British physicist David Deutsch joins me on the show to talk about his life, his work, and his theories on quantum physics and quantum computing. We talk about Richard Feynman, quantum physics, quantum computing, and much more.
00:01:36.500Oh, you're very kind. Let me just read for the people who may not know you a few. You sent me a
00:01:42.820bio that would have taken me a few hours to read, so I distilled it to a few things. You're currently an
00:01:48.740honorary fellow of Wolfson College at Oxford University. You were the first to publish a proof
00:01:54.180of the universality of a universal quantum computer. We'll try to see if we can understand
00:01:59.480what that means. Published the first quantum algorithm proving to be exponentially faster than
00:02:05.340any Turing machine algorithm. Look what I've got here, David. The biography of Alan Turing. And then
00:02:12.380a pioneer in the field of quantum computation. Received the Breakthrough Prize in Physics,
00:02:19.220the Institute of Physics, Isaac Newton Medal and Prize, the Messias Quantum Prize, the Dirac,
00:02:27.140Paul Dirac. Here's his biography. The Paul Dirac Medal and Prize. Holy moly, that's a lot of awards.
00:02:33.700You were a fellow of the Royal Society in 2008. And I'll just mention your two books meant to
00:02:39.220be consumed by the general public, the beginning of infinity, and the fabric of reality. Anything
00:02:46.040else you want to add before we dive deep? No, as you said, that's probably already too much.
00:02:52.120Okay, so I thought the first thing we would do, and by the way, as I was reading your bio, I mean,
00:02:56.880I saw in one of your books, you talk about Popper, you talk about Turing, you talk about Dawkins.
00:03:01.380Those are all guys that I absolutely love. So it's not going to be very hard for us to have a really
00:03:05.980fun conversation. Question one, which I'm very afraid to ask, because Richard Feynman already told
00:03:12.540us that, I can't remember the exact quote, but if you think you understand quantum physics,
00:03:17.640nobody understands quantum physics, or something to that effect. Can you tell us what is quantum
00:03:22.400physics, and then how does that translate into quantum computing?
00:03:26.800Yes. I'm not sure when Feynman said that. It may have been closer to being true when he said it,
00:03:41.860because quantum theory, which is the most fundamental theory we know in physics, took a few wrong turns
00:03:54.300during the early 20th century, and it was in a complete mess philosophically. I mean, it was
00:04:01.180like compulsory to talk nonsense about it, and it's possible that Feynman was referring to that
00:04:08.460when he said that no one understands it. No one could understand it, because it didn't make sense.
00:04:14.740On the one occasion that I was honored to meet Feynman, we spoke, now I was very junior, but I just
00:04:27.140started my work on quantum computers, and he seemed perfectly reasonable. He didn't say anything like
00:04:35.580that. I asked him what he thought of the prevailing so-called interpretation of quantum theory,
00:04:44.740which was nonsense. And he said, oh, Johnny never intended that to be permanent. I thought,
00:04:52.560Johnny? Wait, John von Neumann? Yes. He never intended that to be permanent. It was just a stopgap
00:05:05.880measure to allow them to make computations and make predictions. But it was never meant to be
00:05:13.540actually true of the world. But at that time, by the time I met him, the right way of looking at it
00:05:23.920had already existed for 30 years. It was invented by a physics graduate student of Feynman's mentor,
00:05:35.880John Wheeler, who was also my mentor. Wow. And it is called the Many Worlds, Many Universes Interpretation.
00:05:46.260Though it's not an interpretation. It's simply saying, right, let's go back to the beginning.
00:05:51.440Forget all this nonsense about the observer and one can't distinguish between things that are true
00:06:01.620and things that only look as though they're true and so on. And let's go back to first principles.
00:06:07.060Let's just apply the way of using physical theories that we're all taught when we're undergraduates.
00:06:15.340And he looked at the equations and the equations clearly say that there are many universes and
00:06:23.720usually they're independent of each other. So when you're in one of them, it just looks like the others
00:06:29.240aren't there. But occasionally they affect each other. And that's called quantum interference.
00:06:36.580And that's the most important physical phenomenon in quantum theory. So that's what happens when you
00:06:48.260take quantum theory seriously as a description of the world.
