The Origins Podcast with Lawrence Krauss - John Preskill: From the Early Universe to the Future of Quantum Computing

Episode Date: February 17, 2023

John Preskill is the Richard P. Feynman Professor of Physics at Caltech, a title many physicists would cherish. He is widely known in the field for his work as a theoretical physicist spearheading the field of Quantum Computing, where he is Director of Caltech’s Institute for Quantum Information and Matter, but his expertise and contributions span a far broader spectrum of topics. His background is in theoretical particle physics, gravitation, and cosmology. As a graduate student, his seminal work on the cosmological implications of magnetic monopoles in Grand Unified Theories helped lead Alan Guth to develop his theory of Inflationary Cosmology, in part to resolve a cosmological conundrum John first elucidated. Since that time, John has explored condensed matter systems and the physics of black holes, made a famous bet with Stephen Hawking, and coined the term “quantum supremacy”, to describe a metric that might reveal the first time a quantum computer resolved a problem that a classical computer could not resolve in a feasible human timescale. As Director the Caltech Institute, John leads one of the most vibrant programs exploring quantum information and quantum computation, and I was happy to have the opportunity to connect again with my old friend and colleague to discuss this rapidly evolving field, about which so much is written in the popular press, and which may impact on all of our lives in the 21st century. In our discussion we tried to separate the wheat from the chaff, to discuss the future of the field, its current state, and challenges and opportunities. In addition, we discussed his own scientific career and the physics areas that have excited him, and what helped drive him to become a physicist in the first place. It was a fascinating discussion and I am sure you will be both entertained, and enlightened. As always, an ad-free video version of this podcast is also available to paid Critical Mass subscribers. Your subscriptions support the non-profit Origins Project Foundation, which produces the podcast. The audio version is available free on the Critical Mass site and on all podcast sites, and the video version will also be available on the Origins Project Youtube channel as well. Get full access to Critical Mass at lawrencekrauss.substack.com/subscribe

Hi, I'm Lawrence Krauss and welcome to the Origins podcast. John Preskell is the Richard P. Feynman Professor of Physics at Caltech and also director of its Institute for Quantum Information and Matter, where he directs a program in quantum computing and quantum information, and that's what he's perhaps most well known for now, an incredibly exciting area which we spent a lot of time in this podcast talking about as I'll get to. But John's background is actually in fundamental particle physics, in Cosmology, which is what he was working on.

When I first got to know him, we were together in Boston when he was a graduate student at Harvard and I was a graduate student at MIT. And then we were together at Harvard for several years when he was ultimately an assistant professor at Harvard. And I have to say, we've written one paper together. But beyond that, I've learned more detailed physics from John than perhaps any other collaborator because what he does when he's teaching or working is produces the most amazing set of notes, lecture notes and otherwise, and I've actually found his notes useful in

my own teaching as well. John actually as a graduate student did some work in cosmology, having to do with things called magnetic monopoles that basically changed the future of cosmology because his work on monopoles in the early universe motivated in some sense Alan Gooth to think about a problem that he eventually solved with his theory of inflation. So inflation was partly motivated by resolving a profound problem in cosmology that John had demonstrated in his early work as a graduate student. And John's worked in a variety of areas of fundamental physics, but eventually moved to the area of quantum information and quantum computing, where he has helped establish and lead one of the leading centers in that area. And I wanted to talk to him

about his experience as a scientist early on, but ultimately to focus on quantum computing, a subject you hear about a lot in the press and there's lots of hype and it's hard to separate the hype from reality and I thought it'd be best to go to the horse's mouth and John and I talk about the future of quantum computing its present state the challenges and the opportunities and it's a it's a I believe a fascinating introduction into that field with a very clear thinker and speaker careful to to be accurate at all times it was a real joy to spend some time with John again during this episode I think you'll really enjoy it and I hope you can watch it

add free on our Critical Mass substack site by subscribing to Critical Mass and those subscriptions, those paid subscriptions support the Origins Project Foundation that produces the Origins Podcast and also our public programming and travel adventures and other opportunities to try and connect

science and culture and the most interesting ideas in the 21st century and bring them to the public. You can of course watch it for free on YouTube or listen to it on any standard podcast site. So however you watch it or listen to it, I hope you'll really enjoy this episode with John Presco.

