be very, very difficult indeed for us to fully comprehend that. We can have simulated if we had a super computer. But for us to actually grasp exactly what’s going on inside the simplest insects is in my view a bigger challenge than to understand a black hole or to understand atomic physics.

ANNA MARIA TREMONTI: What do you think - is there a limit to what humans can know?

MARTIN REES: I’m sure there’s a limit to what humans can know. And it could be that the very deep elements of reality, which are as far beyond our grasp as quantum theory is beyond the monkeys grasp. I really think that’s possible. And I think I’d like to distinguish between what we can actually grasp or compute. We could perhaps make a computer that could actually simulate some of these things. But that doesn’t mean we could actually understand it - really grasp what’s going on. I think there’s a real limit to what human brains can actually comprehend and that may be important parts of reality that will be forever beyond that.

ANNA MARIA TREMONTI: Do you not think our brains might be able to keep adapting forever to take in new knowledge?

MARTIN REES: Well we have a long way to go, obviously. The huge opportunities for us to understand things that we don’t yet understand. We’re nowhere near the limits. We are nowhere near hitting the buffers in many sciences. But I think there are almost certainly severe limit to what we know that brains think about a million times slower than even a small computer. We know that for 40 years we’ve had machines can do arithmetic better than us and we have computers that can analyze data millions of times faster than us; can play games like chess and go better than a human. And so we know already that we machines that are better than us and there are certainly going to be limits to the sort of wet hardware inside our skulls.

ANNA MARIA TREMONTI: Well David Deutsch this is a good time to bring you into the conversation. Do you think humans will reach a point where we’ll be at the limit of what we can know?

DAVID DEUTSCH: No. I don’t think there can be such a limit. There are things that are unknowable. Mathematical things - all of them relating to infinity in some way. But I don’t think there can be anything interesting that is anything relevant to humans or human problems that is unknowable. I think that idea is very akin to a belief in the supernatural. I totally agree that the midscale things like living things and brains are probably much harder to understand than the very small or the very large. But brains, especially human brains, have this special relationship with the laws of physics which other brains and other physical processes don’t have. Namely, that they are universal at explaining things. And Martin said that we know that computers will be able to simulate everything, but he doubts that they will be able to comprehend everything or that we will be able to comprehend anything via them. But the thing is, comprehending, understanding, and explaining are all forms of computation and if we can’t do them with our natural hardware, we can do them with artificial hardware as we have been doing for thousands of years. We couldn’t even understand Newton’s laws if it hadn’t been for the invention of writing for example. There is only one type of computation in the universe. Our understanding and everything else must be a type of computation. Anything that goes beyond our abilities must be only because of things like speed and memory capacity, which we can improve artificially.

ANNA MARIA TREMONTI: Martin has compared some of this to monkeys trying to understand Darwinism. What do you think?

DAVID DEUTSCH: Yes. Human brain is a monkey brain with add-ons. The fact that monkeys can’t understand Darwinism - it’s either because they don’t have the physical capacity to hold the program. That is the program for human mind. That seems unlikely because their brains aren’t smaller than ours by such a large factor. It’s more likely human brains just have more computing power available - just more raw memory. Boring things, but which make all the difference when you add them on to the range of computations you can perform. And so I don’t think there’s that much difference in the computation in a monkey brain and the human brain. There can’t be any difference in the computation of human brain and anything further because we know that the our existing computers are already universal.

ANNA MARIA TREMONTI: Can human brains keep evolving to take in new knowledge, David do you think?

DAVID DEUTSCH: It could be done that way. My guess is that like in the past we weren’t artificially make bigger brains like, any science fiction stories with humans with gigantic heads - post-humans with gigantic heads. I think we’ll continue doing it as we have been doing it with technology. We already live largely - many of us anyway - inside our computers or inside our mobile phones and communicate via artificial means more and more. This can continue to whatever extent it’s convenient and to whatever extent we find that it solves our problems. This difference between simulation and understanding is the difference between form and substance. It’s solving problems or not solving problems. And it’s all knowledge worthy of the name is problem solving. That’s what we’ll be using it for.

MARTIN REES: Can I come back and…

ANNA MARIA TREMONTI: Yes absolutely. I was just going to bring you back in. Go ahead Martin.

MARTIN REES: Yes, well I think David [unintelligible] two very different things. One we can compute and one we can understand. Already we’ve had this machine that can play go and the world’s expert go players can’t understand why it made a particular move. That’s an example of something that is computable, but not understandable. And when I say that we humans can’t understand things, I don’t deny that maybe a computer can compute them. But I’m still saying we can’t understand them. But I’d like to make another separate point, which is that doesn’t David Deutsch agree that it could be that the fundamental bedrock of space and time is describable by some 10 or 11 dimensional geometry as in M-theory. And this is what we know; some of the brightest brains in the world are now working on. But isn’t it possible that they will never succeed simply because the maths is beyond the human brain. I think that’s one perspective which we may hit the buffers. The other aspect is that already, as David admits, there are many things that can be computed, but we do have in our heads.

ANNA MARIA TREMONTI: David what do you think?

DAVID DEUTSCH: Well I entirely agree that there may be many things. There already are many things that we can’t do in our heads. In fact the basically the whole of modern science could barely be done in anyone’s head entirely. I agree with that. But I think it’s kind of parochial to care about what we can do with or without help. It’s like saying, we will never run faster than 10 metres per second just because we would never move faster - sorry than 10 meters per second - just because we can run faster. We have a problem of how to get from A to B in a shorter time because it’s more convenient and we can solve that problem provided that the laws of physics allow it. Again I also agree with Martin that a star is a very simple thing compared with a brain. But if you look in detail at the star, the more detail you look at - now we’ve got these pictures of close-ups of a star with lots of structure and curlicues and so on. And to predict that exactly would require a computer the size of the star. But we don’t want to predict that precisely because it’s not interesting. It’s the same is true of Go-playing. If something can beat us at Go-playing just by sheer power and not by there being anything interesting there, that’s not very surprising. And I don’t count that as a limitation of knowledge any more than the details the surface of the sun our limitation on knowledge. But if there is something interesting there that means it affects something human. That means it affects a problem. And just because people don’t understand it now, doesn’t mean they won’t.

ANNA MARIA TREMONTI: David Deutsch you use the example of the hairs on Julius Caesar’s head. Take us through your point there.

DAVID DEUTSCH: Well that’s the same issue really. What are the limits of the science of palaeontology and paleoanthropology? We know an amazing number of things about the daily lives of people thousands of years ago and even hundreds of thousands of years ago. But obviously there are many things there that we cannot know. We cannot know whether the number of hairs on the head of Julius Caesar at the moment when he died was even or odd. And that is not a limitation of human understanding, only a limitation of information. And the distinction I’m trying to make is between understanding in the sense of problem solving and information of which there will always be too much.

ANNA MARIA TREMONTI: In other words, we may not know but we don’t always need to know what we don’t know.

DAVID DEUTSCH: We don’t care. And there’s nothing to make us care. And if there were something to make us care, that very thing would be the lever by which we would discover it.

ANNA MARIA TREMONTI: Martin Rees do you see technology or other organisms becoming smarter than us in the future?

MARTIN REES: Computers are already smarter than us in many respects. But I still think that being a human being I care about what human beings can actually understand and grasp. And I still repeat my point that there are limits to that. And that’s rather sad because I think most of us as scientists want to understand things. We don’t just want to be able to compute something. And I think there are going to be limits to that.

ANNA MARIA TREMONTI: I think David Deutsch would agree with you on that level. Would you not David?

DAVID DEUTSCH: Yes, yes certainly. I think where we’re disagreeing is that I think that machine assisted understanding is possible, as well as machine assisted prediction and machine assisted description. We already do machine assisted understanding. When we understand something about Maxwell’s equations or Einstein’s equations by writing symbols on a piece of paper, we are doing machine assisted understanding, not just prediction.

ANNA MARIA TREMONTI: And so if we can use machine assisted understanding in progressively wider ways, does that mean there is infinite progress in what we can do?

DAVID DEUTSCH: Unlimited progress, yes. Of course we’ll never reach infinity.

ANNA MARIA TREMONTI: And does that mean we will head toward an ideal society? Could this lead us to do that?

DAVID DEUTSCH: No, because this is a different issue. But I believe there will always be problems. And there will always be soluble. But we will never be anywhere near the ideal which is the infinite knowledge or infinite perfection or all problems being solved. It wouldn’t be very nice if all problems were solved because they’d be nothing to think, it would be a bit like death. But nor will we I think reach a stage where there are problems, but they can’t be solved.

ANNA MARIA TREMONTI: Martin Rees if you think humans eventually reach a limit of our knowledge, what happens to progress and everything from science to art when we reach that point?

MARTIN REES: Well I think the certain direction in which we can explore indefinitely - some areas where our brains are limited. But of course I agree with David, that humans are not the sum is in any sense of evolution. We are probably in a halfway stage in evolution. We’ve had four billion years of Darwinian evolution to lead from primordial life to us. And we know that there are billions of years ahead. I think there’s going to be evolution. Whether that’s going to be organic or electronic evolution is unclear. I suspect that the main intelligence evolution will be in the form of electronic entities, not souped up wet human brains. But I think we should remember we’re just beginning and we should also remember that science is an unending quest. And as we settle some questions, then we are able to pose questions we couldn’t even pose before. Science does progress and as it progresses the periphery between the known and the unknown gets longer. And there more questions that come into view that we are going to be challenged by.

ANNA MARIA TREMONTI: Gentlemen, you’ve given us a lot to think about as we begin a new year. Thank you.

MARTIN REES: Thank you very much.

DAVID DEUTSCH: It’s a pleasure.

2017-07-28 The Christian Transhumanist PodcastHumanity's Infinite Reach

01:06:35 - 01:06:37

Very nice and very interesting talking to you.

Micah Redding

01:06:37 - 01:06:43

All right we’ll see you later. Bye bye then. Bye.

Read the article at The Economist Source collection

2017-03-09 The EconomistDavid Deutsch, Father of Quantum Computing

David Deutsch, Father of Quantum Computing

Technology Quarterly · Brain scan · 9 March 2017

A fundamentally new way of harnessing nature

“I OCCASIONALLY go down and look at the experiments being done in the basement of the Clarendon Lab, and it’s incredible.” David Deutsch, of the University of Oxford, is the sort of theoretical physicist who comes up with ideas that shock and confound his experimentalist colleagues—and then seems rather endearingly shocked and confounded by what they are doing. “Last year I saw their ion-trap experiment, where they were experimenting on a single calcium atom,” he says. “The idea of not just accessing but manipulating it, in incredibly subtle ways, is something I totally assumed would never happen. Now they do it routinely.”

Such trapped ions are candidates for the innards of eventual powerful quantum computers. These will be the crowning glory of the quantum theory of computation, a field founded on a 1985 paper by Dr Deutsch. He thinks the widely predicted “quantum supremacy” that eventually puts a quantum computation incontrovertibly ahead of a classical one will be momentous for scientists and laymen alike. He brushes off the fervent debate about whether the commercially available D-Wave computer offers a speed advantage. “If it works, it works in a completely different way that cannot be expressed classically. This is a fundamentally new way of harnessing nature. To me, it’s secondary how fast it is.”

Still, these are steps towards a powerful, universal quantum computer that could solve a lot of thorny problems. To describe such a device properly is to account not only for the states of each of its constituent bits but also for all the couplings between them, for each is entangled with every other. A good-sized one would maintain and manipulate a number of these states that is greater than the number of atoms in the known universe. For that reason, Dr Deutsch has long maintained that a quantum computer would serve as proof positive of universes beyond the known: the “many-worlds interpretation”. This controversial hypothesis suggests that every time an event can have multiple quantum outcomes, all of them occur, each “made real” in its own, separate world.

At the same time, quantum computation, and the quantum-mechanical theory from which it springs, are all subsumed in a newer idea that Dr Deutsch is pursuing. He contends that what he calls his “constructor theory” provides a perspective that will lead to the rewriting of physics altogether. As with classical computer science, quantum computation and even genetics, it is based on the role of information. But rather than letting physical laws define what is and is not possible, as science does now, constructor theory asserts that those laws actually arise from what is and is not possible.

From observed possibilities, a mathematical object called a constructor can be fashioned. Operating with and on these constructors gives rise to what Dr Deutsch reckons is a theory even more fundamental than quantum mechanics. He is enthusiastic about the theory’s potential to upend the very foundations of science, but concedes that testing it experimentally remains a distant possibility. Then again, a few decades ago he would have said the same thing about quantum computers.

This article appeared in the Technology Quarterly section of the 11 March 2017 print edition under the headline “David Deutsch”.

2016-11-16 Making Sense#52 - Finding Our Way in the Cosmos - subscriber version

01:48:18 - 01:48:28

The filmmaker and said you must be very proud okay that’s nothing like what I said well anyway

Sam Harris

01:48:28 - 01:49:17

David it’s once again it’s a thrill to talk to you and I don’t know if we made progress that anyone can measure but I certainly feel like I am a better person every time we talk so thank you so if you enjoyed this podcast there are several ways you can support it you can leave reviews on iTunes or Stitcher or wherever you happen to listen to it you can share it on social media with your friends you can discuss it on your own blog or podcast or you can support it directly at samharris.org/support

2016-03-09 The Unity of the Universe

00:49:41 - 00:51:56

The basic reason that constructor theory can work that sort of magic is that it abstracts away the constructor like the theory of catalysis in chemistry which is another as I said another example of an existing constructor-theoretic theory it’s not this process that is declared to be possible or impossible in constructor theory it’s this just that so the constructor which is usually the macroscopic part of the process is abstracted away and this is what makes constructor theory a natural vehicle for expressing scale invariant laws and substrate invariant laws about quantities like information and heat and work exactly like the perfect cosmological principle which had to be developed into the sophisticated steady-state theory constructor theory will have to be developed quite a bit more before we derive testable predictions from it but unlike steady-state theory constructor theory has already provided a significant unification and illumination of fundamental matters in diverse areas of physics and beyond as I said and I think this already makes it a substantial step towards the unity that Dennis was looking for Okay, thanks.

Event host

00:51:56 - 00:52:04

Okay, and so to mark the closing of this wonderful series of lectures, I’ve been asked to present these to …

2015-12-15 Making Sense#22 - Surviving the Cosmos - subscriber version

01:52:47 - 01:53:00

If you enjoyed this podcast, there are several ways you can support it. You can leave reviews on iTunes or Stitcher or wherever you happen to listen to it. You can share it on social media with your friends. You can discuss it on your own blog or podcast.

2015-10-19 RSA DebateOptimism, Knowledge and the Future of Enlightenment

01:14:02 - 01:14:57

Yeah, well, I agree. And I agree that I also distrust the idea of scientist kings. I think Plato was very wrong about that. I’m not sure it’s fair that I get the first word and the last word. So can I just say that trusting those who provide the best explanations isn’t quite it. The democratization of knowledge via the best explanations would mean, you see, once a process has decided that something is the best explanation, there’s no need to trust anyone. And that’s really the whole point. For knowledge to have effects, it shouldn’t be filtered through any single source. So I agree with that.

Karen O’Brien

01:14:57 - 01:15:19

Well, we’re not giving anyone the last word because the last words will take place over a drink in the Benjamin Franklin room, and what more appropriate room title than Benjamin Franklin for optimism, knowledge and enlightenment. So, I mean, many thanks to the RSA for hosting this event and generously providing us with a drinks reception afterwards. But please join me in thanking our two wonderful speakers. Thank you very much.

Read the Q&A at Nature Source collection

2015-10-07 NatureQ&A: David Deutsch

Q&A: David Deutsch

Interview by Kristin Lynn Sainani · 7 October 2015

Physicist David Deutsch is considered the founding father of quantum computing. In his 2011 book, The Beginning of Infinity, Deutsch argues that there is such a thing as objective beauty.

What is your argument for the existence of objective beauty?

The argument I like best is about why flowers are beautiful. Flowers evolved to attract insects, and insects evolved to be attracted to flowers. But this explanation leaves a massive gap: it only explains why insects like flowers. So how is it possible that something that evolved to attract insects can be attractive to humans too? I conclude that there must be objective beauty — aspects of beauty exist outside cultural fads or sexual selection. And these aesthetic truths are as objective as the laws of physics or maths.

If beauty is objective, why is there so much variation in what people consider beautiful?

Beauty has both a subjective and objective part. Human aesthetic judgment is a complicated mixture of genetic, cultural and objective factors. If you look at paintings from centuries ago, you will find that the women tend to be considerably heavier than what we now consider to be ideal. That can be neither objective nor genetic, so it must be cultural. Our preference for symmetry is probably related to our preference for healthy mates — many diseases and deformities make people less symmetrical. So that one could be genetic.

Our knowledge of the nature of objective beauty is still primitive. We cannot reliably distinguish between subjective and objective beauty, certainly not by just looking. Things that meet aesthetic preferences built into our brains or instilled by culture look just as beautiful to us as those that are objectively beautiful.

Why is it important to acknowledge the existence of objective beauty?

During the twentieth century, some movements denied that there was such a thing as objective truth in science. These movements significantly held back scientific progress. For example, I’m pretty sure quantum computing would have been proposed in the 1950s rather than in the 1980s if it had not been for these beliefs. Because our culture generally denies the existence of objective beauty, research into it is substantially cut down. I’m not aware of any research that looks at the nature of objective beauty.

How do you counter those who insist that beauty is always subjective?

It is remarkable how the arguments against objectivity in aesthetics, and in morality, have exact counterparts in classic arguments against objectivity in science. People say we do not have access to the world; we only have access to the interpretations that we put on the world through our senses. The second part is right, but that does not mean we cannot achieve truth. To think that, is to confuse truth itself with some sort of superhuman, certified, reliable access to the truth. For example, the abolition of slavery was an objective moral improvement. It is not just cultural. It is certainly not genetic. It is not a matter of preference. It would still be true that slavery was wrong even if nobody knew that.

What is the connection between aesthetic beauty and scientific argument?

Beauty in science is called elegance. Physicists will, as a matter of practice, take elegance as a guide. There is the phrase: many a beautiful theory was slain by an ugly fact. This is very true. But when it happens, we inevitably find an underlying theory that is even more beautiful than the theory that was slain. So beauty cannot be used as a criterion of what is true; but it is at the very least useful as a guide to what to try next.

What factors do you believe govern human sexual attraction?

I speculate that human beauty started out just like any other animal beauty — completely biological, and not objective at all. But as humans became intelligent and started making aesthetic judgements, they increasingly tried to improve the aesthetic and other standards by which they chose their mates. And that increasingly led to true standards. So we should find that the common features that have changed in all human populations since our ape ancestors are aspects in which humans have become objectively more beautiful.

Are you saying that humans have steadily made the world more beautiful in the same way that we have achieved scientific progress?

Yes. Objective beauty, like objective truth, is subject to open-ended improvement. For example, our knowledge of physics can contain more and more truth, even though no one theory is ever perfectly true. Newton’s theory contained more truth than what was there before. But it was superseded by Einstein’s theory. And science continues its progress by finding new aspects of reality forever. By contrast, something that is subjective reaches a maximum and then stops.

We discover aesthetic truths in the same way as we discover scientific truths, even if the methods look different. It is conjecture and improvement according to some standard; then improvement of the very standards; then criticism of existing ideas according to these standards; and so on.

Aesthetic progress has been a lot slower than scientific progress because people can only express in words a tiny proportion of what they know about beauty. But humans have achieved an enormous amount. Mozart and Beethoven improved artistic standards in music. And films have become more beautiful in the past century.

Only humans can improve on beauty. When nature achieves beauty it is an accidental by-product of something else. Nature can only get so beautiful, but humans can paint something that is more beautiful than any scene.

This interview has been edited for length and clarity.

2015-06-01 David DeutschPhysics Without Probability

01:05:46 - 01:06:18

Okay, well, you can. If you want to work out the half-life, then you simply have to imagine the situation in which these game players bet on how long it will take and what the rational thing for them to do is and you’ll find that the rational thing for them to do is to bet on exponential decay. And that’s the panel. So, lots of work with the panel. Thank you.

2014-06-01 World Science FestivalWhat Is Quantum Mechanics Really Telling Us

01:22:22 - 01:22:35

Thank you.

Brian Greene

01:23:04 - 01:23:06

Thank you.

Read the response at Edge Source collection

2014-01-01 EdgeWhat Scientific Idea Is Ready for Retirement?

Quantum Jumps

The term “quantum jump” has entered everyday language as a metaphor for a large, discontinuous change. It has also become widespread in the vast but sadly repetitive landscape of pseudo-science and mysticism.

The term comes from physics, and is indeed used by physicists (though rarely in published papers). It evokes the fact that mutually distinguishable states in quantum physical systems are always discrete. Yet there is no such phenomenon in quantum physics as a “quantum jump”: under the laws of quantum theory, change is always continuous in both space and time. OK, maybe some physicists still subscribe to an exception to that, namely the so-called “collapse of the wave function” when an object is observed by a conscious observer. But that nonsense is not the nonsense I am referring to here. I’m referring to misconceptions even about the sub-microscopic world—like: “when an electron in a higher-energy state undergoes a transition to a lower energy level, emitting a photon, it quantum-jumps from one discrete orbit to another without passing through intermediate states”.

Even worse: “when an electron in a tunnel diode approaches the barrier that it does not have enough energy to penetrate (so that under classical physics it would bounce off), the quantum phenomenon of tunneling allows it to appear mysteriously on the other side without ever having been in the region where it would have negative kinetic energy”.

The truth is that the electron in such situations does not have a single energy, or position, but a range of energies and positions, and the allowed range itself can change with time. If the whole range of energies of a tunneling particle were below that required to surmount the barrier, it would indeed bounce off. And if an electron in an atom really were at a discrete energy level, and nothing intervened to change that, then it would never make a transition to any other energy.

Quantum jumps are an instance of what used to be called “action at a distance”: something at one location having an effect, not mediated by anything physical, at another location. Newton called this “so great an Absurdity that I believe no Man who has in philosophical Matters a competent Faculty of thinking can ever fall into it”. And the error has analogues in fields quite distant from classical and quantum physics. For example in political philosophy the “quantum jump” is called revolution, and the absurd error is that progress can be made by violently sweeping away existing political institutions and starting from scratch. In the philosophy of science it is Thomas Kuhn’s idea that science proceeds via revolutions—i.e. victories of one faction over another, both of which are unable to alter their respective “paradigms” rationally. In biology the “quantum jump” is called saltation: the appearance of a new adaptation from one generation to the next, and the absurd error is called saltationism.

Newton was wrong that there is a maximum size of error that competent people can fall into, but right that this particular one is severe. All those versions of it are mistaken for a single reason: they all require information of the requisite kind to appear from nowhere. In reality, the space on the far side of the barrier cannot “know” that an electron, and not a proton or a bison, must appear there, until some physical change, originating at the electron, reaches it. The same holds when it is not a spatial gap but a directly informational one: Political institutions, and biological adaptations, instantiate information—knowledge—about how a complex system can better meet the challenges facing it, and knowledge can be created only by processes of piecemeal variation and selection. And Kuhn’s vision cannot explain how science has in fact been delivering knowledge about physical reality at an ever-accelerating rate.

Quantum jumps in all these fields represent a retreat from explanation, and therefore in effect an appeal to the supernatural. They all have the logic of the Sidney Harris cartoon “Then a Miracle Occurs” (depicting a mathematician with a gap in his proof). As Richard Dawkins puts it, “saltationism is creationism”. And in all cases the reality that fills the gap, the idea that truly explains the phenomenon, is much more interesting and delightful than any faith in its mystery could be.

Read the article at Nautilus Source collection

2013-11-22 NautilusNot Merely the Finest TV Documentary Series Ever Made

Not Merely the Finest TV Documentary Series Ever Made

By David Deutsch · 22 November 2013

A reflection on Jacob Bronowski’s “The Ascent of Man.”

Editorial note: Nautilus’s current page retains three footnote markers but no longer provides the corresponding notes. The orphaned markers have been omitted here.

A range of dark hilltops appears against a dawn sky. On a ridge in the far distance we can discern a human silhouette. It is someone telling us about the uniqueness of Man. “Man …” says the tiny figure in the landscape, “is not a figure in the landscape …”

“He is the shaper of the landscape,” continues Jacob Bronowski, dwarfed by the looming hills that have not been reshaped since the last ice age.

When a narrator expresses a paradox, it should be regarded as a promise to explain something amazing. Bronowski, in this historic 13-part television documentary, delivered on his promises.

He was a one-off. A patrician British academic, he had arrived in Britain at the age of 12 in the aftermath of World War I, speaking no English. He and his family were Jewish refugees from Łódź in the Russian Empire (now Poland). One of the 20th century’s most eloquent advocates of Western values, he was denied senior positions at British universities because (according to his daughter, the historian Lisa Jardine) MI5 mistakenly suspected him of being a communist sympathizer. A powerful intellect of unusual breadth and originality, he was never lucky enough to make the spectacular discovery that would have got him recognized as a great mathematician or scientist, but he made contributions to mathematics, poetry, paleontology, history, moral philosophy, and—during World War II—strategic bombing. Despite his hatred of war in general, he believed it would be pious humbug to stand aside from making the world a better place through winning a war. He wrote books (mostly collections of his electrifying public lectures) and became a respected media pundit. But he is, rightly, best known for his last project, The Ascent of Man (1973), in my view the finest television documentary series ever made.

It came about as a sort of reply to Kenneth Clark’s 1969 series Civilisation, itself a very successful documentary, but one that had been criticized for identifying “civilisation” almost exclusively with art (and European art at that), barely mentioning science. So David Attenborough, at that time the controller of BBC2, who had commissioned Civilisation, looked around for a scientist who could restore the balance by explaining science on television as Clark had explained art, and demonstrate to those steeped in the humanities that scientific knowledge is both thrilling and momentous. Bronowski, with his cross-disciplinary qualifications, was a natural choice, and turned out to be perfectly suited to that task. For in Bronowski’s scheme of things, scientific discovery and artistic achievement—and moral and political progress too—are all integrated in the most detailed, practical way. For example, while the truths of science are morally neutral, the practice of science absolutely depends on certain values. It cannot continue unless independence of thought, the rejection of authoritative sources of truth, and tolerance of dissent are valued and actively practiced by the scientific community. Nor can the wider society make progress in the longer term unless it is able to do as science does: confront its own constitutive ideas with reality and change them creatively when they fail.

Bronowski saw the purpose of art as being the same as that of science: to give meaning and order to our experience by revealing hidden structure beneath the appearances. That conception gives an objective meaning to progress in those fields. Moreover, neither a scientific discovery nor a work of art can be made by merely “copying nature,” for appearances are deceptive, traditions flawed. Hence a system of open-ended self-correction is needed. Creativity and boundless imagination are as essential in the mathematical sciences as in the humanities; criticism and testing in art just as in science.

The Ascent of Man conveys this unity in a beautifully explained account of the pivotal steps in the history of ideas. But Bronowski’s agenda was deeper and more urgent than that. Though he expressed it with characteristic understatement, his was a message of rebellion, a bold attempt to correct one of the great misconceptions of modern times and single-handedly to redirect the great river of intellectual history.

The misconception might—if its perniciousness were generally acknowledged—be called antihumanism: the pattern of ideas that disparages the human species, jeers at its claims to superiority over other species, or to any special entitlement, glories in its cosmic insignificance, and reinterprets its advent as nothing special and its subsequent progress as mostly illusory or fraudulent—and thus in all these senses, denies its ascent. Antihuman ideas would have seemed wicked or insane to the great majority of thinkers since the Enlightenment at least. But during Bronowski’s lifetime, many of them had become mainstream, to the extent of being taken for granted in academic and everyday discourse. So Bronowski’s rebellion is already there in his title: The Ascent of Man. He says that it refers to the “brilliant sequence of cultural peaks”—such as the invention of stone tools, agriculture, cities, and modern science—by which humans have learned how “not to accept the environment but to change it,” thereby improving our lives. And that this progress, despite continual setbacks, has been cumulative for as long as our species has existed.

Think again about those hills of Bronowski’s. Humans really do have the ability to shape them—for instance into fortresses, or temples, or stepped terraces for growing rice. Or to dismantle an entire hill and rebuild it somewhere else as a pyramid or a great wall. Shaping hills (or valleys, or forests) was in fact among the earliest and crudest of our distinctive achievements. If those hills did happen to be unchanged since the ice age it was only because human beings had happened to decide to leave them alone and to shape something else. But what conclusively resolves the paradox is not so much that we can shape our environment (after all, even ants build anthills) but that we shape it into novel forms, with novel functions and purposes, and choose whether and how to make the changes, and invent new purposes (such as scientific research, art, and literature) and new behaviors (such as farming, city-dwelling, and engineering) that enable us to thrive in the formerly lethal environments that we have shaped.

We have ascended in that sense, and improved our ability to do so, incrementally since prehistoric times, and ever more rapidly of late, so that by now we could, if we wished, shape mountains on the moon, and live there. And if we choose to continue our ascent, we shall soon be able to reshape anything from the scale of molecules (nanotechnology) to that of planets (terraforming and geoengineering) for our convenience and delight.

“The Ascent of Man” is also a scholarly pun. It refers to Charles Darwin’s book The Descent of Man (1871), which defended the proposition that humans are descended from other animals. Bronowski of course had no quarrel with the “descent of Man” in that sense: It was essential to his world-view. But he did want to refute the slew of antihuman fallacies and misconceptions that it had inspired. For instance, the fallacy that since we are descended from animals, we should expect to explain something about the roots of warfare by studying the fight-or-flight responses of bison; about adultery by studying the mating behavior of chimpanzees; about nation-states by studying the territorial instincts of birds, and so on. Books claiming to do so, such as Robert Ardrey’s The Territorial Imperative (1966), Konrad Lorenz’s On Aggression (1966) and Desmond Morris’s The Naked Ape (1967), had been enthusiastically received. Their shared approach, of reinterpreting human social and political behaviors as being due to inherited remnants of animal instincts, soon came to dominate both public and academic discourse about human nature.

But in Bronowski’s view, that approach is a nonsense and a travesty. His argument (one of the delights of The Ascent of Man is that it consists entirely of argument) begins by directing our attention away from the common heritage of animals and toward one particular difference between humans and all other present-day species: Every other species is exquisitely adapted by evolution to a certain range of environments, and to a certain range of lifestyles that it is genetically programmed to act out in them. But evolution “has not fitted Man to any specific environment. On the contrary … he has a rather crude survival kit. And yet … one that fits him to all environments.”

How exactly this unique “survival kit” works is still mysterious, but we know that it confers an ability to use imagination, reason, and critical thinking, with which our ancestors could improve their ideas; and it confers the “emotional subtlety and toughness” to apply those ideas to improve the landscape to meet their objectives. And to improve those objectives themselves to create new and better ways of life.

This was of course a biological adaptation, encoded in DNA. But it was fundamentally different from all other adaptations in the biosphere. For it enabled, for the first time, further complex adaptations to be encoded, and improved, independently of DNA. Improving genes by biological evolution requires vast numbers of individuals to die for every scintilla of progress made. But the ability to encode whole new patterns of behavior and to pass them on by cultural means finally disconnected progress from death. As Karl Popper put it, we humans can “let our ideas die in our place.” And this set in motion two radical new forms of evolution which are millions of times more rapid, and infinitely more powerful, than that of genes, namely, the evolution of ideas in individual human minds, and the evolution of culture.

If you want to know how bison, chimpanzees, or birds lived 10,000 years ago you need only study how they live now (in the wild). But for humans, you would learn nothing by that method. Ten thousand years ago we lived in very different surroundings, made a living in different ways, had quite different economic and family relationships, and pursued different objectives. In all these respects, we were worse off—in some of them brutally, almost unimaginably worse—than we are now. In another thousand years, or even a hundred, we may have changed as much again. If we then look back to try to understand the changes, we shall not find the explanation written in our genes.

The fitness “to all environments” that our new adaptation conferred—the universality of the human condition—has disconnected humans from their genetic inheritance. The behavior, lifestyle, abilities, and limitations of all other species are indeed explained entirely through the information in their DNA and the forces of natural selection that created it in their ancestors—that is, through their descent (in Darwin’s sense). But what is important about humans can only be understood through their ascent (in Bronowski’s).

Yet ascent is far from inevitable. Our ancestor and cousin species, who already had our survival kit, can hardly ever have used it for making progress. That is why those species are all extinct. We are the only species left on Earth with the capacity for taking control of our environment and lifestyles, and improving both on less-than-geological timescales. Even we did so only sporadically until the current spate of rapid progress, brought about by the scientific revolution and the Enlightenment.

These fits and starts were possible because, unlike every other survival kit in the biosphere, ours does not come with inborn knowledge of when to use it and what to use it for. It couldn’t, because it includes the ability to change those things too. The propensity to use it to create progress is itself cultural, but most cultures in history have not valued or protected the behaviors that permit progress.

The Ascent of Man depicts many different cultures and societies. Bronowski describes their achievements with admiration and their predicaments with compassion. But his reason for discussing them is always to judge them—and always according to his explanation of whether, and how, they were capable of making progress. Thus, his rebellion defies a further taboo: cultural relativism, the idea that there is no such thing as one culture or society’s values and traditions being better than another’s. And therefore that objectively there is no such thing as progress either—no ascent—only the meeting of some culture’s preferences, which another may loathe.

Aren’t people “basically the same everywhere?” In Bronowski’s view, individual people aren’t basically anything, because their basic aptitude is universal. But in regard to cultures, Bronowski draws an uncompromising distinction between those that facilitate ascent and those that do not.

Filming the last remnants of an ancient horse-based culture in Afghanistan, Bronowski did not regret its decline. On the contrary, instead of applying the customary condescending doublethink towards primitive cultures, Bronowski traced the 5,000-year history from the earliest gangs of horsemen who stole from farmers what they could not produce themselves, to Genghis Khan, the terror of medieval civilization, and does not hesitate to classify that entire strand of cultural history along with Stalinism and Nazism as an archetypal enemy of the ascent of man. And the Nazi tank as the descendant of the war horse. “War,” says Bronowski, “is not a human instinct” (this was a sideswipe at Lorenz and all who wish to see the animal mind inside the human), “it is a highly planned and cooperative form of theft.” Similarly, Bronowski dared to stand in front of the great stone statues of Easter Island and declare that “These frozen faces … mark a civilization which failed to take the first step on the ascent of rational knowledge.” Bronowski said, “I am fond of these ancient, ancestral faces, but in the end, all of them are not worth one child’s dimpled face,” for one human child—any child—has the potential to achieve more than that entire civilization did. Yet “for most of history, civilizations have crudely ignored that enormous potential … children have been asked simply to conform to the image of the adult.” And thus ascent has been sabotaged or frozen.

Continued ascent will depend on adopting Bronowski’s unified understanding that human values are not arbitrary nor science merely utilitarian: that there is a truth of the matter in cultures just as in scientific theories. That these truths are not obvious nor mechanically obtainable in either case, but can be approached only through institutions of relentless error-correction including those of science and of an open society. Bronowski spoke of peaks, but there have been many troughs too. Today, civilization still has enemies. There are still those who devote their creativity to the archaic objective of resisting new ideas. More dangerously, in my view, the many-stranded civilization of reason and science that has so powerfully promoted ascent in recent centuries still sits on the unstable fence between its aspirations for progress and the specious comfort of stasis. Two hundred years after the Luddites, it is still not sure whether labor-saving devices cause wealth or poverty. Two hundred years since Malthus, it has still not realized that the recipe for doom is not “too many people,” but too few ideas. After 300 unprecedented years of sustained progress, it is still tempted to revert to its predecessors’ educational priorities, namely to produce “higher standards” (i.e. more standardized children). And after 50 years of conservationism, it is no longer sure whether it is better to kill a herd of elephants or a child. Most dangerous of all, it is no longer even sure whether its own spread—to other cultures, or to other planets and star systems—would be a good thing.

Go forth, therefore, and watch The Ascent of Man. If you have already done so, watch it again.

About David Deutsch

David Deutsch has done seminal work on quantum computation and is a member of the Centre for Quantum Computation at the Clarendon Laboratory, Oxford University.

This article originally appeared in the Winter 2014 Nautilus Quarterly.

2013-06-30 ABC The Philosopher's ZoneAI - Think Again

00:22:11 - 00:24:03

Yes, creating knowledge within the arena of error. Yes, that’s right. Exactly. If I take an analogy, for example, in the early 19th century when we didn’t know what life was, you know, unless somebody had made the philosophical breakthrough that Darwin made to realize that the shapes of animals are not what the process that made them is about. The process that made them is about these traits, as he called them, or what we would today call genes, which create the shapes of animals as a side effect. So if somebody hadn’t understood that and they had thought that the problem of the origin of species and so on, was why is an elephant the shape it is? Why does it have a trunk and so on? Then they would concentrate their efforts on trying to sculpt elephants on a finer and finer scale in the hope that if they did it on a fine enough scale, the elephant would suddenly start walking away and being exactly like a normal elephant. And they’d be completely wrong because the point of the correct theory of why elephants are the shape that they are is not about the shape. It is about natural selection and replication and variation and differential survival and accurate copying of genes, not of the animal’s body and so on. All those things are hidden when you look at an elephant. But it’s only if you understood those things that you would have the slightest chance of knowing why the elephant has a trunk, let alone making an artificial elephant. My intuition is that it will be very easy to program. It’s the understanding it that’s hard.

Joe Gelonesi

00:24:03 - 00:24:42

Ah, intuition. Let’s not go there. Well, not today anyway. Pioneer of quantum computation, David Deutsch, calling for a new philosophical understanding to break the impasse on AGI, artificial general intelligence, and he’ll take it with or without consciousness. Next week, a fitting finale for season one, renowned philosopher Simon Blackburn on human nature and science. His subject for the inaugural Alan Saunders Memorial Lecture. The Philosopher’s Zone is produced by Diane Dean. Technical production today by Martin Peralta. I’m Joe Gelonesi. See you next week.

00:24:43 - 00:25:13

Will doctors of the future be anti-fat? Looking then at the unconscious results, we found that about 40% of students had a significant anti-obesity bias. Implicit bias in medical students and ways to overcome it. Also, kidney disease, fetal immune programming and a silver bullet for antibiotic resistance. It’s a jam-packed health report with me, Joel Werner. You can download the program from the RN website.

2013-06-25 NautilusWhy It Is Good to Be Wrong

00:26:22 - 00:29:22

You know, what do you think? He was a historian. Charles Parkin was his name. He was a professor of history. And he said, yes, I read it. You know, it seems all right. But this induction stuff, hasn’t that all been disproved by that Popper chappie? And I had only once before even heard of Popper and certainly had no intention of reading him. And it’s really amazing that Charles Parkin even had the idea that inductivism had been refuted by Popper because that was contrary to what the overwhelming majority of professional philosophers thought at the time. They thought that Popper was just ridiculous for having claimed to solve the problem of induction and to have an epistemology that superseded inductivism. By the way, I think even today, the majority of philosophers would say that. But in those days, Popper was just unheard of in that kind of area. But somehow he had heard of it. Somehow he told me about it. Somehow I got the idea of getting a book of Popper’s out of the library. And I was bowled over because the standard of argument was simply head and shoulders above anything I’d read in Bertrand Russell. So, you know, with the hindsight, I could see that Bertrand Russell was trying to kind of fob me off with things that aren’t really arguments. Whereas Popper was getting to grips with the actual thing and finding, yes, this part is right, this part is wrong, this is why, and so on. As far back as I can remember, I wanted to be what I now call a physicist. But for some reason, and I don’t know how this happened, but for some reason I didn’t ever know what physics is until I was 11 and encountered a physics lesson in the school that I entered at the age of 11. Somehow it had passed me by that this thing. And as soon as I realized what that subject is, I knew that I wanted to be a physicist. I can more easily answer the question if I wasn’t allowed to be any kind of scientist, then I would be a film director, yes. But I think it’s not a well formed question in the case of science, because if I was somehow prohibited from being a physicist, I would tend in that direction as close as the prohibition allowed. As I think in fact I have done. I mean, the way that I got to the particular take on physics that I have is by finding that that’s where I tend to, by trying to solve things.

Read the article at Nautilus Watch the separate interview in Deutschland Source collection

2013-05-23 NautilusWhy It’s Good to Be Wrong

Why It’s Good to Be Wrong

By David Deutsch · 23 May 2013

Nothing obstructs access to the truth like a belief in absolute truthfulness.

Editorial note: Nautilus’s current page retains seven footnote markers but no longer provides the corresponding notes. The orphaned markers have been omitted here.

That human beings can be mistaken in anything they think or do is a proposition known as fallibilism. Stated abstractly like that, it is seldom contradicted. Yet few people have ever seriously believed it, either.