00:06:51.700Okay, let me so much to unpack there. So if when you say multiple universes, so and you'll correct me if I'm
00:06:58.600getting some of the numbers wrong right now, we we have a sense that the universe, the outer edges of
00:07:04.380the universe is about 16 billion light years away. Does that sound right? Sounds right.
00:07:09.580So that's the universe as we know it. Okay, whatever. What does it then mean to have multiple universes?
00:07:18.220There is a pathway by which I go from the knowable universe that has a current outer limit of 16
00:07:25.920billion light years away to where what what happens? Where do I go?
00:07:30.260Well, the literal answer to where is right here. But the laws of physics are so I'm already lost.
00:07:36.900I already don't follow what you're saying. But go ahead.
00:07:41.200Suppose, you know, it's like the Twilight Zone or something in Superman comics or something like that.
00:07:46.560It's another world that coexists with ours, roughly speaking, in the same space.
00:07:54.660And because it doesn't interact with us in any way, it's like dark matter, you know, or neutrinos.
00:08:02.740We're deluged with trillions of neutrinos per second.
00:08:07.440But we don't feel them because they don't interact with our kind of matter.
00:08:11.980And it's the same with the other universes of quantum theory, except that unlike neutrinos,
00:08:18.660they do interact strongly with our universe under certain circumstances.
00:08:25.360And these circumstances can be set up in the laboratory.
00:08:29.980And then they give rise to astounding effects.
00:08:33.700So there is an earthly manifestation of that theory that is not just some axiomatic, elegant thing that happens in Booga land in mathematics.
00:08:46.260There is a way for me to interact with that reality to say, there is.
00:08:54.100Yes. And and the the experiments in which this is demonstrated are just astounding.
00:09:03.980I mean, if you think about what's happening there, you just have to go slack jawed and say, you know, how can the world be like this?
00:09:12.420But it is. It is. And the I did a series of lectures once, only got halfway through.
00:09:24.940But but the second one that I did had this experiment.
00:09:29.920I actually went to the to the quantum optics lab in the Clarina laboratory and and and I found a colleague who was willing to set this up, set this experiment up for me, because for them, it's it's like everyday stuff.
00:09:46.740We don't set that up. I mean, we know what will happen.
00:09:49.140But, you know, ordinary people and also theorists like me, who whenever I go into a lab, it's stopped.
00:09:57.340Everything stops working. But but I I I knew what was going to happen.
00:10:03.140But seeing it in person was just astounding.
00:10:08.040It it I mean, to cut a long story short, viewers can go and find it on the Internet.
00:10:14.740And and many other people have put it on the Internet since then as well.
00:10:22.640It's it's it's the experiment is done with a single photon.
00:10:26.420How do you how do you make a single photon?
00:10:28.720Well, you start with a laser with trillions of photons and then you put a dark filter and another dark filter, another dark filter and so on until you can't see anything coming out, of course.
00:10:40.560Yes. But, you know, that there's there's one photon coming out every every I don't know what it is, every millisecond.
00:10:51.340There's one which means that the space between photons is like thousands of miles.
00:10:56.900Each one is like thousands of miles behind the other.
00:10:59.460So you're really only doing an experiment on one photon and then you pass it through through a screen which has many holes.
00:11:08.260And what's coming out of the other side is it it lands on a photo photo multiplier, which is an extremely sensitive detector that can detect individual photons.
00:11:21.500And it it it comes out in a place that is determined by how many holes there are.
00:11:29.320And that's that's the basic which shows that really the the we did this experiment in many universes and the the photon in other universes interacts with our one and they they affect each other.
00:11:43.840And they end up in a place that is not illuminated when when when which way round is it when only one of the slits is open.
00:12:01.300It's it's never illuminated regardless of which one you open.
00:12:04.940Then if you open all of them, it lands in a place that isn't illuminated when just one of them is so that can't happen yet.
00:12:27.500And then jump through a gargantuan bridge to the cosmological level.
00:12:34.080How do we go from that insight at, you know, and before you answer, David, this is maybe an opportunity to bring in Dawkins, which you talk about in one of your books, because he talks about middle world, right?
00:12:52.260I mean, literally in an evolutionary sense, the mind to intuitively understand things at the nano level or at the cosmological level.
00:13:00.540And so that's why for most people, folk physics might make sense, but quantum physics, which defies the way most of us have evolved to understand how you throw a ball or a spear, it's going to break down.
00:14:56.040I mean, of course, all scientific theories are temporary.
00:14:59.980So we will no doubt have better theories in the future.
00:15:03.900But there's no problem with understanding the world insofar as it goes via quantum theory or via general relativity or both.
00:15:13.120The way we know, for example, is that the cosmic microwave background radiation was released into somewhere that we can see when the universe was very small compared with how it is now and very small and very dense.
00:15:36.860And the physical processes affecting it were subject to another quantum phenomenon, quantum fluctuations, which is a bowdlerized or sanitized way of saying interference from other universes.
00:16:00.180But we call it fluctuations as if, you know, it's going fuzzing like boiling water.
00:16:09.520It's just different in different universes and they affect each other.
00:16:13.320And we can work out from that how that microwave background radiation is going to look 13 point something billion years ago, billion years later, when we look up in the sky and we see the microwave background radiation is patchy.
00:16:48.460We see it about 40 percent different from how it actually looks from what we predict, which is a big problem in cosmology at the moment.
00:16:57.680But, you know, that doesn't affect the fact that the smallest level we know about affects the largest level we know about in a testable way.
00:17:13.600And usually the epistemology of what you're talking about is you develop something mathematically, axiomatically, and then you say, if this theory is right.
00:17:29.600If the data over there that I'm looking up looks in pattern A, then the theory must be wrong, hence falsified.
00:17:38.520But if it looks like B, that is unassailable proof that it is correct, right?
00:17:43.880Because the reason I'm asking this question, because I deal with human beings where, I mean, yes, the scientific method is the same in a grand epistemological sense.
00:17:52.820But I'm using a different type of paradigm and testing whether testosterone affects men's behavior when they ride a Porsche.
00:18:02.860So did I get that roughly right in terms of I look at the world, it's either A or B, and then that tells me whether my mathematical model was right?
00:18:12.400This gets right the logic of testing scientific theories, but it doesn't get right the logic and epistemology of the whole of science.
00:18:20.620So this is where you and I probably disagree.
00:18:24.160So before all this testing, before this mathematical writing down of mathematical equations, before that happens, we have a problem.
00:18:35.480This is, a colleague of mine says that this concept of a problem is actually Popper's greatest contribution to philosophy.
00:18:47.300The idea that everything begins with a problem, not with observation, not with conjecture, not with a theory.
00:18:59.320So we have a problem like, how the hell do you explain what this photon does that I described to you earlier?
00:19:06.420Basically, it's inconceivable, you know, it seems to, the photon seems to be everywhere at once, or it seems to be in one place at the same time.
00:19:24.960And they made some conjectures in the early 20th century, then they wrote down the equations, then they tested them.
00:19:33.480And there's this famous experiment by Stern and Gerlach, who some of the physics folklore, I don't know whether this is true, but some of the physics folklore says that Stern and Gerlach were trying to disprove the theory of the pie in the sky theorists.
00:19:58.620That this interference phenomenon would take place with material particles, as well as with light.
00:20:07.300You know, the experiment that I witnessed was done with light, doing it with material particles is much more difficult and so on.
00:21:38.500But a virus is a manifestation of material, isn't it?
00:21:42.840So why is that a third instantiation of the ministry?
00:21:46.560Well, it isn't, you know, according to us theorists, it isn't different.
00:21:53.620But some people were saying, and still do to this day, that the quantum, it's only the microscopic world that sort of constantly divides itself into many universes, but then it collapses back into one.
00:22:08.800And so they would have to predict that when enough atoms are involved in an interference phenomenon, it'll go away.
00:22:19.140The interference phenomenon will go away.
00:22:21.680And so when you aim viruses or something at this grid of holes, you will get just a shadow of the grid, nothing special.
00:22:30.800They'll just land wherever there's a gap and not when there isn't a gap.
00:22:36.820And so that's why experimentalists, you know, they want to have something to do.
00:22:41.480So they are trying to do this with more and more objects.
00:22:45.520On the other hand, there's another way of doing it with more and more objects, which I think is more impressive, and that is with a quantum computer.
00:22:54.420So explain to us before, what is a quantum computer?
00:23:08.220What does it mean to engage in quantum computing?
00:23:11.060So a quantum computer is a computer whose operation relies on distinctively quantum phenomena, especially interference, and the same thing I've been talking about all this time.