That our senses often fail us is a truism; and our self-critical culture has long ago made us familiar with the fact that we can make mistakes of reasoning too. But the type of fallibility that I want to discuss here would be all-pervasive even if our senses were as sharp as the Hubble Telescope and our minds were as logical as a computer. It arises from the way in which our ideas about reality connect with reality itself—how, in other words, we can create knowledge, and how we can fail to.

The trouble is that error is a subject where issues such as logical paradox, self-reference, and the inherent limits of reason rear their ugly heads in practical situations, and bite.

Paradoxes seem to appear when one considers the implications of one’s own fallibility: A fallibilist cannot claim to be infallible even about fallibilism itself. And so, one is forced to doubt that fallibilism is universally true. Which is the same as wondering whether one might be somehow infallible—at least about some things. For instance, can it be true that absolutely anything that you think is true, no matter how certain you are, might be false?

What? How might we be mistaken that two plus two is four? Or about other matters of pure logic? That stubbing one’s toe hurts? That there is a force of gravity pulling us to earth? Or that, as the philosopher René Descartes argued, “I think, therefore I am”?

When fallibilism starts to seem paradoxical, the mistakes begin. We are inclined to seek foundations—solid ground in the vast quicksand of human opinion—on which one can try to base everything else. Throughout the ages, the false authority of experience and the false reassurance of probability have been mistaken for such foundations: “No, we’re not always right,” your parents tell you, “just usually.” They have been on earth longer and think they have seen this situation before. But since that is an argument for “therefore you should always do as we say,” it is functionally a claim of infallibility after all. Moreover, look more closely: It claims literal infallibility too. Can anyone be infallibly right about the probability that they are right?

But wait. Have we now gotten lost in the paradox of being wrong about being wrong? There really is such a thing as existing knowledge—including a vast amount of useful, salutary truth. Parents really do know more than children about everyday dangers; your physician does know more than a passing hobo about your illness. Although, certainly, all concerned are fallible—they could be wrong—isn’t it rational to play the odds? To defer to the opinion of the experts who have a lot more knowledge about the matter? In other words, isn’t it better to act as if one considered them infallible about it, even though they are not?

No. That is not only an irrational answer but, catastrophically, the wrong question. I’ll return to it below. But first, consider infallibility itself.

Ascribing a sphere of infallibility to a parent or expert has the same logic as the Roman Catholic Church’s doctrine about the pope: It likewise considers him infallible only under certain narrowly-defined circumstances, called ex cathedra (metaphorically “from the throne”). So, consider this thought experiment: You seriously believe in papal infallibility. One day, an atheist friend gleefully tells you that the pope has said something which, after due consideration, you decide must be false: “There is no force of gravity.” Immediately, it becomes vital for you to know whether the pope declared this ex cathedra. For if he did, you would have to accept that you are mistaken about gravity, and act accordingly, even if you never managed to understand the mechanics of how that might be so. Because for you, ideas are about something—important precisely because they have consequences for how you think, feel, and act. And so you would have to drop some assumptions that you hitherto considered true incontrovertibly—or even infallibly.

Furthermore, one cannot seriously believe that the pope is infallible while also believing any rival religion, or atheism. So the implications of papal infallibility, even more than parental infallibility, are sweeping. Despite its narrow nominal scope, it is functionally equivalent to the entire gamut of Roman Catholic doctrine. But there is another class of implications—even more sweeping—in the opposite direction.

Consider the steps you are obliged to follow, from hearing of an ex cathedra declaration to believing its content.

A passing hobo tells you that he saw the pope making the declaration ex cathedra. Do you therefore accept that there is no force of gravity? Obviously not: That would involve assuming that the hobo was infallible—which would contradict the church’s teachings. And the same would hold even if an archbishop were to visit you and swear that he had witnessed it too, and stated his expert opinion that it met the requirements for being ex cathedra. Since the doctrine does not ascribe infallibility to archbishops, you would still not be required to accept the claim about gravity. Thus the doctrine of infallibility has made you take the fallibility of archbishops more seriously than you otherwise might. Even if the pope himself were to swear that his claim about gravity was strictly ex cathedra, you would not be forced, by your faith, to believe it. The doctrine of papal infallibility does not say that the reminiscences of a pope are infallible—unless they are ex cathedra reminiscences.

So your very faith in papal infallibility has led you to within touching distance of one of the cornerstones of scientific rationality: nullius in verba—“take no one’s word for it”—the motto of the Royal Society.

But now, what if you personally witnessed the ex cathedra statement?

So, there you were, visiting the Vatican and you took a wrong turn and found yourself witnessing the pope as he solemnly declared that there is no force of gravity. You happened to have purchased, from the souvenir shop, a checklist of the official requirements for a declaration to count as ex cathedra, and you took the trouble to verify that each one was met. None of this constitutes direct observation of what you need to know. Did you observe infallibly that it was the pope? Did you do a DNA test? Can you be certain that souvenir checklists never contain typos? And how is your church Latin? Was your translation of the crucial phrase “no force of gravity” infallible? Have you never mistranslated anything?

The fact is, there’s nothing infallible about “direct experience” either. Indeed, experience is never direct. It is a sort of virtual reality, created by our brains using sketchy and flawed sensory clues, given substance only by fallible expectations, explanations, and interpretations. Those can easily be more mistaken than the testimony of the passing hobo. If you doubt this, look at the work of psychologists Christopher Chabris and Daniel Simons, and verify by direct experience the fallibility of your own direct experience. Furthermore, the idea that your reminiscences are infallible is also heresy by the very doctrine that you are faithful to.

I’ll tell you what really happened. You witnessed a dress rehearsal. The real ex cathedra ceremony was on the following day. In order not to make the declaration a day early, they substituted for the real text (which was about some arcane theological issue, not gravity) a lorem-ipsum-type placeholder that they deemed so absurd that any serious listener would immediately realize that that’s what it was.

And indeed, you did realize this; and as a result, you reinterpreted your “direct experience,” which was identical to that of witnessing an ex cathedra declaration, as not being one. Precisely by reasoning that the content of the declaration was absurd, you concluded that you didn’t have to believe it. Which is also what you would have done if you hadn’t believed the infallibility doctrine.

You remain a believer, serious about giving your faith absolute priority over your own “unaided” reason (as reason is called in these contexts). But that very seriousness has forced you to decide first on the substance of the issue, using reason, and only then whether to defer to the infallible authority. This is neither fluke nor paradox. It is simply that if you take ideas seriously, there is no escape, even in dogma and faith, from the obligation to use reason and to give it priority over dogma, faith, and obedience.

The real pope is unlikely to make an ex cathedra statement about gravity, and therefore you may be lucky enough never to encounter this particular case of the dilemma. Also, the real pope doesn’t just pull ex cathedra statements out of a hat. They’re hammered out by a team of expert advisors trying their best to weed out mistakes, a process structurally not unlike peer review. But if your faith in papal infallibility depends on reassuring yourself of things like that, then that just goes to show that for you, reason takes priority over faith.

It is hard to contain reason within bounds. If you take your faith sufficiently seriously you may realize that it is not only the printers who are fallible in stating the rules for ex cathedra, but also the committee that wrote down those rules. And then that nothing can infallibly tell you what is infallible, nor what is probable. It is precisely because you, being fallible and having no infallible access to the infallible authority, no infallible way of interpreting what the authority means, and no infallible means of identifying an infallible authority in the first place, that infallibility cannot help you before reason has had its say.

A related useful thing that faith tells you, if you take it seriously enough, is that the great majority of people who believe something on faith, in fact believe falsehoods. Hence, faith is insufficient for true belief. As the Nobel-Prize-winning biologist Peter Medawar said: “the intensity of the conviction that a hypothesis is true has no bearing on whether it is true or not.”

You know that Medawar’s advice holds for all ideas, not just scientific ones, and, by the same argument, to all the other diverse things that are held up as infallible (or probable) touchstones of truth: holy books; the evidence of the senses; statements about who is probably right; even true love.

How should this no-exceptions fallibilism play out when the physician suggests a treatment? The right question is not “who is more likely to be right, the physician or I?” but “has this idea been judged rationally, by its content?” Which means, in particular, has it been subjected to sufficiently severe attempts to detect and eliminate errors—both by explanatory argument and by rigorous experiment? If you think it has, then your opinion and the physician’s should become the same, and the issue of deference should not arise, nor should the need for anyone to claim effective infallibility.

On the other hand, if you suspect that the physician has not given enough thought to some feature that makes your case unusual, it would be irrational to defer. The physician’s greater knowledge is irrelevant until you are satisfied with the way that idea has been taken into account. And whether the idea was originally suggested to you by a passing hobo or a physicist makes no difference, either.

This logic of fallibility, discovered and rediscovered from time to time, has had profound salutary effects in the history of ideas. Whenever anything demands blind obedience, its ideology contains a claim of infallibility somewhere; but wherever someone believes seriously enough in that infallibility, they rediscover the need for reason to identify and correctly interpret the infallible source. Thus the sages of ancient Judaism were led, by the assumption of the Bible’s infallibility, to develop their tradition of critical discussion. And in an apparently remote application of the same logic, the British constitutional doctrine of “parliamentary sovereignty” was used by 20th-century judges such as Lord Denning to develop an institution of judicial review similar to that which, in the United States, had grown out of the opposite doctrine of “separation of powers.”

Fallibilism has practical consequences for the methodology and administration of science, and in government, law, education, and every aspect of public life. The philosopher Karl Popper elaborated on many of these. He wrote:

The question about the sources of our knowledge … has always been asked in the spirit of: ‘What are the best sources of our knowledge—the most reliable ones, those which will not lead us into error, and those to which we can and must turn, in case of doubt, as the last court of appeal?’ I propose to assume, instead, that no such ideal sources exist—no more than ideal rulers—and that all ‘sources’ are liable to lead us into error at times. And I propose to replace, therefore, the question of the sources of our knowledge by the entirely different question: ‘How can we hope to detect and eliminate error?’

It’s all about error. We used to think that there was a way to organize ourselves that would minimize errors. This is an infallibilist chimera that has been part of every tyranny since time immemorial, from the “divine right of kings” to centralized economic planning. And it is implemented by many patterns of thought that protect misconceptions in individual minds, making someone blind to evidence that he isn’t Napoleon, or making the scientific crank reinterpret peer review as a conspiracy to keep falsehoods in place.

Popper’s answer is: We can hope to detect and eliminate error if we set up traditions of criticism—substantive criticism, directed at the content of ideas, not their sources, and directed at whether they solve the problems that they purport to solve. Here is another apparent paradox, for a tradition is a set of ideas that stay the same, while criticism is an attempt to change ideas. But there is no contradiction. Our systems of checks and balances are steeped in traditions—such as freedom of speech and of the press, elections, and parliamentary procedures, the values behind concepts of contract and of tort—that survive not because they are deferred to but precisely because they are not: They themselves are continually criticized, and either survive criticism (which allows them to be adopted without deference) or are improved (for example, when the franchise is extended, or slavery abolished). Democracy, in this conception, is not a system for enforcing obedience to the authority of the majority. In the bigger picture, it is a mechanism for promoting the creation of consent, by creating objectively better ideas, by eliminating errors from existing ones.

“Our whole problem,” said the physicist John Wheeler, “is to make the mistakes as fast as possible.” This liberating thought is more obviously true in theoretical physics than in situations where mistakes hurt. A mistake in a military operation, or a surgical operation, can kill. But that only means that whenever possible we should make the mistakes in theory, or in the laboratory; we should “let our theories die in our place,” as Popper put it. But when the enemy is at the gates, or the patient is dying, one cannot confine oneself to theory. We should abjure the traditional totalitarian assumption, still lurking in almost every educational system, that every mistake is the result of wrongdoing or stupidity. For that implies that everyone other than the stupid and the wrongdoers is infallible. Headline writers should not call every failed military strike “botched;” courts should not call every medical tragedy malpractice, even if it’s true that they “shouldn’t have happened” in the sense that lessons can be learned to prevent them from happening again. “We are all alike,” as Popper remarked, “in our infinite ignorance.” And this is a good and hopeful thing, for it allows for a future of unbounded improvement.

Fallibilism, correctly understood, implies the possibility, not the impossibility, of knowledge, because the very concept of error, if taken seriously, implies that truth exists and can be found. The inherent limitation on human reason, that it can never find solid foundations for ideas, does not constitute any sort of limit on the creation of objective knowledge nor, therefore, on progress. The absence of foundation, whether infallible or probable, is no loss to anyone except tyrants and charlatans, because what the rest of us want from ideas is their content, not their provenance: If your disease has been cured by medical science, and you then become aware that science never proves anything but only disproves theories (and then only tentatively), you do not respond “oh dear, I’ll just have to die, then.”

The theory of knowledge is a tightrope that is the only path from A to B, with a long, hard drop for anyone who steps off on one side into “knowledge is impossible, progress is an illusion” or on the other side into “I must be right, or at least probably right.” Indeed, infallibilism and nihilism are twins. Both fail to understand that mistakes are not only inevitable, they are correctable (fallibly). Which is why they both abhor institutions of substantive criticism and error correction, and denigrate rational thought as useless or fraudulent. They both justify the same tyrannies. They both justify each other.

I must now apologize for trying to trick you earlier: All the ideas that I suggested we might know infallibly are in fact falsehoods. “Two plus two” of course isn’t “four” as you’d discover if you wrote “2+2” in an arithmetic test when asked to add two and two. If we were infallible about matters of pure logic, no one would ever fail a logic test either. Stubbing your toe does not always hurt if you are focused on some overriding priority like rescuing a comrade in battle. And as for knowing that “I” exist because I think—note that your knowledge that you think is only a memory of what you did think, a second or so ago, and that can easily be a false memory. (For discussions of some fascinating experiments demonstrating this, see Daniel Dennett’s book Brainstorms.) Moreover, if you think you are Napoleon, the person you think must exist because you think, doesn’t exist.

And the general theory of relativity denies that gravity exerts a force on falling objects. The pope would actually be on firm ground if he were to concur with that ex cathedra. Now, are you going to defer to my authority as a physicist about that? Or decide that modern physics is a sham? Or are you going to decide according to whether that claim really has survived all rational attempts to refute it?

About David Deutsch

David Deutsch, internationally acclaimed for his seminal publications on quantum computation, is a member of the Centre for Quantum Computation at the Clarendon Laboratory, Oxford University.

2013-05-08 Closer To TruthCan Science Provide Ultimate Answers

00:06:57 - 00:07:17

That’s exactly right. Science has limitations, reason however does not and there will be ultimate limits of science beyond which progress will only ever be made with philosophy broadly speaking. We don’t know what those are, they are almost certainly beyond the current limits of science.

2013-05-08 Closer To TruthMany Worlds of Quantum Theory

00:07:38 - 00:08:37

One has to understand it in this way. Well, first of all, yes, they have existed since the Big Bang. There was an older version of the quantum multiverse in which they only appeared as branches, but that has now been abandoned. And in fact, we now think of it in terms of happening for all time. But remember that there’s a lot more in the multiverse than just universes. And in fact, universes are an emergent property of the multiverse. So some types of phenomena like human thought and the outcomes of scientific experiments are expressible in terms of universes. But a lot of other phenomena such as quantum computers or the interior of molecules are not. And for those, it’s the multiverse rather than individual universes.

2013-05-08 Closer To TruthIs the Cosmos a Computer

00:05:35 - 00:05:43

Because almost all ways that the laws of physics could be do not have the property that a universal computer could exist.

2013-05-08 Closer To TruthWhat Does Quantum Theory Mean

00:07:54 - 00:08:11

They’re all existing on the same basis as each other. None of them is privileged relative to the other. As you rightly say, the flow of time is an emergent property of this thing, and we can’t think of the multiverse as being in time. It’s more that time is a feature of the multiverse.

2013-05-08 Closer To TruthWhat Is Ultimate Reality

00:08:28 - 00:08:55

Well I think it has made me think of the world in a much more unified way. And that’s why I eventually came around to writing the second book, which is applying this fabric of reality to various issues which on the face of it, like the four strands themselves, don’t look as though they have anything to do with each other or with the four strands. And yet the closer you look, the more integrated they are and the more they do have to do with each other.

2013-05-08 Closer To TruthWhy Is the Quantum So Strange

00:07:46 - 00:08:43

Yes, or even our mobile phones. In other words, a very, very weak, comparatively weak quantum computer could perform more computations simultaneously than could be performed by the entire visible universe if it was all made into computers. In fact, when I say more, that’s an understatement, exponentially more than that. But only for certain types of computation, and that’s a token of the fact that the whole notion of computation is different in quantum computers. It’s not that, like with all classical computers, you can say that one computer is ten times as fast as the other. With quantum computers, they are vastly faster than classical computers for some computations and the same for others. And interestingly, they’re not slower for any computations because a quantum computer, among its abilities, is to simulate a classical computer.

2013-05-08 Closer To TruthWhich Laws of Nature Are Fundamental

00:13:27 - 00:13:28

Exactly.

2012-10-22 EdgeConstructor Theory and the Real Reality

00:43:33 - 00:46:48

So one of the central uh philosophical motivations for why I do fundamental physics is that I am interested in what the world is like that is uh not just the world of our observations what we see but the invisible world the invisible processes and objects that bring about the visible because the visible is only the tiny superficial and parochial sheen on top of the real reality and the amazing thing about the world and our place in it is that we can discover the real reality we can discover what is at the center of stars even though we’ve never been there we can find out that those cold tiny objects in the sky that we call stars are actually million kilometer white hot gaseous spheres now they don’t look like that they look like cold dots but we know different we know that the invisible reality is there giving rise to our visible perceptions now that science has to be about that has been for many decades a minority and unpopular view among philosophers and to a great extent regrettably even among scientists they have taken the view that science just because it is characterized by experimental tests has to be only about experimental tests but that’s a trap that means if that was so it would mean that science is only about humans and not even about humans but about human experience only it’s solipsism it’s purporting to have a rigorous objective world view that only observations count but it ends up by its own inexorable logic as saying that only human experience is real which is solipsism so I think it’s important to regard science not as an enterprise for the purpose of making predictions but as an enterprise for the purpose of discovering what the world is really like what is really there how it behaves and why which is tested by observation but it is absolutely amazing that the tiny little parochial and weak and error-prone access that we have to observations is capable of testing theories and knowledge of the whole of reality that has tremendous reach far beyond our experience and yet we know about it that’s the amazing thing about science that’s the aspect of science that I want to pursue

Read the essay at Aeon Source collection

2012-10-03 AeonCreative Blocks

Creative Blocks

By David Deutsch · Edited by Ed Lake · 3 October 2012

The very laws of physics imply that artificial intelligence must be possible. What’s holding us up?

David Deutsch is a physicist at the University of Oxford and a fellow of the Royal Society. His latest book is The Beginning of Infinity.

It is uncontroversial that the human brain has capabilities that are, in some respects, far superior to those of all other known objects in the cosmos. It is the only kind of object capable of understanding that the cosmos is even there, or why there are infinitely many prime numbers, or that apples fall because of the curvature of space-time, or that obeying its own inborn instincts can be morally wrong, or that it itself exists. Nor are its unique abilities confined to such cerebral matters. The cold, physical fact is that it is the only kind of object that can propel itself into space and back without harm, or predict and prevent a meteor strike on itself, or cool objects to a billionth of a degree above absolute zero, or detect others of its kind across galactic distances.

But no brain on Earth is yet close to knowing what brains do in order to achieve any of that functionality. The enterprise of achieving it artificially — the field of ‘artificial general intelligence’ or AGI — has made no progress whatever during the entire six decades of its existence.

Why? Because, as an unknown sage once remarked, ‘it ain’t what we don’t know that causes trouble, it’s what we know for sure that just ain’t so’ (and if you know that sage was Mark Twain, then what you know ain’t so either). I cannot think of any other significant field of knowledge in which the prevailing wisdom, not only in society at large but also among experts, is so beset with entrenched, overlapping, fundamental errors. Yet it has also been one of the most self-confident fields in prophesying that it will soon achieve the ultimate breakthrough.

Despite this long record of failure, AGI must be possible. And that is because of a deep property of the laws of physics, namely the universality of computation. This entails that everything that the laws of physics require a physical object to do can, in principle, be emulated in arbitrarily fine detail by some program on a general-purpose computer, provided it is given enough time and memory. The first people to guess this and to grapple with its ramifications were the 19th-century mathematician Charles Babbage and his assistant Ada, Countess of Lovelace. It remained a guess until the 1980s, when I proved it using the quantum theory of computation.

Babbage came upon universality from an unpromising direction. He had been much exercised by the fact that tables of mathematical functions (such as logarithms and cosines) contained mistakes. At the time they were compiled by armies of clerks, known as ‘computers’, which is the origin of the word. Being human, the computers were fallible. There were elaborate systems of error correction, but even proofreading for typographical errors was a nightmare. Such errors were not merely inconvenient and expensive: they could cost lives. For instance, the tables were extensively used in navigation. So, Babbage designed a mechanical calculator, which he called the Difference Engine. It would be programmed by initialising certain cogs. The mechanism would drive a printer, in order to automate the production of the tables. That would bring the error rate down to negligible levels, to the eternal benefit of humankind.

Unfortunately, Babbage’s project-management skills were so poor that despite spending vast amounts of his own and the British government’s money, he never managed to get the machine built. Yet his design was sound, and has since been implemented by a team led by the engineer Doron Swade at the Science Museum in London.

Here was a cognitive task that only humans had been able to perform. Nothing else in the known universe even came close to matching them, but the Difference Engine would perform better than the best humans. And therefore, even at that faltering, embryonic stage of the history of automated computation — before Babbage had considered anything like AGI — we can see the seeds of a philosophical puzzle that is controversial to this day: what exactly is the difference between what the human ‘computers’ were doing and what the Difference Engine could do? What type of cognitive task, if any, could either type of entity perform that the other could not in principle perform too?

One immediate difference between them was that the sequence of elementary steps (of counting, adding, multiplying by 10, and so on) that the Difference Engine used to compute a given function did not mirror those of the human ‘computers’. That is to say, they used different algorithms. In itself, that is not a fundamental difference: the Difference Engine could have been modified with additional gears and levers to mimic the humans’ algorithm exactly. Yet that would have achieved nothing except an increase in the error rate, due to increased numbers of glitches in the more complex machinery. Similarly, the humans, given different instructions but no hardware changes, would have been capable of emulating every detail of the Difference Engine’s method — and doing so would have been just as perverse. It would not have copied the Engine’s main advantage, its accuracy, which was due to hardware not software. It would only have made an arduous, boring task even more arduous and boring, which would have made errors more likely, not less.

For humans, that difference in outcomes — the different error rate — would have been caused by the fact that computing exactly the same table with two different algorithms felt different. But it would not have felt different to the Difference Engine. It had no feelings. Experiencing boredom was one of many cognitive tasks at which the Difference Engine would have been hopelessly inferior to humans. Nor was it capable of knowing or proving, as Babbage did, that the two algorithms would give identical results if executed accurately. Still less was it capable of wanting, as he did, to benefit seafarers and humankind in general. In fact, its repertoire was confined to evaluating a tiny class of specialised mathematical functions (basically, power series in a single variable).

Thinking about how he could enlarge that repertoire, Babbage first realised that the programming phase of the Engine’s operation could itself be automated: the initial settings of the cogs could be encoded on punched cards. And then he had an epoch-making insight. The Engine could be adapted to punch new cards and store them for its own later use, making what we today call a computer memory. If it could run for long enough — powered, as he envisaged, by a steam engine — and had an unlimited supply of blank cards, its repertoire would jump from that tiny class of mathematical functions to the set of all computations that can possibly be performed by any physical object. That’s universality.

Babbage called this improved machine the Analytical Engine. He and Lovelace understood that its universality would give it revolutionary potential to improve almost every scientific endeavour and manufacturing process, as well as everyday life. They showed remarkable foresight about specific applications. They knew that it could be programmed to do algebra, play chess, compose music, process images and so on. Unlike the Difference Engine, it could be programmed to use exactly the same method as humans used to make those tables. And prove that the two methods must give the same answers, and do the same error-checking and proofreading (using, say, optical character recognition) as well.

But could the Analytical Engine feel the same boredom? Could it feel anything? Could it want to better the lot of humankind (or of Analytical Enginekind)? Could it disagree with its programmer about its programming? Here is where Babbage and Lovelace’s insight failed them. They thought that some cognitive functions of the human brain were beyond the reach of computational universality. As Lovelace wrote, ‘The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform. It can follow analysis; but it has no power of anticipating any analytical relations or truths.’

And yet ‘originating things’, ‘following analysis’, and ‘anticipating analytical relations and truths’ are all behaviours of brains and, therefore, of the atoms of which brains are composed. Such behaviours obey the laws of physics. So it follows inexorably from universality that, with the right program, an Analytical Engine would undergo them too, atom by atom and step by step. True, the atoms in the brain would be emulated by metal cogs and levers rather than organic material — but in the present context, inferring anything substantive from that distinction would be rank racism.

Despite their best efforts, Babbage and Lovelace failed almost entirely to convey their enthusiasm about the Analytical Engine to others. In one of the great might-have-beens of history, the idea of a universal computer languished on the back burner of human thought. There it remained until the 20th century, when Alan Turing arrived with a spectacular series of intellectual tours de force, laying the foundations of the classical theory of computation, establishing the limits of computability, participating in the building of the first universal classical computer and, by helping to crack the Enigma code, contributing to the Allied victory in the Second World War.

Turing fully understood universality. In his 1950 paper ‘Computing Machinery and Intelligence’, he used it to sweep away what he called ‘Lady Lovelace’s objection’, and every other objection both reasonable and unreasonable. He concluded that a computer program whose repertoire included all the distinctive attributes of the human brain — feelings, free will, consciousness and all — could be written.

This astounding claim split the intellectual world into two camps, one insisting that AGI was none the less impossible, and the other that it was imminent. Both were mistaken. The first, initially predominant, camp cited a plethora of reasons ranging from the supernatural to the incoherent. All shared the basic mistake that they did not understand what computational universality implies about the physical world, and about human brains in particular.

But it is the other camp’s basic mistake that is responsible for the lack of progress. It was a failure to recognise that what distinguishes human brains from all other physical systems is qualitatively different from all other functionalities, and cannot be specified in the way that all other attributes of computer programs can be. It cannot be programmed by any of the techniques that suffice for writing any other type of program. Nor can it be achieved merely by improving their performance at tasks that they currently do perform, no matter by how much.

Why? I call the core functionality in question creativity: the ability to produce new explanations. For example, suppose that you want someone to write you a computer program to convert temperature measurements from Centigrade to Fahrenheit. Even the Difference Engine could have been programmed to do that. A universal computer like the Analytical Engine could achieve it in many more ways. To specify the functionality to the programmer, you might, for instance, provide a long list of all inputs that you might ever want to give it (say, all numbers from -89.2 to +57.8 in increments of 0.1) with the corresponding correct outputs, so that the program could work by looking up the answer in the list on each occasion. Alternatively, you might state an algorithm, such as ‘divide by five, multiply by nine, add 32 and round to the nearest 10th’. The point is that, however the program worked, you would consider it to meet your specification — to be a bona fide temperature converter — if, and only if, it always correctly converted whatever temperature you gave it, within the stated range.

Now imagine that you require a program with a more ambitious functionality: to address some outstanding problem in theoretical physics — say the nature of Dark Matter — with a new explanation that is plausible and rigorous enough to meet the criteria for publication in an academic journal.

Such a program would presumably be an AGI (and then some). But how would you specify its task to computer programmers? Never mind that it’s more complicated than temperature conversion: there’s a much more fundamental difficulty. Suppose you were somehow to give them a list, as with the temperature-conversion program, of explanations of Dark Matter that would be acceptable outputs of the program. If the program did output one of those explanations later, that would not constitute meeting your requirement to generate new explanations. For none of those explanations would be new: you would already have created them yourself in order to write the specification. So, in this case, and actually in all other cases of programming genuine AGI, only an algorithm with the right functionality would suffice. But writing that algorithm (without first making new discoveries in physics and hiding them in the program) is exactly what you wanted the programmers to do!

Traditionally, discussions of AGI have evaded that issue by imagining only a test of the program, not its specification — the traditional test having been proposed by Turing himself. It was that (human) judges be unable to detect whether the program is human or not, when interacting with it via some purely textual medium so that only its cognitive abilities would affect the outcome. But that test, being purely behavioural, gives no clue for how to meet the criterion. Nor can it be met by the technique of ‘evolutionary algorithms’: the Turing test cannot itself be automated without first knowing how to write an AGI program, since the ‘judges’ of a program need to have the target ability themselves. (For how I think biological evolution gave us the ability in the first place, see my book The Beginning of Infinity.)

And in any case, AGI cannot possibly be defined purely behaviourally. In the classic ‘brain in a vat’ thought experiment, the brain, when temporarily disconnected from its input and output channels, is thinking, feeling, creating explanations — it has all the cognitive attributes of an AGI. So the relevant attributes of an AGI program do not consist only of the relationships between its inputs and outputs.

The upshot is that, unlike any functionality that has ever been programmed to date, this one can be achieved neither by a specification nor a test of the outputs. What is needed is nothing less than a breakthrough in philosophy, a new epistemological theory that explains how brains create explanatory knowledge and hence defines, in principle, without ever running them as programs, which algorithms possess that functionality and which do not.

Such a theory is beyond present-day knowledge. What we do know about epistemology implies that any approach not directed towards that philosophical breakthrough must be futile. Unfortunately, what we know about epistemology is contained largely in the work of the philosopher Karl Popper and is almost universally underrated and misunderstood (even — or perhaps especially — by philosophers). For example, it is still taken for granted by almost every authority that knowledge consists of justified, true beliefs and that, therefore, an AGI’s thinking must include some process during which it justifies some of its theories as true, or probable, while rejecting others as false or improbable. But an AGI programmer needs to know where the theories come from in the first place. The prevailing misconception is that by assuming that ‘the future will be like the past’, it can ‘derive’ (or ‘extrapolate’ or ‘generalise’) theories from repeated experiences by an alleged process called ‘induction’. But that is impossible. I myself remember, for example, observing on thousands of consecutive occasions that on calendars the first two digits of the year were ‘19’. I never observed a single exception until, one day, they started being ‘20’. Not only was I not surprised, I fully expected that there would be an interval of 17,000 years until the next such ‘19’, a period that neither I nor any other human being had previously experienced even once.

How could I have ‘extrapolated’ that there would be such a sharp departure from an unbroken pattern of experiences, and that a never-yet-observed process (the 17,000-year interval) would follow? Because it is simply not true that knowledge comes from extrapolating repeated observations. Nor is it true that ‘the future is like the past’, in any sense that one could detect in advance without already knowing the explanation. The future is actually unlike the past in most ways. Of course, given the explanation, those drastic ‘changes’ in the earlier pattern of 19s are straightforwardly understood as being due to an invariant underlying pattern or law. But the explanation always comes first. Without that, any continuation of any sequence constitutes ‘the same thing happening again’ under some explanation.

So, why is it still conventional wisdom that we get our theories by induction? For some reason, beyond the scope of this article, conventional wisdom adheres to a trope called the ‘problem of induction’, which asks: ‘How and why can induction nevertheless somehow be done, yielding justified true beliefs after all, despite being impossible and invalid respectively?’ Thanks to this trope, every disproof (such as that by Popper and David Miller back in 1988), rather than ending inductivism, simply causes the mainstream to marvel in even greater awe at the depth of the great ‘problem of induction’.

In regard to how the AGI problem is perceived, this has the catastrophic effect of simultaneously framing it as the ‘problem of induction’, and making that problem look easy, because it casts thinking as a process of predicting that future patterns of sensory experience will be like past ones. That looks like extrapolation — which computers already do all the time (once they are given a theory of what causes the data). But in reality, only a tiny component of thinking is about prediction at all, let alone prediction of our sensory experiences. We think about the world: not just the physical world but also worlds of abstractions such as right and wrong, beauty and ugliness, the infinite and the infinitesimal, causation, fiction, fears, and aspirations — and about thinking itself.

Now, the truth is that knowledge consists of conjectured explanations — guesses about what really is (or really should be, or might be) out there in all those worlds. Even in the hard sciences, these guesses have no foundations and don’t need justification. Why? Because genuine knowledge, though by definition it does contain truth, almost always contains error as well. So it is not ‘true’ in the sense studied in mathematics and logic. Thinking consists of criticising and correcting partially true guesses with the intention of locating and eliminating the errors and misconceptions in them, not generating or justifying extrapolations from sense data. And therefore, attempts to work towards creating an AGI that would do the latter are just as doomed as an attempt to bring life to Mars by praying for a Creation event to happen there.

Currently one of the most influential versions of the ‘induction’ approach to AGI (and to the philosophy of science) is Bayesianism, unfairly named after the 18th-century mathematician Thomas Bayes, who was quite innocent of the mistake. The doctrine assumes that minds work by assigning probabilities to their ideas and modifying those probabilities in the light of experience as a way of choosing how to act. This is especially perverse when it comes to an AGI’s values — the moral and aesthetic ideas that inform its choices and intentions — for it allows only a behaviouristic model of them, in which values that are ‘rewarded’ by ‘experience’ are ‘reinforced’ and come to dominate behaviour while those that are ‘punished’ by ‘experience’ are extinguished. As I argued above, that behaviourist, input-output model is appropriate for most computer programming other than AGI, but hopeless for AGI. It is ironic that mainstream psychology has largely renounced behaviourism, which has been recognised as both inadequate and inhuman, while computer science, thanks to philosophical misconceptions such as inductivism, still intends to manufacture human-type cognition on essentially behaviourist lines.

Furthermore, despite the above-mentioned enormous variety of things that we create explanations about, our core method of doing so, namely Popperian conjecture and criticism, has a single, unified, logic. Hence the term ‘general’ in AGI. A computer program either has that yet-to-be-fully-understood logic, in which case it can perform human-type thinking about anything, including its own thinking and how to improve it, or it doesn’t, in which case it is in no sense an AGI. Consequently, another hopeless approach to AGI is to start from existing knowledge of how to program specific tasks — such as playing chess, performing statistical analysis or searching databases — and then to try to improve those programs in the hope that this will somehow generate AGI as a side effect, as happened to Skynet in the Terminator films.

Nowadays, an accelerating stream of marvellous and useful functionalities for computers are coming into use, some of them sooner than had been foreseen even quite recently. But what is neither marvellous nor useful is the argument that often greets these developments, that they are reaching the frontiers of AGI. An especially severe outbreak of this occurred recently when a search engine called Watson, developed by IBM, defeated the best human player of a word-association database-searching game called Jeopardy. ‘Smartest machine on Earth’, the PBS documentary series Nova called it, and characterised its function as ‘mimicking the human thought process with software.’ But that is precisely what it does not do.

The thing is, playing Jeopardy — like every one of the computational functionalities at which we rightly marvel today — is firmly among the functionalities that can be specified in the standard, behaviourist way that I discussed above. No Jeopardy answer will ever be published in a journal of new discoveries. The fact that humans perform that task less well by using creativity to generate the underlying guesses is not a sign that the program has near-human cognitive abilities. The exact opposite is true, for the two methods are utterly different from the ground up. Likewise, when a computer program beats a grandmaster at chess, the two are not using even remotely similar algorithms. The grandmaster can explain why it seemed worth sacrificing the knight for strategic advantage and can write an exciting book on the subject. The program can only prove that the sacrifice does not force a checkmate, and cannot write a book because it has no clue even what the objective of a chess game is. Programming AGI is not the same sort of problem as programming Jeopardy or chess.

An AGI is qualitatively, not quantitatively, different from all other computer programs. The Skynet misconception likewise informs the hope that AGI is merely an emergent property of complexity, or that increased computer power will bring it forth (as if someone had already written an AGI program but it takes a year to utter each sentence). It is behind the notion that the unique abilities of the brain are due to its ‘massive parallelism’ or to its neuronal architecture, two ideas that violate computational universality. Expecting to create an AGI without first understanding in detail how it works is like expecting skyscrapers to learn to fly if we build them tall enough.

In 1950, Turing expected that by the year 2000, ‘one will be able to speak of machines thinking without expecting to be contradicted.’ In 1968, Arthur C. Clarke expected it by 2001. Yet today in 2012 no one is any better at programming an AGI than Turing himself would have been.

This does not surprise people in the first camp, the dwindling band of opponents of the very possibility of AGI. But for the people in the other camp (the AGI-is-imminent one) such a history of failure cries out to be explained — or, at least, to be rationalised away. And indeed, unfazed by the fact that they could never induce such rationalisations from experience as they expect their AGIs to do, they have thought of many.

The very term ‘AGI’ is an example of one. The field used to be called ‘AI’ — artificial intelligence. But ‘AI’ was gradually appropriated to describe all sorts of unrelated computer programs such as game players, search engines and chatbots, until the G for ‘general’ was added to make it possible to refer to the real thing again, but now with the implication that an AGI is just a smarter species of chatbot.

Another class of rationalisations runs along the general lines of: AGI isn’t that great anyway; existing software is already as smart or smarter, but in a non-human way, and we are too vain or too culturally biased to give it due credit. This gets some traction because it invokes the persistently popular irrationality of cultural relativism, and also the related trope that: ‘We humans pride ourselves on being the paragon of animals, but that pride is misplaced because they, too, have language, tools …

… And self-awareness.’

Remember the significance attributed to Skynet’s becoming ‘self-aware’? That’s just another philosophical misconception, sufficient in itself to block any viable approach to AGI. The fact is that present-day software developers could straightforwardly program a computer to have ‘self-awareness’ in the behavioural sense — for example, to pass the ‘mirror test’ of being able to use a mirror to infer facts about itself — if they wanted to. As far as I am aware, no one has done so, presumably because it is a fairly useless ability as well as a trivial one.

AGIs will indeed be capable of self-awareness — but that is because they will be General: they will be capable of awareness of every kind of deep and subtle thing, including their own selves. This does not mean that apes who pass the mirror test have any hint of the attributes of ‘general intelligence’ of which AGI would be an artificial version. Indeed, Richard Byrne’s wonderful research into gorilla memes has revealed how apes are able to learn useful behaviours from each other without ever understanding what they are for: the explanation of how ape cognition works really is behaviouristic.

Ironically, that group of rationalisations (AGI has already been done/is trivial/ exists in apes/is a cultural conceit) are mirror images of arguments that originated in the AGI-is-impossible camp. For every argument of the form ‘You can’t do AGI because you’ll never be able to program the human soul, because it’s supernatural’, the AGI-is-easy camp has the rationalisation, ‘If you think that human cognition is qualitatively different from that of apes, you must believe in a supernatural soul.’

‘Anything we don’t yet know how to program is called human intelligence,’ is another such rationalisation. It is the mirror image of the argument advanced by the philosopher John Searle (from the ‘impossible’ camp), who has pointed out that before computers existed, steam engines and later telegraph systems were used as metaphors for how the human mind must work. Searle argues that the hope for AGI rests on a similarly insubstantial metaphor, namely that the mind is ‘essentially’ a computer program. But that’s not a metaphor: the universality of computation follows from the known laws of physics.

Some, such as the mathematician Roger Penrose, have suggested that the brain uses quantum computation, or even hyper-quantum computation relying on as-yet-unknown physics beyond quantum theory, and that this explains the failure to create AGI on existing computers. To explain why I, and most researchers in the quantum theory of computation, disagree that this is a plausible source of the human brain’s unique functionality is beyond the scope of this essay. (If you want to know more, read Litt et al’s 2006 paper ‘Is the Brain a Quantum Computer?’, published in the journal Cognitive Science.)

That AGIs are people has been implicit in the very concept from the outset. If there were a program that lacked even a single cognitive ability that is characteristic of people, then by definition it would not qualify as an AGI. Using non-cognitive attributes (such as percentage carbon content) to define personhood would, again, be racist. But the fact that the ability to create new explanations is the unique, morally and intellectually significant functionality of people (humans and AGIs), and that they achieve this functionality by conjecture and criticism, changes everything.

Currently, personhood is often treated symbolically rather than factually — as an honorific, a promise to pretend that an entity (an ape, a foetus, a corporation) is a person in order to achieve some philosophical or practical aim. This isn’t good. Never mind the terminology; change it if you like, and there are indeed reasons for treating various entities with respect, protecting them from harm and so on. All the same, the distinction between actual people, defined by that objective criterion, and other entities has enormous moral and practical significance, and is going to become vital to the functioning of a civilisation that includes AGIs.