00:23:27.240So if you imagine a photon, for example, going through this grid of holes and splitting up into, sorry, not splitting up, differentiating so that in different universes it goes through different holes.
00:23:43.420Now, suppose you put something in front of each of those holes that performs a computation.
00:23:47.460Well, you can't perform much of a computation with one photon.
00:23:52.480You can perform perhaps the knot operation, you know, flip it from polarization down to polarization up.
00:23:59.740But still, you can have them all doing different ones, and they come together, and you can make the result depend on all of them.
00:24:12.620So you have a single photon doing the result of n different computations.
00:24:21.440If you were to send through two photons, which you can't, it's not feasible, so you have to do it differently with electrons in a crystal, but it can be done.
00:24:36.560Now, if you had two of them that were also interacting with each other in every individual universe, but were also interacting with their counterparts in other universes, then you could do a whole load of computations that depend on, that act on two bits.
00:24:59.100So let me stop you right there, David.
00:25:01.560So I'm going to harness right now my undergraduate degree in mathematics and computer science and hopefully not make a fool of myself.
00:25:08.900So I remember I took a course in analysis of algorithms where you are basically calculating.
00:25:17.060I hope if any of my former professors are watching this, this is 40 years later, and I would probably still get an A-plus in the exam.
00:25:24.020So in analysis of algorithms, you're saying, okay, this can be done in order, whatever, n log n, whatever.
00:25:32.880What quantum computing does effectively is it takes that computational speed and increases it.
00:25:42.720That's the net final conclusion of what we're talking about.
00:26:30.780So then I would imagine that for things that require great computational burden, searches, that's where you're most likely to see this application?
00:26:43.880A general search is not speedable up by much.
00:26:48.980The, well, it depends what you count as much.
00:26:53.780It can, the speed goes like the square root of n.
00:26:59.200If the classical computer would take n steps because you're looking at n different possibilities, let's say you're searching something, then the quantum computer can do it in the square root of n, which is not, you know, the complexity theorists don't regard that as very impressive.
00:27:18.440But in real life, you know, converting a trillion into a million is a million-fold improvement.
00:27:41.900All they've built so far is quantum computers, special purpose quantum computers, like the code-breaking computers that they had in World War II.
00:28:28.440In 2008, I blogged that a particular method that had been invented in 2008 and which would greatly impress me will deliver universal quantum computer in 10 years.
00:28:43.060In 2018, it hadn't done anything like that.
00:28:57.760Okay, I want to talk a bit about Turing.
00:29:00.740Then I want to come back to, so in looking at your work, and maybe many physicists have this kind of synthetic thing where they want to put everything under a universal mechanism.
00:29:12.720I would call it, and I'm not sure if, tell me if you're familiar with the term.
00:29:16.020Are you familiar with the term consilience, David, of E.O. Wilson?
00:29:26.560Oh, so we'll come back to that because that, I think, is going to be a wonderful overlap between our mutual interests.
00:29:32.680But let me mention my own interaction with Turing's work.
00:29:38.040So I've often said that of all my studies, whether it be as an actual student in university and subsequently as a professor for 30 years, nothing blew my mind in a true mystical sense as much as when I took the course.
00:29:55.320Well, the course was called Formal Languages.
00:29:58.080I even remember, and I didn't even check this before our meeting.
00:30:10.160Okay, so it's a book on formal languages where it introduces all the Turing stuff, and I've gone back to it maybe a couple of years ago.
00:30:17.660It's still in my university office where I'm looking at the margins where I've written stuff.
00:30:24.600Half of the symbols I no longer recognize, but what I do take away from it, that it seemed as though it was a level of intellectual insight that is almost difficult to imagine, right?
00:30:39.100I mean, I've interacted with a lot of brilliant people, but when you kind of interact with the Turing stuff, it's almost as if he existed in another plane.
00:30:47.060So, number one, as someone who is a theoretical physicist, so you may even have more access to such folks, is my evaluation that Turing was a unique beast?
00:31:01.900That's why I dedicated, he was one of the people I dedicated my first book to.
00:31:07.900So, yes, but it's interesting that you should talk about complexity theory as an example of this, because he was, like, at the very beginning, he killed himself.
00:32:41.960And, and, and, uh, by the way, it seems that Ada Lovelace almost had it.