For example, the mere fact that it is not the computer but the running program that is a person, raises unsolved philosophical problems that will become practical, political controversies as soon as AGIs exist. Once an AGI program is running in a computer, to deprive it of that computer would be murder (or at least false imprisonment or slavery, as the case may be), just like depriving a human mind of its body. But unlike a human body, an AGI program can be copied into multiple computers at the touch of a button. Are those programs, while they are still executing identical steps (ie before they have become differentiated due to random choices or different experiences), the same person or many different people? Do they get one vote, or many? Is deleting one of them murder, or a minor assault? And if some rogue programmer, perhaps illegally, creates billions of different AGI people, either on one computer or on many, what happens next? They are still people, with rights. Do they all get the vote?

Furthermore, in regard to AGIs, like any other entities with creativity, we have to forget almost all existing connotations of the word ‘programming’. To treat AGIs like any other computer programs would constitute brainwashing, slavery, and tyranny. And cruelty to children, too, for ‘programming’ an already-running AGI, unlike all other programming, constitutes education. And it constitutes debate, moral as well as factual. To ignore the rights and personhood of AGIs would not only be the epitome of evil, but also a recipe for disaster: creative beings cannot be enslaved forever.

Some people are wondering whether we should welcome our new robot overlords. Some hope to learn how we can rig their programming to make them constitutionally unable to harm humans (as in Isaac Asimov’s ‘laws of robotics’), or to prevent them from acquiring the theory that the universe should be converted into paper clips (as imagined by Nick Bostrom). None of these are the real problem. It has always been the case that a single exceptionally creative person can be thousands of times as productive — economically, intellectually or whatever — as most people; and that such a person could do enormous harm were he to turn his powers to evil instead of good.

These phenomena have nothing to do with AGIs. The battle between good and evil ideas is as old as our species and will continue regardless of the hardware on which it is running. The issue is: we want the intelligences with (morally) good ideas always to defeat the evil intelligences, biological and artificial; but we are fallible, and our own conception of ‘good’ needs continual improvement. How should society be organised so as to promote that improvement? ‘Enslave all intelligence’ would be a catastrophically wrong answer, and ‘enslave all intelligence that doesn’t look like us’ would not be much better.

One implication is that we must stop regarding education (of humans or AGIs alike) as instruction — as a means of transmitting existing knowledge unaltered, and causing existing values to be enacted obediently. As Popper wrote (in the context of scientific discovery, but it applies equally to the programming of AGIs and the education of children): ‘there is no such thing as instruction from without … We do not discover new facts or new effects by copying them, or by inferring them inductively from observation, or by any other method of instruction by the environment. We use, rather, the method of trial and the elimination of error.’ That is to say, conjecture and criticism. Learning must be something that newly created intelligences do, and control, for themselves.

I do not highlight all these philosophical issues because I fear that AGIs will be invented before we have developed the philosophical sophistication to understand them and to integrate them into civilisation. It is for almost the opposite reason: I am convinced that the whole problem of developing AGIs is a matter of philosophy, not computer science or neurophysiology, and that the philosophical progress that is essential to their future integration is also a prerequisite for developing them in the first place.

The lack of progress in AGI is due to a severe logjam of misconceptions. Without Popperian epistemology, one cannot even begin to guess what detailed functionality must be achieved to make an AGI. And Popperian epistemology is not widely known, let alone understood well enough to be applied. Thinking of an AGI as a machine for translating experiences, rewards and punishments into ideas (or worse, just into behaviours) is like trying to cure infectious diseases by balancing bodily humours: futile because it is rooted in an archaic and wildly mistaken world view.

Without understanding that the functionality of an AGI is qualitatively different from that of any other kind of computer program, one is working in an entirely different field. If one works towards programs whose ‘thinking’ is constitutionally incapable of violating predetermined constraints, one is trying to engineer away the defining attribute of an intelligent being, of a person: namely creativity.

Clearing this logjam will not, by itself, provide the answer. Yet the answer, conceived in those terms, cannot be all that difficult. For yet another consequence of understanding that the target ability is qualitatively different is that, since humans have it and apes do not, the information for how to achieve it must be encoded in the relatively tiny number of differences between the DNA of humans and that of chimpanzees. So in one respect I can agree with the AGI-is-imminent camp: it is plausible that just a single idea stands between us and the breakthrough. But it will have to be one of the best ideas ever.

2011-11-22 TEDxBrusselsThe Unknowable and How to Prepare for It

Duration: 00:17:00

Transcript

David Deutsch

00:00:00 - 00:02:49

Thank you. We have a bit of feedback. No? Okay, good. Right. So what can we know about what this conference is calling a day in the deep future, 50 years hence. I guess we’re all curious about what life will be like by then, what new opportunities will have opened up for people, and what problems they’ll be thinking about, including maybe about their own future. Also, what we may be able to do today to improve the chances of the outcomes and try to prevent the less favorable ones. But before we can even think relevantly about any of that, there’s a more fundamental issue that we need to address first, namely, what kinds of things are knowable in advance and what is not. I want to tell you about a fundamental limit on the power of science and technology and reason to predict the future, and also about what we’re going to have to do about that. So, prophecy, fortune telling, has always been popular to date. People want to know whether their plans are going to succeed or fail, and there’s always been prophets claiming to be able to find out. All they really do is just either tell people, tell the punters what they wanted to hear and take their money, or they threaten them with doom unless they behave righteously in some sense. And either way, they’re just making stuff up. They have never had genuine knowledge of the phenomena that they purport to predict. And for most of history, nor did anyone else.

00:02:49 - 00:05:29

People just used wishful thinking or superstition, or at best, they used traditions and rules of thumb that had worked in the past, or were thought to have worked. Often they hadn’t. Often structures collapsed, crops failed, ships sank, people died of treatments that were supposed to be cures. Nowadays, such tragedies are much rarer because we have scientific knowledge which is predictive. That’s the defining characteristic of science. And that allows mistaken scientific theories to be corrected when they make false predictions, which allows for cumulative improvement in our ability to predict. And it’s not just cumulative. Progress is exponential. And that’s basically because scientific knowledge is not just testable, it’s explanatory. It derives its predictions from an understanding of how things actually work. So we don’t predict the maximum safe load of a bridge from past experience anymore. We predict it by understanding what makes structures in general stand up. And that allows for new designs of bridge that are not covered by any existing rules of thumb. And it also applies to new ways of creating knowledge, like new scientific instruments and computers and communication and sheer wealth. So one of the things that explanatory knowledge provides is wealth, which allows for the creation of new knowledge in the sense of increasing the repertoire of physical transformations that human beings are capable of causing or preventing also. And as this repertoire expands, and we increasingly cause the outcomes that we prefer, our ability to predict those outcomes, those same outcomes correspondingly increases.

00:05:29 - 00:07:42

Some things remain technically unpredictable because they’re random, or because they’re so complicated that they can’t be distinguished from random. But that’s no impediment to predictability in the practical sense, because if science predicts that something is random, we can play the odds. And again, technology allows us to change the odds in our favor. So I can, if there’s a random process such as weather, then I can use an umbrella not only to change the probability that I will get wet, but also to increase my probability of predicting whether I will get wet. And so the faster we gain new understanding and new technology, the faster our predictions also improve. And so you might expect that our ability to predict the future would be greater now than ever before in history. And that isn’t so, because there’s one thing in our future that nothing whatever is capable of predicting. And that is the future course of knowledge itself. A simple example, if there were a method of predicting today, some scientific truth that is only going to be discovered next year, then by using that method, we would have gained that knowledge today. And that’s a contradiction. The future is not random. The more we know, the more we are able to correct errors. And so the less random the future becomes, what we are going to discover is determined by what we shall want, and by what the laws of physics, the laws of nature allow, and by what ideas are true. And none of that is random. But none of that is knowable in advance either.

00:07:42 - 00:10:03

We can’t even predict for very long what future discoveries are even going to be about. The possibilities that are opened up by a new discovery are very often new discoveries in their own right. For instance, on this day, 50 years in the past, nobody had ever connected together two computers in different buildings. Nobody had a computer in their home. Nobody could operate a computer except a handful of specialists. And no one predicted the consequences of computer networking, the internet. By today, home computers would be a global network, which is a major channel of trade and communication and entertainment and education. And nor did they know when they connected those first networking cables together, that they were igniting a process which by today would be instrumental in the overthrow of dictators. Which brings me to what people will want. 50 years ago, some far sighted people realized that the new technology of jet airliners was soon going to make it normal for ordinary people to travel abroad for jobs and education and holidays. And they were right. That revolution happened. But they also thought that by today, all that jet travel would be supersonic. And that has not happened. People at some point decided that supersonic travel is morally unacceptable. And also the moon colony, the Mars expedition, nuclear power. By even the late 1970s, people no longer wanted what people had wanted by the early 60s. And because of that unpredictable change, their prediction of what our lives were going to be like was false.

00:10:03 - 00:12:04

That’s part of a wider impediment to prediction, namely, unforeseen problems, the fads and fallacies and blunders that are going to seem like a good idea at any time in the next 50 years. By definition, we can’t foretell now. So predicting our future is nothing like predicting the strength of a bridge. Every significant innovation has unpredictable effects, and they have knock on effects. And after a few steps of that, the consequences and their consequences come to be the major component of what is happening. And as knowledge grows faster, the time for that to happen becomes shorter and shorter. The growth of knowledge is the only impediment to our ability to predict the future. But in this respect, it’s decisive, it will impose an ever closer planning horizon, beyond which we are blind to the most important determinants of what is going to happen. So we face a paradox. The more we create knowledge, the less we know about our future. And yet, as ideas are transforming our lives ever faster, it becomes ever more necessary to plan and prepare for that effect. This is a new predicament for our species. Historically, knowledge was very sparse, it changed very rarely, and the little that people thought they knew was also mostly wrong. And so prediction of the type that I’ve been discussing was correspondingly easy. One could predict what life was going to be like in 50 years time, because it was going to be much like one’s own life.

00:12:04 - 00:14:20

Something like a plague might strike unpredictably, but the means of coping with plagues would be unchanged. There’d be a different king, but the role of a king would be unchanged. The laws of the land, people’s means of making a living, and the patterns of their personal lives, their values, their common sense, none of that would have changed. Our situation is profoundly different. All those things are already changing on a time scale much faster than a human lifetime. Now, mistakes are inevitable, and the exponential growth of knowledge causes an exponentially increasing rate of mistakes, unforeseen mistakes. And so we need to prepare for contingencies that we don’t know about today, because they’re going to be dominated by knowledge that we do not yet have, or by ignorance that we do not yet know is relevant. How do we do that? Well, fortunately, the answer to that is knowable. The only thing that can solve this problem is the same thing that caused it in the first place, the power of explanatory knowledge, because some explanations do reach beyond the planning horizon. One example is the argument that I’ve just been giving, that the planning horizon has got to come ever closer. And another is that although the content of our explanations will be changing, the nature of explanation itself will not. Even beyond the planning horizon, problems will be soluble by conjecture and criticism, creative thought, error correction, and that will itself give rise to new problems and so on. So therefore, consider what kind of knowledge is needed to cope with unforeseen dangers and disasters, both unintended and also the malevolent use of technology by criminals or enemies of civilization.

00:14:20 - 00:16:20

The only possible defense against that, against the unforeseen, is general purpose knowledge. It’s like to prepare for an epidemic of a known disease, you can stockpile known cures and known preventive means, but to cope with unknown diseases, one needs general explanations of how diseases and healthy processes work. The same holds for all unknown dangers and unknown opportunities. We can only prepare for them with explanatory knowledge that was not created for the specific purpose that it will eventually be used for, but for its own sake. We also need the means of implementing unforeseen solutions. That’s technological knowledge and again, as powerful and general as possible and wealth, the ability to cause transformations as large as possible. Knowledge notoriously can be used for good or evil, but the growth of knowledge is good. If it ever slows down, we’re already in danger. Among other things, the world’s bad guys need only get ahead in one unpredictable way, and they’ll threaten unlimited harm. The good guys, because they try to improve their values and objectives too and also their means of making decisions, have what it takes to stay ahead and discover and implement defenses in time. If they choose to make fast enough progress, they’re going to be able to do that. So, what is the future of progress? Martin Rees has a metaphor for the unknowable future, that we are playing Russian roulette with science as the gun and technology as the bullets. But that would be random.

00:16:20 - 00:16:51

What we’re actually facing is almost the opposite, basically just the structure of the physical world and its immutable laws. We’re going into a world of unknown dangers and opportunities, and our only defense against that is the universality of explanatory knowledge and the desire to create it and to apply it. Thank you very much.

2011-10-01 What Now with Ken Rose

Transcript

00:48:08 - 00:48:11

Very interesting questions thanks for having me on the show I hope to speak to you again

Ken Rose

00:48:11 - 00:48:29

All the best okay bye bye then David Deutsch in Oxford, Great Britain the author of The Beginning of Infinity: Explanations That Transform the World also author of a …

2011-09-28 Sci-Fi London Fringe Event with Chris Patmore

00:43:16 - 00:43:22

It’s pretty much blown my mind for the rest of the day.

2011-09-20 Bloggingheads TV with John Horgan

01:07:40 - 01:07:43

Well, that’s great, and I’m glad that you’re not one.

John Horgan

01:07:43 - 01:07:48

I’m working on it. Thank you very much, David. It was really a pleasure.

David Deutsch

01:07:48 - 01:07:53

Okay. Nice talking to you. All right. Same here.

2011-09-19 KPCC AirTalkFear Not, Problem-Solving Children of the Enlightenment

00:17:01 - 00:17:30

And if you have any theories of your own or explanations about how things are the way they are, please share them with us online at kpcc.org. Click on through to the Air Talk page. We will segue in just a moment from quantum mechanics to, well, television. There you go. A natural segue as we’re joined by Mary McNamara of the LA Times to give us a recap of what happened at the Emmys. I’m David Lazarus in for Larry Mantle. This is Air Talk on 89.3 KPCC. Let’s head straight into the newsroom now and check in with Hetty Lynn Hurdy.

2011-09-07 Groks Science ShowInfinite Beginnings

00:25:09 - 00:25:16

And I’m Charles Lee. Make sure you also see us on the web at www.groks.net. Have a great afternoon and keep on grokkin’.

Read the article at Project Syndicate Source collection

2011-09-02 Project SyndicateEinstein the Realist

Einstein the Realist

By David Deutsch · 2 September 2011

With the discovery that the universe’s expansion is accelerating, not slowing, as was previously thought, newspaper headlines are proclaiming that Albert Einstein was right, after all. But Einstein’s most important legacy is his belief that a physical world, existing in reality, accounts for all of our experience.

OXFORD – It was recently discovered that the universe’s expansion is accelerating, not slowing, as was previously thought. Light from distant exploding stars revealed that an unknown force (dubbed “dark energy”) more than outweighs gravity on cosmological scales.

Unexpected by researchers, such a force had nevertheless been predicted in 1915 by a modification that Albert Einstein proposed to his own theory of gravity, the general theory of relativity. But he later dropped the modification, known as the “cosmological term,” calling it the “biggest blunder” of his life.

So the headlines proclaim: “Einstein was right after all,” as though scientists should be compared as one would clairvoyants: Who is distinguished from the common herd by knowing the unknowable – such as the outcome of experiments that have yet to be conceived, let alone conducted? Who, with hindsight, has prophesied correctly?

But science is not a competition between scientists; it is a contest of ideas – namely, explanations of what is out there in reality, how it behaves, and why. These explanations are initially tested not by experiment but by criteria of reason, logic, applicability, and uniqueness at solving the mysteries of nature that they address. Predictions are used to test only the tiny minority of explanations that survive these criteria.

The story of why Einstein proposed the cosmological term, why he dropped it, and why cosmologists today have reintroduced it illustrates this process. Einstein sought to avoid the implication of unmodified general relativity that the universe cannot be static – that it can expand (slowing down, against its own gravity), collapse, or be instantaneously at rest, but that it cannot hang unsupported.

This particular prediction cannot be tested (no observation could establish that the universe is at rest, even if it were), but it is impossible to change the equations of general relativity arbitrarily. They are tightly constrained by the explanatory substance of Einstein’s theory, which holds that gravity is due to the curvature of spacetime, that light has the same speed for all observers, and so on.

But Einstein realized that it is possible to add one particular term – the cosmological term – and adjust its magnitude to predict a static universe, without spoiling any other explanation. All other predictions based on the previous theory of gravity – that of Isaac Newton – that were testable at the time were good approximations to those of unmodified general relativity, with that single exception: Newton’s space was an unmoving background against which objects move. There was no evidence yet, contradicting Newton’s view – no mystery of expansion to explain. Moreover, anything beyond that traditional conception of space required a considerable conceptual leap, while the cosmological term made no measurable difference to other predictions. So Einstein added it.

Then, in 1929, Edwin Hubble discovered that the universe is expanding, consistently (within the observational accuracy of the day) with unmodified general relativity. So Einstein dropped the cosmological term. His doing so had nothing to do with Hubble being less blunder-prone; nor was Einstein deferring to Hubble’s superior prophetic abilities. It was just that the problem that the term was intended to solve no longer existed.

The new observations did not refute the existence of a cosmological term. They merely made it a bad explanation. Then, in 1998, came those new observations of a universe whose expansion is accelerating. As a result, the cosmological term that has been “reinstated” to account for the new observations is not quite the one that Einstein proposed and retracted. It is larger, for it now has to explain not just why the universe isn’t collapsing, but why its expansion is accelerating.

Einstein’s remark about having “blundered” is as misleading as the idea that he is “right after all.” The cosmological term is not something that should never have been proposed. Its introduction represented progress in understanding reality – as did its abandonment in light of Hubble’s discovery and its reinstatement in revised form to account for the new observations.

Likewise, the mid-twentieth century “Bohr-Einstein debate” about quantum theory is often misinterpreted as a personal clash between wizards. So counter-intuitive are quantum theory’s predictions that, under the leadership of one of its pioneers, Niels Bohr, a myth grew that there is no underlying reality that explains them. Particles get from A to B without passing through the intervening space, where they have insufficient energy to exist; they briefly “borrow” the energy, because we are “uncertain” about what their energy is. Information gets from A to B without anything passing in between – what Einstein called “spooky action at a distance.” And so on.

What these paradoxical interpretations have in common is that they abandon realism, the doctrine that a physical world, existing in reality, accounts for all of our experience. Anti-realism remains popular and appears in various guises in textbooks and popular accounts of quantum theory. But Einstein insisted that physical phenomena have explanations in terms of what he called “elements of reality.”

Fortunately, a minority of physicists, myself included, likewise side unequivocally with realism, by adopting Hugh Everett’s multiple-universes interpretation of quantum theory. According to this view, no particles exist where they have insufficient energy to be; it is simply that in some universes they have more energy than average, and in others, less. All alleged “paradoxes” of quantum theory are similarly resolved.

So, while most accounts say that Bohr won the debate, my view is that Einstein, as usual, was seeking an explanation of reality, while his rivals were advocating nonsense. Everett’s interpretation doesn’t make Einstein a demigod. But it does make him right.

About David Deutsch

David Deutsch is Visiting Professor of Physics and a founder member of the Centre for Quantum Computation at the Clarendon Laboratory, University of Oxford, and author of The Fabric of Reality.

2011-08-29 KERA ThinkExplanations That Transform the World

00:47:35 - 00:47:37

Thank you for having me on the program.

Krys Boyd

2011-04-10 AV Referendum

Duration: 00:14:50

Transcript

David Deutsch

00:00:00 - 00:04:00

The referendum on whether Britain should change its voting system to the alternative vote system, called AV, is now approaching. And I’m worried that in the debate about this, the criterion that seems to me to be the decisive one in choosing a voting system isn’t even being mentioned by either side in this debate. But first, why is this issue of what is a better or a worse voting system, why has that come up at all? As opposed to debates about which policies and which politicians are better, which is what political debate is usually about in an advanced democratic political system. Well, it’s come up because the existing voting system, which is usually known as first past the post, has a strong tendency to give the two most popular parties a larger proportion of the seats in parliament than the proportion of the votes they get in the country as a whole. And correspondingly, the smaller parties all get a lower proportion of seats than their proportion of the votes. This non-proportional response of the first past the post system is deemed to be unfair and biased by some people. And the point of AV is to reduce this bias, this supposed bias, against the smaller parties. But even aside from that, a lot of the debate has involved the two sides talking completely past each other on that issue, what you might call the arithmetic of fairness. The AV side, as I’ve just said, it says, look, the existing system isn’t proportional and so it’s biased against the smaller parties. And the proponents of the existing system point to a completely different issue, which is also about arithmetical fairness, namely that the AV system and actually any system that is any more proportional than first past the post gives the smaller parties and especially the third largest party a disproportionate political power, even though it gives it a more proportionate number of seats. And the reason that happens is that, well, in any system it’s very rare for one political party to be the favourite party of more than 50% of the electorate. And so any system that reduces this alleged unfairness to the smaller parties will statistically tend to prevent any party from having an overall majority of MPs either. And so it will prevent them from forming a government and so they will have to form a coalition with other parties, usually the third largest party. And that is what gives the third largest party in particular enormously disproportionate political power. That happens in two ways.

00:04:00 - 00:06:32

The first way is that that party then chooses who is to be the Prime Minister, because the leader of the third largest party can choose either the largest party or the second largest party to form the government with. And there’s nothing in AV or in any other system that will make the leader of that third largest party lean in the same direction as the electorate leant. So the third party leader can choose to put the second party leader into office as Prime Minister or the leader of the leading party. So it’s the kingmaker. And that power of the third largest party leads to another disproportionate power that it then has, which is that it can make conditions for whom it will pick to put into government. And through these conditions it can insert its own policies into government policy, no matter how unpopular those policies are with the electorate. Now, I should say that even the existing system, the first past the post system, does produce hung parliaments and therefore coalition governments every so often, every couple of decades. And so by this criterion of arithmetical fairness, that it shouldn’t give disproportionate political power to the smaller parties, even our existing system is actually biased towards the third largest party. And on that view, AV would do nothing except exacerbate what is already quite a serious injustice in the system. Instead of just doing so occasionally, it would put the third largest party into government in most future elections forever. And in that case, there would be almost nothing that the voters could do to reduce that power of the third party to override their wishes. For instance, let’s suppose that the third largest party leader is in a coalition and is therefore in charge of some ministry or other, and is doing things that you consider to be bad as a voter.

00:06:32 - 00:08:39

Well, at the next election there’s no way you can deploy your vote to remove that person from power. You can vote against them as much as you like. In fact, in that situation, most voters are already doing that. They’re already voting against the third largest party. But whether you cast your vote for the first party or the second party in your protest against what the third party is doing in government, that minister will probably still be in office after the election, whether the first two parties or the electorate like it or not, because that will be the price of joining any coalition. The third largest party just isn’t going to join the coalition unless it gets ministerial jobs as a result and policy as a result. So, as I said, both sides in the referendum debate are focusing on different aspects of arithmetic fairness, and that they’re talking completely past each other, and neither of them is addressing the issue that therefore logically comes up of what criterion should be used to distinguish between good and bad political institutions. Well, fortunately, this issue has been addressed by the great philosopher of reason in the 20th century, Sir Karl Popper. And his solution comes in two parts. The first part is to avoid a certain irrational fantasy, namely that there exists some way of setting up a government that knows in advance what is right, that will not enforce mistakes and bad laws and impose incompetent leaders on people, because it has knowledge in advance of what will work, what will be right, and what will not.

00:08:39 - 00:11:03

Now, this is the fantasy which in times past put aristocracies or hereditary monarchies into power, like the rule by the best is what aristocracy literally meant, and it is the fantasy that has often put dictators into power, and it is just impossible because mistakes are inevitable. Mistakes are the human condition. Mistakes are legion. The difference between a rational political system which is capable of making progress and irrational systems that are not capable of making progress is entirely in their ability to correct errors that have already happened, not in some magical ability to prevent errors from happening at all. And therefore, Popper’s criterion for political institutions is this. A rational institution of government, and that includes electoral systems, is one that makes it as easy as possible to remove bad policies and bad rulers without violence. OK, so how do First Past the Post and AV fare when they are judged by Popper’s criterion? Well, transferable vote systems, including AV, have the following argument that they can make in that regard, in their own favour. Suppose that in a particular constituency there is a consensus that the sitting MP is either incompetent or is going to vote for very bad policies in Parliament. And suppose that there is also a consensus that a particular rival of that MP would actually be unequivocally better. The constituents needn’t agree that that rival is the best candidate, but just that that candidate is better than the incumbent.

00:11:03 - 00:13:25

If there is such a consensus, then in an alternative vote system they can all put that candidate as second choice, and then, if they’re lucky, the arithmetic will work out in such a way that the second choice votes will add up enough to remove the incumbent MP. The trouble with that is that a system that, even if it can eliminate a particular legislator who is deemed to be either incompetent or have bad policies and so on, that system is completely ineffective if it adds up to a system which, at the level of what laws are actually passed, can’t correct errors. If it can’t correct errors at that level, then it doesn’t make any difference whether one can eliminate particular legislators who are deemed to be bad. And it is at that level of lawmaking and choosing leaders that all systems other than first-past-the-post fall down badly. Their effect is to entrench the third party in power. The leader of the third-largest party doesn’t fear the electorate, nor the first two parties or their supporters. He or she may fear the rise of the fourth-largest party, but that is a really bad tail by which to wag the dog of the policy of the entire country. In fact, if I were a supporter of the third party, and therefore if I was tempted by the prospect of my party being put permanently into government by an electoral system, nevertheless I would fear this even more, because as stultifying as the AV system would be for the country, it would be a catastrophe for the third party itself, which would then have no incentive to create political knowledge, create better policies.

00:13:25 - 00:14:49

Its entire functioning would be in wheeling and dealing for a fixed set of policies for a fixed constituency to influence the other parties, regardless of the wishes of the electorate. A rational political system, therefore, is one in which the politicians in power are constantly afraid that if they don’t put things right, the electorate will put them out. Conversely, we need to avoid systems like AV in which politicians can more easily relax about how the electorate feels, because if they’re the third party, they know that it would take enormous unlikely swings in opinion to force them out, while the two top parties know that even if they lose some significant support among the population, provided they make the right deal with the minority parties after the election, they can still stay in power. In fact, this almost happened even under the existing system at the last British election. Under AV, it would practically be a fixture, and that would be a catastrophe for political progress.

2011-03-10 Oxford TranshumanistsHow to Think About the Future

01:00:40 - 01:00:42

Thank you.

2009-07-01 TEDA New Way to Explain Explanation

Duration: 00:17:15

Transcript

David Deutsch

00:00:00 - 00:03:29

[Music]

Read the interview at Der Spiegel Source collection

2007-02-11 Der Spiegel"Ich existiere unendlich oft"

”Ich existiere unendlich oft”

English title: “I Exist Infinitely Many Times”

Für den Physiker David Deutsch ist unser Universum nur eines von unendlich vielen, in denen er parallel lebt. Was Laien absurd erscheint, folgert Deutsch aus der Quantentheorie. Im Interview wirft er seinen Physikerkollegen vor, die objektive Realität zu ignorieren.

English: For physicist David Deutsch, our universe is only one of infinitely many in which he lives in parallel. What seems absurd to laypeople, Deutsch derives from quantum theory. In this interview, he accuses his fellow physicists of ignoring objective reality.

11.02.2007, 08.00 Uhr

Frage: Herr Deutsch, Sie sagen, dass es ein Universum gibt, in dem die Erde anders als in unserem Universum vor 60 Millionen Jahren nicht von einem Kometen getroffen wurde. Sind Sie sich sicher?

English: Mr Deutsch, you say there is a universe in which, unlike in ours, Earth was not struck by a comet 60 million years ago. Are you certain?

David Deutsch: Ja, ganz sicher.

English: Yes, absolutely certain.

Frage: Gibt es dann auch Universen, in denen die Dinosaurier überlebt und sich zu intelligenten Lebewesen weiterentwickelt haben?

English: Are there also universes in which the dinosaurs survived and evolved into intelligent beings?

Deutsch: Das folgt aus der Quantentheorie, einer der fundamentalsten Theorien der Physik. Diese Folgerung, dass es viele verschiedene Universen gibt, ist übrigens nicht meine Idee. Sie ist seit 50 Jahren bekannt und wurde von Hugh Everett als Erstes publiziert.

English: That follows from quantum theory, one of the most fundamental theories in physics. Incidentally, the conclusion that there are many different universes is not my idea. It has been known for 50 years and was first published by Hugh Everett.

Frage: Wo sind denn die ganzen anderen Universen?

English: Where are all the other universes, then?

Deutsch: Wir müssen berücksichtigen, dass die Frage nach dem Wo sich immer schon auf ein bestimmtes Universum bezieht. Zum Beispiel, ich wohne in Oxford, Sie nicht. Wir beziehen uns auf verschiedene Orte im gleichen Universum. Wenn Sie fragen, wo das Universum ist, geraten Sie in Gefahr, Ihre Vorstellung irrezuführen, indem Sie versuchen, sich das andere Universum irgendwo in unserem Universum vorzustellen, so wie einen anderen Ort.

English: We must bear in mind that the question of where always refers to a particular universe. For example, I live in Oxford; you do not. We are referring to different places in the same universe. If you ask where the universe is, you risk misleading your imagination by trying to picture the other universe somewhere in our universe, like another place.

Frage: Wie kann man sich das dann besser vorstellen?

English: How can one picture it more accurately, then?

Deutsch: Es ist besser, am anderen Ende zu beginnen. Nach der Quantentheorie besteht die physikalische Realität aus einem viel größeren Ding als der Gesamtsumme der Dinge, die wir sehen, der Sterne, der Galaxien. Die Realität ist ein viel größeres Ding, und das nennen wir Multiversum. Es hat Regionen, die sich fast autonom von den anderen Regionen verhalten. Und das sind eben die verschiedenen Universen.

English: It is better to begin at the other end. According to quantum theory, physical reality consists of something much larger than the sum total of the things we see, the stars and the galaxies. Reality is something much larger, and we call that the multiverse. It has regions that behave almost autonomously from the other regions. Those are the different universes.

Frage: Sie sagen, wir alle würden in vielen dieser Universen existieren. In wie vielen dieser Universen würden wir Sie treffen?

English: You say that all of us exist in many of these universes. In how many of them would we meet you?

Deutsch: Wir sprechen hier über exponentiell große Zahlen. Es gibt unter den Universen solche, die sich deutlich voneinander unterscheiden und solche, die vollkommen identisch sind. Welche wollen Sie zählen?

English: We are talking about exponentially large numbers. Among the universes are some that differ markedly from one another and some that are completely identical. Which do you want to count?

Frage: Nur die verschiedenen, in denen es trotzdem David Deutsch gibt.

English: Only the different ones in which David Deutsch nevertheless exists.

Deutsch: Das wäre immer noch eine extrem große Zahl. Wenn wir alle, also auch die identischen Universen zählen, wäre die Antwort sogar unendlich.

English: That would still be an extremely large number. If we count all of them, including the identical universes, the answer would even be infinite.

Frage: Wann beginnen Universen denn, sich unterschiedlich zu entwickeln?

English: When do universes begin to develop differently?

Deutsch: Wenn Sie eine Münze werfen, werfen Sie die in unendlich vielen Universen, die identisch sind, dazu noch in ein paar anderen. Konzentrieren wir uns für dieses Beispiel auf die identischen: Sie teilen sich in zwei Gruppen auf, die sich unterschiedlich entwickeln, entsprechend den zwei möglichen Ergebnissen Kopf oder Zahl. Jede Gruppe enthält immer noch unendlich viele Universen.

English: When you toss a coin, you toss it in infinitely many identical universes, as well as in a few others. For this example, let us concentrate on the identical ones. They divide into two groups that develop differently, corresponding to the two possible outcomes, heads or tails. Each group still contains infinitely many universes.

Frage: Für Sie sind alle Möglichkeiten, die nach der Quantentheorie eintreffen, Realität, wenn auch in verschiedenen Universen. Nach der von den meisten Physikern vertretenen Kopenhagener Interpretation kann aber nur eine Möglichkeit real sein. Welche, Kopf oder Zahl, entscheidet sich bei der Messung. Wann haben Sie begonnen, daran zu zweifeln?

English: For you, all the possibilities that occur according to quantum theory are real, though in different universes. According to the Copenhagen interpretation accepted by most physicists, however, only one possibility can be real. Which one, heads or tails, is decided at measurement. When did you begin to doubt that?

Deutsch: Ich fühlte mich schon immer unzufrieden damit. Ich denke, das geht allen Physikern so, sogar den Anhängern der Kopenhagener Deutung. Auch ich habe sie ohne ernsthafte Kritik akzeptiert, bis ich als Graduiertenstudent in den 1970ern versucht habe, sie zu verbessern. Ich habe damals an der Frage geforscht, welche Mechanismen, zum Beispiel welche Aspekte des Bewusstseins, ein Universum auswählen und real machen und dafür sorgen, dass die anderen Universen reine Möglichkeiten bleiben. Als ich das versucht habe, wurde mir klar, dass es nicht geht. Die einzige funktionierende Alternative ist die Hypothese, dass alle Universen real und immer noch vorhanden sind.

English: I had always felt dissatisfied with it. I think all physicists feel that way, even adherents of the Copenhagen interpretation. I too accepted it without serious criticism until, as a graduate student in the 1970s, I tried to improve it. At the time, I researched which mechanisms, for example which aspects of consciousness, select a universe and make it real while ensuring that the other universes remain mere possibilities. When I tried, I realised that it could not be done. The only workable alternative is the hypothesis that all universes are real and still present.

Frage: Warum ist es falsch anzunehmen, dass der Akt der Messung entscheidet, was real ist?

English: Why is it wrong to assume that the act of measurement decides what is real?

Deutsch: Die Frage ist, ob man Ergebnisse interpretieren kann, indem man sagt, der Akt der Messung ändert unvermeidlicherweise das, was herauskommt. Wenn Sie das zu Ende denken, kommen Sie darauf, dass diese Lösung nicht funktioniert. Denn sie verleugnet die Realität der Dinge, bevor sie gemessen oder wahrgenommen werden.

English: The question is whether one can interpret results by saying that the act of measurement inevitably changes what comes out. If you think that through to its conclusion, you find that this solution does not work. For it denies the reality of things before they are measured or perceived.

Frage: Vor der Messung sieht man die Dinge ja auch nicht. Warum soll auch real sein, was man nicht sieht?

English: Before measurement, one does not see the things either. Why should what one cannot see be real?

Deutsch: Diese Argumentation, dass etwas nicht real ist, weil man es nicht sieht, kann gegen alle wissenschaftlichen Theorien verwendet werden. Man kann sie immer benutzen, um die Realität der Gegenstände und Größen abzustreiten, über die in der Wissenschaft gesprochen wird.

English: This argument that something is not real because one cannot see it can be used against all scientific theories. It can always be used to deny the reality of the objects and quantities discussed in science.

Frage: Warum folgen so viele Physiker immer noch der Kopenhagener Interpretation und nur vielleicht zehn Prozent der Multiversum-Theorie?

English: Why do so many physicists still follow the Copenhagen interpretation, while perhaps only ten percent follow multiverse theory?

Deutsch: Das ist eine schlimme Erscheinung. Leider habe ich auch keine vollständige Antwort darauf. Es hängt mit der Philosophie des 20. Jahrhunderts zusammen, die versuchte, die Existenz objektiver Realität zu bestreiten. Das ist Gift für Wissenschaft und für Philosophie.

English: That is a terrible phenomenon. Unfortunately, I do not have a complete answer either. It is connected with twentieth-century philosophy, which tried to deny the existence of objective reality. That is poison for science and for philosophy.

Frage: Hat es denn einen praktischen Effekt, für welche Theorie man sich entscheidet?

English: Does it have any practical effect which theory one chooses?

Deutsch: Mit allen Theorien, die man kennt und versteht, kann man richtige Vorhersagen machen unabhängig davon, ob man sagt, etwas sei real oder nicht. Aber wenn man eine Theorie verbessern will, ist das anders. Mit dem Ansatz, dass eine wissenschaftliche Theorie nur ein Satz Regeln ist, mit dem man etwas vorhersagt, würde man die Regeln nie finden. Denn die Regeln werden nur entdeckt, indem man postuliert, dass etwas real ist und reale Eigenschaften hat, von denen wir die Beobachtungen ableiten können. Wenn man diesen Schritt auslässt, dann friert man den Fortschritt ein. Und genau das ist auch geschehen.

English: With all the theories one knows and understands, one can make correct predictions regardless of whether one says that something is real or not. But if one wants to improve a theory, it is different. With the approach that a scientific theory is only a set of rules for making predictions, one would never find the rules. For the rules are discovered only by postulating that something is real and has real properties from which we can derive the observations. If one omits this step, one freezes progress. And that is precisely what happened.

Frage: Inwiefern?

English: In what way?

Deutsch: Zum Beispiel in meinem engeren Forschungsfeld. Die Entdeckungen, die ich und Kollegen in den 1980ern über den universalen Quantencomputer gemacht haben, hätten gut und gern auch schon in den 1950ern und selbst in den 1930ern gemacht werden können. Das geschah nur nicht, glaube ich, weil niemand Quantentheorie als eine Theorie der Realität ernst nahm. Man nahm sie nur als Methode ernst, mit der man Vorhersagen macht. So kann man nicht über neue reale Objekte nachdenken, über die man noch nie nachgedacht hatte, und die Quantentheorie in einer Art befolgen, über die man auch noch nicht nachgedacht hat, so wie es bei den Quantencomputern geschieht. Ich sollte sagen, dass zwar nicht alle, die daran forschen, an die Multiversum-Interpretation glauben. Aber ich denke, es ist bemerkenswert, dass es auf dem Gebiet mehr Anhänger für die Multiversum-Theorie gibt als in anderen Feldern der Physik.

English: For example, in my own research field. The discoveries that my colleagues and I made in the 1980s about the universal quantum computer could easily have been made in the 1950s, or even in the 1930s. That did not happen, I believe, because no one took quantum theory seriously as a theory of reality. It was taken seriously only as a method for making predictions. In that way, one cannot think about new real objects that have never been considered before, or follow quantum theory in a way that has never been considered before, as happens with quantum computers. I should say that not everyone researching them believes in the multiverse interpretation. But I think it is remarkable that the field has more adherents of multiverse theory than other areas of physics.

Frage: Glauben Sie, dass Sie noch mehr Kollegen von der Richtigkeit der Multiversum-Theorie überzeugen können?

English: Do you believe you can convince more colleagues that multiverse theory is correct?

Deutsch: Ich glaube schon. Jüngere Menschen sind offener dafür. Außerdem, und das ist wichtiger, ändert sich das philosophische Klima zurück dahin, die Realität zu respektieren. Ich denke, es ist eher das als die direkte Überzeugungsarbeit, was der Multiversum-Theorie mehr Anhänger bringen wird.

English: I think so. Younger people are more open to it. In addition, and more importantly, the philosophical climate is changing back towards respecting reality. I think it is that, rather than direct persuasion, that will bring multiverse theory more adherents.

Das Gespräch führte Michael Fuhs.

English: Interview by Michael Fuhs.

Technology Review, Heise Zeitschriften Verlag, Hannover

2007-01-27 Why Are Flowers Beautiful

01:05:37 - 01:06:14

Well David, thank you very, very much on behalf of the Irish Museum of Modern Art. It’s a brilliant talk today, so thank you very, very much. Thank you. Great audience, thanks for coming. Just to tell everybody that Lunar Reggae continues until the 18th and we also have two shows opening, Alex Katz and Thomas Demand, and that’s coming, so do look to our programme for that. Thank you very much. David, thanks.

David Deutsch

01:06:15 - 01:06:18

Thank you. Bye everyone.

Rachel Thomas

01:06:19 - 01:06:21

Hi David, we’ll be in touch as well.

Read the Q&A at New Scientist Source collection

2006-12-21 New ScientistReaders’ Q & A with David Deutsch

Readers’ Questions and Answers

Published 21 December 2006.

Thank you for sending so many excellent questions for David Deutsch. His answers to the best questions are below.

Joseph Tanega

Question: I’ve read your Fabric of Reality and very much liked how you have taken the basic ideas of some theories very seriously as forms of explanation. I’m wondering whether we might take Von Neumann’s concept of the Universal Constructor very seriously. That is, if quantum computing comes from a fundamental discreteness of the universe, is there a fundamental limit to a Von Neumann Universal Constructor?

David Deutsch: I think that there are indeed insights to be gained by applying quantum theory to the theory of constructors, and it’s one of the things I have been trying to do. However, there is no reason to believe that quantum physics makes universal constructors impossible. Quite the contrary. As with computation, if anything, the discreteness of observables in nature makes universality easier, not harder, to attain.

John Newson

Question: As I understand it there are not even enough particles in this universe – even if all were capable of computation – to perform even a modest quantum computation. Deutsch says therefore that the computation must take place in serial universes. What about the quid pro quo? In these serial universes many similar situations must exist, leading (as I see it) to a similar shortage of computational capacity in ALL computation-capable universes.

David Deutsch: No. The way the sharing of computational capacity happens is that there is always a swathe of universes in the multiverse in which we are doing any given thing, including any particular quantum computation. The universes in that swathe cooperate, and they can do so precisely because they are all interested in getting the same result, and by the same means.

Anders Lotsson

Question: You write about factorizing a 10,000-digit integer, and how long it would take on a quantum computer as opposed to a regular computer. What about the energy requirements?

David Deutsch: This would depend on the particular hardware implementation of the universal quantum computer – of which none have yet been achieved. But roughly speaking, the time would be about the same as that required for a classical computer built with similar technology to perform one of the parallel strands. And the energy requirement would be determined entirely by the inefficiency of the machine. Physics imposes no minimum of energy needed to perform a given factorisation.

Peter Hayward

Question: “Truth in science must mean correspondence to reality, or it means nothing.” Would you agree we still have to decide if something is real or not?

David Deutsch: Yes. We should decide that something is real when it appears unavoidably in our explanations of the world.

Wes Wilson

Question: For me, the following sentence is intriguing. “The way the quantum computer works is: the universe differentiates itself into multiple universes and each one performs a different sub-computation.” My question is, does this computation create a measurable affect in the other universes? Or perhaps the converse is more interesting, does the operation of quantum computers in other universes create a measurable affect in our universe?

David Deutsch: No. Only the effects of the computation itself. That is because of the fact that the only universes that cause interference are the ones that were identical initially and are again identical at the end.

Tarik Ozkanli

Question: My question is about the realization of the quantum computer. I think the realization process itself is a unique quantum phenomena. I wonder if is it meaningful to search for an existence proof of this realization. I mean the proof of the possibility of such a quantum task.

David Deutsch: It is indeed meaningful and worthwhile. But it’s a job for quantum constructor theory – which does not exist yet.

Daniel Taghioff

Question: Does the Afshar Experiment throw into question your version of the many worlds interpretation, and if not, why not?

David Deutsch: It does not, because the outcome of the experiment is as predicted by quantum theory, and the many-universes ‘interpretation’ is nothing more than the assertion that the theory describes reality.

Stephen T. Horn

Question: What sort of personal identity do we have if we exist in infinite copies in a multiverse system?

David Deutsch: The same as we do as a result of existing in slightly different versions at an infinity of different times.

Jeff L Jones

Jeff L Jones: I’m a graduate student working on my PhD in high energy physics, and David Deutsch has been a great inspiration for me.

David Deutsch: Thank you.

1. Question: What can young physicists do to challenge the logical positivist ideas that are so widespread in the physics community today?

David Deutsch: Argue: don’t let colleagues get away with denying that you exist! (Also, urge them to read The Fabric of Reality.)

2. Question: Do you think there is a chance the Many Worlds Interpretation of quantum mechanics can be tested using the anthropic principle in conjunction with fine tuning problems in cosmology such as the cosmological constant problem?

David Deutsch: In principle, yes, though we do not understand the relevant processes in anything like enough detail yet. Actually this would be a test of the anthropic principle, not the many universes interpretation, which is nothing other than quantum theory itself. By the way, to the best of my knowledge this test was first suggested by Dennis Sciama in the 1970s, but he didn’t publish it and so he doesn’t get the credit.

3. Question: What is your favourite scenario for an experimental test of the Many Worlds Interpretation?

David Deutsch: The question has to be: against what? I know of no other interpretations that make sense. One day, no doubt, there will be a viable rival theory to quantum theory. Even then, testing it would only be a test of the many-universes interpretation if it turned out to be a single-universe theory. But I find that very implausible, given the evidence that we already have.

4. Question: Do you think we will ever have conscious computers, whether quantum or classical? If so, how long do you think it will be?

David Deutsch: I am sure we will have them, I expect they will be purely classical, and I expect that it will be a long time in the future. Significant advances in our philosophical understanding of what consciousness is, will be needed.

Charles Goodwin

Question: How do you reply to people who agree that the universes splits during a superposition, but claim that afterwards the other branches disappear?

David Deutsch: I assume that if I successfully ignore them, this has caused them to disappear. After all, that is what they themselves are claiming.

Follow-up: Admittedly this is a more complex theory than ‘many-worlds’ in that a mechanism has to be proposed for this process, but […] can these views be distinguished aside from arguing from simplicity?

David Deutsch: It’s not ‘simplicity’. It’s arguing for explanations, and refusing to take seriously things which, like fairies at the bottom of the garden, do not figure in any explanation of anything.

Tom Williamson

Question: Can the Everett-DeWitt-Deutsch multiverse explain the ‘fine-tuned for life’ appearance of our universe? If not, do we need a polyverse – a multiverse of multiverses?

David Deutsch: I don’t know, but I suspect that whatever can be explained by the anthropic principle (and that certainly can’t be the laws of physics as a whole) might be explained sufficiently well by the quantum multiplicity, without having to appeal to further multiplicities for whose existence there is no independent evidence.

John L. Quel

Question: Physicist Victor Stenger (Timeless Reality) is a proponent of the equivalence of the forward and backward directions of time in physics as a resolution to the many quantum mysteries. Couldn’t quantum computation be explained by the computational/experimental setup simply looping backwards and forward in time until it completes?

David Deutsch: Like many purported alternative ‘interpretations’, the plausibility of this one relies on equivocating about whether these ‘loops backwards and forwards in time’ exist in reality or not. If they don’t, then the ‘interpretation’ suffers from the inconsistency of attributing observable events to causes that never actually happened. If they do, then the loops are simply parallel universes in a pointless disguise.

Fifi Lamour

Question: Are you proposing the sort of multiple realities that the movie What the Bleep Do We Know? proposes? Do you think we shape the objective world with only our thoughts? (By this I don’t mean our subjective experience of the world, or thoughts that lead to actions, but more along the lines of if we wish really hard what we wish for will manifest out of an “alternative” dimension or world.)

David Deutsch: Definitely not.

2005-07-01 TEDChemical Scum That Dream of Distant Quasars

Duration: 00:19:45

Transcript

David Deutsch

00:00:00 - 00:02:39

[Music]

We’ve been told to go out on a limb and say something surprising. So I’ll try and do that, but I want to start with two things that everyone already knows. And the first one, in fact, is something that has been known for most of recorded history. And that is that the planet Earth, or the solar system, or our environment, or whatever, is uniquely suited to sustain our evolution, or creation as it used to be thought, and our present existence, and most important, our future survival. And nowadays this idea has a dramatic name, Spaceship Earth. And the idea there is that outside the spaceship, the universe is implacably hostile, and inside is all we have, all we depend on, and we only get the one chance. If we mess up our spaceship, we’ve got nowhere else to go. Now the second thing that everyone already knows is that contrary to what was believed for most of human history, human beings are not, in fact, the hub of existence. As Stephen Hawking famously said, we’re just a chemical scum on the surface of a typical planet that’s in orbit around a typical star, which is on the outskirts of a typical galaxy, and so on. Now the first of those two things that everyone knows is kind of saying that we’re at a very untypical place, uniquely suited, and so on. And the second one is saying that we’re at a typical place, and especially if you regard these two as deep truths to live by and to inform your life decisions, then they seem a little bit to conflict with each other. But that doesn’t prevent them from both being completely false. And they are. So let me start with the second one, typical.

00:02:39 - 00:04:53

Well, is this a typical place? Well, let’s look around, you know, look in a random direction, and we see a wall and chemical scum, and that’s not typical of the universe at all. All you’ve got to do is go a few hundred miles in that same direction and look back, and you won’t see any walls or chemical scum at all. All you see is a blue planet. If you go further than that, you’ll see the sun, the solar system, and the stars, and so on. But that’s still not typical of the universe because stars come in galaxies, and most places in the universe, a typical place in the universe, is nowhere near any galaxy. So let’s go out further till we’re outside the galaxy, and look back, and yeah, there’s the huge galaxy with spiral arms laid out in front of us. At this point, we’ve come 100,000 light years from here, but we’re still nowhere near a typical place in the universe. To get to a typical place, you’ve got to go a thousand times as far as that into intergalactic space. And so what does that look like? Typical. What does a typical place in the universe look like? Well, at enormous expense, Ted has arranged a high-resolution, immersion virtual reality rendering of intergalactic space, the view from intergalactic space. Can we have the lights off, please, so we can see it? Well, not quite. Not quite perfect. You see, intergalactic space is completely dark, pitch dark. It’s so dark that if you were to be looking at the nearest star to you, and that star were to explode as a supernova, and you were to be staring directly at it at the moment when its light reached you, you still wouldn’t be able to see even a glimmer.

00:04:53 - 00:06:51

That’s how big and how dark the universe is. And that’s despite the fact that a supernova is so bright, so brilliant an event that it would kill you stone dead at a range of several light years. And yet, from intergalactic space, so far away, you wouldn’t even see it. It’s also very cold out there, less than three degrees above absolute zero. And it’s very empty. The vacuum there is one million times less dense than the highest vacuum that our best technology on earth can currently create. So that’s how different a typical place is from this place. And that is how untypical this place is. So can we have the lights back on, please? Thank you. Now, how do we know about an environment that’s so far away and so different and so alien from anything we’re used to? Well, the earth, our environment in the form of us, is creating knowledge. Well, what does that mean? Well, look out even further than we’ve just been, I mean from here with the telescope, and you’ll see things that look like stars. They’re called quasars. Quasars originally meant quasi-stellar object, which means things that look a bit like stars. But they’re not stars, and we know what they are. Billions of years ago and billions of light years away, the material at the center of a galaxy collapsed towards a supermassive black hole, and then intense magnetic fields directed some of the energy of that gravitational collapse and some of the matter back out in the form of tremendous jets which illuminated lobes with the brilliance of, I think, it’s a trillion suns.

00:06:51 - 00:08:42

Now, the physics of the human brain could hardly be more unlike the physics of such a jet. We couldn’t survive for an instant in it. Language breaks down when trying to describe what it would be like in one of those jets. It would be a bit like experiencing a supernova explosion, but at point-blank range and for millions of years at a time. And yet, that jet happened in precisely such a way that billions of years later on the other side of the universe, some bit of chemical scum could accurately describe and model and predict and explain, above all (there’s your reference), what was happening there in reality. The one physical system, the brain, contains an accurate working model of the other, the quasar. Not just a superficial image of it, though it contains that as well, but an explanatory model embodying the same mathematical relationships and the same causal structure. Now, that is knowledge. And if that weren’t amazing enough, the faithfulness with which the one structure resembles the other is increasing with time. That is the growth of knowledge. So the laws of physics have this special property that physical objects, as unlike each other as they could possibly be, can nevertheless embody the same mathematical and causal structure and to do it more and more so over time. So we are a chemical scum that is different. This chemical scum has universality. Its structure contains, with ever increasing precision, the structure of everything.

00:08:42 - 00:10:49

This place, and not other places in the universe, is a hub which contains within itself the structural and causal essence of the whole of the rest of physical reality. And so far from being insignificant, the fact that the laws of physics allow this, or even mandate that this can happen, is one of the most important things about the physical world. Now, how does the solar system, our environment in the form of us, acquire this special relationship with the rest of the universe? Well, one thing that’s true about Stephen Hawking’s remark, I mean it is true, but it’s the wrong emphasis, one thing that’s true about it is that it doesn’t do it with any special physics. There’s no special dispensation, no miracles involved. It does it simply with three things that we have here in abundance. One of them is matter, because, well, the growth of knowledge is a form of information processing. Information processing is computation. Computation requires a computer. There’s no known way of making a computer without matter. We also need energy to make the computer, and most important, to make the media, in effect, onto which we record the knowledge that we discover. And then, thirdly, less tangible, but just as essential for the open-ended creation of knowledge, of explanations, is evidence. Now, our environment is inundated with evidence. We happen to get around to testing, let’s say, Newton’s law of gravity about 300 years ago, but the evidence that we used to do that was falling down on every square meter of the Earth for billions of years before that, and will continue to fall for billions of years afterwards. And the same is true for all the other sciences. As far as we know, evidence to discover the most fundamental truths of all the sciences is here just for the taking on our planet.

00:10:49 - 00:12:33

Our location is saturated with evidence and also with matter and energy. Out in intergalactic space, those three prerequisites for the open-ended creation of knowledge are at their lowest possible supply. As I said, it’s empty, it’s cold, and it’s dark out there. Or is it? Now, actually, that’s just another parochial misconception, because imagine a cube out there in intergalactic space, the same size as our home, the solar system. Now, that cube is very empty by human standards, but that still means that it contains over a million tons of matter. And a million tons is enough to make, say, a self-contained space station on which there’s a colony of scientists that are devoted to creating an open-ended stream of knowledge, and so on. Now, it’s way beyond present technology to even gather the hydrogen from intergalactic space and form it into other elements and so on. But the thing is, in a comprehensible universe, if something isn’t forbidden by the laws of physics, then what could possibly prevent us from doing it other than knowing how? In other words, it’s a matter of knowledge, not resources. Well, if we could do that, we’d automatically have an energy supply, because this transmutation would be a fusion reactor. And evidence? Well, again, it’s dark out there to human senses. But all you’ve got to do is take a telescope, even one of present-day design, look out, and you’ll see the same galaxies as we do from here. And with a more powerful telescope, you’ll be able to see stars and planets in those galaxies.

00:12:33 - 00:14:19

You’ll be able to do astrophysics and learn the laws of physics. Locally there, you could build particle accelerators and learn elementary particle physics and chemistry and so on. Probably the hardest science to do would be biology field trips, because it would take several hundred million years to get to the nearest life-bearing planet and back. But I have to tell you, and sorry, Richard, but I never did like biology field trips much. I think we can just about make do with one every few hundred million years. So, in fact, intergalactic space does contain all the prerequisites for the open-ended creation of knowledge. Any such cube anywhere in the universe could become the same kind of hub that we are if the knowledge of how to do so were present there. So we’re not in a uniquely hospitable place. If intergalactic space is capable of creating an open-ended stream of explanations, then so is almost every other environment, so is the Earth, so is a polluted Earth. And the limiting factor there and here is not resources, because they’re plentiful, but knowledge, which is scarce. Now, this cosmic knowledge-based view may, and I think ought to, make us feel very special. But it should also make us feel vulnerable, because it means that without the specific knowledge that’s needed to survive the ongoing challenges of the universe, we won’t survive them. All it takes is for a supernova to go off a few light years away, and we’ll all be dead.

00:14:19 - 00:16:47

Martin Rees has recently written a book about our vulnerability to all sorts of things, from astrophysics to scientific experiments gone wrong, and most importantly, to terrorism with weapons of mass destruction. And he thinks that civilization has only a 50% chance of surviving this century. I think he’s going to talk about that later in the conference. Now, I don’t think that probability is the right category to discuss this issue in, but I do agree with him about this. We can survive, and we can fail to survive. But it depends not on chance, but on whether we create the relevant knowledge in time. The danger is not at all unprecedented. Species go extinct all the time. Civilizations end. The overwhelming majority of all species and all civilizations that have ever existed are now history. And if we want to be the exception to that, then logically, our only hope is to make use of the one feature that distinguishes our species and our civilization from all the others, namely our special relationship with the laws of physics, our ability to create new explanations, new knowledge, to be a hub of existence. So let me now apply this to a current controversy, not because I want to advocate any particular solution, but just to illustrate the kind of thing I mean. The controversy is global warming. Now, I’m a physicist, but I’m not the right kind of physicist. In regard to global warming, I’m just a layman. And the rational thing for a layman to do is to take seriously the prevailing scientific theory. And according to that theory, it’s already too late to avoid a disaster, because if it’s true that our best option at the moment is to prevent CO2 emissions with something like the Kyoto Protocol, with its constraints on economic activity and its enormous cost of hundreds of billions of dollars or whatever it is, then that is already a disaster by any reasonable measure. And the actions that are advocated are not even purported to solve the problem, merely to postpone it by a little. So it’s already too late to avoid it, and it probably has been too late to avoid it ever since before anyone realized the danger.

00:16:47 - 00:18:29

It was probably already too late in the 1970s when the best available scientific theory was telling us that industrial emissions were about to precipitate a new ice age in which billions would die. Now the lesson of that seems clear to me, and I don’t know why it isn’t informing public debate. It is that we can’t always know. When we know of an impending disaster and how to solve it at a cost less than the cost of the disaster itself, then there’s not going to be much argument, really. But no precautions and no precautionary principle can avoid problems that we do not yet foresee. Hence we need a stance of problem fixing, not just problem avoidance. It’s true that an ounce of prevention equals a pound of cure, but that’s only if we know what to prevent. If you’ve been punched on the nose, then the science of medicine does not consist of teaching you how to avoid punches. If medical science stopped seeking cures and concentrated on prevention only, then it would achieve very little of either. The world is buzzing at the moment with plans to force reductions in gas emissions at all costs. It ought to be buzzing with plans to reduce the temperature and with plans to live at the higher temperature, and not at all costs, but efficiently and cheaply. Some such plans exist, things like swarms of mirrors in space to deflect the sunlight away and encouraging aquatic organisms to eat more carbon dioxide. At the moment, these things are fringe research.

00:18:29 - 00:19:39

They’re not central to the human effort to face this problem or problems in general. And with problems that we are not aware of yet, the ability to put right, not the sheer good luck of avoiding indefinitely, is our only hope, not just of solving problems, but of survival. So take two stone tablets and carve on them. On one of them, carve, problems are soluble. And on the other one, carve, problems are inevitable. Thank you.

Read the interview at Der Spiegel Source collection

2005-03-13 Der Spiegel»Die Welt ist bizarr«

Searchable English notes

English title: “The World Is Bizarre”

SPIEGEL: Herr Deutsch, Sie behaupten, unser Universum sei nur eines von vielen.

English: Mr Deutsch, you claim that our universe is only one among many.

Deutsch: Von sehr, sehr vielen sogar. Alles, was physikalisch möglich ist, geschieht in mindestens einem dieser Universen.

English: One among very, very many, in fact. Everything that is physically possible happens in at least one of these universes.

SPIEGEL: Das heißt, es gibt auch ein Universum, in dem Saddam Hussein glücklich mit Laura Bush verheiratet ist?

English: Does that mean there is also a universe in which Saddam Hussein is happily married to Laura Bush?

Deutsch: Wenn wir die Quantentheorie ernst nehmen, ja – wenngleich diese Ehe vermutlich nur in einer winzigen Zahl von Universen geschlossen wurde.

English: If we take quantum theory seriously, yes, although that marriage presumably occurred in only a tiny number of universes.

SPIEGEL: Und andernorts haben die Dinosaurier überlebt und grübeln nun an Stelle von uns über die Gesetze der Physik?

English: And elsewhere the dinosaurs survived and are now pondering the laws of physics instead of us?

Deutsch: Ja, absolut.

English: Yes, absolutely.

SPIEGEL: Könnte auch in unserer Welt plötzlich ein Dino durchs Fenster gucken?

English: Could a dinosaur suddenly peer through the window in our world too?

Deutsch: Theoretisch sogar das – allerdings nur mit einer astronomisch geringen Wahrscheinlichkeit.

English: Even that is theoretically possible, although only with an astronomically small probability.

SPIEGEL: Dann müssten die Dinos ja 65 Millionen Jahre lang überlebt haben …

English: But then the dinosaurs would have had to survive for 65 million years …

Deutsch: … ohne bemerkt zu werden? Nun, ein Dino könnte ja auch spontan ins Leben gerufen worden sein, plötzlich dem Boden entstiegen. Es gibt schließlich einen physikalischen Prozess, der Dinosaurier in Erde verwandelt – also muss auch die Umkehrung möglich sein. Aber das ist natürlich astronomisch unwahrscheinlich – wobei der Begriff »astronomisch« noch viel zu schwach ist, um auszudrücken, wie unwahrscheinlich es ist.

English: … without being noticed? Well, a dinosaur could also have been brought into existence spontaneously, suddenly rising out of the ground. After all, there is a physical process that turns dinosaurs into earth, so the reverse must also be possible. But that is, of course, astronomically unlikely, although the word “astronomically” is far too weak to express just how unlikely it is.

SPIEGEL: Aber Universen mit eckigen Atomen und durchsichtigen Planeten, die wenigstens gibt es nicht, oder?

English: But universes with square atoms and transparent planets, at least those do not exist, do they?

Deutsch: Nein. Die Naturgesetze gelten in allen Universen. Die Masse des Elektrons ist in allen Universen gleich. Auch einen Urknall gab es überall.

English: No. The laws of nature apply in every universe. The mass of the electron is the same in every universe. There was a Big Bang everywhere too.

SPIEGEL: Und Zauberei gibt es auch nicht?

English: And there is no magic either?

Deutsch: Nein – außer in dem, was ich Harry-Potter-Universum nenne.

English: No, except in what I call the Harry Potter universe.

SPIEGEL: Wie bitte?

English: I beg your pardon?

Deutsch: Wie gesagt, in allen Universen gelten die Naturgesetze, und die schließen Zauberei aus. Es gibt allerdings einige wenige Universen, in denen Zauberei zu funktionieren scheint, weil zufälligerweise ein Hase genau in dem Moment auftaucht, in dem jemand den Zauberstab schwingt.

English: As I said, the laws of nature apply in every universe, and they rule out magic. There are, however, a few universes in which magic appears to work because, by sheer coincidence, a rabbit appears at precisely the moment someone waves a magic wand.

SPIEGEL: Und woran erkennt man dann, dass es keine echte Zauberei ist?

English: Then how can you tell that it is not real magic?

Deutsch: Stellen Sie sich das Multiversum einfach als eine Ansammlung von Universen vor.

English: Simply imagine the multiverse as a collection of universes.

SPIEGEL: Nichts einfacher als das.

English: Nothing could be easier.

Deutsch: Wenn Sie nun zu einem bestimmten Moment alle Universen betrachten, in denen es Anhaltspunkte dafür gibt, dass dort Harry-Pottersche Physik regiert, dann werden die allermeisten davon im nächsten Moment schon aufhören, so zu erscheinen.

English: Now consider, at a particular moment, all the universes in which there is evidence that Harry Potter physics governs. By the very next moment, the vast majority will already have ceased to look that way.

SPIEGEL: Weil beim nächsten Zauberstabschwingen kein Hase mehr aus dem Zylinder springt.

English: Because the next time the wand is waved, no rabbit jumps out of the hat.

Deutsch: Ganz genau.

English: Exactly.

SPIEGEL: Die Vorstellung all dieser unendlich vielen Universen ist ja zweifellos faszinierend. Aber sie klingt eher nach Science-Fiction. Sind Sie wirklich überzeugt, dass all diese vielen Universen – diese Dinowächst-aus-dem-Erdboden-Universen, diese Hussein-heiratet-Laura-Bush-Universen und diese Harry-Potter-Universen –, dass die genauso real sind wie dasjenige, in dem wir hier sitzen und uns unterhalten?

English: The idea of all these infinitely many universes is undoubtedly fascinating. But it sounds more like science fiction. Are you really convinced that all these universes, these dinosaur-grows-out-of-the-ground universes, these Hussein-marries-Laura-Bush universes and these Harry Potter universes, are just as real as the one in which we are sitting here having this conversation?

Deutsch: Absolut. Die Physik sagt es uns.

English: Absolutely. Physics tells us so.

SPIEGEL: Aber sehen können wir all diese Universen nicht?

English: But we cannot see all these universes?

Deutsch: Zumindest nicht direkt.

English: At least not directly.

SPIEGEL: Wie können Sie dann so sicher sein, dass es sie gibt?

English: Then how can you be so certain that they exist?

Deutsch: Dass wir etwas, das wir nicht sehen können, für real halten, ist ja gar nichts Ungewöhnliches. Wir können auch keine schwarzen Löcher und keine Atome und Quarks sehen, und doch sind wir überzeugt, dass es sie gibt. Das Entscheidende ist: Die Multiversumstheorie ist die beste Erklärung der Welt, die wir haben.

English: There is nothing unusual about regarding something we cannot see as real. We cannot see black holes, atoms or quarks either, yet we are convinced that they exist. The crucial point is this: the multiverse theory is the best explanation of the world that we have.

SPIEGEL: Welche experimentellen Beweise gibt es denn dafür?

English: What experimental evidence is there for it?

Deutsch: Jetzt kommen wir zu dem, was ich als größten Skandal der Physik des 20. Jahrhunderts bezeichne. Bei der Quantentheorie verzichten die Physiker nämlich, anders als bei allen anderen physikalischen Theorien, völlig darauf, sie als eine wahre Beschreibung der Welt zu betrachten. Sie stellen nur die Frage nach der experimentellen Testbarkeit und reden dann von der »Interpretation« der Quantentheorie. Das ist ein fundamentaler Fehler! Er hat die Physik sehr lange aufgehalten und viel Sand ins Getriebe der Wissenschaftsphilosophie gestreut.

English: Now we come to what I call the greatest scandal in twentieth-century physics. Unlike with every other physical theory, physicists completely refrain from treating quantum theory as a true description of the world. They ask only whether it can be tested experimentally and then speak of the “interpretation” of quantum theory. That is a fundamental mistake. It held physics back for a very long time and threw a great deal of sand into the machinery of the philosophy of science.

SPIEGEL: Ist »Skandal« nicht trotzdem ein wenig hochgegriffen?

English: Even so, is “scandal” not putting it a little strongly?

Deutsch: Überhaupt nicht. Die unlogische Trennung von experimenteller Testbarkeit und Erklärung ist genau das, was die Inquisition von Galileo zu erzwingen suchte: Sie wäre bereit gewesen, die praktischen Auswirkungen seines heliozentrischen Weltbildes zu akzeptieren, wenn er nur darauf verzichtet hätte zu behaupten, dass die Erde wirklich um die Sonne kreist. Wenn man derart die Verbindung zur Wirklichkeit leugnet, erfährt man auch nichts über die Wirklichkeit.

English: Not at all. The illogical separation of experimental testability from explanation is precisely what the Inquisition tried to force upon Galileo. It would have been willing to accept the practical consequences of his heliocentric view if only he had refrained from claiming that the Earth really revolves around the Sun. If you deny the connection to reality in this way, you learn nothing about reality either.

SPIEGEL: Wenn die Menschen die Wirklichkeit des Multiversums akzeptieren würden, würde das irgendetwas praktisch ändern?

English: If people accepted the reality of the multiverse, would it make any practical difference?

Deutsch: Aber ja! Das Leugnen der physikalischen Wirklichkeit hat den Fortschritt schwerstens behindert. Die von mir gefundene Theorie des Quantencomputers hätte 30 bis 50 Jahre früher entdeckt werden können. Außerdem halte ich es für fatal, dass die Leute inzwischen glauben, die Naturwissenschaft könne die Wirklichkeit nicht mehr beschreiben. Es ist ehrbar geworden anzunehmen, es gebe gar keine objektive Wirklichkeit. Und die größten Physiker verleihen dieser üblen Philosophie dann auch noch die Weihen der Wissenschaft.

English: Certainly. Denying physical reality has severely obstructed progress. The theory of the quantum computer that I discovered could have been discovered 30 to 50 years earlier. I also think it is disastrous that people have come to believe that science can no longer describe reality. It has become respectable to suppose that there is no objective reality at all. And the greatest physicists then go so far as to bestow the blessing of science upon this pernicious philosophy.

SPIEGEL: Einer davon ist der Nobelpreisträger Steven Weinberg. Der spottet, er sei »zu beschäftigt, um sich mit Kram wie der Viele-Welten-Interpretation zu befassen«.

English: One of them is the Nobel laureate Steven Weinberg. He scoffs that he is “too busy to concern himself with stuff like the many-worlds interpretation.”

Deutsch: Das ist schon richtig, die Viele-Welten-Theorie wird von nicht mehr als etwa 10 Prozent der Physiker vertreten. Die anderen 90 Prozent sind immer noch gefangen in dieser – ja, wie soll ich es nennen? – Verleugnung der Wahrheit ihrer eigenen Theorie.

English: That is true: the many-worlds theory is supported by no more than about ten per cent of physicists. The other 90 per cent remain trapped in this, well, what shall I call it, denial of the truth of their own theory.

SPIEGEL: Aber warum sind sie das? Vielleicht weil sie die Vorstellung, dass da so verdammt viele Kopien von ihnen in all diesen vielen Universen herumlungern, allzu abenteuerlich finden?

English: But why are they? Perhaps because they find the idea that so many damned copies of themselves are loitering around in all these universes rather too outlandish?

Deutsch: So schwer ist diese Idee doch gar nicht zu begreifen, jedenfalls viel leichter als etwa die allgemeine Relativität. Ich glaube, dass jeder intuitiv eine Vorstellung davon hat, was die Parallele-Universen-Theorie bedeutet, zumindest sehr grob. Aber was meint einer, wenn er sagt, dass die Gleichzeitigkeit relativ ist und der Raum gekrümmt – ich glaube, sehr wenige Leute können das begreifen.

English: The idea is not really that difficult to grasp, certainly far easier than general relativity, for example. I think everyone has at least a very rough intuitive sense of what the parallel-universes theory means. But what does someone mean when he says that simultaneity is relative and space is curved? I think very few people can comprehend that.

SPIEGEL: Nun haben Sie uns aber immer noch nicht gesagt, welche Experimente die Existenz eines Multiversums beweisen.

English: But you still have not told us which experiments prove that a multiverse exists.

Deutsch: Oh, im Grunde alle Interferenzexperimente.

English: Oh, essentially all interference experiments.

SPIEGEL: Das müssen Sie uns genauer erklären.

English: You will have to explain that in greater detail.

Deutsch: Also, wenn Sie Licht durch ein winziges rundes Loch werfen, dann wird auf einem Schirm dahinter ein heller Fleck erscheinen, das Abbild des Lochs. Und wenn das Licht durch zwei Löcher fällt, entstehen zwei Flecken. Wenn Sie nun aber die beiden Löcher immer näher aneinander heranrücken, dann werden die Abbilder der Löcher irgendwann aufhören auszusehen wie Flecken. Stattdessen erscheint ein kompliziertes geometrisches Muster, das viel komplizierter als die Form der Löcher ist.

English: If you shine light through a tiny round hole, a bright spot will appear on a screen behind it: an image of the hole. If the light passes through two holes, two spots appear. But if you move the two holes closer and closer together, their images eventually stop looking like spots. Instead, a complicated geometrical pattern appears, one far more complex than the shape of the holes.

SPIEGEL: Sie reden von Interferenzmustern.

English: You are talking about interference patterns.

Deutsch: Richtig. Sie werden verursacht, weil das Licht, das durch das eine Loch fällt, das Licht, das durchs Nachbarloch fällt, beeinflusst.

English: Correct. They arise because the light passing through one hole affects the light passing through the neighbouring hole.

SPIEGEL: Und das soll jetzt das Multiversum beweisen?

English: And this is supposed to prove the multiverse?

Deutsch: Warten Sie. Stellen Sie sich nun vor, Sie schicken das Licht aus Ihrer Lichtquelle durch so viele dunkle Filter, dass am Ende nur noch ein einziges Photon pro Minute durch die Lochblende geht. Was passiert? Eigentlich würden Sie doch erwarten, dass dieses Photon, das ja jetzt nicht mehr gestört werden kann durch etwaige Nachbar-Photonen, entweder auf dem Abbild des einen oder auf dem Abbild des anderen Lochs landet.

English: Wait. Now imagine sending the light from your source through so many dark filters that, in the end, only a single photon per minute passes through the aperture. What happens? You would expect this photon, which can no longer be disturbed by any neighbouring photons, to land either on the image of one hole or on the image of the other.

SPIEGEL: Sie meinen, die Interferenz müsste verschwinden.

English: You mean the interference ought to disappear.

Deutsch: Genau. Aber das Erstaunliche ist: Sie tut es nicht. Auch wenn nur ein einziges Photon durch unsere Lochblende fliegt, landet es auf einem Platz, der dem komplexen Interferenzmuster entspricht! Was folgt daraus? Offensichtlich muss doch, während unser Photon durch das eine Loch geschossen ist, irgendetwas anderes durch das andere Loch gelangt sein, um mit unserem sichtbaren Photon zu interferieren.

English: Exactly. But the astonishing thing is that it does not. Even when only a single photon flies through our aperture, it lands at a position corresponding to the complex interference pattern. What follows from that? Evidently, while our photon was passing through one hole, something else must have passed through the other in order to interfere with our visible photon.

SPIEGEL: Und was ist dieses Etwas?

English: And what is this something?

Deutsch: Nun, experimentell lässt sich nachweisen, dass sich dieses Etwas exakt so verhält wie ein Photon. Zum Beispiel können wir ihm Spiegel in den Weg stellen und Linsen, irgendwelches optisches Gerät – das Ding, das die Bahn unseres sichtbaren, messbaren Photons beeinflusst, wird sich benehmen, als wäre es selbst ein Photon.

English: Well, experiments can establish that this something behaves exactly like a photon. For example, we can place mirrors, lenses or any sort of optical equipment in its path. The thing influencing the trajectory of our visible, measurable photon will behave as though it were itself a photon.

SPIEGEL: Und Sie meinen jetzt, jenes unsichtbare Photon, das das sichtbare aus der Bahn wirft, stammt aus einem anderen Universum?

English: And you are now saying that the invisible photon which deflects the visible one comes from another universe?

Deutsch: Sie sagen es. Und damit haben Sie Ihren experimentellen Beweis für die Existenz des Multiversums.

English: You said it. And there you have your experimental evidence for the existence of the multiverse.

SPIEGEL: Und warum interferieren die beiden Universen ausgerechnet in dem Moment, in dem Sie Ihr Interferenzexperiment machen?

English: And why do the two universes interfere at precisely the moment when you perform your interference experiment?

Deutsch: Sie tun’s die ganze Zeit, nicht nur, wenn wir Experimente machen. Deshalb sind wir auch überall umgeben von indirekten Hinweisen auf parallele Universen. Nehmen Sie die Tatsache, dass Materie fest ist. Wenn die klassische Physik wahr wäre und Atome wirklich nur in einem einzigen Universum existieren und einander anziehen würden, dann könnte man beweisen, dass es so etwas wie feste Materie gar nicht gibt. Es gäbe dann weder Ihren Stift noch Ihr Tonband, und Sie selbst gäbe es auch nicht. Jede Menge Eigenschaften der Welt sind verursacht durch Quanteninterferenz, durch das Zusammenwirken vieler paralleler Universen.

English: They do it all the time, not only when we perform experiments. That is why we are surrounded everywhere by indirect evidence of parallel universes. Consider the fact that matter is solid. If classical physics were true, and atoms really existed in only one universe and attracted one another, it could be proved that there could be no such thing as solid matter. Neither your pen nor your tape recorder would exist, and neither would you. A great many properties of the world are caused by quantum interference, by the interaction of many parallel universes.

SPIEGEL: Sie haben vorhin gesagt, Ihre Theorie des Quantencomputers verdankten Sie letztlich der Multiversumsidee. Gibt es sonst noch praktische Folgerungen aus der Viele-Welten-Theorie?

English: You said earlier that you ultimately owed your theory of the quantum computer to the idea of the multiverse. Are there any other practical consequences of the many-worlds theory?

Deutsch: Zeitreisen zum Beispiel – wobei die eher den theoretischen als den praktischen Anwendungen zuzurechnen sind. Allerdings wendet sich die Quantentheorie nur der Frage zu, was passieren würde, wenn wir eine Zeitmaschine hätten. Vielleicht lautet die für einen Laien interessantere Frage: Wie machen wir eine?

English: Time travel, for example, although that belongs more among the theoretical than the practical applications. Quantum theory, however, addresses only the question of what would happen if we had a time machine. Perhaps the more interesting question for a layperson is: how do we make one?

SPIEGEL: Also, verraten Sie es uns: Wie machen wir eine?

English: Then tell us: how do we make one?

Deutsch: Das berührt ein komplett anderes Gebiet der Physik – das gehört ins Reich der allgemeinen Relativität. Einen Pfad in die Vergangenheit können wir nur bauen, wenn wir irgendeine brutale Unterbrechung der Raumzeit nutzen. Dazu müsste es gelingen, ein schwarzes Loch in sehr schnelle Rotation zu versetzen.

English: That touches upon a completely different area of physics: it belongs to the realm of general relativity. We can build a path into the past only by exploiting some violent disruption of spacetime. To do that, we would have to set a black hole spinning extremely rapidly.

SPIEGEL: Und das soll möglich sein?

English: And that is supposed to be possible?

Deutsch: Irgendwann vermutlich schon.

English: Presumably, one day.

SPIEGEL: Aber was passiert, wenn Sie dann in die Vergangenheit reisen und etwas schaffen wie das Großvater-Paradox: Sie töten Ihren Großvater, noch bevor er Ihren Vater zeugen konnte?

English: But what happens if you then travel into the past and create something like the grandfather paradox: you kill your grandfather before he could father your father?

Deutsch: Vergessen Sie nicht: Wer aus der Zeitmaschine heraustritt, betritt ein anderes Universum. Dort können Sie verursachen, was Sie wollen – es wird nicht das Universum sein, aus dem Sie stammen.

English: Do not forget: whoever steps out of the time machine enters another universe. There you can cause whatever you like. It will not be the universe from which you came.

SPIEGEL: Das heißt also, der Mord an meinem Großvater hätte nur zur Folge, dass ich in diesem anderen Universum nie geboren würde.

English: So murdering my grandfather would merely mean that I was never born in this other universe.

Deutsch: Ja, richtig. Aber ich finde, diese Mordgeschichten verwirren das Ganze nur. Nehmen wir an, Sie töten ihn nicht; dann wird er einen Sohn zeugen, der wiederum ein Kind zeugt. Und Sie werden dieses Kind – Ihr jüngeres Selbst – treffen. Es wird zwei Kopien von Ihnen geben.

English: Yes, that is right. But I think these murder stories only confuse matters. Suppose you do not kill him. He will then father a son, who in turn has a child. And you will meet that child, your younger self. There will be two copies of you.

SPIEGEL: Und Sie könnten dann Ihrer jüngeren Kopie sagen: »Ach, hör doch auf mit deiner Theorie des Quantencomputers, das habe ich alles schon erledigt«?

English: And you could then tell your younger copy, “Oh, stop working on your theory of the quantum computer. I have already taken care of all that”?

Deutsch: Ja. Ich glaube, wenn wir je Zeitmaschinentechnik beherrschen sollten, werden wir sie genau zu diesem Zweck nutzen: zur Informationsweitergabe – und nicht, um selbst auf Reisen zu gehen.

English: Yes. I think that if we ever master time-machine technology, we will use it for precisely this purpose: to transmit information, not to travel ourselves.

SPIEGEL: Warum sollte man aufs Reisen verzichten?

English: Why refrain from travelling?

Deutsch: Schon allein, weil niemand von einer Reise durch die Zeit zurückkehren kann. Sie würden für immer aus diesem Universum verschwinden. Alle Ihre Freunde würden Sie unwiderruflich verlieren – und Sie wiederum Ihre Freunde. Wobei Sie jüngere Ausgaben von ihnen treffen würden, allerdings in Begleitung einer jüngeren Kopie Ihrer selbst.

English: If only because no one can return from a journey through time. You would disappear from this universe forever. All your friends would lose you irrevocably, and you would lose your friends in turn. You would meet younger versions of them, though they would be accompanied by a younger copy of you.

SPIEGEL: Nun gut, wir würden also eher Informationen, Daten in die Vergangenheit schicken. Und wozu soll das gut sein?

English: Very well, then: we would be more likely to send information, data, into the past. What good would that do?

Deutsch: Zeitmaschinen werden die Informationsverarbeitung revolutionieren – mindestens so grundlegend, wie es die Erfindung des Computers getan hat.

English: Time machines will revolutionise information processing, at least as fundamentally as the invention of the computer did.

SPIEGEL: Was, glauben Sie, würde man konkret damit anfangen?

English: What do you think we would actually do with them?

Deutsch: Wir hätten viel mehr Zugang zu Informationen. Wir könnten zum Beispiel etwas über Universen erfahren, in denen eine bestimmte Wirtschaftspolitik ausprobiert wurde, und so herausfinden, was dort ein Jahrzehnt später los war. Solche Daten können von heute aus gesehen extrem wertvoll sein. Wir wären also bald angelangt in einer Ära, in der wir Dinge, die wir heute durch Versuch und Irrtum lösen müssen, dann durch Informationsverarbeitung klären könnten.

English: We would have far greater access to information. We could, for example, learn about universes in which a particular economic policy had been tried and discover what was happening there a decade later. From our present perspective, such data could be extremely valuable. We would soon enter an era in which matters that we must now resolve through trial and error could instead be settled through information processing.