00:32:50.640Uh, I, I always thought that people who, uh, you know, Babbage, Babbage was first with a universal computer design for a universal computer.
00:33:01.320He, he, he, he, for all sorts of reasons, he messed it up.
00:33:05.360And, and I always thought that, um, Ada was like, um, his, his, um, upper class muse.
00:33:15.200Um, and, and I thought that when people called her his collaborator and, and the first programmer and so on,
00:33:21.980they were just doing the usual thing of saying that, that we're not giving enough credit to the woman in the story.
00:33:31.060Um, and, and so I didn't take it seriously, but then, uh, I, I saw a documentary, which, which presented, um, cast iron evidence that not only did she get universality, but, but Babbage didn't quite get it.
00:33:48.560He, he, he only realized that the, the, um, uh, analytical engine that he had designed would be able to do all mathematics, but, uh, he didn't think of the idea of, of other uses.
00:34:05.060It's such as simulating a physical object or composing music or, uh, uh, analyzing images or whatever, but Lovelace did.
00:34:16.220And, and, um, she, she, she wrote this down and, um, then it was forgotten.
00:34:23.060You know, nobody thought of this for like almost a hundred years and Turing in his.
00:34:31.680Epoch making paper on computational universality, he refers to Lovelace and I, and yes.
00:34:40.520So the one thing that she, that she couldn't get her mind round universality extending to was thinking.
00:34:48.140So she wrote, um, after listing these things that, that it might be able to do, then, then she wrote, but one thing it can't do, it can't initiate an idea.
00:35:02.120It can only obey orders, not initiate an idea.
00:35:05.700Remember this is before, before anything like a computer had been made.
00:36:25.780Well, my, our, we have some good friends who are actually neighbors of ours.
00:36:29.940The husband is a, uh, cyber security, uh, expert and they just, about a year ago, they had, uh, their first child, a daughter, and they named her Ada in, in, in honor of, uh, Lovelace.
00:36:46.980So I guess the other person that I could think of that reaches that level of, let's call it kind of mystical intellectual, uh, you know, abilities would be maybe, uh, girdle with his incompleteness theorem.
00:37:02.500That to me seems like in the same stratospheric, uh, range.
00:37:16.700They, they didn't think of it like that at the time, you know, as, as often happens, uh, one, one only sees things in perspective and the originators of an idea usually don't see the full ramifications.
00:38:46.300Now, is there, do we know why one would have taken the, what we can do versus we can't do?
00:38:54.020Is there something in their temperament, their personality, their background that would have caused them to attack fundamentally, roughly the same thing, but one in a positive frame, one in a negative one?
00:39:12.260But Turing, and all mathematicians are going to disagree with me on this, because they have what I call the mathematician's misconception.
00:39:24.620But they, Turing was physically focused.
00:39:30.300He, when, when, when he was trying to answer Hilbert's question, like, he was trying to answer the question, well, what could mathematicians do?
00:39:42.940That, that was, that was, that was Turing's, and I think his whole formalism that he set up is, is, has the form, what can physical objects do?
00:46:49.520Turing thought that there'd be an AGI by the year 2000, and that it would require two megabytes of memory.
00:46:55.880Now, he's obviously wrong about the year 2000, but two megabytes of memory, I reckon that's what it'll be.
00:47:04.940In other words, these large language models and all this massive computer power is going in entirely the wrong direction.
00:47:14.160The answer will be a philosophical breakthrough, which will allow, once we understand what we're trying to make,
00:47:24.140it will be relatively easy to make it with relatively few computational resources.
00:47:29.780But, okay, I've gone off a tangent, so I've forgotten what you actually asked.
00:47:34.260No, no, but I was just trying to link E.O. Wilson's concept of consilience to your and many other physicists' desire for synthesis,
00:47:45.280for universality through different mechanisms.
00:47:47.780And in an epistemological sense, it's the same general project.
00:47:52.340You're heretofore taking things that appear to be fragmented and orphaned, separate, and trying to find a way to link them.
00:48:00.480Yes. Well, yes, I'm not sure that these connections between things were done by trying to find a way to link them.
00:48:12.660For example, quantum theory and computation, which I did, I wasn't looking for a way to unify them.
00:48:23.580The way to unify them emerged as a consequence of the work I did.
00:48:31.640The first work I did, which had a quantum computer in it, I didn't know that it was a quantum computer.