SPIEGEL: Wann rechnen Sie mit dem ersten Prototyp?

English: When do you expect the first prototype?

Deutsch: Na ja, wir reden hier über Science-Fiction, die vielleicht Jahrmillionen in der Zukunft liegt. Ich habe so leichthin gesagt, man müsse ein schwarzes Loch in Rotation versetzen – in Wahrheit müsste es mindestens die dreifache Masse unseres Sonnensystems besitzen. Und es müsste mit der Präzision eines Quantencomputers manipuliert werden.

English: Well, we are talking about science fiction that may lie millions of years in the future. I said rather casually that one would have to set a black hole spinning. In reality, it would have to possess at least three times the mass of our solar system, and it would have to be manipulated with the precision of a quantum computer.

SPIEGEL: Würden Quantencomputer oder Zeitmaschinen es dem Menschen ermöglichen, die Welt noch tiefer zu verstehen? Und hätten angesichts der damit verbundenen Einsichten die Quanten- und die Multiversumstheorie überhaupt noch Bestand?

English: Would quantum computers or time machines allow humanity to understand the world more deeply still? And, in light of the insights they brought, would quantum theory and the multiverse theory continue to stand?

Deutsch: Ich glaube nicht, dass irgendetwas in der Welt unerklärbar ist. Und ich glaube auch nicht, dass das Wachstum des Wissens je zu einem Ende kommen wird. Deswegen wird die Quantentheorie auch nicht die letzte Theorie sein – es gibt keine letztgültige Erklärung.

English: I do not believe that anything in the world is inexplicable. Nor do I believe that the growth of knowledge will ever come to an end. That is why quantum theory will not be the final theory either. There is no ultimate explanation.

SPIEGEL: Aber es wird jedenfalls immer neue, immer unbegreiflichere Erklärungen geben?

English: But there will always be new and increasingly incomprehensible explanations?

Deutsch: Davon bin ich überzeugt. Die Physik hat stets neue, immer bessere Erklärungen hervorgebracht. Und es wäre sehr überraschend, wenn sich das, was am Ende an die Stelle der heutigen Multiversumstheorie tritt, als weniger merkwürdig erweisen würde.

English: I am convinced of that. Physics has always produced new and ever better explanations. And it would be very surprising if whatever eventually replaces today’s multiverse theory turned out to be less strange.

SPIEGEL: Je merkwürdiger, desto besser?

English: The stranger, the better?

Deutsch: Nein. Merkwürdigkeit ist ja kein Selbstzweck. Aber zum einen werden grundlegende physikalische Erklärungen immer merkwürdiger, weil sie sich immer weiter entfernen von unserer Alltagserfahrung. Und zum anderen wären Erklärungen, die weniger sonderbar sind, längst entdeckt worden.

English: No. Strangeness is not an end in itself. But fundamental physical explanations become increasingly strange because they move ever farther away from our everyday experience. And explanations that were less peculiar would have been discovered long ago.

SPIEGEL: Also müssen wir uns damit abfinden, dass uns die Physiker immer noch bizarrere Ideen präsentieren werden?

English: So we must resign ourselves to physicists presenting us with ever more bizarre ideas?

Deutsch: Ja. Das Bizarre muss man einfach akzeptieren – weil die Welt bizarr ist.

English: Yes. We simply have to accept the bizarre, because the world is bizarre.

SPIEGEL: Herr Deutsch, wir danken Ihnen für dieses Gespräch.

English: Mr Deutsch, thank you for speaking with us.

Das Gespräch führten die Redakteure Johann Grolle und Rafaela von Bredow.*

English: The interview was conducted by editors Johann Grolle and Rafaela von Bredow.*

* Im Wintergarten seines Hauses in Oxford.

English: In the conservatory of his home in Oxford.

2005-01-01 Discovery Ultra ScienceTime Travel

00:04:27 - 00:04:52

Now could that mean that there are time travelers here now in our world? Visitors from the future who’d been trapped in their travels through history? Some scientists believe such visits may have been crucial to our very existence.

Read the response at Edge Source collection

2004-01-01 EdgeWhat’s Your Law?

Deutsch’s Law

Every problem that is interesting is also soluble.

Corollary #1

Inherently insoluble problems are inherently boring.

Corollary #2

In the long run, the distinction between what is interesting and what is boring is not a matter of subjective taste but an objective fact.

Corollary #3

The problem of why every problem that is interesting is also soluble, is soluble.

2003-01-01 Quantum Computation Lecture 6Grover's Search Algorithm

Duration: 00:41:05

Transcript

David Deutsch

00:01:00 - 00:03:06

Today, I’m going to describe a remarkable quantum algorithm which was invented by Lov Grover in 1996. It’s a search algorithm suitable for a very broad category of computational task known as algorithmic searching. Algorithmic searching is just exhaustive searching for a number with a given mathematical property. It’s a programmer’s last resort in addressing computational tasks where there’s a quick way of verifying that a given number is the answer once you have it, but no easy way of constructing the answer. In other words, the task is we’re given a criterion for whether a number is the answer or not. We’re given it in the form of an algorithm $F$ that delivers an output $+1$ or $-1$ where minus one means that the input meets the criterion and plus one that it doesn’t. And we have to find a value $t$ , the target value, such that $F(t)=-1$ . Suppose there are $N$ possible values that have to be searched. And suppose for simplicity that $N=2^L$ for some integer $L$ so that we can encode each possible value uniquely in a register of $L$ qubits. By the way, note that the conventional representation for binary numbers stored in qubits is that the eigenvalue $+1$ state represents the binary digit zero and the $-1$ state represents the digit one. Anyway, suppose that there’s exactly one value that meets the criterion in the range that we want to search.

00:03:06 - 00:05:12

That’s the target value $t$ with $F(t)=-1$ . For all other values $x\ne t$ , $F(x)=+1$ . And we’re given a classical reversible algorithm for evaluating $F$ on an arbitrary $L$ -bit input. Like last time, we’re given this in the form of an oracle. It’s some sort of computer operating coherently on a certain number of qubits that we can give it as input and then get out again after a fixed period. But we’re not allowed to look inside it. Now that models the fact that we’re only doing this search in the first place because we’ve exhausted all ways of simplifying the problem analytically. And we’re looking for an exhaustive search algorithm, one that doesn’t depend on our knowing anything about the structure of the function $F$ . So the oracle computes $F$ on an $L$ -bit input and gives us a one-bit output, whether it meets the criterion or not. But again, because of reversibility, it must actually have the same number of outputs as inputs. So it must operate something like this. For inputs $x$ and $y$ , where $x$ can take one of $2^L$ values and $y$ can take one of two values, $+1$ or $-1$ , it gives an output $(x,yF(x))$ . So the oracle as a whole operates on $L+1$ qubits. And it may also contain an unknown number of internal qubits that we never see. How many evaluations of $F$ , how many invocations of the oracle, are we going to need to find the unique value of $t$ such that $F(t)=-1$ ?

00:05:12 - 00:07:31

Well, by a similar argument to last time, it’s clear that we may need to invoke the oracle as many as $N-1$ times. It’s $N-1$ rather than $N$ , because if the conditions of the task say that there’s exactly one value meeting the criterion and we’ve checked all but one of the possible values, then we know that the answer is the remaining value. But that’s the only break we get. We could check the values randomly and then on average we’d probably need about $N/2$ oracle invocations, probably. Using quantum computation in which the oracle receives different inputs in different universes, we can do a great deal better than that and that’s what Grover’s algorithm does. Grover’s algorithm uses three simple subroutines, so I’ll describe those separately first and then put them together. The first subroutine involves the Hadamard gate that I introduced last time. It consists of applying a Hadamard gate to each of $L$ qubits. In our case, it’s the $L$ qubits in our $L$ -qubit register, like this. I’ll call this operation $\mathbf{H}$ to distinguish it from a single qubit Hadamard gate. The algorithm begins, as I advocated last time, with all $L$ of these qubits in their blank initial state where they all have sharp values plus one and I’ll call that state $|0\rangle$ . It’s a state of $L$ qubits all in state plus one. The effect of H on the blank state is $\mathbf{H}|0\rangle=2^{-L/2}(|+\rangle+|-\rangle)^{\otimes L}$ .

00:07:31 - 00:11:04

$L$ factors in the tensor product. This state of the $L$ qubits plays an important role in the analysis of this algorithm, so let me give it a name, $\mu$ . So $\mathbf{H}|0\rangle=|\mu\rangle$ and since the Hadamard gate is its own inverse, it must also be true that $\mathbf{H}|\mu\rangle=|0\rangle$ . If we multiply out mu, we see that it consists of a superposition of states with all $N=2^L$ possible sequences of plus and minus values representing the $2^L$ possible values of $x$ . Hence, if $|x\rangle$ is any computation basis state, the scalar product is $\langle x|\mu\rangle=1/\sqrt{N}$ . Okay, the next subroutine involves the oracle with its $L$ qubits to hold the argument of $F$ and its one auxiliary qubit. I’ll call this the marking subroutine, $M$ . You’ll see why in a moment. It starts with the auxiliary bit being put through a NOT gate and then a Hadamard gate. That’s just to prepare it in the state $\frac{1}{\sqrt{2}}(|+\rangle-|-\rangle)$ . With that as the input for the auxiliary qubit, the effect of the oracle on a sharp state of the first $L$ qubits with the value $x$ is, well as always, $x$ passes through unchanged and the value of $F(x)$ is exclusive-ORed into the auxiliary qubit, which in this case just multiplies the overall state by $F(x)$ . That means that the auxiliary qubit is left unchanged. $M$ could run repeatedly just recycling the same qubit which never changes. So it never receives any information, which means in turn that the first $L$ qubits never receive any information from it. So these first $L$ qubits by themselves evolve autonomously, which means as I said in lecture four, that they undergo coherent evolution during the process $M$ . And the effect of $M$ is that of a unitary matrix acting on the state of those $L$ qubits alone. So we don’t have to mention the auxiliary qubit explicitly when we analyse algorithms involving $M$ , just as we don’t explicitly mention the unseen qubits in the oracle when it computes $F$ .

00:11:11 - 00:13:50

The net effect of $M$ on the $L$ qubits holding $x$ can be summarised as $M|x\rangle=F(x)|x\rangle$ . Explicitly the unitary matrix that is $M$ is $I-2|t\rangle\langle t|$ . That’s because the computation basis states are orthonormal. So $M|t\rangle=-|t\rangle$ and $M$ operating on any other computation basis state leaves the state unchanged. Hence also the effect of $M$ on the state $\mu$ is $M|\mu\rangle=\frac{1}{\sqrt N}M\sum_x|x\rangle=\frac{1}{\sqrt N}\sum_x F(x)|x\rangle$ . Now for all but one value of $x$ , $F(x)$ is $+1$ , only $F(t)$ is $-1$ . Now you can see why I called this the marking operation. It marks the target term in the superposition by changing its sign. And we can write the effect of $M$ in terms of states I’ve already named like this. The third subroutine, which I’ll call $B$ , is just marking the blank state. Or for convenience actually, we’re going to mark everything but the blank state. So this operation is like $M$ , except that instead of the oracle, it uses a network that performs the NAND operation on all $L$ inputs and exclusive-ORs the result into the auxiliary qubit. That means that for computation basis state inputs, the auxiliary qubit is flipped unless the first $L$ inputs are all plus one. And we’ll use the same trick of using a Hadamard gate to prepare the auxiliary qubit in a state $\frac{1}{\sqrt{2}}(|+\rangle-|-\rangle)$ , giving us an $L$ -qubit operation $B$ .

00:13:51 - 00:16:38

The effect of $B$ is that $B|0\rangle=|0\rangle$ , and $B$ on any other computation basis state is minus that state. So $B=2|0\rangle\langle0|-I$ . Those are the ingredients of Grover’s algorithm. All of them are operations on $L$ qubits. There’s the operation $\mathbf{H}$ , which transforms the blank state to the grand superposition $\mu$ and vice versa. $M$ , which marks the target state, and $B$ , which marks all computation basis states except the blank state. Grover’s algorithm consists of starting with a blank initial state, perform $\mathbf{H}$ , which makes the state $\mu$ , and then perform a certain number of iterations of the sequence $M$ , then $\mathbf{H}$ , then $B$ , then $\mathbf{H}$ again. So the net effect of such an iteration called the Grover iteration on an arbitrary state is given by the unitary matrix $\mathbf{H}B\mathbf{H}M$ . They’re in reverse order to the order they’re performed in because it’s the rightmost factor that hits the ket first. We’ll see in a moment how many of these Grover iterations you have to do. That’s it. Now what does all that do and why does it work? There’s a beautiful geometric interpretation of what it’s doing. Let’s take a look at a two-dimensional plane through the $2^L$ -dimensional Hilbert space of these $L$ qubits. The plane is defined as the one containing both the target state $|t\rangle$ and the superposition $|\mu\rangle$ .

00:16:39 - 00:19:03

Now $|t\rangle$ and $|\mu\rangle$ are very nearly orthogonal for large $N$ . But never quite orthogonal. Their scalar product is $1/\sqrt{N}$ . Let’s define a state that is orthogonal to $|t\rangle$ here. Call it $\phi$ . And now consider a family of states all lying in this plane. $\cos\theta|\phi\rangle+\sin\theta|t\rangle$ , where $\theta$ is a real parameter. The state that we want to end up in, namely $|t\rangle$ , is in this parameterized family with $\theta=\pi/2$ . And at the moment after our initial Hadamard operation, just before we start the iterations, we’re in the state $|\mu\rangle$ , which is also of this form, but with $\theta=\arcsin(1/\sqrt{N})$ . For a general state of this form, what’s the effect of one Grover iteration? Well, the Grover iteration begins with an $M$ operation, and we know what the effect of $M$ is. It changes the sign of the coefficient of the target $|t\rangle$ , and it leaves all other computation basis states unchanged. Hence, the effect of $M$ on a general state in this plane is to reflect it in the horizontal axis. After $M$ , we do $\mathbf{H}B\mathbf{H}$ , and we can simplify that if we just substitute what $\mathbf{H}$ and $B$ are as unitary matrices. Substituting the unitary matrices gives the reflection formula.

00:19:03 - 00:21:14

We get $\mathbf{H}B\mathbf{H}=2|\mu\rangle\langle\mu|-I$ , which has no effect on the state $|\mu\rangle$ , but changes the sign of any state perpendicular to $\mu$ . In other words, the effect of $\mathbf{H}B\mathbf{H}$ on an arbitrary state in this plane is to reflect it in the line $\mu$ . The product of these two reflections, namely the Grover iteration as a whole, is a rotation. You can easily prove that it’s a rotation through an angle $2\arcsin(1/\sqrt{N})$ . Exactly twice this angle, and in the anti-clockwise direction. In other words, towards $|t\rangle$ . So, each Grover iteration rotates the state a little closer to our target. Until we go past it and then get further away from it again. So, it’s vital to choose the right number of iterations. And the right number will be $\pi/2$ divided by the angle that the rotation rotates the state by. Except that because we start the rotation at half that angle away from the horizontal, that means that the state will be closest to the target $|t\rangle$ after the integer part of that number of operations. Namely, $\left\lfloor\frac{\pi}{4\arcsin(1/\sqrt N)}\right\rfloor$ .

00:21:14 - 00:23:19

Roughly speaking, that’s a constant times $\sqrt{N}$ . So, Grover’s algorithm completes its search using about the square root of the number of oracle calls required using classical computation. That’s a tremendous saving. You can search a million possibilities in only about a thousand iterations. A trillion possibilities in only a million iterations. Now, just a small detail. The state at the end won’t have exactly hit the target state. Except when $N=4$ , by the way. You might like to check that interesting special case where a single oracle call is actually enough to do the whole search. But for all other values, the output of Grover’s algorithm is not a computation basis state at all. It’s a state of the $L$ qubits that’s close to but not equal to the target state. Does that matter? It doesn’t because, well, suppose Grover’s algorithm delivers a state $\cos\epsilon|t\rangle+\sin\epsilon|\phi\rangle$ . Well, from the argument I’ve just given, we can expect $\epsilon$ to be less than about $1/\sqrt{N}$ , which is very small. Well, suppose we then measure the output value of the $L$ qubits and check whether it’s correct, whether it’s the value such that $F(t)=-1$ . We can do that very easily by using the oracle one more time, this time with the auxiliary qubit initially blank, with the sharp value $+1$ , like this. And then running the oracle will flip the sign of the component that goes with $|t\rangle$ and leave the other one alone.

00:23:19 - 00:25:38

What is then the expectation value of the $Z$ component of that auxiliary qubit? If the output had been sharply at the target value, then $Z$ for the auxiliary qubit would also be sharp with value minus 1. If its expectation value is anything above that, that means that the answer was wrong in some universes, and we can work out how many. The expectation value of $Z$ is $\operatorname{tr}(\rho Z)$ . And $\rho$ for this pure state gives the expectation value of $Z$ . So we get the expectation value $-\cos^2\epsilon+\sin^2\epsilon$ , which is $-1+O(\epsilon^2)$ . So the probability of seeing a wrong answer, that is, the proportion of universes in which the answer will be wrong, is of order $O(1/N)$ . Very small. In practice, many other small errors will be introduced by physical factors such as noise, or the fact that the computer doesn’t perfectly match its ideal design. But it’s in the nature of algorithmic searching that we can’t be led astray by such errors. We check the result, and in the rare cases when it’s wrong, we just run the algorithm again. You’ll see in the worked examples that Grover’s algorithm can also be used in cases where more than one value satisfies the target criterion, and that it speeds up such searches by the same factor.

00:25:41 - 00:27:33

It has been proved that the algorithm is optimal. That is, no algorithm, quantum or classical, can ever do an exhaustive algorithmic search faster than Grover’s. Though bear in mind that the proof of this applies only to the oracle version of the searching task, which is an idealization. In real life algorithmic searches, we often do know something about the structure of the function we’re investigating, and this knowledge can be used to speed up searching. The possibility is still open that in some cases it speeds up quantum searching by more than classical searching. In other cases though, it offsets the benefit of quantum searching altogether. I once mentioned in a popular article about quantum computers that one day, Grover’s algorithm will be used to make super powerful chess-playing machines. My thought was that since the best classical chess-playing algorithms work by exhaustively searching all possible continuations of the game from a given position, Grover’s algorithm would allow one to search the square of the number of possibilities in a given number of steps, a huge improvement. But no, as was pointed out by Richard Cleve, the nature of that kind of search, which has a tree-like structure, is that most of the work of computing any of the final positions is shared with many other final positions. And under those circumstances, the advantage of Grover’s algorithm over conventional tree searching disappears. And the same seems to be true of game playing in general.

00:27:33 - 00:29:37

Though I should add that that doesn’t rule out the possibility that there may be other quantum algorithms for playing certain games. Grover’s paper in which he first published his algorithm was called A Fast Quantum Mechanical Algorithm for Database Search. That’s a slightly misleading name because the algorithm is probably at its least useful in searching databases. First of all, if your database is a custom-built physical object like an oracle containing essentially a read-only memory of $N$ bits or $N$ qubits, then it would surely be easier just to store the single value $t$ and it could just tell you that value on request and you wouldn’t have to do any searching. On the other hand, if the database is not known in advance or if the query you’re going to make is not known in advance, suppose the data was some recently gathered data from the search for extraterrestrial intelligence and you want to search it for a given pattern, well then the act of reading the data into a quantum computer memory or imprinting it permanently on an oracle is itself a task that takes of the order of $N$ operations. So Grover’s algorithm would only speed up database searching by perhaps a constant factor and even that factor might be swamped by the extra technological difficulty of doing coherent quantum computations as opposed to incoherent classical ones. Or it might not, but in any case, once there are quantum computers, Grover’s algorithm will be unrivaled when it comes to algorithmic searching which includes a wide variety of very important tasks.

00:29:37 - 00:31:38

There’s a large number of tasks, some of which currently take up a lot of computer time where there’s at present no known alternative but the hard slog of try $F(1)$ , try $F(2)$ , try $F(3)$ and so on and where having tried any number of such guesses gives you little or no useful information about where the target may be among the remaining possibilities. Perhaps the archetypal case of this is cryptanalysis, the science of reading encrypted messages. One of the basic computational tasks of cryptanalysis comes up when we know the cryptographic system that was used by the writer of an encrypted message but we don’t know the key that they used for that session. We know the decoding algorithm, but not the key that would make it decode the ciphertext correctly. Say we have an encrypted message, ciphertext C and we know that the original plain text was written in English, say. To solve that problem by algorithmic search means to try one value of k after another until we find one for which the decoding algorithm yields a message that seems to be an English text. More generally, the problems for which algorithmic search is useful are in the complexity class called NP, basically problems where it’s easy to recognise a solution once you have it. And they especially include the important subset of NP called NP-complete which includes problems such as the travelling salesman problem.

00:31:38 - 00:33:43

Grover’s algorithm will speed up the solution of such problems by a factor of $\sqrt{N}$ . That doesn’t make such problems tractable on a quantum computer in the language of complexity theorists because the technical definition of a tractable task is one which for very large $N$ requires only some power of $\log N$ steps. But if you’re a programmer contemplating an algorithmic search of a trillion possibilities to solve some vital design problem or whatever, it’ll be a great help to have a way of doing that that uses only a million function calls instead of a trillion. With cryptanalysis, the numbers are much larger still and in general, the harder the task is, the greater the advantage Grover’s algorithm will confer. Let me try to give a feel for the numbers involved. Imagine some future quantum computer that’s performing Grover’s algorithm. Say it has a quantum processor capable of performing a hundred million calls of criterion $F$ per second. And suppose it’s engaged in a particularly arduous search through a space of $10^{30}$ possible solutions looking for the one with $F(t)=-1$ . How long will that take? Well, it’ll require about $10^{15}$ calls of $F$ which at 100 million calls per second will take about $10^7$ seconds or about 4 months. A classical computer with the same processor speed would require about $5\times10^{29}$ calls of $F$ which would take it $5\times10^{21}$ seconds or very nearly the age of the universe.

00:33:43 - 00:35:39

That’s the size of the practical benefit of the quantum search algorithm over the classical. But the theoretical implication is even more mind-boggling because we know how Grover’s algorithm works. It isn’t just doing $10^{15}$ evaluations of $F$ in parallel which you could mimic classically by, say, making all the computers on earth work on nothing but this problem. The quantum processor is doing $10^{30}$ possible computations of $F$ in parallel every time it’s invoked. If all the silicon in the whole of our planet were made into microchips of 1 cubic millimetre each and they were all set performing different computations there’d still be fewer distinct computations going on in that giant parallel computer than there would be in a single quantum processor performing Grover’s algorithm. That’s a measure of the complexity of structure and process that exists in ordinary matter just beyond our perception because quantum processes are of course going on all the time everywhere. In a quantum computer, some small part of that complexity is put to good use. Thank you.

2003-01-01 Quantum Computation Lecture 5A Quantum Algorithm

Duration: 00:45:59

Transcript

David Deutsch

00:01:14 - 00:06:08

An algorithm is a hardware independent specification of a computation by hardware independent I mean that it doesn’t specify what the computer should be made of only the effect that it should have on information that’s provided to it as input so that if you had access to any technology that allowed you to build a computer a universal computer you could translate the algorithm into a program for that computer which would perform the information processing that the algorithm specifies an algorithm is a way of performing a computational task like sorting a list or factorizing a number in general to specify a computational task you specify properties that the output information should have as a function of the input information so a computational task is a problem to which an algorithm is the solution the algorithm that I’m going to describe today is probably the simplest of all quantum algorithms so I’ll first describe the problem that it solves you’re given a black box containing a computer that’s dedicated to computing a particular function $F$ using reversible classical computation it operates coherently though because part of our task is going to be to use this black box as a component in a quantum computational network you’re not allowed to look inside the box all you’re allowed to do is feed qubits into it at one end and then retrieve them at the other end after a fixed time that’s independent of the input such a box is known as an oracle the idea of having an oracle in the specification of a computational task is a trick much beloved by complexity theorists both classical and quantum because it can greatly simplify the analysis of algorithms that’s because often looking inside the oracle doesn’t help you to perform the task yet it’s quite hard to prove that it doesn’t why that’s so often the case is quite a deep question it’s beyond the scope of these lectures but I’ll explain why it’s plausible in this case in a moment. Okay, the oracle in this problem computes a boolean function $F$ with one Boolean parameter, that is to say $F:\{-1,+1\}\to\{-1,+1\}$ since the oracle performs a reversible computation it can’t just do this $x\mapsto F(x)$ that wouldn’t be reversible the oracle must have an auxiliary input and output and when it computes $F$ what it must really be doing is something like this $(x,y)\mapsto(x,yF(x))$ that’s a reversible computation and if you want to use the oracle to compute $F(x)$ then you just have to make sure that $y$ is initialized to the value $+1$ now there are only four functions that map a single bit to a single bit the identity the not operation or delivering a constant output minus one or one

00:06:08 - 00:07:13

Internally the oracle may be computing something arbitrarily complex or rather one of two arbitrarily complex things depending on the input for instance it might be doing the traveling salesman problem or some other hard problem for one of two graphs depending on the input and then reporting some boolean property of the answer such as whether the shortest path has an even or odd number of steps but however complex the oracle is internally it will be computing one of these four functions that’s why it’s plausible that looking inside the oracle won’t in general help the oracle might contain a complicated network with a large number of qubits doing things in parallel so it might still be a lot faster to run it twice than to work out what it does once

00:07:13 - 00:09:24

Anyway a simple computational task given the oracle would be find out what $F$ is. Well, the only way to do that without looking inside is to run the oracle once for each possible input. The resulting four pairs of outputs tabulate $F$ . Is it possible to perform that task using only one run of the oracle? Well, we can prove that it isn’t. If you’re only allowed to run the oracle once, then whatever else you do you’ll have to pick a particular pair of values for its inputs $x$ and $y$ , and you’ll get output values $x$ , which you already knew, and $yF(x)$ . Only the second of those two outputs even depends on $F$ , so that can’t fully specify what $F$ is because $F$ can be one of four different functions. It turns out that that is true in the quantum case too. You can’t work out what $F$ is without invoking the oracle twice, so quantum computation doesn’t help with that task, which illustrates the general fact that quantum computation doesn’t speed up all computational tasks, only some of them. That’s one reason why quantum complexity theory is fundamentally different from classical complexity theory.

00:09:24 - 00:11:20

Anyway the problem that we are going to solve with a quantum algorithm is not to find out what $F$ is but to determine a property of $F$ , specifically whether $F(1)=F(-1)$ or not or more concisely the task is to compute the product $F(1)F(-1)$ without looking inside the oracle knowing the quantity $F(1)F(-1)$ certainly doesn’t tell you everything about $F$ ; it just tells you one bit of information about it which is half the information in the table of $F$ , so the proof I gave that the previous task requires two calls of the oracle doesn’t hold for this task so is there any way of computing the single boolean value $F(1)F(-1)$ using only one invocation of the oracle for the case of classical computation it’s again easy to prove that there is no way even though it’s only one bit of information this time because again there are only four possible ways of invoking the oracle in a classical computation and in all four the first output is again independent of $F$ and the second depends on only one of the two values $F(1)$ or $F(-1)$ never both so $F(1)F(-1)$ can never be deduced from it

00:11:20 - 00:12:47

Here’s a classical network that finds the answer using two instances of the oracle this is a controlled not gate which we’re using in its capacity as an exclusive OR gate alternatively we could find the answer by invoking one instance of the oracle twice doubling the running time like this the input bits both start at one they pass through the oracle making them $+1$ and $F(1)$ then they each go around a loop the first one passes through a not gate that flips it to minus one and the second goes in again as the auxiliary bit so the final output is $F(1)F(-1)$ strictly speaking this network runs forever so I’ll leave it as an exercise to modify it so that it delivers an output after exactly two oracle invocations

00:12:47 - 00:13:45

What the quantum algorithm does in short is that it uses only one instance of the oracle and it invokes it only once but it invokes it with a different input in different universes which means that the oracle performs different computations in different universes yielding potentially different outputs and the single qubit that holds those two outputs in different universes combines them in an interference phenomenon it exclusive-ORs them a bit like this except that the invocations of the oracle occur in different universes

00:13:45 - 00:16:38

To describe the quantum algorithm that does that as with most quantum algorithms it’s simplest to work in the Schrödinger picture where as you’ll recall the observables are constant matrices and the state changes with time in this problem we can make the further simplification of considering only pure states remember that a quantum system is said to be in a pure state when its density matrix takes the form $|\psi\rangle\langle\psi|$ for some $|\psi\rangle$ psi is then said to be the state vector of the system and we often refer to the state vector as just the state of such a system the effect of a quantum gate operating between times $t$ and $t+1$ on an arbitrary pure state is $\psi(t+1)=U\psi(t)$ where $U$ is a constant unitary matrix characteristic of the gate like the reversible classical algorithms for performing this task this algorithm involves two qubits that’s not counting whatever qubits might be inside the oracle combining two quantum systems in the Schrödinger picture is much the same as doing it in the Heisenberg picture any tensor product of observables one from each system is an observable of the combined system and if both systems are in pure states say $|\psi\rangle$ and $|\phi\rangle$ then the combined system is in a pure state too and its state vector is the tensor product $|\psi\rangle|\phi\rangle$ we write the tensor product of kets without any explicit multiplication symbol like this and that’s a ket of the combined system consider a system of two qubits then in an eigenstate of both their $Z$ observables $Z_1$ and $Z_2$ the four such eigenstates $|a\rangle|b\rangle$ where a and b range over the eigenvalues plus and minus one form an orthonormal basis in the four-dimensional vector space of all pure states of the combined system

00:16:38 - 00:23:51

The space of all pure states that a system could be in is technically a Hilbert space basically that’s just a vector space with a norm and we usually refer to it as the Hilbert space of the given system strictly speaking pure states are always unit vectors that follows from the definition of a pure state that I’ve given because the trace of the density matrix has to be one but it’s no big deal some people like to work with unnormalized vectors and they insert a normalization factor in this formula so unnormalized kets are usually called states too and eigenvectors of observables are called eigenstates the normalized pure states lie on a unit sphere in the Hilbert space and all the non-zero states along any straight line through the origin represent the same physical state when working with pure states we usually pick a fixed basis in the system’s Hilbert space and express the evolving state of the system as a linear combination with time varying coefficients time varying complex coefficients of those basis states a linear combination of states is called a superposition in quantum computation we call whatever basis we choose for doing that the computation basis we can choose any basis as the computation basis but usually some choices are more convenient than others usually the most convenient one is determined by the classical operations that form part of the computation I’ll explain in a classical computer the state of the computation at each step is specified by the sharp values of a set of observables the computational variables so when a quantum computer simulates a classical computer there’s a set of observables which are sharp at the beginning and end of every step and that sharpness is maintained because as I mentioned last time the elementary classical operations in a quantum computer have the property that a particular set of observables evolve autonomously we usually use a notation in which these are the Z observables of the qubits so that means that during periods when our quantum computer is executing only classical operations the Z observables of the qubits are distinguished by being the observables on which those gates perform classical operations note that in general the Z observables will not be sharp during those classical operations because they need not have been sharp when the classical part of the computation began some earlier quantum operations will in general have made them unsharp nevertheless the Z observables form an autonomous evolving system under classical operations and that’s why the basis of eigenstates of those Z observables is often the most convenient one to use as the computation basis I’ll explain classical reversible computations starting in a computation basis state proceed from computation basis state to computation basis state for this and other reasons the dynamics of simple quantum gates are particularly easy to represent in the Schrödinger picture it’s enough to specify the behavior of each member of a set of basis states such as the computation basis for instance the NOT gate is a single qubit gate a single qubit has a two-dimensional Hilbert space so we can completely specify the behavior of the NOT gate by giving its effect on just two states like this $|A\rangle\mapsto|-A\rangle$ where $A$ is $+1$ or $-1$ and these are eigenstates of the qubit’s $Z$ observable this statement means that if the state at time zero is the eigenvalue- $A$ eigenstate of $Z$ then the state at time one after the gate has acted is the eigenvalue- $-A$ eigenstate compare that with the description of the NOT gate in the Heisenberg picture and you begin to see why the Schrödinger picture is preferred for most calculations in the quantum theory of computation another important single qubit gate that I haven’t mentioned before and which is used in the algorithm I’m going to describe is the so-called Hadamard gate named after the mathematician Jacques Hadamard for historical reasons it has this effect $|1\rangle\mapsto \frac{1}{\sqrt{2}}(|1\rangle+|-1\rangle)$ and $|-1\rangle\mapsto \frac{1}{\sqrt{2}}(|1\rangle-|-1\rangle)$ the Hadamard operation is like the NOT operation a square root of the unit operation it’s its own inverse but unlike NOT it doesn’t have a classical analog by the way the square root of two factor is just there to normalize the states

00:23:51 - 00:24:50

Now here’s the definition of the controlled NOT gate in the Schrödinger picture look how simple it is it evolves a system of two qubits in the eigenstate $|x,y\rangle$ of the observables $Z_1$ and $Z_2$ into the state $|x,xy\rangle$ the $|x,y\rangle$ is just another way of writing the tensor product $|x\rangle|y\rangle$ and again compare this with the Heisenberg picture definition of the controlled NOT gate the operation of the oracle is as follows $|x,y\rangle\mapsto |x,yF(x)\rangle$

00:24:50 - 00:29:20

Now for the algorithm our two qubits start in the state where they’re both sharp with the value $+1$ we could think of that as the blank state of our two qubit computer memory by the way when analyzing algorithms in general the algorithm should always start with a state where all the qubits that are not given as part of the task are blank that way your analysis will automatically account for the resources required for any other initialization that you might want to do the quantum network that solves our problem contains a NOT gate three Hadamard gates and the oracle at time 0 the beginning of the computation the state of each qubit is the eigenvalue- $+1$ eigenstate of its $Z$ observable so the overall state at time 0 is the tensor product of those which we can write $|1,1\rangle$ then the second qubit encounters a NOT gate which flips its $Z$ observable to $-1$ next both qubits pass through Hadamard gates and we can just substitute from the definition of the Hadamard gate what the state will be at time 2: $\frac12(|1\rangle+|-1\rangle)(|1\rangle-|-1\rangle)$ then the oracle acts on the qubits once but neither of the $Z$ observables in the input is now sharp so we’re presenting the oracle with four different inputs in different universes next well the oracle will in general take a long time to run but the running time is a constant independent of the input and we’re not interested in what it is for the moment so let’s just call the time when it’s complete time 3 and again we can fill in what $\psi(3)$ will be at this point the computation isn’t quite finished yet we still have that final Hadamard gate to go but let’s just look at this state suppose that $F(1)=F(-1)$ so we can call them both $f$ then the state at time 3 will take this form which factorizes into a tensor product so we see that the qubits are still individually in pure states and the first qubit is in a state proportional to $|1\rangle+|-1\rangle$ okay, that’s if $F(1)=F(-1)$ if they’re unequal then $F(1)=-F(-1)$ which we can call $f$ again and again the qubits are each in a pure state this time qubit 1 is in a state proportional to $|1\rangle-|-1\rangle$

00:29:20 - 00:31:11

If we can distinguish those two states of qubit 1 we shall have determined whether $F(1)$ and $F(-1)$ are the same or different and that’s just what the final Hadamard gate does because Hadamard acting on $\frac{1}{\sqrt{2}}(|1\rangle+|-1\rangle)$ is $|1\rangle$ and Hadamard acting on $\frac{1}{\sqrt{2}}(|1\rangle-|-1\rangle)$ is $|-1\rangle$ and we can summarize that as at time 4 the state is proportional to $|F(1)F(-1)\rangle$ times some state of qubit number 2 so the $Z$ observable of qubit 1 now contains information that depends logically on the outcomes of the computations of both $F(1)$ and $F(-1)$ and it’s sharp although those computations were performed in different universes between times 2 and 3 the interference phenomenon effected by the Hadamard gate between times 3 and 4 combined those values and caused the answer to appear in qubit 1 in all the universes $Z_1$ is sharp at time 4

00:31:11 - 00:31:51

The problem solved by this algorithm has come to be known as the Deutsch problem, and the algorithm I’ve just described is known as the Deutsch algorithm. I should say that that gives me slightly too much credit. The algorithm that I originally proposed was somewhat less elegant and significantly less powerful than this one. I refer you to the worked examples to see in what way it was less powerful. The version I’ve shown you was published in 1998 by Richard Cleve, Artur Ekert, Chiara Macchiavello, and Michele Mosca.

00:31:51 - 00:34:56

The way this algorithm works is typical of how quantum algorithms work in general at least typical of those that have been discovered so far namely we start out with the state of the system being an element of the computational basis that is to say we prepare the Z observables with particular initial values and the first thing we do is unsharpen those observables so that they contain many possible values in our case all four possible values that unsharpening is a quantum operation then we perform a classical reversible computation using coherent quantum components in our case that’s the computation of $F$ by the oracle but this computation is being done with different inputs in different universes giving different outputs too and then these outputs are somehow combined using another quantum operation which gives a sharp answer in this case and more generally as sharp as possible an answer as the output and that means that these computations are interference phenomena observables which are sharp become unsharp and then sharp again an algorithm which performs multiple classical computations on unsharp inputs and then combines them is said to be using quantum parallelism that’s because as I’ve shown the process is reminiscent of classical parallel computation with the difference that you don’t need a second copy of the oracle you use the parallel universe counterparts of the one you’re given. Okay, so by using this quantum algorithm we’ve gained a factor of two in speed but the real significance of the existence of this algorithm is not its speed it’s the fact of being able to manage with just one evaluation of $F$ to obtain information that depends logically on both values it’s the fact that during that function evaluation something is going on that cannot be analyzed as a sequence of states in which each computational observable has a single value. This is the characteristic of this new mode of information processing which is not the implementation of any classical algorithm and performs a task that no classical algorithm can perform

00:34:56 - 00:35:41

The ability to perform this particular task is unlikely to have any practical application though well in some very contrived circumstances it just might say you have a time limit by which you have to perform a very important computation that consists of evaluating $F(1)F(-1)$ for some complex algorithm $F$ as I said the real significance is theoretical but next time I’ll describe an amazing quantum algorithm again using quantum parallelism that is both theoretically interesting and likely to be useful in a wide range of practical applications

2003-01-01 Quantum Computation Lecture 4The Schrödinger Picture

Duration: 00:48:23

Transcript

David Deutsch

00:01:00 - 00:03:25

So far, whenever I’ve described the dynamics of quantum systems to you, and they’ve all been quantum gates, I’ve always just told you the laws of motion of each gate by fiat. A set of equations for how the representative observables of the qubits are changed by passage through that particular gate. Now I’ll go to the other extreme and show you the most general framework for quantum dynamics and tell you in principle which sets of such equations describe processes that occur in nature and which don’t. Laws of motion are about how things change with time. But as I said in the first lecture, the notion of a system changing itself implies that some things about it are invariant. That’s what I called the constitution of the system. The laws of motion, the dynamics, are part of the constitution. The other part is the algebra of the observables at any one time, which in a system of $n$ qubits is always the same as the algebra of all Hermitian matrices of dimension $2^n$ . So for instance, you’ll recall that this Pauli algebra is one way of summarizing the algebra of 2 by 2 Hermitian matrices representing the observables of a single qubit at any given time. Now since the algebra is invariant, the laws of motion have to be such that even though the observables change, their algebra does not. And that’s obviously a constraint on what the laws of motion can be. For instance, take this equation for the commutator of the observables $X$ and $Y$ of a qubit.

00:03:25 - 00:05:54

Rearrange it and you get a quantity which the laws of motion have to prohibit from ever changing at all. It has to stay fixed at zero for all time. Differentiate it with respect to time. That too has to be zero. So there we have a relationship between the time derivatives of observables. And we know that this relationship must be not just compatible but deducible from the law of motion of any qubit that exists anywhere in nature. And we can get similar constraints from any other equation holding among the observables at any given time. It may sound as though it’s going to be tricky to find any law of motion that enforces all of these intricate constraints on how observables can change with time. But no, it’s easy to find one. Here’s one. Pick any observable $H$ of a quantum system at a given time. If the system is a qubit, then like all observables, $H$ can be expressed as a linear combination with real coefficients of the representative observables $X$ , $Y$ , and $Z$ at that time and the unit observable. Do the same for every other time. In other words, pick a whole one parameter family of observables, $H_t$ , one for each time. The reason why I’m not calling that $H(t)$ with the $t$ in parentheses is that we’re already using that notation to refer to the time evolution of a single observable, whereas in this construction we’re allowed to choose a different observable at each time. Now given that one parameter family of observables, consider the following law of motion. For each observable $A(t)$ of the system, $\frac{dA}{dt}=i[H_t,A(t)]$ .

00:05:54 - 00:08:42

We can express that law in a form that’s more like the specific laws of motion I’ve told you in previous lectures by just integrating it over an infinitesimal period dt. So we get $A(t+dt)=A(t)+i\,dt[H_t,A(t)]$ . In that form it expresses each observable at one time, t plus dt, in terms of observables at another time t. $H_t$ is by construction some observable at time t, and so is $A(t)$ . For gates we were only interested in the relationship between observables before and after the gate acted. So we used laws of motion referring to a unit time instead of an infinitesimal time. Such laws could be obtained by integrating this differential equation of motion between t and t plus 1. Okay, now does our new law of motion preserve all the algebraic relationships that may exist between observables? Well consider any such relationship at time t. We can always write it in the form $f(A(t),B(t),\ldots)=0$ , where $A$ and $B$ and so on are observables at time t, and the function f uses any of the operations of matrix algebra, namely matrix addition, matrix multiplication and multiplication by a constant. It’s sufficient to prove that f itself obeys the law of motion. Well what can f be? Suppose f is a real multiple of something that does obey the law of motion. Then $F$ obeys it too. Suppose f is the sum of two things that each obey the law of motion. Then again f also obeys it.

00:08:48 - 00:11:13

What if f is the matrix product of two things that obey the law of motion? Well then $\frac{d}{dt}(PQ)=\frac{dP}{dt}Q+P\frac{dQ}{dt}$ , and since p and q obey the laws of motion, multiply that out and you’ll see that it’s $i$ times the commutator of $H$ with the product $PQ$ , which is $i$ times the commutator of $H$ with $F$ again. So the upshot is, however $F$ is composed out of observables obeying our equation of motion, $F$ obeys it too, and hence it remains zero over time. So any equation of motion of this form leaves any algebra of observables invariant. And actually we’re done, because as you can prove in the worked examples, the converse is also true. Any law of motion that leaves the algebra of observables of some quantum system invariant takes this form. $H$ is called the Hamiltonian of the system. This is called the Heisenberg equation of motion, and the way of representing observables that I’ve been using is called the Heisenberg picture of quantum theory. Consider the special case where the Hamiltonian is actually the same observable at all times. So then $H_t$ does equal some $H(t)$ , where $H(t)$ is a bona fide observable. In such cases we say that the Hamiltonian has no explicit time dependence. How does this observable change with time? Well, as you can see, it doesn’t. When the Hamiltonian has no explicit time dependence, it has no time dependence at all. So under those circumstances, it’s not only an observable, it’s a conserved quantity.

00:11:13 - 00:13:59

This is a conservation law, and in fact the Hamiltonian is up to a constant of proportionality the energy observable. You can also see that any other observable that commutes with the Hamiltonian is a conserved quantity too. And if an observable commutes with a conserved Hamiltonian at any one instant, even one instant, then because the algebra is invariant, it will commute at all instants and will be conserved too. And again, the converse is also true. In a system with a conserved Hamiltonian, every conserved observable commutes with that Hamiltonian. Now, in the case where the Hamiltonian is not only the same observable at all times, but is one of the system’s own observables, the system is said to be isolated. The idea of an isolated system is an idealisation. There are no physical systems that remain unaffected by the outside world for all possible states of the outside world. Remember that the idea of a physical system itself is not perfectly well defined because there is no physical system that even remains in existence in all states. And for both these reasons, the idea of the Hamiltonian for a given physical system is always an idealisation or an approximation. Except perhaps for the multiverse as a whole, that’s not interacting with anything outside itself. So the multiverse is the only truly isolated system. Nevertheless, in real physical phenomena, it’s often a very good approximation to say that a system is isolated for a period. For instance, I said that the qubit in our interference experiment in Lecture 2 was isolated during periods when the photon was travelling between gates. In fact, it was well described as having a Hamiltonian of zero during those periods, even though its real Hamiltonian, ultimately the Hamiltonian of the multiverse, would have contained terms describing how the photon would interact with a passing bumblebee that just happened to fly through the apparatus at the same time, and indeed how it did interact with one in some universes.

00:14:00 - 00:17:16

Now, even when a system isn’t isolated, it is sometimes possible to encode the whole influence of the outside world on that system into a law of motion that involves only that system’s observables, but with a time-dependent Hamiltonian. Such systems are said to be coherent. Another way of stating the definition of coherent is that a coherent quantum system is one in which the observables at one time are functions only of the observables of the same system at another time. That would be in a given state or in a given class of states. In summary, in states where the dynamics of a system can be well described by a Hamiltonian that’s simply one of its own observables, it’s said to be isolated. When it can be well described by a Hamiltonian built out of its own observables, but possibly with explicit time dependence, it is said to be coherent. So an isolated system is also coherent. When it can only be described by a Hamiltonian involving other systems, it is said to be decoherent. But ultimately, all these terms are approximative. Short of the multiverse as a whole, there are no isolated systems and every system is decoherent to some extent. And there are no time-dependent Hamiltonians. Every time-dependent Hamiltonian is just an approximate way of taking into account the effects of other systems in states where the decoherence that they cause in the system of interest is negligible. In our single photon interference experiment, for example, the qubit was effectively isolated during its motion between the beam splitters and the mirrors, but as it interacted with those, its Hamiltonian changed briefly, as I described, but only to some function of the qubit’s own observables. And then it changed back to zero. So the qubit remained coherent throughout the whole experiment until the moment when it interacted with the detector at the end. At that point, the qubit’s Hamiltonian would have depended on observables of the detector and vice versa, thus causing decoherence. And also, as we saw, if the qubit interacted with another qubit during the experiment, then its Hamiltonian would depend on observables of both qubits, and we saw that the interference was then reduced or eliminated. And it’s true in general that for an interference phenomenon to occur, or for a quantum computation to be performed, the system doesn’t have to be isolated, but it does have to be coherent.

00:17:25 - 00:19:58

Okay, now, for an isolated system, we can solve the equation of motion in closed form. It’s just $A(t)=e^{iHt}A(0)e^{-iHt}$ . So that gives A at any time in terms of A at time zero, and $H$ . We can also express the constant Hamiltonian $H$ in terms of observables at time zero. And so this equation would express A at an arbitrary time t in terms of observables of the system at time zero. For a general coherent system, the solution will be a slight generalization of that. It’ll be $A(t)=U_t^\dagger A(0)U_t$ , where $U_t^\dagger$ is the Hermitian adjoint of $U_t$ . $U_t$ is a one parameter family of unitary matrices. Unitary means $U^\dagger U=I$ , whose own equation of motion is $\frac{dU}{dt}=-iUH$ . U is called the evolution matrix between time zero and time t. The state of a system is defined by giving its expectation value function, which is a linear function mapping its observables to real numbers. I’ve used this notation to denote the expectation value of an observable $A(t)$ . But because it’s a linear function, it must also be possible to write it in any given matrix representation like this, where $\alpha$ and $\beta$ are matrix indices and $\rho$ is some matrix that doesn’t change with time. In other words, the expectation value is $\operatorname{tr}(A(t)\rho)$ . Since the expectation value of the unit observable is always one, $\operatorname{tr}(\rho)$ has got to be one. $\rho$ is called the density matrix of the system.

00:19:58 - 00:21:57

It has to satisfy certain conditions to make sure that the expectation value can never be higher than the highest eigenvalue of A nor lower than the lowest. You can find out what those conditions are in the worked examples. Now look at the expression for the expectation value of an arbitrary observable. It’s often more convenient to write this in the form $\operatorname{tr}(U_t^\dagger A(0)U_t\rho)$ . Why? Because in this expression, the only quantity that changes with time is the unitary evolution matrix U, which is the same in the corresponding expression for any observable. Therefore, to track everything that the system is doing over time, we don’t have to solve the equations of motion for all the observables in terms of their values at their previous times. We need only solve this one equation for the evolution matrix $U_t$ , given the Hamiltonian. And with that, we can read off the time evolution of any observable of the system evolving under that Hamiltonian. Actually, we can do better than that. The way the dynamics of quantum systems are encoded in these unitary matrices allows for a whole alternative way of describing quantum systems that’s often much more efficient. What I’ve described so far has been the Heisenberg picture, and the alternative way I’ll show you now is called the Schrödinger picture. First, look again at this expression for the expectation value of a general observable at an arbitrary time.

00:21:57 - 00:24:22

We can rewrite it using the cyclic invariance of the trace, like this. $\langle A(t)\rangle=\operatorname{tr}(U_t\rho U_t^\dagger A(0))$ . Let’s call this quantity $\rho_t$ . $\rho_t$ is called the density matrix in the Schrödinger picture. So to evaluate an arbitrary expectation value, we need only know this density matrix in the Schrödinger picture and all the observables at any one time, say t equals 0. So for each observable $A$ , we only need to know one matrix, $\hat{A}(0)$ , which in the Schrödinger picture we just call $\hat{A}$ . To summarise the Schrödinger picture then, each observable is represented by a constant matrix, while the density matrix changes with time. This is to be compared with the Heisenberg picture, where the observables are functions of time and the density matrix is a constant. From this expression, we can read off the law of motion for the Schrödinger picture density matrix. It’s $\frac{d\rho}{dt}=-i[H,\rho]$ . This has almost the same form as the law of motion for an observable in the Heisenberg picture, but with the opposite sign. I’ll come back to that sign in a moment. Now, consider the eigenvectors of the Schrödinger density matrix at time t. The standard notation for vectors in quantum theory is called the Dirac notation. It uses a symbol called a ket, which looks like this, to denote vectors. We typically write inside the ket symbol the information specifying which vector it is. For instance, we can call the nth eigenvector of $\rho(t)$ the ket with label n and t.

00:24:25 - 00:27:00

The bra vector, that is the Hermitian transpose or dual of a given ket, say with label $\alpha$ , is written like this. This symbol is called a bra. The origin of this terminology is that the scalar product of a bra with a ket is written like this, $\langle\beta|\alpha\rangle$ , the whole thing being a bracket. So the left half of it is a bra and the right half a ket. The scalar product of a bra with the corresponding ket, say $\langle\alpha|\alpha\rangle$ , is by definition the squared magnitude or norm of $|\alpha\rangle$ . The vector space of kets is therefore a Hilbert space, basically a vector space with a norm, and it’s usually called the Hilbert space of the given system. Observables represented by matrices are linear operators on the Hilbert space. The set of all eigenkets of the density matrix at any given time, or of any observable at any time, constitutes a basis for the system’s Hilbert space. So we can expand the density matrix in terms of its eigenvalues and eigenvectors like this. $\rho(t)=\sum_n p_n |n(t)\rangle\langle n(t)|$ . And you can prove in the worked examples that the eigenvalues $p_n$ remain constant and that the equation of motion for the eigenvectors is this. $\frac{d}{dt}|n\rangle=-iH|n\rangle$ . Or more generally, if $\psi$ is any eigenvector of the density matrix, $\frac{d\psi}{dt}=-iH\psi$ . This is called the Schrödinger equation. Over a period t, its general solution will have this form where the $U_t$ are the same unitary matrices as in the solution of the Heisenberg equation of motion.

00:27:03 - 00:29:25

Now the significance of that sign in the equation of motion of the density matrix. The density matrix is not an observable. Both in the Schrödinger and Heisenberg pictures, it’s equal to a different observable at each instant. So why does it have a similar equation of motion to an observable but with the opposite sign? Well, in the Schrödinger picture, the state changes and the observables are constant matrices. In the Heisenberg picture, the state is invariant and the observables change. But in both of them, the expectation value of an observable A at a given time is $\operatorname{tr}(\rho A)$ . And the Heisenberg $A(t)$ we know is $U^\dagger$ $A(0)U$ . Remember, the Schrödinger $A$ equals Heisenberg $A(0)$ and also Heisenberg $\rho$ equals Schrödinger $\rho(0)$ . What’s happening is that the unitary transformation $U_t$ that defines the motion amounts to a rigid rotation in Hilbert space. It’s rigid in that it preserves the scalar product between any two kets that it acts on. When $|\alpha\rangle$ goes to $U|\alpha\rangle$ , $\langle\alpha|$ goes to $\langle\alpha|$ $U^\dagger$ . And so any scalar product $\langle\alpha|\beta\rangle$ is unchanged and that’s rigid rotation. The density matrix represents the state of the world. In the Heisenberg picture, the state is constant and the observables rotate in the sense that their eigenvectors rotate in Hilbert space. In the Schrödinger picture, the observables are fixed.

00:29:31 - 00:31:56

The state rotates rigidly in the opposite sense, thus ensuring that both pictures make the same predictions for the physical quantities, the expectation values of observables. The Schrödinger equation is the Schrödinger picture way of defining the dynamical evolution of a quantum system. In the general case, we have to solve the equation for each eigenvector of the density matrix, from which we can reconstitute the density matrix itself. But if $\rho$ is ever sharp, well $\rho$ is not an observable, but what I mean is if it’s ever equal at some instant to a sharp observable, then the system is said to be in a pure state. And we’re often particularly interested in systems in pure states. For instance, if the $Z$ observables of a set of qubits are all sharp, so the qubits together are like a register of a classical computer holding a single integer, then the density matrix is sharp and the system is in a pure state. You can prove that in the worked examples. And you can also prove that in a pure state, in the eigenvector representation of the density matrix, all the coefficients vanish except for one, which takes the value one. And since the eigenvalues of the density matrix don’t change with time, if a system is in a pure state at any instant, then it remains in a pure state so long as the evolution remains coherent. And the density matrix in such a case just takes this form where $\psi(t)$ is a single one parameter family of kets obeying the Schrödinger equation with respect to t. $\psi(t)$ is then called the Schrödinger state vector. For a quantum system in a pure state, all the motion of the system is summed up in the motion of this state vector.

00:31:56 - 00:33:51

Okay, this equation for pure states and this one for non-pure states are in the majority of cases the most convenient ways of analysing the motion of quantum systems, and in particular the computations of quantum computers. And I’ll mostly use it from now on. But beware, as we’ve just seen, this economy of calculation comes at the expense of additional layers of abstraction. Heisenberg observables, which are the dynamical quantities in the Heisenberg picture of quantum physics, have a lot in common with the variables of classical physics. Yes, they’re matrices rather than real numbers, but at least there’s one such matrix corresponding to each quantity we can observe. And the particular number corresponding to what we do directly observe is in there somewhere, it’s one of the eigenvalues. And in the Heisenberg picture, the changing quantities, the observables, are located in particular systems at particular locations. They obey an equation of motion that expresses how they affect other observables and are affected by them, and thereby carry information from one place to another. The Schrödinger state doesn’t have any of that. There’s only one of them. It’s not located anywhere, but refers to the whole system or ultimately to the whole multiverse. What is the Hamiltonian of the whole multiverse? Fundamentally, fundamental physics has been about what types of systems exist in nature, what their observables are, and what their Hamiltonians are.

00:33:51 - 00:36:04

That’s what elementary particle physicists call the theory of everything. But there is another way of looking at what is elementary or fundamental. Instead of asking what types of Hamiltonian are found in natural systems, which means what sorts of changes can occur in elementary systems over an infinitesimal time, one could go all the way to the bottom line and ask which transformations can be realized in nature by some quantum system evolving for some time and which cannot. And the short answer is all changes in which observables of the system undergo unitary evolution with every observable undergoing the same transformation so as to preserve their algebra, every one of those can occur in nature and nothing else can. Every unitary matrix is the evolution matrix for some quantum system evolving over some time. Or, at least, we think it is, because it turns out that the vast majority of these possible evolutions can only be realized in a very special type of physical system, a universal quantum computer. They don’t occur naturally because they require a complex computer program to bring them about. So here’s a fascinating situation. In terms of Hamiltonians, the laws of physics are very finely tuned. The multiverse has its Hamiltonian and subsystems of the multiverse only have very special Hamiltonians. Most Hermitian matrices cannot be realized in nature as Hamiltonians. But when we ask a slightly different question, which unitary evolution operators can be realized in nature, the answer is all of them, if a universal quantum computer can exist.

00:36:04 - 00:36:39

This special type of object, the universal quantum computer, in a sense contains within itself all the diversity in nature. No other system does, except perhaps systems that are capable of constructing a universal quantum computer. Suddenly we find ourselves unavoidably playing a role at the deepest level of the structure of physical reality.

2003-01-01 Quantum Computation Lecture 3Measurement

Duration: 00:43:09

Transcript

David Deutsch

00:01:05 - 00:03:22

Today I’m going to talk about the quantum theory of measurement. Measurement is an important type of quantum computation. It’s also, of course, what links quantum theory with experiment, but for historical reasons, unfortunately, the quantum theory of measurement also plays a prominent role in what we could call quantum mechanics folklore: the informal misunderstandings of quantum physics that range from careless remarks in textbooks through bad philosophy and all the way down to complete nonsense. You may have heard that the measurement process in quantum theory is deeply mysterious and spooky, that it has a special role for the conscious observer. You may have heard that when you make a measurement in one place, there’s an instantaneous effect on quantum systems in a different place. You may have heard that measurements are inherently irreversible processes, even though the laws of motion in fundamental physics are all reversible. As you’ll see in this lecture and later lectures, none of that is true. In fact, quantum measurement theory is pretty well understood and has been a powerful tool for understanding quantum physics. Nowadays, it’s part of the quantum theory of computation. The most elementary but important thing to understand about measurements is that they are physical processes that we understand using the same theory and laws of physics as we do for all other physical processes. A measurement is a process in which one physical system, a measuring instrument, interacts with another physical system that’s being measured. Some observable of the measuring instrument is affected by some observable of the system being measured and ends up ideally having the same value.

00:03:25 - 00:06:28

Let’s confine our attention for the moment to the useful idealization of a perfect measurement. We can define a perfect measurement process by the effect it has when the observable being measured is sharp, has a single value. First, the outcome of a perfect measurement of a sharp observable is the value of that observable, and second, at the end of the measurement, the observable being measured is still sharp with the same value it had before. So a perfect measurement records sharp values accurately and leaves them unchanged. The simplest possible measurement is the measurement of a Boolean observable where the outcome is stored in the second Boolean observable. Let’s think about that first in the case of a classical computation. Let’s make it a reversible classical computation partly to substantiate what I said just now about measurement not being inherently irreversible, but mainly because in fact classical reversible computations are a special case of quantum computation. Imagine that we have a single bit that we’re going to measure, and the bit has an unknown value that’s either plus one or minus one. Call that unknown value a, and we have a second bit somewhere in an apparatus in which we want to record a copy of the value of the first bit. The second bit is called the target bit. So the target bit has to end up with the value a, and the bit being measured has to keep the value a. Considered as a computation, perfect measurement is a process of faithfully copying information, so that where there was one copy of the sharp value of the observable, there are now two, the second one being in the measuring apparatus. As far as the definition of a perfect measurement is concerned, we’re not interested in what the initial value of the target bit was so long as the final value is a, but in a reversible computation, different inputs must always produce different outputs. So it follows that if the computation only involves those two bits, it can only be a perfect measurement for one initial value of the target bit.

00:06:28 - 00:09:01

Let’s say for the value $+1$ . So it’s not a perfect measurement when the target bit starts at minus one. What does happen if it’s minus one? It depends on the law of motion. One interesting law of motion is that of a very useful operation called controlled not. The ordinary not operation is the single bit or single qubit operation that figured in the interference experiment. I discussed last time. It flips plus one to minus one and vice versa. Control not means flip the value of the target bit only if the first bit is minus one. The first bit is now called the control bit if the control bit is plus one the target bit remains unaffected. So we can summarize this controlled not operation like this: if the input values of the control and target bits are a and B, then the outputs are a and AB. A computer like this that performs a simple computation on a fixed number of bits and completes it in a fixed time is called a computational gate also known as a logic gate, and in this case, it’s a reversible logic gate because it’s performing a reversible computation: the controlled not operation. Let me just summarize some of the ways in which we can think about what the controlled not gate does. First of all, if B equals plus one, then considering it as a physical process, it’s a perfect measurement: the first bit starts with an unknown value a, and the second bit ends up holding the measured value of the first bit which should also be a. Second, considered as a computation, if B equals plus one then the process copies information, the information a, which starts out in only one of the bits ends up in both of them.

00:09:04 - 00:12:56

Third, for an arbitrary B, well, that’s the controlled not operation. The target bit is flipped if and only if the control bit is minus one. A fourth interpretation: this gate is the reversible version of the classical exclusive or gate. If we consider plus one as standing for false and minus one for true, then a times B is the same thing as the exclusive or of A and B, and the target bit ends up as the exclusive or of the two inputs. And as I said like all classical reversible gates, the controlled not gate has a quantum implementation that works on two qubits instead of two bits. That’s the quantum controlled not gate. In some implementations there’s a physical object you can identify as the gate as with the mirror in the interference experiment last time which was a not gate and the beam splitter which was a fractional power of not but speaking more precisely the computational gate is not the object that makes the process happen, but the dynamical process itself that’s undergone by the qubit or qubits that physically realize the process and that actually satisfies a rule that defines each output in terms of the inputs and nothing else. Now to describe the quantum physical system consisting of two qubits passing through a quantum controlled not gate, I’m going to make the Z observable of the control qubit control the Z observable of the target qubit. To describe that quantum system, I have to tell you as always the static constitution, the dynamics, and on each particular occasion when the experiment is done the state which summarizes how the qubits were prepared as the input. I’ll start as usual with the static constitution. For this I have to tell you the algebra of all the observables at any one time, which will then be the same as the algebra at any other time because the algebra summarizes the time invariant features of a quantum system. So here we have a two qubit quantum system. We already know what the algebra of the observables of either one of the two qubits is because that algebra is itself an invariant. It’s the same for all qubits and regardless of what interactions the qubit happens to be undergoing so as in the one qubit interference experiment, the first qubit will have observables x y and z which have the same algebra as the Pauli matrices, together with the unit observable and linear combinations of those with constant coefficients. Let me call these observables $X_1$ , $Y_1$ , and $Z_1$ now, where the suffixes indicate that they are observables of qubit number one the control qubit of the quantum controlled not gate. We don’t have to put a suffix on the unit observable because, as you will remember, the unit observable can be measured without even referring to the system.

00:12:56 - 00:23:29

The second qubit the target qubit will have different observables $X_2$ , $Y_2$ , $Z_2$ and so on but they’ll also have the Pauli algebra. Now I have to specify all the additional algebraic relations that may hold involving observables from both qubits. And here’s a universal rule that defines the algebra of any composite quantum system given the algebras of its constituent systems. It just says that the observables of different quantum systems commute with each other. That is to say if a1 is an observable of one system and b2 is an observable of another, then $A_1B_2=B_2A_1$ . I told you last time that two by two matrix representations of a qubit algebra wouldn’t always be enough. Here’s why: this set of algebraic relations can’t be faithfully represented by two by two matrices. This set by itself can, so can this set. But because this commutation relation for the combined system has to be represented as well, qubits one and two can’t use the same set of matrices. Take the $Z_2$ for instance. It has to commute with every observable of qubit one but the only two by two matrices that commute with every two by two matrix are multiples of the unit matrix, and $Z_2$ can’t be represented by a multiple of the unit matrix because it’s a Boolean observable. It has to have two distinct eigenvalues and the multiples of the unit matrix have only one eigenvalue. So there is no two by two matrix representation of this algebra as a whole. The simplest representation turns out to be a four by four representation, and the way it is constructed applies to any two quantum systems not just a pair of qubits that you want to consider as a single system. It uses the tensor product of matrices. The tensor product of two matrices of dimensions M and N is an M N dimensional matrix consisting of all possible products of pairs of elements one from the first matrix and one from the second like this: A B C D with E F G H is the four by four matrix A E A F and so on. That symbol denotes the tensor product. So the tensor product of a pair of two by two matrices is a matrix of all 16 possible products consisting of an element of the first matrix multiplied by an element of the second. You can verify in the worked examples that the tensor product of a pair of Hermitian matrices is a Hermitian matrix. So the tensor product of two observables one from each system is an observable of the combined system. What observable is it? Well, if you have two systems and a is a matrix representing an observable of system one by itself and B is a matrix representing an observable of system two by itself, then the tensor product $A\otimes B$ is an observable that you measure by measuring a on the first system and B on the second and then multiplying the results together. You can verify that the tensor product operation is associative. But it’s not commutative. In other words $A\otimes B$ is not the same as $B\otimes A$ . You can also verify that the tensor product has the following property with regard to ordinary matrix multiplication: $(A\otimes B)(C\otimes D)=AC\otimes BD$ . Now it follows from all that that, using the tensor product, we can make a four by four representation of the algebra of observables of qubit one as follows. Just multiply each matrix of any two by two representation by the two by two unit matrix on the right. These are now four by four matrices, and because of this property, it’s easy to show that they do indeed represent the standard algebra for one qubit. For qubit two we can do a similar thing this time multiplying by the unit matrix on the left. They too form a four by four representation of the standard algebra for a single qubit. The object of the exercise: any observable of the first qubit commutes with any observable of the second. For instance X1 Y2 at time zero is $\sigma_X\otimes I$ times $I\otimes\sigma_Y$ , which equals $\sigma_X\otimes\sigma_Y$ and taking it the other way around $Y_2X_1=(I\otimes\sigma_Y)(\sigma_X\otimes I)$ , which also equals $\sigma_X\otimes\sigma_Y$ by the way. Like in the single qubit two by two case, we need never work with the actual components of these matrices. Every four by four Hermitian matrix can be expressed as a linear combination of tensor products of the form $\sigma\otimes\sigma$ , $\sigma\otimes I$ , $I\otimes\sigma$ , and $I\otimes I$ which is the four by four unit matrix. So again, we can just express all four by four matrices and so all observables in terms of Pauli matrices and do all matrix arithmetic in terms of the algebra of Pauli matrices. So I’ve told you the static constitution of a pair of qubits, which would be the same for any two qubit gate. Now the dynamics of the controlled not gate in particular. As usual we’ll imagine that it performs its operation in one unit of time one computational step. For the moment, we’re considering the gate as an elementary quantum computation, so we’re not interested in its internal workings. We’re only interested in how the output depends on the input. So for present purposes the dynamics of the gate just means how the observables at time one after the gate has acted depend on the observables at time zero. Well, the dynamics of a controlled not gate are defined by a set of equations. Most of whose details are not relevant for present purposes, but I’ll put them on screen anyway, just so that I can point out the properties of the gate being a controlled not gate and highlight the parts that are relevant. These six equations tell us how six representative observables change: three of them from one qubit and three from the other. We don’t have to list the unit observable explicitly because it never changes. Using these equations, we can find out how any other observable behaves as well by expressing it at time zero in terms of these representative observables and then using the fact that algebraic relations between observables at a given time don’t change with time. These equations are actually redundant. We could make do with just four of them because for instance the static constitution for any qubit already defines its observable $Z$ at any given time in terms of its observables X and Y at that time. Now take a look at the equations for the time evolution of just the $Z$ observables. You can see that $Z_1$ and $Z_2$ at time one depend only on $Z_1$ and $Z_2$ at time zero. The X’s and Y’s don’t affect the evolution of the $Z$ observables. So the $Z$ observables almost form a physical system in their own right a little subsystem of the two qubit system that evolves independently of the rest. It doesn’t quite count as a separate physical system because you can never change $Z$ without changing X or Y. Nevertheless in a controlled not computation, the $Z$ observables are autonomous.

00:23:29 - 00:26:52

They’re evolving independently of all other observables. I’ve drawn your attention to that because it will come up again in later lectures, but for the moment its only significance is that it makes it easy for us to check that this gate really does perform a perfect measurement. Remember a perfect measurement interaction is defined by what it does when the observable being measured is sharp. So now it’s time for me to specify a convenient state for our two qubits in which the $Z$ observables of both qubits are sharp. There are only four possible cases of this, so we may as well deal with all four of them wholesale. Let the observable $Z_1$ be sharp with the value a and let $Z_2$ be sharp with the value B. A and B are each plus or minus 1. So to specify the state or actually four possible states, we have to specify the expectation values of all the observables. And I’ve just said that the expectation value of $Z_1(0)=a$ and the expectation value of $Z_2(0)=b$ . As I mentioned in the last lecture and you will have proved in the worked examples $Z$ being maximally sharp implies that X and Y are both minimally sharp. In other words, that their expectation values are both 0. This in turn tells us how to find expectation values of Pauli matrices in this state. In the representation we’ve chosen for the observables, we have expectation value of $\sigma_Z\otimes I$ equals a and expectation value of $I\otimes\sigma_Z$ equals B. By linearity this tells us all the expectation values of matrices of the form something cross 1 and 1 cross something. You’ll see in the worked examples that these relations also tell us that in this state the expectation values of any tensor product of the form $A\otimes B$ where a and B are two by two matrices is the product of the expectation values. With these rules, we can find expectation values of general matrices and hence of general observables in this state. So what do we predict for the output of the controlled not gate where the inputs were sharp? Well, the control observable $Z_1$ has the same expectation value at time 1 as it did at time 0, namely a, and since a is plus or minus 1, that means that $Z_1$ is still sharp with the value a, which is correct. $Z_2$ is sharp as well, but its value has changed.

00:26:52 - 00:29:09

It’s now $aB$ , just like the target bit of a classical control not gate. Okay, the next obvious thing to work out is what happens if we perform a perfect measurement of an observable that’s not sharp. Well, you can prove in the worked examples that if $Z_1$ isn’t sharp at time 0, it’ll be just as unsharp at time 1, and also $Z_2$ will have become just as unsharp as well. So the measurement interaction propagates unsharpness from one system to the other. But there’s more. Consider the observable $Z_1Z_2$ . That is an observable because $Z_1$ and $Z_2$ commute at any one time So $Z_1Z_2$ is Hermitian, and from what I’ve said its operational meaning is that it’s the observable for what you’d get if you measured $Z_1$ and $Z_2$ and multiplied the outcomes together In other words. It’s a Boolean observable whose eigenvalue minus 1 means that the outcomes of measuring $Z_1$ and $Z_2$ would be different, and plus 1 means that those outcomes would be the same. Now here’s a remarkable thing Starting in a state where $Z_1$ is not sharp at time 0 and hence both $Z_1$ and $Z_2$ are unsharp at time 1, calculate the expectation value of $Z_1Z_2$ . You’ll find it’s plus 1, which means that the observable $Z_1Z_2$ is sharp In other words $Z_1$ and $Z_2$ are equal at the end of the measurement, sharply equal even though neither of them has a sharp value? How can two things be perfectly equal without either of them being sharp?

00:29:09 - 00:33:17

Well like this, of course Finally Let’s look at the case where this measurement of an unsharp observable Takes place in the middle of an attempted interference experiment Let’s work out what happens when we combine The analysis of the previous lecture of the interference experiment with this time’s analysis of the dynamics of measurement Let’s take a qubit based on the photon’s direction of motion and Pass it through a beam splitter to make $Z_1(1)$ unsharp So from the equations of motion $Z_1$ of 1 takes this form In the 2 by 2 representation of last time But now we’re going to measure the direction of motion So we need a second qubit and therefore a 4 by 4 representation like this The second qubit will be a subsystem of some instrument that detects the direction of motion in principle That could be another subatomic particle But equally it could be a pair of photon detectors like the ones we actually used in the interference experiment The essence of any such instrument is that somewhere in there is an observable That’s going to be the target of a controlled not or perfect measurement operation a boolean observable Measuring the Z observable as we defined it for the photon’s direction of motion So We can just analyze those two qubits and forget all the other degrees of freedom just as we did before The details again are in the worked examples. We find that the expectation value of $Z_1$ The photon direction of motion at time 2 just after the measurement is 0 just as it was in the original experiment Because in half the universes the photon travels on one path and in the other half it travels on the other and So the expectation value of $Z_2$ of 2 is also 0 Just as we would expect from something that has measured the correct value of $Z_1$ in each universe then we let the photon go on and do exactly what it did before namely bounce off the mirrors and Strike the second beam splitter Which previously had the effect of making the direction of motion sharp again But look not in this experiment The expectation value of $Z_1$ is still 0 at the end of the experiment The fact of having made a measurement Even a perfect measurement of the value of $Z_1$ at an intermediate stage of the experiment has spoiled the interference phenomenon This is why we don’t see very clear examples of quantum interference in everyday life It’s because undergoing an interaction in which even one qubit from the outside is affected by a physical system is enough to suppress interference and It doesn’t have to be a measurement interaction.

00:33:17 - 00:35:08

You’ll see that almost any interaction affecting an outside qubit will do Specifically what prevents interference is when something carries off information about the system and The reason why that makes a difference is that any process that transfers information out of a system always changes the system itself if The process is a perfect measurement. It doesn’t change the observable being measured The observable being measured say Z but it changes other observables like X or Y that are inextricably linked with it by the uncertainty principle and by the dynamics of quantum physics and that can make the system subsequently behave differently a Process in which information is carried off in this way is known as a decoherence process in practical implementations Decoherence is the great enemy of quantum computation and I’ll be saying more about that So far When I’ve described the dynamics of quantum systems to you They’ve all been quantum gates I’ve always just told you the laws of motion of the gate by fiat Just a set of equations for how the representative Observables of the qubits are changed by passage through the gate in the next lecture I’ll show you a general framework for quantum dynamics and I’ll tell you in principle Which sets of such equations describe processes that can occur in nature and which can’t?

2003-01-01 Quantum Computation Lecture 2Interference

Duration: 00:41:52

Transcript

David Deutsch

00:01:00 - 00:03:13

In the last lecture I gave an overview of how physical systems in general are described in quantum theory via entities called observables which can be represented by matrices and via expectation value functions that map observables to real numbers. I also told you about the simplest kind of observables, Boolean observables, which are those whose spectrum contains exactly two elements. And I told you about the simplest kind of quantum system, the qubit, a physical system all of whose non-trivial observables are Boolean. But I didn’t describe any phenomena. In this lecture I’ll use that theory of observables and expectation values to analyze the archetypal quantum phenomenon that of quantum interference. It can be defined as a phenomenon in which an easily measurable observable is first sharp and then becomes unsharp and then becomes sharp again. To put that in parallel universes terms, an interference phenomenon is one in which the observed outcome depends on what has been happening in more than one universe. Quantum interference can be demonstrated in a classic set of experiments involving a single qubit. In the case I’ll describe, the qubit is a subsystem of a single photon or particle of light. What I mean by a subsystem of a single photon is that only some of the observables of the photon are involved and these observables evolve independently of all the others during these experiments. So it’s permissible to regard them as constituting a quantum physical system in their own right.

00:03:13 - 00:05:30

Like all experiments, this one can be regarded as a computation on a single qubit. Now a classical computer with only a single bit of memory couldn’t do very much. In fact there are exactly four possible computations that one can do with a single bit. Set it to one value, set it to its other possible value, flip the value, that’s the not computation, or leave it alone, that’s the null computation, or there’s any sequence of those, though the net effect of any sequence would still be the same as one of those four. That’s classical computations, but a single qubit already allows for quite a variety of quantum computations. Some of them are useful as computations, some of them are interesting in the theory of quantum computation, and some of them are interesting as physical processes. The single qubit experiment I’m going to describe has a bit of all three. Now I’ll specify a qubit within an individual photon using the method I introduced in the previous lecture, namely I’ll pick a Boolean observable of the photon and then I’ll define our qubit as the minimal physical system containing that Boolean observable. The observable in question is which of two particular directions the photon is traveling in. We’ll make sure that in this experiment it never does travel in any direction other than very close to one or other of these two directions. So measuring which one it is, we’re measuring a Boolean observable, and I’ll call that observable $\hat{Z}$ with eigenvalues plus and minus one, and standing for the two possible directions of travel.

00:05:30 - 00:07:49

Here’s a source of photons, a laser. Here are some of them streaming out of it, about three times $10^{15}$ of them per second in fact. If we made this beam a meter long say, well the speed of light is about three times ten to the eight meters per second, so at any one instant there’d be about ten to the seven or ten million photons present in the beam. If we want to experiment on one photon at a time, we put a dark filter in front of the laser which absorbs most of the photons. If the filter reduced the intensity of light by a factor of ten million, then there’d be only one photon at a time on average on this meter long path. In the actual experiment I’m going to show you in a moment, we’re going to make the intensity even lower than that, so the effective distance between successive photons entering the apparatus will be not one meter but about thirty-five meters. This is the quantum optics lab where the experiment is going to be done. This is Manuel and this is Eran. Hi. This is a source of photons, it’s a laser and it’s emitting about five milliwatts in the form of photons here. Now we want to do this experiment on one photon at a time, so we pass the beam through dark filters which consist of glass with embedded metal in it. There’s one of them and then the light passes into this fiber optic cable, around here and into this, past this dark filter whose effect you can see here, making the beam far too weak for the naked eye to see.

00:07:49 - 00:09:57

The net effect of these filters is to attenuate the beam by a factor of $10^{10}$ so that here there are only $10^6$ photons entering the apparatus per second, one per microsecond. I said that a photon is a particle. What I mean is that in a single universe it has many of the properties that particles would have in classical physics. For instance, the photons here are moving through space at a constant speed in a straight line, just as Newton’s laws would require, or actually a very slightly curved line because of gravity. And here they are bouncing off a barrier with the neat specular reflection that the conservation laws of classical mechanics would predict. Crossing two beams shows that to the extent that they resemble classical particles, photons must be really tiny, point particles to a good approximation because the beams just don’t notice each other. Even with $10^{15}$ odd photons passing by each other every second here, we don’t notice any absorption or deflection out of the beam through collisions. But we’ll see in a moment that in other ways they don’t resemble classical particles and they’re not located at just one point. We could verify that none of them is deflected by putting a sensitive photon detector like this one somewhere outside the two beams. This is a solid state photon detector. If a photon goes in here it creates a small electric current which is then amplified and can be displayed on the oscilloscope. So in an experiment involving only one photon, this device measures a Boolean observable. Does the photon strike a given location where we put the detector or not?

00:09:57 - 00:12:07

Incidentally this is a destructive sort of measurement. The photon, including the qubit we’re interested in, is always destroyed in the course of detecting its presence with this device. That’s a limitation that won’t matter in these experiments but in future lectures when I discuss more sophisticated quantum computations we’ll need qubits that remain in existence after each computational step so they can participate in subsequent steps. So we can only use this device as the last step of a computation reading the output. It wouldn’t do as an internal component of a quantum computer but this device would. Here’s the key component in the little one qubit quantum computer that we’re going to build. It’s called a beam splitter. When a photon with a sharp direction of motion strikes it, its direction of motion becomes unsharp. What happens is that in half the universes it just carries on straight through and in the other half it bounces off as if from a mirror. It turns out that it can do the reverse of that too. If we were to put ordinary mirrors here and here to reflect the photon back the way it came, whichever way that was in each universe, then after it strikes the beam splitter again its direction of motion is sharp again. Let me just remind you that this isn’t a case of two photons merging or becoming alike. This is a case of a single photon whose direction of motion is not sharp striking the beam splitter and its direction of motion becoming sharp again. So the beam splitter is also a beam joiner. That phenomenon of an unsharp direction of motion becoming sharp is our quantum interference phenomenon.

00:12:07 - 00:14:24

Both parts of the photon that have traveled on different paths in different universes participate in the final interference process. How can we tell? Well if we were to remove one of the mirrors so that whenever the photon takes this path it never returns then the photon never would become sharp. In fact it would end up with a three-fold unsharpness going in that direction in some universes and then the beam splitter would have nothing to join so in other universes it would end up going in this direction and in others in that direction. If we want to experiment on a single qubit we have to keep things simpler than that. A qubit has only Boolean observables so we have to make sure that none of the observables of the system or subsystem that we’re investigating ever take on more than two values at a time. So the basic form of all the single qubit experiments that I’m going to describe is this. We start with a photon, we pass it through a beam splitter after which it’s traveling on two paths, we define a Boolean observable of the photon as being which of the two paths it’s traveling on. Then using the method I introduced in the previous lecture we’ll define our qubit as the minimal physical system containing that Boolean observable. I’ll call that observable $\hat{Z}$ with spectrum $\{+1,-1\}$ , the eigenvalue minus 1 standing for one path and plus 1 for the other. Since $\hat{Z}$ has spectrum $\{+1,-1\}$ it’ll be represented by a Hermitian matrix which may change with time but always in such a way that its eigenvalues are plus or minus 1. To give a full description of the qubit and what happens to it in these experiments we need to specify as always three things.