00:48:41.660I wasn't looking for a quantum computer, let alone a universal one.
00:48:45.000I needed it for an experiment to test the many universes theory in quantum theory.
00:48:54.380And when I say I needed it, I needed the theory of it.
00:48:59.140Actually doing the experiment is not really relevant.
00:49:02.800And then it was several years later that I began to think that there's a new mode of computation enabled by this machine that I had designed for a completely different purpose.
00:49:25.300And then Popper, he called his epistemology, he eventually called it evolutionary epistemology because of its relationship with the theory of evolution.
00:49:39.580But he didn't call it that when he first invented it.
00:49:59.920Well, universal Darwinism, which is a term that I'm sure you're familiar with.
00:50:05.180I mean, what excites me about that term is that, well, first, so you're right that there are elements that are bottom up.
00:50:14.820In your case, it was organic that you saw the link between the quantum work you were doing and then the application and quantum computing.
00:50:22.200But there is, I think, also a top down mechanism whereby some thinkers and scientists are inherently or they score higher on the quest for synthetic thinking.
00:50:40.640So my doctoral dissertation was in psychology of decision making.
00:50:44.600I was looking at the specific problem I was studying, David, was when is it that people have collected enough information to stop and make a decision?
00:50:56.160So it's a search problem, but it's a stopping decision.
00:51:10.940So it's a iterative sequential sampling mechanism.
00:51:13.860What excited me about that process, which I could then apply in marketing or political behavior or mate choice, is precisely that, that it could cause me or allow me to traverse into an, I don't want to say an infinite number, but a boundless number of intellectual ecosystems.
00:51:33.160The same itch was able to be scratched when I became an evolutionist, which is I could now apply evolutionary thinking to study certain psychiatric disorders or study gift giving behavior that are very, very different.
00:51:48.360And the capacity to have this key that allows me to unlock all sorts of problems is what allowed me, in my view, to have an exciting career that that can then permit me to have, I hope, an intelligent conversation with David Deutsch.
00:52:04.060But tomorrow I could speak with, and that polymath ability is very important to me, but it's exactly what is not rewarded in academia where you have to be a hyper-specialist.
00:52:16.660So maybe that's the next segue of our conversation.
00:52:19.180What are your thoughts about this perennial battle between what our graduate students are expected to be, which is hyper-specialist versus the truly big thinkers, the John von Neumanns, are defined by the fact that they can make contributions across a bewildering number of fields?
00:52:37.420Yeah, so I'm not sure which aspect of the appalling mess that is the academic world at the moment, where I should start.
00:52:55.100Maybe you won't agree with my starting point.
00:52:58.240I think that universities should not regard themselves primarily as teaching institutions.
00:53:05.180I think what's needed in the world, what's needed for human progress in the large scale is creative problem-solving, research.
00:53:18.800And it's very beneficial for research to gather the researchers into one place, and that could be a university, it could be like the Princeton Institute.
00:53:30.580I was going to say, that's the Princeton Institute, yeah.
00:53:33.560It's also the Oxford and Cambridge model, where you go into lunch in the university cafeteria, and you're sitting opposite somebody who is an authority on something that's unrelated, apparently unrelated.
00:53:47.960And you have a conversation with somebody who is very knowledgeable about something not in your field.
00:53:53.980So that is, I think, the primary function of academic institutions.
00:54:01.400But that cannot survive unless there is also scholarship.
00:54:08.740That is, if there is a built-up tradition of people who learn the stuff and who are experts on it,
00:54:16.300so that when the researcher finds something that they want to understand, they will go, like ideally, they will go and find in the university,
00:54:28.420or nowadays on the internet, of course, but it's the same idea,
00:54:32.940somebody who has spent their life knowing about this stuff and knowing that, for example,
00:54:43.620what most people think is true of this field is not true, and so on.
00:54:49.560And so then, that's the second thing you need.
00:54:52.940The third thing you need, if you're still with me, is that in order to have researchers and scholars,
00:55:03.760you need to have a tradition where young people are inducted into this system of scholarship and research.
00:55:11.500And they have to be, I hesitate to use the word taught,
00:55:19.840because it's really joining a culture rather than learning facts or techniques.
00:55:29.300They need contact with the existing culture so that they can become scholars and researchers themselves.