00:14:24 - 00:16:28

Its static constitution which is what its observables are and what the algebra is at any one time. The dynamics is how the observables change with time, remember they change in such a way as to keep their algebra at any given time constant. And the state which means how the system is programmed, how it’s prepared on a particular occasion and that’s all summed up in the expectation value function. So static constitution first. As I indicated last time the set of all observables of a qubit at any one instant has the same algebra as the set of all two by two Hermitian matrices. Now that algebra is most straightforwardly represented by two by two Hermitian matrices but there are higher dimensional matrix representations of the same algebra and it turns out that in most experiments we’d have to use one of those. There’s a simple reason for that which I’ll get to in the next lecture. Anyway for our present purposes it’s sufficient to represent the observables of our qubit as two by two matrices. There’s a nice way of working with two by two Hermitian matrices and more generally $2^n$ by $2^n$ ones as well where we don’t have to bother getting our hands dirty with the actual components, the complex numbers. Instead we use four fixed standard Hermitian matrices namely the unit matrix I and three other matrices known as the Pauli matrices $\sigma_X$ $\sigma_Y$ and $\sigma_Z$ . Every two by two Hermitian matrix can be expressed as a linear combination with real coefficients of these four matrices.

00:16:28 - 00:18:48

Here are some more properties of Pauli matrices. First you can easily verify that the eigenvalues of a Pauli matrix are plus and minus one and that the square of each Pauli matrix therefore is the unit matrix. Another property $\sigma_X\sigma_Y=i\sigma_Z$ and the other two cyclic permutations of that and also the Hermitian conjugates of those three $\sigma_Y\sigma_X=-i\sigma_Z$ and so on. These formulae completely define the algebra of two by two Hermitian matrices in the sense that if we express matrices as linear combinations of these we don’t need to know the rules for adding or multiplying matrices or finding their inverses. That’s all encoded in these formulae. So I’ll express the observables of our qubit in terms of Pauli matrices. By the way, Pauli matrices are often called the Pauli spin matrices because they’re useful in the quantum mechanics of spin and rotation but nothing in today’s experiment has anything to do with spin or rotation. Nowadays when I think of Pauli matrices I don’t think of spin I think of qubits. Well now I can tell you the matrix representation of the observable $\hat{Z}$ that I’ve defined. $\hat{Z}$ is represented at time zero by the matrix $\sigma_Z$ and the unit observable $\hat{1}$ is as always represented by the two by two unit matrix I. These equal signs should really be is represented by not equals but we save ourselves a lot of useless mathematical baggage if we gloss over that difference when we’re doing calculations.

00:18:48 - 00:21:22

Now all the other two by two Hermitian matrices also represent observables of the qubit at time zero or at any other time but in particular at time zero. So there is an observable whose representation at time zero is $\sigma_X$ . I’ll call that observable $\hat{X}$ and I’ll call the observable represented by $\sigma_Y$ at time zero $\hat{Y}$ . What are $\hat{X}$ and $\hat{Y}$ physically? It’ll become clear in the course of this analysis what they are how one would measure them but from this definition we already know what their algebra is at any time. Since $\hat{X}$ , $\hat{Y}$ , and $\hat{Z}$ have the same algebra as the Pauli matrices at time zero and since the algebra of observables is an invariant feature of any quantum system they must have the same algebra at any other time $t$ as well. So for instance $\hat{X}(t)^2$ must just equal $\hat{1}$ , and the same for $\hat{Y}$ and $\hat{Z}$ , and $\hat{X}(t)\hat{Y}(t)=i\hat{Z}(t)$ and so on. All other observables must be linear combinations of $\hat{X}(t)$ , $\hat{Y}(t)$ , and $\hat{Z}(t)$ and the unit observable with real coefficients and the coefficients must not change with time. You can check the worked examples to verify that. Well next I’ll specify the state of our qubit. As I said we’re going to start each of these experiments by shining the laser at the beam splitter in the $Z=+1$ direction. So at time zero Z is sharp with the value $+1$ and so its expectation value must be one. In terms of matrices that just says the expectation value of $\sigma_Z$ is one and in that way we define the expectation value function for matrices as well as observables.

00:21:22 - 00:24:08

You’ll see in the worked examples that it follows from this that the expectation value of $\sigma_X$ and the expectation value of $\sigma_Y$ are both zero and as always we have that the expectation value of the unit observable is one so the expectation value of the unit matrix I is one as well. Since we can express any observable of the qubit as a linear combination of Pauli matrices and since the expectation value function is a linear function we can use these formulae to find the expectation value of any observable at any time. So I’ve completely specified the state of the qubit. Note that these formulae refer to one particular state the one with $Z$ initially sharp with the value $+1$ . They’re not inherent properties of Pauli matrices in a different state these expectation values would be different. Okay now the dynamics. The dynamics are the laws of motion of all the qubit’s observables. If we think of the apparatus as a quantum computer with our qubit as its sole working register the laws of motion are the rules defining what the computer does step by step to an arbitrary single qubit input. Well the first thing that happens is that the photon passes through the beam splitter. As in classical computation it’s convenient to analyze computations as proceeding in steps each of which is a relatively simple computation. For present purposes we’re not interested in what happens during the operation only in its net effect on observables. So here’s the law of motion in that sense for the photon or for the qubit passing through a beam splitter. $\hat{Z}(t+1)=\hat{X}(t)$ , $\hat{Y}(t+1)=\hat{Y}(t)$ , and $\hat{X}(t+1)=-\hat{Z}(t)$ .

00:24:08 - 00:26:50

The unit observable of course never changes. By taking linear combinations of these we can determine the law of motion for any observable of the qubit. So $\hat{Z}(1)$ we said was equal to $\hat{X}(0)$ which equals $\sigma_X$ and the expectation value of $\hat{Z}(1)$ equals the expectation value of $\sigma_X$ which I said earlier is zero. So Z at time 1 is no longer sharp. Its expectation value goes to zero which means physically that the photon goes straight through on path plus 1 in half the universes and bounces back on path minus 1 in the other half. Okay next the photon bounces off an ordinary mirror on either path and you can see what that does. The operation of bouncing off either of these mirrors is just the quantum generalization of the simplest single bit classical operation namely the NOT operation converting minus 1 to plus 1 and vice versa. The law of motion for the NOT operation is, just as for the beam splitter: $\hat{Z}(t+1)=-\hat{Z}(t)$ , $\hat{Y}(t+1)=\hat{Y}(t)$ , and $\hat{X}(t+1)=-\hat{X}(t)$ . In general I’d have to describe the dynamics of one further elementary operation before I could describe the experiment as a whole. It’s simply the operation of the photon traveling unimpeded for a certain distance in a straight line in between bits of apparatus. Well the observables change periodically with distance so to make the calculation easy I’m going to pretend that all the distances traveled are a whole number of periods a whole number of wavelengths so that our whole qubit will remain unchanged in between mirrors.

00:26:50 - 00:29:36

Now here’s the experiment it starts at time 0 with the photon entering the apparatus traveling in the plus 1 direction. It strikes a beam splitter and we’ll call some instant after that time 1. So at time 1 it’s traveling in these two different directions plus 1 and minus 1 in different universes. Then regardless of which of those directions it’s traveling in it strikes an ordinary mirror and that brings us to time 2. And then again regardless of its direction of motion it again strikes the beam splitter from which there are two possible directions of exit again plus 1 and minus 1. Finally at time 3 the photon enters one or other of two photon detectors which together constitute an instrument for measuring the Boolean observable Z. Let’s use quantum theory and the quantum mechanical description I’ve given of the qubit to predict the outcome of this measurement. Now at time t equals 0 the observable $\hat{X}$ is represented by $\sigma_X$ , the observable $\hat{Y}(0)$ by $\sigma_Y$ , and the observable $\hat{Z}$ by $\sigma_Z$ . Now the first thing that happens is that our photon strikes the beam splitter and so what happens to $\hat{X}$ , $\hat{Y}$ , and $\hat{Z}$ ? Well by time 1 it will have hit the beam splitter and so the observables of the qubit will have changed according to the formulae that I’ve given for the effect of a beam splitter. So we can read off $\hat{X}(1)=-\hat{Z}(0)$ which equals minus $\sigma_Z$ and similarly for $\hat{Y}(1)$ and $\hat{Z}(1)$ .

00:29:36 - 00:32:00

Well then the photon strikes the mirrors and the effect is the not operation according to these formulae that I gave from which we read off the observables at time 2. Finally the photon hits the second beam splitter and again we can read off $\hat{X}(3)=-\hat{Z}(2)$ which makes it $\sigma_X$ and so on. It turns out that all the observables of the qubit at time 3 are the same as they were initially at time 0. In particular $\hat{Z}(3)=\hat{Z}(0)$ and so $\hat{Z}(3)$ is also sharp. So finally the photon is traveling only in the plus one direction and there’s our prediction. Remember there’s only one photon participating in this experiment. Consider the moment just before it strikes the second beam splitter. In different universes it’s coming from different directions but just consider the universes in which it’s coming along here say in the northwards direction. It’s approaching the beam splitter. Usually when a photon approaches a beam splitter well it goes through in half the universes and bounces off in the other half. So usually if we observe what happens half the time we detect that it’s passed straight through. Not in this experiment. In this experiment the photon necessarily takes a sharp right turn and is never observed to pass straight through and similarly in universes where the photon is approaching from the eastward direction it ignores the possibility of being reflected and just passes straight through. And the net effect is that in all the universes in which the experiment is done the photon is observed at the east detector and never at the north detector.

00:32:00 - 00:34:15

Now I’ll show you this experiment in action. Before I do I’d better explain a detail that might be confusing otherwise. The experimentalists like to use the same physical beam splitter for both ends of the experiment. That way they don’t have to bother with finding two of them with precisely matched properties and making them exactly parallel to each other and so on. It’s hard enough aligning everything as it is. They eliminate the need for one of the beam splitters by slightly changing the geometry like this. Now this is still the plus one direction and this is the minus one direction for Z and there are the two corresponding detectors and our prediction again is that all the photons will come out here into this detector and none into this one. And when we do the experiment that’s exactly what we see. Each spike here on the oscilloscope represents one photon. The upper trace, the yellow one, is from the detector at the plus one position and the lower trace, the green one, from the minus one position. The few stray spikes on the minus one trace are due to various imperfections in the apparatus, especially the detector. What is it that has made the photon in the universes where it approaches the beam splitter northwards turn right? It is the presence of its counterparts in the other universes striking the same point at the same time going east and we can verify that because if we put an opaque barrier on the eastward path and let some photons come through suddenly both detectors will be firing.

00:34:15 - 00:36:59

Open up this path again, the north detector goes quiet. Close it off again and the photons again start reaching the north detector. Now let’s consider this whole process as a computation. The input was one. The computation as a whole consisted of three elementary computations. The middle one was a not operation. The outer two were the same as each other. Let me call it B for beam splitter and the output was again one. And you can see from here that the whole computation is an elaborate way of performing the unit operation, otherwise known as doing nothing at all. Is that an interesting computation? Yes, because just look what’s happened here. We’ve performed B followed by not followed by B and the result is the unit operation. So $B\,\mathrm{NOT}\,B=I$ . Dot here just means performed after. It’s easy to prove that there is no classical computation using just one bit with this property. No computation such that if you perform it and then flip the value of the bit, then perform it again, the value of the bit will be unchanged. So B is an example of an elementary computation, an elementary quantum computation that has no classical analog. B also has an inverse which formally you can see from here is the square root of NOT. I’ll show you in later lectures that this isn’t just a manner of speaking. In quantum computation not really does have a square root. Many of the new modes of computation that quantum computers are capable of make use of the roots of not. In this experiment we only ever measured Z.

00:36:59 - 00:39:14

How would we measure X or Y or the qubit’s other observables? We’ve already seen the answer there. Look at the effect of beam splitter. $\hat{Z}(1)$ equals $\hat{X}(0)$ . So if we measure $\hat{Z}$ at time 1, that’s the same as measuring $\hat{X}$ at time 0. So the act of passing a photon through a beam splitter and then measuring $\hat{Z}$ amounts to a measurement of $\hat{X}$ . So the beam splitter followed by the two detectors constitute a measuring instrument for measuring $\hat{X}$ . Here again I’ve been making the simplifying assumption that our qubit’s observables don’t change spontaneously with time, but they do according to this. Don’t worry about the details, but it follows from this that if we slightly change the path lengths in one of these experiments, and in the lab the fine adjustments always involve doing that anyway, we can easily convert an instrument for measuring $\hat{X}$ into one for measuring Y. That’s how to measure $\hat{Y}$ , but you may still be wondering what is $\hat{Y}$ in the sense that $\hat{Z}$ is the Boolean observable for whether the particular photon is traveling in a particular direction. What is the Boolean observable $Y$ ? Well I could answer Y is the Boolean observable for which of the two directions the photon would travel in if it first passed through a certain beam splitter. Here we’re coming up against the fact that since quantum systems are far richer than those that everyday language and intuition are adapted to describing, there isn’t always a satisfactory terminology for describing quantum systems in everyday language.

00:39:14 - 00:41:48

There’s a lot more to be said on this issue, but it would take us far beyond the scope of this lecture. You’ll see in the worked examples that similar methods using phase shifts as well as mirrors and beam splitters allow us to measure any of the other observables of the qubit in a similar way. Note that in the general case where the observable being measured is not sharp, we have to make sure that all the instances of the photon in all universes strike a beam splitter say at the same moment at the same spot from their two possible directions. Otherwise the photon will be on four possible paths and the observable for which path it’s on is no longer Boolean and so we’re no longer talking about a qubit. I described this as a single qubit experiment but to perform it we did need other physical systems aside from the qubit itself. For a start the photon has other subsystems with observables such as its polarization or components of its velocity in directions that we made sure it never traveled in. Another physical system we used was the laser to generate the photon and then the filter to make sure that only one photon at a time was in the apparatus and then there were the mirrors and the beam splitters. These are all physical systems involving vast numbers of atoms and even larger numbers of observables all engaged in intricate interactions. This experiment was carefully arranged so that the net effect of all those interactions on the qubit during the computation could be expressed as a law of motion for the qubit alone. For reasons that I’ll explain next time this would not have been possible for instance with the detectors. As I’ve said that didn’t matter in this experiment because we’re not interested in what happens to the photon after its Z observable has been measured at the end of the experiment. But to understand experiments involving repeated measurements of the same quantum system within a quantum computation we need a quantum analysis of the measurement process and that’s what the next lecture will be about.

2003-01-01 Quantum Computation Lecture 1The Qubit

Read the response at Edge Source collection

Duration: 01:01:56

Transcript

David Deutsch

00:00:30 - 00:02:41

Quantum theory and Einstein’s general theory of relativity are the two great fundamental theories of contemporary physics. Between them, they provide the conceptual framework and the mathematical language in which we express all other theories in physics, and they provide the basic principles to which all known laws of nature conform. The deeper and more general a theory is, the further away it tends to be from everyday experience. So it’s not surprising that our deepest theories involve some very unfamiliar counter-intuitive phenomena, not least of which are the phenomena of quantum computation, the subject of these lectures. But in this first lecture, I won’t describe any phenomena. I’ll give an overview of how quantum theory describes the world and physical processes, and then I’ll introduce you to the simplest of all quantum systems, which is also the centerpiece of quantum computation, the qubit or quantum bit. Quantum computation isn’t something that existing microchips do, even if they rely on quantum mechanical phenomena. Today’s computers don’t count as quantum computers because their repertoire of computations is still the same as that of the abstract universal Turing machine, which was the prototype of all classical computers devised by the mathematician Alan Turing in 1936. Quantum computers will be capable of new modes of computation, which classical computers are incapable of even in principle.

00:02:41 - 00:04:46

The equations that predict the outcomes of quantum mechanical experiments are all uncontroversial, but what the underlying explanation is, what’s happening physically to bring about those outcomes is still very controversial. The various rival explanations are called interpretations of quantum theory. The one I’ll be using sounds like science fiction at first. It was proposed by Hugh Everett in 1957, and it’s called the many universes interpretation. It says that the universe, the space we see around us with all the galaxies and stars and matter, doesn’t constitute the whole of reality. In fact, it’s just a small slice of physical reality as a whole, where there are, among other things, vast numbers of coexisting universes similar to ours. If you’re new to this idea and skeptical, that’s good, but I ask you to go along with it for the purpose of learning the theory. In the course of that, I expect to persuade you that this very fruitful way of understanding quantum theory makes sense. In fact, that it’s the only way that makes sense. Anyway, if there are many universes, we need a new word to denote physical reality as a whole, and that word is, instead of universe, multiverse. Our universe, then, is to some approximation a self-contained entity within the multiverse. This approximation is called classical physics, pre-quantum physics, and in computation theory, it’s called classical computation, that is to say Turing-type computation. As we’ll see, Turing’s theory is a complete model for computations that happen within individual universes.

00:04:46 - 00:07:30

The quantum theory of computation is the full theory, which has the multiverse as its arena. In many physical phenomena, especially on microscopic scales, the classical approximation just breaks down, because in reality, physical objects aren’t confined to just one universe. They have a certain extension across the multiverse, or to put that in another way, every object in one universe has counterparts in a range of other universes, and these counterparts can behave differently from each other, and they can affect each other. Such effects are called quantum interference. They constitute our evidence of the existence of a reality beyond our universe. Under certain circumstances, they permit fundamentally new modes of information processing, which we call quantum computation and quantum communication. The theory of computation was originally conceived of as a branch of pure mathematics. It has been incorporated into physics via the quantum theory of computation, which is now the theory of computation. The previous abstract theory developed by Turing and others lives on only as the classical approximation, though as I said, that’s good enough to describe what all computers currently on the market do. With the benefit of hindsight, we can see that the theory of computation always did have a lot in common conceptually with physics. When a computer performs a computation, it starts with some input information, which it modifies according to definite rules which are characteristic of the hardware of that computer. So the output depends on the input and on the rules by which the computer operates.

00:07:30 - 00:10:06

A physical system is roughly speaking some part of nature that could in principle be experimented on, such as this. Physical systems undergo motion or change. In other words, we can pick any two times and say that the system has changed from an initial state to a final state between those two times, according to laws of motion, which are the laws of physics as specialized to that system. Experiment and measurement are just forms of motion. They involve both a system we’re experimenting on and some measuring instrument or observer. We find the system in an initial state or we prepare it in some way and we prepare a measuring instrument. The system and the measuring instrument then interact according to the laws of physics, which makes the measuring instrument display the outcome of the experiment. You can see that everything in the left-hand column here is a special case of the corresponding thing on the right. But you can also think of that the other way around. Any final state contains information about the system’s initial state and about what has happened to it since. So the motion of any physical system, because it obeys definite laws, can be regarded as information processing. In this first lecture, I’m going to describe the simplest of all quantum physical systems, the qubit, short for quantum bit. To do that, I first have to explain how physical systems are described in quantum theory.

00:10:06 - 00:12:04

The central idea of computer science is that of a computational variable or memory location, a place where information can be stored at one time and perhaps processed and later retrieved. Classical physics has a very similar concept to that, a degree of freedom. The degrees of freedom of a classical system are the real numbers that specify its configuration. For instance, a point particle would have three degrees of freedom because three real numbers are necessary to specify its position in space at a given instant. In quantum physics, the closest thing to a degree of freedom is an observable. Just as in classical physics, any attribute of a physical system that could in principle be prepared with a value that could in principle be measured is a quantum observable. But a quantum observable can’t be summed up as a mere number like the value of a degree of freedom at a given time. There’s a lot more to it than that. It takes a while to get to grips with this concept. The word observable might even be misleading since what we see of a physical object is part of a larger object extending across many universes. A quantum observable refers to what we see and its counterparts in other universes and it contains information about the structure of the multiversal object. So the angle between these two rods, which would be deemed a degree of freedom if this system were described in classical physics, is in fact a quantum observable.

00:12:04 - 00:14:03

This protractor is a measuring instrument that can be used either to prepare that observable with a given value or to measure its value to some degree of accuracy. Let me call that observable $\hat{\theta}(t)$ . I’ll always use this hat symbol or caret for observables to stress that they’re not numbers. If I were to measure the angle at time t and the outcome was say 37 degrees, it would still be quite false to write $\theta(t)=37$ . That’s an observable. It refers to the whole multiversal object in many universes. That’s a number. It just refers to the universes in which the outcome was 37. Strictly speaking, the measuring instrument also includes the light which I use to align the protractor with the rods. That’s because for any measurement to work, something has to be affected by the physical system in question and in this case it’s light that’s affected by the rods and then by the protractor and carries information about them to my eye. For each memory location in a computer, there’s a finite set of possible values that can be stored in it. For instance, one bit can hold two possible values. A byte consisting of eight bits can hold any one of $2^8$ or 256 different values. In the quantum theory similarly, each observable $\hat{X}$ is associated with a set of possible ways in which it could be prepared and which could later be distinguished from each other by measuring x.

00:14:03 - 00:16:27

Each of these ways of preparing x or possible outcomes of measuring x is given a distinct label which is a real number. This set of labels is called the spectrum of x and it’s written like this. For example, the spectrum of the angle $\hat{\theta}$ measured in degrees might be the set of all real numbers between say 10 degrees and 170 degrees. So $\operatorname{spec}(\hat{\theta})=\{x:10\le x\le170\}$ . Now that’s quite a big set in terms of number of members. It has uncountably many members. No real experiment could prepare theta with an arbitrary value in that set or measure precisely what it was. What we could really measure is something like the angle between the rods rounded down to the nearest degree at time t. Let’s call that observable $\hat{\phi}$ . Its spectrum is a discrete set, not a continuum. It’s the integers from 10 to 170. Many quantum observables have discrete spectra from the outset. They’re not approximations to anything continuous at all. That’s what gave quantum theory its name. Quantum means discrete chunk. Notice that a physical system has to be envisaged as having two contrasting properties. On the one hand, it can undergo motion, including being experimented on and measured. It can undergo changes over time. But on the other hand, for the very idea of a physical system that changes to make sense, it must retain its identity over time, which means that it has some characteristics that remain invariant and identifiable throughout any possible change.

00:16:27 - 00:19:23

This angle can change with time, but the fact that the system has an angle observable, that it’s defined in a certain way in terms of these rods, is an invariant feature of the system. If this were disassembled into components, it might still be possible to define theta and to measure it. But ultimately, if the object were melted down, say, it could become physically impossible to identify which atom was which, and then theta would definitely no longer exist, and nor would this physical system. To some extent, it’s arbitrary where we draw the line between a physical system having merely changed its configuration and having ceased to exist. But what’s important is that in quantum physics, it’s always possible to analyze phenomena in terms of physical systems that undergo changes and interactions with other systems, but nevertheless retain their identity over time because some things about them remain invariant. In particular, what observables they have, how they can be measured, and what their spectra are. What else is invariant? Well, the system’s laws of motion. Its actual motion can be different on different occasions, but the laws that determine how the system behaves in isolation or how it would behave under any given external circumstances, these too are invariant features of a physical system. Let me call that which is invariant about a physical system its constitution. Its static constitution is all its invariant characteristics apart from laws of motion, and its dynamics are its laws of motion. To complete the description of a quantum physical system, we need to specify a third thing, namely its state, what it’s actually doing during a particular experiment in all the universes in which it exists. In classical physics, the analog of specifying the state would be specifying which trajectory the system is on. So in the quantum case, the state specifies which of its trajectories in the multiverse, the whole multiversal object is on. In this sense of the word state, the state of a system doesn’t change.

00:19:23 - 00:21:59

Quantum observables change under the laws of motion, but the state is timeless. This sort of state is known as the Heisenberg state of the system. I’m just telling you that in case you come across an alternative way of specifying what a quantum system is doing called the Schrödinger state, which does change with time. I’ll be using the Schrödinger state a lot in later lectures, but for the moment by state, I mean the constant Heisenberg state. In a moment, I’ll tell you how we specify all three things, the static constitution of a system, its laws of motion and its Heisenberg state. But first, some more about observables. We can express the relationship that I defined between the observables theta and phi like this. $\hat{\phi}=\lfloor\hat{\theta}\rfloor$ . Now, floor is the function that takes any real number to the greatest integer less than or equal to it. Since this function is initially defined for real numbers, we have to be careful about what we mean by applying it to an observable. This brings us to our first encounter with the algebra of observables. Given any observable, say X, and any function f that’s defined for every element in the spectrum of x, let’s say x1, x2, up to xn, quantum theory says that there exists another observable, $F(X)$ , whose spectrum consists of elements which are f of elements in the spectrum of x, up to f of xn. For example, we can multiply x by a real number, say lambda, and get another observable, $\lambda X$ , whose spectrum consists of lambda times elements in the spectrum of x.

00:21:59 - 00:23:59

Okay, this is the spectrum of $F(X)$ , but what is $F(X)$ physically? Well, given a way of measuring x, there is guaranteed to be at least one way of measuring $F(X)$ . One way is to measure x and then to compute the function f of the outcome of that measurement. That whole operation is a measurement of $F(X)$ . Another way of measuring the observable $F(X)$ would be to relabel the measuring instrument that’s used to measure x. For example, if we took this protractor and relabeled it so that where it now says 10, 20, 30, and so on, we would write 20, 40, 60, and so on, then the act of lining up that protractor with these rods and reading off the number on that scale would constitute measuring the observable $2\hat{\phi}$ instead of $\hat{\phi}$ . In general, there are also other ways of measuring the observable $F(X)$ that don’t involve x at all. For instance, my measurement of $\hat{\phi}$ here certainly does not involve first measuring $\hat{\theta}$ and then rounding the outcome to the nearest integer, nor does it involve recalibrating an instrument that measures $\hat{\theta}$ . My whole reason for introducing phi was that there just isn’t an instrument that’s capable of literally measuring $\hat{\theta}$ .

00:24:07 - 00:27:16

Now, let g be the function that maps any real number to a constant, say 1. Consider the observable g of x. Its spectrum just contains a single element, spectrum of g of x equals the set containing just 1. And therefore, to measure g of x, you don’t even have to do an experiment. All you have to do is write down the outcome was 1. We call this trivial observable the unit observable, $\hat{1}$ . $\operatorname{spec}(\hat{1})=\{1\}$ . Similarly, $\lambda\hat{1}$ , where lambda is any real number, must be another trivial observable, the one where the only possible outcome of measuring it is lambda. I said that the spectrum of an observable is part of the system’s invariant identity. We can express this invariance in an algebraic way. Let the values in the spectrum of some observable, x of t, be x1, x2, and so on, up to xn. These values don’t change with time. Consider this function, p, maps x to x minus x1 times x minus x2 times, and so on, up to x minus xn. This is a polynomial. It’ll have an expression of x1, x2, and so on. It’ll have an expansion like this. $p(x)=a_0+a_1x+\cdots+a_nx^n$ . The degree n of the polynomial is the same as the number of elements in the spectrum of x. Now, since p is a function that’s defined for all elements of the spectrum of x, in fact, it’s defined for all real numbers, $p(X(t))$ must be an observable. And it’s easy to find out what observable it is if we first work out its spectrum.

00:27:16 - 00:30:19

The elements of the spectrum are p of x1 and p of x2 and so on, up to p of xn. And those values are all 0, because whenever little x is in the spectrum, there’s a term in this product that’s 0. So the spectrum of $p(X(t))$ only contains one element, namely 0. Measure it, and you’ll always get the outcome 0. And therefore, $p(X(t))$ itself must be 0 times the unit observable, which we also write just 0. So although the observable x itself changes with time, we found an algebraic equation that it satisfies at all times. Let me write it down explicitly. $a_0\hat{1}+a_1X(t)+\cdots+a_nX(t)^n=0$ . Therefore, this algebraic statement about x of t is a statement about the static constitution of the quantum system that x belongs to. And here’s a general truth about quantum systems. Every algebraic relation among observables at one time is true of the same observables at every other time too. And so is part of the static constitution of the system. In fact, the set of all true algebraic relations among observables at any one time defines the static constitution of the system. So suppose you found an algebraic equation that related some of the observables of the system at time t. Say $f(A(t),B(t),\ldots)=0$ . Then no matter how much the system may change or interact with other systems, so long as it remains in existence at all, quantum theory says that its observables will only change in ways that keep that equation and all such equations true.

00:30:20 - 00:32:47

The set of all true algebraic equations among the observables of a system is called the algebra of the observables. The algebra of the observables at any one time, like this, is what specifies the system’s static constitution. The algebraic relationships between observables at different times specify the system’s dynamics or laws of motion. These can usually be summarized as differential equations, in other words, algebraic relationships between observables at infinitesimally different times. So in summary, the whole constitution of a quantum system is defined by the whole algebra of its observables, including observables at different times. What about the state? To specify the state of a quantum system, you have to specify a function called the expectation value function that maps each observable x to a real number called its expectation value, which is written like this. The expectation value of the observable x of t, or the expectation value of x at time t. Quantum theory places certain constraints on this function, namely that the expectation value of an observable is never lower than the lowest element in its spectrum, and never higher than the highest element. So the $\min\operatorname{spec}(\hat{X})\leq\langle\hat{X}\rangle\leq\max\operatorname{spec}(\hat{X})$ . And the other condition is that it has to be a linear function. So the expectation value of $\lambda\hat{X}+\mu\hat{Y}$ , say, equals $\lambda\langle\hat{X}\rangle+\mu\langle\hat{Y}\rangle$ . So $\langle\lambda\hat{X}+\mu\hat{Y}\rangle=\lambda\langle\hat{X}\rangle+\mu\langle\hat{Y}\rangle$ .

00:32:55 - 00:34:58

I’ll give an example of an expectation value function in a moment, but first let me give an indication of what expectation values mean. If you do an experiment, you’re doing it in a range of universes. You and the system and the measuring instrument are all multiversal objects. In general, all the outcomes in the spectrum of x actually occur in different universes. Therefore, it is in general impossible to predict a specific outcome for a measurement. That’s where expectation values come in. They are what quantum theory makes predictions about. For instance, suppose you perform the same experiment repeatedly. What’s meant by the same experiment? Well, you keep preparing the system in the same way again and again, and on each occasion you measure x at a time t later. Then the average of all the outcomes of all those measurements tends towards the expectation value of x of t. In the limit of an infinite sequence of preparations and measurements, the average outcome would be exactly the expectation value of x of t. That’s one rough and ready operational meaning of the expectation value. Another rough and ready operational meaning is this. Suppose you make many copies of the system and prepare them all in identical ways and then measure x and its counterparts in the copies, then the average outcome tends to the expectation value of x as the number of copies tends to infinity.

00:34:58 - 00:36:55

These meanings cease to be operational meanings as soon as you insert the qualification infinitely, but they cease to be strictly true if you leave it out. Well, the real meaning of the expectation value is that it’s the average value of x over a region of the multiverse, the region where the system is prepared and measured in the given way. But I’m getting ahead of myself. In due course, I’ll explain what a region of the multiverse is and how one averages over it and how that relates to these more operational meanings of the expectation value of an observable. Historically, the first formulation of what we would today call quantum theory was proposed by Werner Heisenberg with the help of Max Born and Pascual Jordan in 1925. It was called matrix mechanics because in it, each observable is represented by a Hermitian matrix. Check the accompanying notes if you want a summary of the basic properties of Hermitian matrices. The eigenvalues of the matrix, which are real numbers, constitute the spectrum of the observable represented by the matrix. We’re going to be using a lot of matrices, but it’s important to keep at the back of your mind that it’s not the matrices but the algebra of the matrices, the algebra of the observables that defines a physical system. If we found any set of matrices that satisfy the same algebraic relationships as the observables of a given physical system, then those matrices would do as a description of the constitution of the system. So there’s a lot of freedom to choose quite different sets of matrices to describe the same physical system.

00:36:55 - 00:39:14

One constraint is that all the matrices representing observables of a given system must have the same dimension. Otherwise we couldn’t add and subtract them and do arithmetic with them as I’ve described. For each physical system there’s a minimum dimension of matrix required. That’s determined by the largest spectrum: if an observable has a spectrum with $n$ elements, then the matrix representing that observable has $n$ different eigenvalues, so it must be at least an $n\times n$ matrix. Furthermore, not only is every observable of the system represented by an $n\times n$ Hermitian matrix, every such matrix represents an observable of the system. So that’s how quantum theory describes the world. Now what is the simplest possible quantum physical system? In classical computation, the simplest possible memory location is one that can hold either one of exactly two values. That’s called a bit. Considered as a variable, a bit is something like a degree of freedom that has only two possible values, though such things don’t fit comfortably into the scheme of classical dynamics, which is based on continuously varying degrees of freedom. The simplest possible quantum observable is a Boolean observable, defined as an observable with exactly two eigenvalues. Any observable simpler than that, having only one eigenvalue, would be trivial. It would be a multiple of the unit observable. Every physical system has Boolean observables. For instance, whether theta is less than 90 degrees or greater than or equal to 90 degrees is a Boolean observable.

00:39:14 - 00:41:28

It’s the observable $\lfloor\hat{\theta}(t)/90\rfloor$ , whose eigenvalue 0 stands for less than 90 and 1 stands for greater than or equal to 90. The fact that an observable has exactly two eigenvalues is physically much more significant than what those eigenvalues actually are. Remember that the eigenvalues are just labels for possible outcomes of measurements, and we can always relabel those. So for any observable, say X, with eigenvalues A and B, there exists another observable of the form $\alpha\hat{X}+\beta\hat{1}$ that has any other two eigenvalues we like. And measuring that observable isn’t going to be much different from measuring X. So consider any physical system, let’s say S. And let’s choose a Boolean observable of S, call it $\hat{Z}(t)$ , that has eigenvalues plus or minus 1. That turns out to be slightly more elegant for our purposes than choosing 1 and 0. Now measuring $Z(t)$ has only those two possible outcomes. We could measure it by first measuring some more complicated observable of S and then evaluating some function of the outcome that ranges over plus and minus 1. But usually there are easier and more direct ways of measuring Z, which involve ignoring most of S. In other words, we need only interact with the subsystem of S, in which case $Z(t)$ is also an observable of that subsystem. How simple could that subsystem be? In other words, what’s the simplest kind of physical system that could hold one bit of information?

00:41:28 - 00:43:35

Well if $Z(t)$ is represented by a certain matrix, it has to be at least a 2 by 2 matrix because it must have exactly two eigenvalues. Then as I’ve said, every other Hermitian matrix of the same dimension also represents an observable of the same system. So we know that any physical system that has $Z(t)$ as an observable also has a whole continuum of other observables, one for every 2 by 2 Hermitian matrix. And that’s the minimum set of observables that a physical system can have. Every 2 by 2 matrix has either one or two distinct eigenvalues. Therefore, every observable of this minimal type of physical system is either a Boolean observable or a multiple of the unit observable. A physical system with that property that every one of its non-trivial observables is a Boolean observable is called a qubit. Bear in mind the difference between a qubit, a Boolean observable and a bit. A bit is a degree of freedom that can take one of two possible values. A Boolean observable is the quantum generalization of that. In any one universe it resembles a bit, but it can take two different values simultaneously in different universes. A qubit is a physical system, a minimal physical system that contains Boolean observables. And I’ll describe several such systems in future lectures. Now consider a qubit at a given time, say t equals 0. We won’t consider other times, so I can drop the t. This qubit will have many Boolean observables.

00:43:35 - 00:46:09

Let’s pick one that has eigenvalues plus and minus 1 and is represented by a diagonal matrix. Let’s call that observable Z. Now I’ll define a state that this qubit could be in during some experiment. To do that, remember, I have to specify a function on the set of all its observables, which means on all 2 by 2 Hermitian matrices. And this is the function I’ll specify. The expectation value of $\begin{pmatrix}A&B\\B^*&C\end{pmatrix}$ equals just A. In other words, the expectation value of any observable is the top left element of the matrix representing that observable. In this state that I’ve defined. If you look at the worked examples for this lecture, you can verify that this function has all the properties that I specified for expectation value functions, and therefore that it specifies a possible state. So what does this tell us about what’s happening to our qubit? Well, what’s happening to its observable Z? Well, the expectation value of Z is minus 1. So for instance, if we repeated the whole experiment many times, repeated it including the initial preparation and everything, then the average value of all the outcomes would be minus 1. OK, but look at the spectrum of Z. It only contains the values minus 1 and 1. So each individual outcome must be one of those two values. And so if the average over many outcomes is minus 1, it follows that every single outcome must in fact be minus 1. Therefore, in this case, we don’t have to bother with repeating the experiment many times.

00:46:09 - 00:48:29

We don’t have to set up an ensemble of identically prepared copies of the system. We don’t have to worry about what’s happening in other universes. Quantum theory predicts that the outcome of a measurement of Z will be minus 1. In a given state, when an observable has this property, that if it were measured, the same outcome would occur in all universes, then the observable is said to be sharp in that state. Now let’s think about measuring a different observable of this same qubit at time 0. Let’s say this one. $X=\begin{pmatrix}0&1\\1&0\end{pmatrix}$ . So we can read off the expectation value. It’s the top left corner. The expectation value of X is 0. So if we make many copies of this experiment measuring X, the average value of the outcome will be 0. Same with repeating it many times. Same is true of the average value over all the universes in which X is measured. No particular outcome will be 0, though. That’s because, as you can verify in the worked examples, the eigenvalues of X are also plus and minus 1. So each outcome will be either plus 1 or minus 1. If their average is going to tend to 0, well, you can see that half of them are going to have to be plus 1 and the other half minus 1 in the long run. This translates to a probabilistic prediction. The probability of each of these two outcomes is one half. So the observable X is not sharp. In fact, it’s as far from being sharp in this state as a Boolean observable can get. Z was sharp, X isn’t. And that’s no accident.

00:48:29 - 00:50:09

And Why?

“What Questions Have Disappeared…And Why?” Funny you should ask that. “And why?” could itself be the most important question that has disappeared from many fields.

“And why?”: in other words, “what is the explanation for what we see happening?” “What is it in reality that brings about the outcome that we predict?” Whenever we fail to take that question seriously enough, we are blinded to gaps in our favoured explanation. And so, when we use that explanation to interpret regularities that we may observe, instead of understanding that the explanation was an assumption in our analysis, we regard it as the inescapable implication of our observations.

“I just can’t feel myself split”, complained Bryce DeWitt when he first encountered the many-universes interpretation of quantum theory. Then Hugh Everett convinced him that this was the same circular reasoning that Galileo rejected when he explained how the Earth can be in motion even though we observe it to be at rest. The point is, both theories are consistent with that observation. Thanks to Everett, DeWitt and others, the “and why” question began gradually to return to quantum theory, whence it had largely disappeared during the 1930s. I believe that its absence did great harm both in impeding progress and in encouraging all sorts of mystical fads and pseudo-science. But elsewhere, especially in the human philosophies (generally known as social sciences), it is still largely missing. Although behaviourism — the principled refusal to ask “and why?” — is no longer dominant as an explicit ideology, it is still widespread as a psychological attitude in the human philosophies.

Suppose you identified a gene G, and a human behaviour B, and you undertook a study with 1000 randomly chosen people, and the result was that of the 500 people who had G in their genome, 499 did B, while of the 500 who lacked G, 499 failed to do B. You’d conclude, wouldn’t you, that G is the predominant cause of B? Obviously there must be other mechanisms involved, but they have little influence on whether a person does B or not. You’d inform the press that all those once-trendy theories that tried to explain B through people’s upbringing or culture, or attributed it to the exercise of free will or the logic of the situation or any combination of such factors — were just wrong. You’ve proved that when people choose to do B, they are at the very least responding to a powerful influence from their genes. And if someone points out that your results are perfectly consistent with B being 100% caused by something other than G (or any other gene), or with G exerting an influence in the direction of not doing B, you will shrug momentarily, and then forget that possibility. Won’t you?

About David Deutsch

David Deutsch’s research in quantum physics has been influential and highly acclaimed. His papers on quantum computation laid the foundations for that field, breaking new ground in the theory of computation as well as physics, and have triggered an explosion of research efforts worldwide. He is a member of the Centre for Quantum Computation at the Clarendon Laboratory, Oxford University and the author of The Fabric of Reality.

Read the interview at Philosophy Now Source collection

2000-12-01 Philosophy NowInterview with David Deutsch

Interview

David Deutsch is a distinguished quantum physicist and a member of the Centre for Quantum Computation at the Clarendon Laboratory, Oxford University. He has received the Paul Dirac Prize and Medal from the Institute of Physics for ‘outstanding contributions to theoretical physics’. He recently talked with Filiz Peach about his work and hopes.

David Deutsch’s book The Fabric of Reality offers a startling new worldview which combines quantum physics, evolution, epistemology and computation. He also deals with quantum computation, a new field of physics in which he has been a pioneer. His explanation of the nature of the universe in terms of quantum physics is inspiring and thought-provoking. However, his favoured interpretation of quantum theory in terms of there being many parallel universes (or a ‘multiverse’ as he calls it) is not widely accepted in the scientific community, or at least not yet. But it may well be part of a new unifying theory of the universe in the 21st century. The Fabric of Reality is a clearly-written book, intelligible even for those of us who are not scientists. It was shortlisted for the 1997 Los Angeles Times Book Prize, and the 1998 Rhône-Poulenc Prize for Science Books.

FILIZ PEACH: Professor Deutsch, could you please tell our readers why you became interested in quantum physics?

DAVID DEUTSCH: I am interested in anything that is fundamental. Quantum physics and the General Theory of Relativity are the two most fundamental theories that physics has. They are the theories within which other theories are formulated; they provide the framework for all of physics.

FILIZ PEACH: So how did you first become involved?

DAVID DEUTSCH: When I was a graduate student, for my thesis I studied quantum field theory in curved space-time – a topic that is on the boundary between quantum theory and the General Theory of Relativity. It was hoped – it still is hoped – that one day these two theories will be unified. Logically, they are in deep conflict with each other, and this conflict is not within the reach of present day experiments to resolve. We know that a unification isn’t going to be easy. That unified theory would be called quantum gravity. The reason for studying quantum field theory in curved space-time was that it was hoped that when we understood that well, it would provide a clue to quantum gravity. We did eventually understand it well, and it did not provide a clue to quantum gravity. But it did convince me that quantum theory is at present the deeper of the two, and also, for the moment at any rate, provides more promising lines of research. Peter Medawar said once that science is the art of the soluble. You cannot necessarily solve the most profound problems right away. You have to go for the most profound soluble problem. And in that respect, I thought that quantum theory was the more promising.

FILIZ PEACH: So you now believe that quantum mechanics will provide a unifying theory of the universe?

DAVID DEUTSCH: ‘Provide’ is not quite the right word. Quantum theory will be a pathway, a component of some future more unifying theory which will involve among other things the General Theory of Relativity. But also I think it will involve areas which are now not even considered part of physics. Certain areas of epistemology, certain parts of philosophy and mathematics, and the theory of evolution will also be part of the new unifying theory, of which we do have glimpses but which has not yet been formulated.

FILIZ PEACH: Quantum mechanics is very complex. And there are still unresolved areas. Do you think the mystery of it may be resolved, say, within 20 years or so? Or is that too optimistic?

DAVID DEUTSCH: One hears a lot about the ‘mysteries’ of quantum mechanics but I do not think that there are any. Although there are still open areas of research within quantum mechanics I do not think that they are fundamental mysteries provided that one adopts the many-worlds interpretation of quantum physics [see page 22]. There are mysteries in physics, principally the unification of quantum theory with General Relativity. We really have only clues at the moment, and I would be rash to predict that this would be solved in the next 20 years, although this is one of those areas where the solution could come at any time. And then there would be a frantic rush to work out its meaning. Even that frantic rush might take decades. So, I do not know.

FILIZ PEACH: How will this theory help to explain man’s existence in the world?

DAVID DEUTSCH: Again, we don’t know yet. We only have some tantalising clues. It seems likely to me that the 400 year old consensus in science that human beings are insignificant in the fundamental scheme of things in the universe has to break down. It is not that we know what the true role of humans is. It is that the arguments that humans don’t have a fundamental role in the scheme of things, which used to seem so self-evidently true, have all fallen away. I mean, it is no longer true that human beings are necessarily destined to have a negligible effect on physical events, because there is the possibility that humans will spread and colonize the galaxy. If they do, they will necessarily have to affect its physical constitution in some ways. It is no longer true that the fundamental quantities of nature – forces, energies, pressures – are independent of anything that humans do, because the creation of knowledge (or ‘adaptation’ or ‘evolution’ and so on) now has to be understood as one of the fundamental processes in nature; that is, they are fundamental in the sense that one needs to understand them in order to understand the universe in a fundamental way. So, in this and other ways, ‘human’ quantities – human considerations, human affairs and so on – are fundamental after all. But we do not yet understand the details of how they fit in with the more familiar fundamental processes that we know about from physics.

FILIZ PEACH: What scientists or philosophers have most influenced your own work?

DAVID DEUTSCH: Let us deal with the philosophers first because that is a shorter list. I think it is principally Karl Popper, and to a lesser extent Jacob Bronowski (through The Ascent of Man) and William Godwin, who is a very underrated 18th century philosopher, with a broader, more integrated and more sophisticated perspective than, say, Locke or Hume. He is underrated because he made serious mistakes too. For instance, he completely misunderstood economics and that led him to advocate a sort of communistic lifestyle. Yet many of his political ideas are actually spot on, and very modern.

As far as the scientists go, one can divide them into two categories, that is scientists who personally influenced me, and those whose work influenced my work. The ones who personally influenced me were Dennis Sciama, the cosmologist and astrophysicist who sadly died last year, and John Wheeler. Both had the very rare attribute of being able to choose and nurture excellent students. Sciama, for example, was the supervisor of Martin Rees, Stephen Hawking and, in all, over a dozen of the foremost physicists and cosmologists in Britain. And the same is true in America with John Wheeler. The third person I should mention is Bryce DeWitt, who I worked under when I was in Texas as a student. He was the one who introduced me to Everett’s many-worlds interpretation of quantum mechanics, and to the wider implications of quantum field theory, and it was because of his take on both the formalism and interpretation of quantum mechanics that I got interested in quantum computers.

FILIZ PEACH: In the context of the current interest in human consciousness how do you see the relationship between the material explanation of the human being and consciousness? How does consciousness fit into the quantum world?

DAVID DEUTSCH: First of all, I do not believe in the supernatural, so I take it for granted that consciousness has a material explanation. I also do not believe in insoluble problems, therefore I believe that this explanation is accessible in principle to reason, and that one day we will understand consciousness just as we today understand what life is, whereas once this was a deep mystery.

FILIZ PEACH: Are you saying that human consciousness can be reduced to neural activities in the brain?

DAVID DEUTSCH: No, no. ‘Reduction’ to an underlying level is just one possible mode of explanation. For instance, although we know that living processes, at the reductionist level are nothing more than physical and chemical processes, we also know that their explanation cannot be made at that underlying level. That is, although the physics of life is not different from the physics of anything else, the explanation of life requires a substantive new theory, namely the theory of evolution. That is the kind of relationship, I think, that consciousness has with physics; the explanation of consciousness again needs a different mode of explanation, except that for consciousness it has not yet been invented, that is the problem. I am completely unsatisfied with modes of explanation such as Daniel Dennett’s which try to say that the problem is already solved. In general I think that it is rare for a situation to exist where a lot of people think there is a problem and in fact it is already solved. In the case of consciousness I think that there are genuine problems, for instance the problem of what are qualia (such as the subjective experience of seeing red). This is clearly unsolved and Dennett’s proposals don’t solve it.

FILIZ PEACH: In your book, The Fabric of Reality, you are challenging the single universe conception of reality. In Chapter II, you clearly explain quantum theory which tells us about the behaviour of microscopic particles. You also explain the ‘single particle interference’ experiment and argue that there are intangible shadow particles, and then that there are parallel universes each of which is similar to the tangible one. This is a difficult step for many of us. Could you please clarify how you proceed from intangible particles to many universes (or multiverse as you call it)?

DAVID DEUTSCH: Let’s start with the microscopic world, because it is only at the microscopic level that we have direct evidence of parallel universes. The first stage in the argument is to note that the behaviour of particles in the single slit experiment reveals there are processes going on that we do not see but which we can detect because of their interference effects on things that we do see. The second step is to note that the complexity of this unseen part of the microscopic world is much greater than that which we do see. And the strongest illustration of that is in quantum computation where we can tell that a moderate-sized quantum computer could perform computations of enormous complexity, greater complexity than the entire visible universe with all the atoms that we see, all taking place within a quantum computer consisting of just a few hundred atoms. So there is a lot more in reality than what we can see. What we can see is a tiny part of reality and the rest of it most of the time does not affect us. But in these special experiments some parts of it do affect us, and even those parts are far more complicated than the whole of what we see. The only remaining intermediate step is to see that quantum mechanics, as we already have it, describes these other parts of reality, the parts that we don’t see, just as much as the parts we do see. It also describes the interaction of the two, and when we analyse the structure of the unseen part we see that to a very good approximation, it consists of many copies of the part that we can see. It is not that there is a monolithic ‘other universe’ which is very complicated and has different rules or whatever. The unseen part behaves very like the seen part, except that there are many copies.

It is rather like the discovery of other planets or other galaxies. Having previously known only the Milky Way, we did not just find that there are vast numbers of stars out there, far more than in the Milky Way. There are more galaxies out there than there are stars in the Milky Way. We also found that most of the stars outside the Milky Way are actually arranged in other little Milky Ways themselves. And that is exactly what happens with parallel universes. It is of course only an analogy but quite a good one; just like the stars and galaxies, the unseen parts of reality are arranged in groups that resemble the seen part. Within one of these groups, which we call a parallel universe, the particles all can interact with each other, even though they barely interact with particles in other universes. They interact in much the same way as the ones in our seen universe interact with each other. That is the justification for calling them universes. The justification for calling them parallel is that they hardly interact with each other, like parallel lines that do not cross. That is an approximation, because interference phenomena do make them interact slightly. So, that is the sequence of arguments that leads from the parallelism, which by the way is much less controversial at the microscopic level than the macroscopic level, right up to parallel universes. Philosophically, I would like to add to that that it simply does not make sense to say that there are parallel copies of all particles that participate in microscopic interactions, but that there are not parallel copies of macroscopic ones. It is like saying that someone is going to double the number of pennies in a bank account without doubling the number of Pounds.

FILIZ PEACH: But couldn’t this interference phenomenon be due to a yet unknown law of physics within this universe?

DAVID DEUTSCH: Well, there are very sweeping theorems that tell us that no single-universe explanation can account for quantum phenomena in the same way that the full quantum theory does. Quantum theory explains all these phenomena to the limits of present day experiment perfectly, and it is, according to some measures anyway, the best corroborated theory in the history of science. And there are no rival theories known except slight variants of quantum theory itself. We know that an alternative explanation could not be made along single-universe lines, unless perhaps it is a completely new kind of theory. So, the answer is ‘no’.

FILIZ PEACH: A few years ago, BBC Horizon did a documentary on time travel in which you explained the parallel universes theory and suggested that there was ‘hard evidence’ for it. Well, it is a controversial theory and is accepted only by a minority of physicists, as you yourself acknowledge in your book. Why do you think there is such a strong reaction to this theory in the scientific community? And how do you reply to their criticism?

DAVID DEUTSCH: I must confess that I am at a loss to understand this sociological phenomenon, the phenomenon of the slowness with which the many universes interpretation has been accepted over the years. I am aware of certain processes and events that have contributed to it. For instance Niels Bohr, who was the inventor of the Copenhagen interpretation, had a very profound influence over a generation of physicists and one must remember that physics was a much smaller field in those days. So, the influence of a single person, especially such a powerful personality as Niels Bohr, could make itself felt much more than it would be today. So that is one thing – that Niels Bohr’s influence educated two generations of physicists to make certain philosophical moves of the form “we must not ask such and such a question.” Or, “a particle can be a wave and a wave can be a particle,” became a sort of mantra and if one questioned it one was accused of not understanding the theory fully. Another thing is that quantum theory happened to arise in the heyday of the logical positivists. Many physicists – perplexed by the prevailing interpretations of quantum physics – realised that they could do their day-to-day job without ever addressing that issue, and then along came a philosophy which said that this day-to-day job was, as a matter of logic, all that there is in physics. This is a very dangerous and stultifying approach to science but many physicists took it and it is a very popular view within physics even to this day. Nobody will laugh at you if, in reply to the question “are there really parallel universes or not?”, you answer “that is a meaningless question; all that matters is the shapes of the traces in the bubble chamber, that is all that actually exists.” Whereas philosophers have slowly realised that that is absurd, physicists still adopt it as a way out. It is certainly no more than ten percent, or probably fewer, of physicists talking many universes language. But it is heartening that the ones who do tend to be the ones working in fields where that question is significant, which are quantum cosmology and quantum theory of computation. By no means all, even in those fields, but those are the strongholds of the many-worlds interpretation. Those also tend to be the physicists who have thought most about that issue. But why it has taken so long, why there is such resistance, and why people feel so strongly about this issue, I do not fully understand.

FILIZ PEACH: I know that you are also working on a quantum computer. Given the counterintuitive character of the quantum world, it must be a very challenging project.

DAVID DEUTSCH: It is a very hard technical task, and the science is in its infancy. I am not involved in any of the experimental work, except as a spectator. I work only on the theory. I can only say that I am extremely impressed by the power of the experimental techniques that are now available. These people routinely manipulate individual atoms and individual photons, and engineer interactions between them and measure them with extraordinary precision, and they are very optimistic about the possibility of building working quantum computers. At the moment the most powerful quantum computer in the world probably has 3 or 4 qubits. One would probably need several hundred to perform any quantum computation that was useful as such.

FILIZ PEACH: How close are you to achieving your objective?

DAVID DEUTSCH: There are many intermediate objectives, but speaking of the objective of a quantum computer that can actually perform useful quantum computations, we are decades away. But there are many intermediate objectives of great theoretical and philosophical interest which will happen before that.

FILIZ PEACH: Could you perhaps tell us how a quantum computer can contribute to our understanding of quantum mechanics? And what kind of effect can it have, if any, on our everyday lives?

DAVID DEUTSCH: Those are two questions. For the first one, I think quantum computers will contribute in two separate ways. One is that the theory of quantum computation appears to be a very elegant and powerful way of looking at quantum mechanics in general, and quantum mechanics in general is arguably the deepest theory in physics along with General Relativity. Expressing the theories of physics in the language and notation of quantum computation makes them clearer and gives us a deeper understanding of what they mean. The other way that it helps us understand physics is by helping us to understand the many universes theory. Before quantum computation the prototype experiments which would demonstrate the existence of parallel universes were things like the two-slit experiment where the number of universes involved is small. The interaction between them is very crude. A particle is deflected into another direction, and not much else happens. When you finish the interference has ended. But in quantum computation the complexity of what is happening is very high so that philosophically, it becomes an unavoidable obligation to try to explain it. It is not just a correction to something else; it is the overwhelmingly dominant effect. It is not just crude; the outcome is a complex and subtle function of how the experiment is set up, and of what happens in the hidden parts of the multiverse. One can then take those results and as with any other computation one can put them into a further quantum computation and the second one will work only if the first one produced all the right results in all the universes. It really cries out for explanation rather than simply prediction. This will have philosophical implications in the long run, just in the way that the existence of Newton’s laws profoundly affected the debate on things like determinism. It is not that people actually used Newton’s laws in that debate, but the fact that they existed at all coloured a great deal of philosophical discussions subsequently. That will happen with quantum computers I am sure.

In our everyday lives, that is still an open question, because that rather depends on how feasible it is to build quantum computers and how cheap they will be when we do build them. It also depends on, theoretically, how many useful types of quantum algorithm are invented. The only general-purpose useful algorithm so far is Grover’s algorithm, which is a search algorithm. If quantum computers can be built economically then they will have an impact because of Grover’s algorithm. Search is a component of almost every computer program because searching through a list of possibilities is what you do in every case where there is not a clever mathematical algorithm to get what you want. An obvious example is chess playing; there is no formula for the best chess move given a certain position. All you do is search through all the possibilities of how the given position can continue. And the fastest known algorithms are simply search algorithms. They take one move after another and just search down to whatever depth they can in a given time.

FILIZ PEACH: Can an ordinary computer do the same job?

DAVID DEUTSCH: Yes, ordinary computers can perform searches; the best existing chess computers are ordinary computers which do normal searches. But Grover’s algorithm does searching much faster than any classical algorithm could do. It is a feature of classical searching that if you are searching through n possibilities, the time taken is proportional to n – that is, basically it is n times the time taken to look at one possibility. Quantum computing using Grover’s algorithm uses the square root of n steps, so it needs only the time to look at the square root of the total number of possibilities, and it shares the work among the square root of n universes. To put it another way, in the time a classical computer can perform a thousand search steps, a quantum computer can perform a million. In the time the classical one can perform a million, a quantum computer can perform a trillion. You soon get to the region where the classical computer is outclassed even if the quantum computer is slower in terms of the actual steps. I think the existing computers perform hundreds of millions of search steps per second and to play a chess move takes a few seconds. So, a quantum computer doing the same kind of thing would be able to perform some trillions of times more analyses and therefore would completely outclass Deep Blue, the best existing chess machine. But it is not just chess, it is any problem where one has to search through possible solutions: cryptography, design, where you are trying different wing shapes for an aeroplane or whatever. Anywhere where there is not a formula to the answer. Probably most of computer time that is currently devoted to solving problems, is devoted to searching of some kind or other.

FILIZ PEACH: In your research do you get support from your colleagues or is there a general scepticism around?

DAVID DEUTSCH: I would say that I am sceptical myself about, for instance, the speed of progress that we can expect in quantum theory and experiments. I am sceptical but optimistic at the same time. As regards the subject of quantum computers, it is generally regarded as an exciting growth area. The Centre for Quantum Computation has been formed at the Clarendon Laboratory in Oxford and it is attracting world class researchers and they seem to get some outstanding research students too. We are making a remarkable progress and the field is regarded by the physics community at large as very promising. Of course, we cannot predict the future growth of knowledge, as Popper would say. So, we do not know that our progress in the future will continue to be as exciting and as rapid as it has been. That is the way it is looking at present.

FILIZ PEACH: What is the most frustrating part of your research?

DAVID DEUTSCH: I think perhaps, if I am to pick out some frustrating part, it is that the field has now grown so much and become so complex that I cannot follow it all. There are whole areas, for instance, in the mathematical theory of quantum computers, quantum complexity theory, where I just do not know enough to follow the latest research in detail. So I have to pick and choose. For many years I was in the fortunate position of being in a very, very new field, and everybody knew everybody. Everybody understood everyone else’s research. That is no longer the case. It is just too big and too diverse. We do still have, though, the atmosphere of camaraderie where we all help each other that we had originally. I think what tends to happen when fields get big is that competition and rivalry set in, and people tend to hide their results from each other. So far, that is not happening in our field and it is wonderful.

FILIZ PEACH: In view of scientific developments in areas like biochemistry, DNA research, genetic engineering, information technology, are you optimistic about the 21st century? Or do you see a dark side?

DAVID DEUTSCH: Oh yes. I am optimistic about technological progress, but there is bound to be a dark side. There are bound to be many horrible unintended consequences of new knowledge. That always happens. I think rationalism, the whole philosophical stance of advocating reason and progress would do much better to glorify problems than theories. It is problems that are inherently wonderful; solutions are merely useful. And the fact that solutions always create new problems is not, on balance, a drawback but their most useful attribute. Science ought to be regarded as a transition from one problem situation to the next. The theory – the means by which we make a transition – is secondary. It is the problem that is primary. In fact, I even sometimes say, only half jokingly, that theories ought to be renamed ‘misconceptions’, and that progress consists of moving from one misconception to a preferable misconception. That is, from a misconception that contains a great deal of falsehood to one that contains less falsehood. Then perhaps we would not be tempted to hubris when we make a great discovery. Also the public would not gain the mistaken impression that science claims to know everything and to solve everything and to insulate the human race against uncertainty or error. That is something science cannot do. But the other side of the coin is that we ought to be embracing new problem situations as good; we have to accept that bad things will happen, but we ought to expect to solve them in turn. Because the only alternative is to stick with the bad things that we have and then we might as well be dead.

FILIZ PEACH: In explaining the world, do you think science and philosophy are compatible? Can they interact?

DAVID DEUTSCH: Absolutely. In fact science and philosophy have both gone through a bad period in the 20th century, philosophically speaking. Many blind alleys were explored, many steps for the worse were taken, not in the predictive part of science but in the explanatory part, and in philosophy generally. I think that in the last years of the 20th century people began to realise this and do what is necessary to cure philosophy of these ills. I think it is now basically taken for granted once again that philosophy is about understanding things, questioning things and that logic makes sense and that theories have to be coherent. There are genuine philosophical problems, not just word games; there are such things as solutions even though they are very hard to come by, though perhaps in line with my earlier comment we should really rename the solutions ‘misconceptions’ just so that we understand what they really are. We have a set of misconceptions and we are trying to move to a better set of misconceptions. Scientists ironically do drag their feet, there is still a lot of positivism, a lot of instrumentalism, a lot of not taking philosophy seriously, but things are going in the right direction.

FILIZ PEACH: Professor Deutsch, thank you very much. It has been a pleasure talking with you.

You can find out more about quantum computation at the website of the Centre for Quantum Computation: www.qubit.org

Filiz Peach is working on a PhD on Existentialist perspectives on death. She lives in London.

Read the interview at Edge Source collection

2000-11-19 EdgeIt’s a Much Bigger Thing Than It Looks

However useful the theory [of quantum computation] as such is today and however spectacular the practical applications may be in the distant future, the really important thing is the philosophical implications — epistemological and metaphysical — and the implications for theoretical physics itself. One of the most important implications from my point of view is one that we get before we even build the first qubit [quantum bit]. The very structure of the theory already forces upon us a view of physical reality as a multiverse. Whether you call this the multiverse or ‘parallel universes’ or ‘parallel histories’, or ‘many histories’, or ‘many minds’ — there are now half a dozen or more variants of this idea — what the theory of quantum computation does is force us to revise our explanatory theories of the world, to recognize that it is a much bigger thing than it looks. I’m trying to say this in a way that is independent of ‘interpretation’: it’s a much bigger thing than it looks.

Introduction

In 1998 Oxford physicist David Deutsch was awarded the Paul Dirac Prize “For pioneering work in quantum computation leading to the concept of a quantum computer and for contributing to the understanding of how such devices might be constructed from quantum logic gates in quantum networks.”

“Quantum computing,” Deutsch says, “is information processing that depends for its action on some inherently quantum property, especially superposition. Typically we would superpose a vast number of different computations — potentially more than there are atoms in the universe — and then bring them together by quantum interference to get a result. Other quantum computations, notably quantum cryptography, couldn’t be done by classical computers even in theory.”

Deutsch’s work on quantum computation has led him into two important areas of research concerning (a) “the structure of the multiverse — making precise what we mean by such previously hand-waving terms as ‘parallel’, ‘universes’ and ‘consists of’. It turns out that the structure of the multiverse is largely determined by the flow of quantum information within it, and I am applying the techniques we used in that paper to analyse that information flow”; and (b) “a generalization of the quantum theory of computation, to allow it to describe exotic types of information flow such as we expect to exist in black holes and at the quantum gravity level. This is all in the context of my growing conviction that the quantum theory of computation is quantum theory.”

According to Deutsch, one spinoff from the quantum theory of computation is that “it provides the clearest and simplest language, and mathematical formalism, for setting out quantum theory itself.”

— JB

DAVID DEUTSCH’S research in quantum physics has been influential and highly acclaimed. His papers on quantum computation laid the foundations for that field, breaking new ground in the theory of computation as well as physics, and have triggered an explosion of research efforts worldwide. His work has revealed the importance of quantum effects in the physics of time travel, and he is an authority on the theory of parallel universes.

Born in Haifa, Israel, David Deutsch was educated at Cambridge and Oxford universities. After several years at the University of Texas at Austin, he returned to Oxford, where he now lives and works. He is a member of the Centre for Quantum Computation at the Clarendon Laboratory, Oxford University. He is the author of The Fabric Of Reality.

David Deutsch’s Edge Bio Page

It’s a Much Bigger Thing Than It Looks

EDGE: In what direction are you asking the most questions at the moment?

DEUTSCH: The direction of even deeper connections between physics and the theory of computation. We’ve got the quantum theory of computation — which, by the way, is THE theory of computation. As I always say, we have to regard the Turing theory (the traditional theory of computation) as being just the classical approximation to the real, quantum theory of computation. We already know of a few issues in theoretical physics (like the Maxwell Demon question, and the relationship of thermodynamics with statistics) which it is useful to regard as computational questions — questions about how information can or cannot be processed. What I am aiming for now is a new kind of theory, quantum constructor theory, which is the theory of what can be built, or more generally, the theory of what can be done, physically.

We build computers and skyscrapers and space ships, and we clone animals, and so on. At root you can regard all of these too as computations, because when you build a space ship and fly it to a different place, you get new information, or rather a different perspective on the same information, which is just what happens when you input information into a computer and look at the output. However, flying in a spaceship is not quite the same, even computationally speaking, as putting a camera on the space ship and letting it go somewhere, and watching, because, for instance, there’s a time delay, so the machine gets harder to interact with if it’s far away. Experience is inherently interactive, so there’s a fundamental difference, imposed by the laws of physics, between the information processing you can do by going there vicariously using a robot and what you can do going there in person.

I’ve been thinking about those questions; that is, what sorts of computations do physical processes correspond to; which of these ‘computations’ can be arranged with what resources? And which sorts can’t be arranged at all? What little we know about this new subject consists of a few broad limitations such as the finiteness of the speed of light. The theory of computability and complexity theory give us more detail on the quantum side. But a big technological question in my field at the moment is, can useful quantum computers actually be built? The basic laws of physics seem to permit them. We can design them in theory. We know what physical operations they would have to perform. But there is still room for doubt about whether one can build them out of actual atoms and make them work in a useful way. Some people are still pessimistic about that, but either way, that debate is not really a scientific one at the moment, because there is no scientific theory about what can and can’t be built. Similar questions are raised by the whole range of nanotechnology that has been proposed in principle. So that’s where a quantum constructor theory is needed.

EDGE: Why specifically a quantum constructor theory?

DEUTSCH: Because quantum theory is our basic theory of the physical world. All construction is quantum construction.

EDGE: What is distinctive about a quantum computer, compared to the computers we know today?

DEUTSCH: Quantum computing is information processing that depends for its action on some inherently quantum property, especially superposition. Typically we would superpose a vast number of different computations — potentially more than there are atoms in the universe — and then bring them together by quantum interference to get a result. Other quantum computations, notably quantum cryptography, couldn’t be done by classical computers even in theory.

EDGE: What is the importance of this work?

DEUTSCH: I think that in future, quantum mechanics textbooks will use quantum computations as their introductory examples, rather than calculating the energy levels of the hydrogen atom and suchlike, which contain a high proportion of irrelevant stuff. Quantum computation gets down to basics, because quantum computation is the basics.

EDGE: But for you, the main application of the theory is to change our sense of the nature of reality?

DEUTSCH: Yes. However useful the theory as such is today and however spectacular the practical applications may be in the distant future, the really important thing is the philosophical implications — epistemological and metaphysical — and the implications for theoretical physics itself. One of the most important implications from my point of view is one that we get before we even build the first qubit [quantum bit]. The very structure of the theory already forces upon us a view of physical reality as a multiverse. Whether you call this the multiverse or ‘parallel universes’ or ‘parallel histories’, or ‘many histories’, or ‘many minds’ — there are now half a dozen or more variants of this idea — what the theory of quantum computation does is force us to revise our explanatory theories of the world, to recognize that it is a much bigger thing than it looks. I’m trying to say this in a way that is independent of ‘interpretation’: it’s a much bigger thing than it looks.

EDGE: What do you mean by ‘bigger’?

DEUTSCH: What I mean is — suppose we were to measure ‘amounts’ of reality, the sizes of things, in terms of the amount of information needed to describe them. To specify the positions of the atoms in this room, I need three numbers for each atom. The more atoms I want to describe, the more numbers I need. The more accurately I want to do it, the more decimal places I need to give. So that requires a certain amount of information. I can think of doing that for the whole universe. That may sound a lot of information, because there are 10^80-odd atoms in the known universe, not to mention the other degrees of freedom. So it may seem unimaginably vast. Yet it is minuscule compared to the amount of information that would be needed to specify the computational state of a single quantum computer, sitting on some future laboratory bench. So in terms of world view, or conceptual model, a quantum computer is a much bigger object than the whole of the classical universe. This fact forces quite a change in our world view.

EDGE: So the theory tells us that a quantum computer is in itself a universe.

DEUTSCH: It would be an object far more complex than the whole of the classical universe. The whole of physical reality is like that too, of course, and we sometimes call it the multiverse. We see, very roughly, a classical universe out there because most of the multiverse is not directly accessible. You can only infer the existence of hidden quantum information indirectly, as in the entanglement experiments I mentioned.

To many people this conclusion was already compelling even before quantum computers. The many-universes interpretation was proposed in 1957. But you can construe all the earlier arguments as being computational arguments too. The people making them didn’t think of them as such, but that’s what they were. They were saying: we look around us and we see something that’s approximately a classical universe, and we might expect that if you take quantum mechanics into account, that might add a certain amount of extra ‘stuff’ — like relativity did — which behaves differently but there’s still roughly the same ‘amount’ of reality as we thought there was. But that’s not what happens when you take quantum mechanics into account. Reality becomes a vastly, exponentially bigger and more complex thing than it was under classical physics.

EDGE: How can we tell that there’s so much of this ‘hidden information’ in a quantum system?

DEUTSCH: Apart from quantum cryptography, it’s unlikely to have practical applications in the near or medium-term future. It’s theoretical. But as such it does give us some immediate benefits. One is the benefit of looking backwards. Let me give you a recent example from my own work.

Quantum mechanics, in the traditional formulation, seems to have a ‘non local’ character: that is, things you do HERE instantaneously affect things that happen THERE. It has been known from the beginning that this ‘non locality’ can’t be used to send signals or anything. But still, philosophically, what are we to make of it? What sort of reality is quantum mechanics telling us we live in? And of course it’s hard not to wonder: “well, if something gets there instantaneously, it is going faster than light. So in another reference frame it’s travelling into the past. So it could create paradoxes; couldn’t that solve the problem of consciousness, explain telepathy, summon up ghosts…?” — you name it. This non-locality idea is one of the things that’s helped to fuel the appalling mysticism and double-talk that’s grown up around quantum mechanics over the decades.

But once you understand that this is all about information processing, it becomes much easier to stop hand-waving and start calculating where the information actually goes in quantum phenomena. That’s what Patrick Hayden and I did. The results (recently published in Proceedings of the Royal Society — click here) blow the ‘quantum non-locality’ misconception clean out of the water. Doing things HERE can only affect things THERE (visibly or invisibly) once the information about what you’ve done here has travelled there in some information-carrying physical object. Nothing instantaneous; nothing non local, nothing mystical.

EDGE: What about the famous experiments that demonstrate quantum non-locality in the lab?

DEUTSCH: They don’t. They demonstrate quantum entanglement: one of the fundamental quantum phenomena, but a local one. It turns out that when it looks as though there’s a non-local effect — as in Bell-inequality experiments — what’s really happening is that some of the information in quantum objects has become inaccessible to direct observation. And in our analysis we actually track how this information travels during entanglement phenomena. It never exceeds the speed of light, and it always interacts in a purely local way.

By the way, the presence of such not-directly-accessible information can be seen as the very thing that’s responsible for the power of quantum computers. The insights we gained from that work are leading in other very promising directions too.

EDGE: Such as?

DEUTSCH: Well, I am currently working on two spin-offs of that paper. One is work on the structure of the multiverse — making precise what we mean by such previously hand-waving terms as ‘parallel’, ‘universes’ and ‘consists of’. It turns out that the structure of the multiverse is largely determined by the flow of quantum information within it, and I am applying the techniques we used in that paper to analyse that information flow. The other is a generalization of the quantum theory of computation, to allow it to describe exotic types of information flow such as we expect to exist in black holes and at the quantum gravity level. This is all in the context of my growing conviction that the quantum theory of computation is quantum theory.

Speaking of that, another spinoff from the quantum theory of computation is that it provides the clearest and simplest language, and mathematical formalism, for setting out quantum theory itself. I’m planning a series of lectures on video which I think will be quite revolutionary. They will constitute a course in quantum theory for an audience that has no previous knowledge of it — say, university-entry level — all the way to leading-edge issues in quantum computation, in just twelve lectures (we’re currently looking for a sponsor for them, by the way!).

DEUTSCH: If the system is a quantum computer, we can tell because of the answers that it gives us. Take Grover’s quantum search algorithm, for instance. It works like this: Let’s say you’re writing a chess program; you’re searching through all the possible continuations from a given position. From one position there might be 20 possible moves, and from each of those there are 20 for the other player, and so on, so after N moves there are 20 to the power N different possible continuations. And you want to program the computer to search through all those continuations to evaluate a given move. Say you want it to search through a trillion continuations. It is a trivial theorem of classical computation, that if you want to search through a trillion unknown things, you generally have to do a trillion physical operations of some kind. You might be able to do some of them in parallel, but a given computer will only be able to do a fixed number at a time in parallel. One way or another you have to do a trillion things, so if you want to use the same computer to search through two trillion things it must take at least twice as long, and so on.

But with a quantum computer, you could do better: First of all, to search through a list of a trillion things you need only do a million operations. In general, in order to search through N possibilities one need only do the square root of N physical operations. And then, if you let your quantum chess machine think for twice as long, it will examine four times as many continuations. Three times as long, nine times as many, and so on. The explanation of this, in terms of many universes, is very simple. It’s just that there are the square root of N universes collaborating on such a task. But again, never mind the question of interpretation as such. If we just think of what this computation implies for the reality we find ourselves in, again, the answer is that reality is much bigger than it looks. The winning move, when we find it, logically depends on all the positions we searched. So as a matter of logic, those positions must all have existed somewhere, and been compared with the answer we got.

EDGE: There seems to be a gap here, between abstract information on the one hand, and physical objects such as computers and stars and universes on the other. What’s the connection?

DEUTSCH: Ultimately, information has got to have a physical realization; that’s why it does come down to atoms, or stars, or whatever, in the end. But because of the universality of computation you don’t have to think in terms of specific implementations. I don’t have to know whether my information is going to be stored in magnetic disc, or whatever. I just know that more information means a bigger object.

EDGE: Where is work on quantum computation being done?

DEUTSCH: More and more places every day, it seems. In the US alone there are about a dozen very high quality research groups working flat out on quantum computation, theoretical and experimental. Probably another dozen in Europe. Also Japan, Australia, Israel…

EDGE: Let’s talk about practical things. You’re at a Microsoft, an Intel, a Sun Microsystems, and you read about David Deutsch and his theories about quantum computing. How will it impact your business? What measures would you take?

DEUTSCH: If computers are going to continue to become more powerful, processors and memory devices must become smaller. For that reason alone, quantum processes must be harnessed. Whether to make quantum computers or not doesn’t really matter. Even to make classical computers out of atomic-scale components you’d have to use quantum physics and ultimately the quantum theory of computation. And once you’re making those, the same technology could probably also make quantum computers. And the incentive would be there because of the various inherent advantages of quantum computation.

EDGE: How would you build one?

DEUTSCH: Proposed technologies for building them are at present competing. We don’t know which way it’s going to go. It could be ion traps or it could be quantum dots, or other solid state devices, or it could be superconducting loops. It could be molecules, or something we don’t know about yet.

At present the biggest quantum computer in the world has about 3 qubits. Not much practical use, and it requires quite a large apparatus to make it work. Yet with three qubits you can already implement quantum algorithms that no classical computer using three bits could mimic.

Quantum cryptographic devices already exist in the laboratory. Eventually that’s going to give perfectly secure communication. No longer will cryptography depend on the difficulty, or the intractability, of guessing an unknown key. It will simply be physically impossible to discover the key if you don’t have the relevant physical object. So that is the ultimate in cryptography.

EDGE: We know that historically, advances in cryptography have been suppressed by governments. Could it be that quantum computers will never come on the market because people will make sure they don’t?

DEUTSCH: If so, I know nothing about it. Both in Britain and America there are government agencies working on quantum cryptography, and as far as I can tell, they participate in much the same way as they would if they were academic institutions. Presumably they have their secrets — I hope they do! — but I’m not aware of them having tried to prevent any of these technologies from being developed, let alone theoretical advances. But I do find it a bit surprising, now that you come to mention it, that there isn’t already a quantum cryptographic device on the market.

EDGE: For e-business?

DEUTSCH: No. The trouble is that at the moment quantum cryptography is severely limited in range. It can’t be done through open air. It’s got to be done through fiber-optic cable, and I think the world record is about 100 kilometers. But still, you could wire up the City of London, or central Washington DC, with absolutely secure communications. I don’t know why that hasn’t been done. I doubt that it has anything to do with sinister machinations by the government, though. It’s probably just that it takes a long time for an idea to become genuinely commercially viable.

EDGE: What if there was a critical situation such as a war which required security?

DEUTSCH: In that case we already know how to build absolutely secure communications if we want to, at ranges of a few kilometers. Longer ranges would present a problem, but at least one group at Los Alamos is working on a system that would allow you to bounce quantum-encrypted messages off a satellite, and that would essentially solve the problem.

In the long run the problem could also be solved by quantum repeating stations. Unfortunately they would require much more sophisticated quantum computation than the raw cryptography does. They will come along eventually, perhaps in a decade or two.

Another thing that will come along — probably after more than a decade or two — is quantum cryptanalysis, where you would use a quantum computer to decrypt existing codes. Quantum decryption machines would render existing cryptographic systems obsolete.

EDGE: Ten years from now will I have any quantum-computer technology on my desk?

DEUTSCH: I guess not. But who knows? I’ve been surprised repeatedly by how well the experimentalists have been able to implement theoretical concepts in quantum computing. But apart from quantum cryptography I’d be amazed if anything technologically useful comes out in ten years, 20 years, even longer. But I’ve been amazed before.

1995-02-26 Channel 4Reality on the Rocks - Part 3 - Beyond Our Ken

00:50:09 - 00:50:34

He said, the human race is just a chemical scum on a moderate sized planet, orbiting around a very average star. In the outer suburb of one among a hundred billion galaxies. We are so insignificant that I can’t believe the whole universe exists for our benefit. That would be like saying that you would disappear if I closed my eyes.

Ken Campbell

00:50:34 - 00:50:36

Close my eyes.

1995-01-01 NoorderlichtThe Multiverse Documentary

00:21:57 - 00:24:31

The fact that we can’t directly detect parallel universes is to some degree an accident of the way we’re built. If you wanted to detect the motion of the Earth with your human senses, you wouldn’t be able to do so either. The way we detect the motion of the Earth is, for instance, by taking a long pendulum, Foucault’s pendulum, and letting it hang for a while. And in the course of a few hours, we see the plane of its rotation changing. And this is because the plane actually stays the same, and the Earth is rotating underneath it. So this is an instrument which detects the rotation of the Earth. If, by chance, our senses had contained an instrument equivalent to this pendulum, we would be able to feel the rotation of the Earth. In a similar way, if our senses happened to work through quantum interference, then we would be able to feel directly the influence of parallel universes. So saying that you don’t feel other universes is like the Inquisition saying to Galileo that they don’t feel the Earth moving beneath their feet. Well, the reason why they don’t feel the Earth moving beneath their feet is because Galileo’s theory explains why they won’t feel it moving. But if they do a different experiment with a Foucault’s pendulum or with telescopes, they will indirectly be able to tell that the Earth is moving. The core of every scientific advance is a better explanation. Now, quantum mechanics and the many-universes interpretation are our best explanation of the physical world at the fundamental level. In fact, I would say they are the only current explanation of the physical world at the fundamental level. There are no real rival explanations. We don’t understand it fully yet, and I don’t think one ever understands a theory fully until one has its successor. One day there will be a successor to quantum theory, just as quantum theory itself is the successor to classical theories of physics. And then we will understand the full implications of quantum theory and where it went wrong. And I would expect, again, on the basis of what usually happens in science, we would expect that future theory to be even stranger than quantum theory.

Noorderlicht narrator/interviewer

00:24:56 - 00:25:01

[end credits]

1992-07-06 Antenna BBCTime Travel

00:26:51 - 00:28:40

At the moment, all routes to time travel are far beyond our technological capabilities, but more physicists than ever are doing research in this area, and all the traditional objections to time travel have turned out to be invalid. My gut feeling is that if technological obstacles are the only ones in our way, then one day time machines will be built. It will be a pity if it turns out that time travel is not possible. It would have been exciting to explore the universe in time as well as space. Well, the subject of time travel is certainly very, very speculative, as indeed is the whole issue of how you combine quantum theory and general relativity. But in these sorts of topics, however far they may appear to be from ordinary, daily, common sense, you really are trying to address the fundamental ingredients of reality at the really sort of basic physical level. In that sense, I think it is a very important program and well worth pursuing, even though it is, I say, very speculative. My personal belief is that we know nothing to definitely rule out the possibility of time travel. We’ve been doing intensive research on this subject for many years, and we’ve studied carefully laws of physics. We see no reason for thinking that time travel is forbidden. Almost all of the discussion about time travel is beyond the present realm of experimental testing. But I have faith in the theory, and time travel is allowed by the theory. Maybe I’m an optimist, but there’s an old saying in physics, whatever is not forbidden is compulsory.

BBC continuity announcer

00:29:21 - 00:29:28

Next week in Antenna, Professor Bob Williamson explores what has been described as the next revolution in medicine.