Well, John, thank you so much for joining me. It's been quite a while since we've been together, but it's great to see you again. It's a great pleasure. And we go back. We go back. In fact, you were, you won't remember this, but I will. Oh, I bet I do.

Okay, me. The first, I remember the very first time I met you. I think I do too, but we'll see if our stories align. You were in the office, you shared it with my friend Ian Affleck. Exactly. And I had just been accepted to MIT, and I came down and stayed with Ian, and then visited Harvard, where you were there. And I remember how intimidating you seemed at the time to me.

You were kidding. Because you were working. We were talking, and you were just at your desk working, and I thought, my gosh, this guy can work through all this. I really, it's really important. I don't remember the part. about working but I do remember being introduced by Ian and I I gathered that the two of you had a rapport of being fellow Canadians. Fellow Canadians. Although Ian was, actually Ian grew up in the city I went to

college and in Ottawa and had it, you know, my, neither, we'll get to your background. Neither of my parents sort of finished high school. Ian's came from a different family. He told me once, I don't know if you ever knew this, but he bought a poem when he was in kindergarten or grade one, which says, when I grow up, I want to be a doctor of philosophy. And I thought, wow, I never even knew what that was until much later. Anyway, I remember you then. And, of course, you know, when I was a student coming down up the river all the time. And then we were in the Society of Fellows for, I think, a year or two together before you

moved, before you became on the faculty at Harvard. You were 80 to 83 or something? Yeah, one year I was in the society. Oh, that's right. You got promoted or seduced away from the good life. Yeah, yeah, yeah. And where you actually had to work for a living. Yeah.

I remember, you know, Sidney Coleman called to offer me this assistant professorship. And I said, do you happen to know what the salary is? And he said, no, I have no idea. But I'm sure it's absolutely pathetic. compared to anything but the salary of a junior fellow. And he was right. He was right. And you got to,

you had to teach on top, which you, which the next thing I want to point out to remember is which you were, you are a, a fantastic teacher, which one of the reasons I wanted to talk to you about physics today. I, your lecture,

I remember actually sitting in one, while I was in the society, sitting in a class that you taught on, on sort of advanced quantum field theory. It was a class on, basically the work of a tuft as far as I can remember. Because it was sort of all of the, it just sort of led a whole series of in late 70s,

middle late 70s, for new ways of thinking about quantum field theory. And I relate, of course, like many people, I borrowed from your lecture notes when I was later on teaching it. It's the fact that you cruise these immaculate lecture notes just helped me time and time again, including, in fact, in preparing for this podcast. So you've done the world and your,

fellow physicists a great, a great service. But you know who learned the most from those lecture notes, me. Yeah, and well, I think, you know, but that's the point of teaching after one of the points is you finally, when you have to write it down and you think, when you have to try and explain it, you realize how many things you don't understand. Yeah. I think it must, it was probably 1982 when I first. Probably 82.

I thought that topic. And I think I still draw on the knowledge and insights I got from that experience. I worked very hard to, you know, digest and synthesize a lot of material. You worked incredibly hard. It was an amazing course. And when you think back, maybe you still do this, but if you think back, you say, wow, it's amazing. I worked that hard.

I mean, the energy of a young faculty member to work to learn things is really impressive. I don't, maybe you've maintained that and you still. Well, it's hard to know for sure, but I do have the impression. I have a little less stamina now than. I did 40 years ago. You have a great, well, the subject, one I want to talk to you about, which we'll get to will be quantum computers and quantum computing. And you have been teaching class in that for a long time, and I've been looking at those

lecture notes as they've been developing and just superb. So in any case, you did move from the Society of Fellows. But before that, I want to go back. So I want to go back, because this is an origins podcast, as you know, to your own origins, because I don't know a lot about it. I knew you grew up in Highland Park in Illinois. Is that right? That's right, yeah.

I was born there. I went to the public schools there all the way through high school. Is it one of these good suburban public high schools? Yeah, it was a very good school system. And I got a pretty good education there. Yeah, it seemed like it. And not only that, you did pretty well. You were the valedictorian, as my research seems to have indicated.

All right, your staff did their homework. Yes, my staff. That's right. But I am, and yeah, the good big suburban schools, when I taught a case, you know, there was Shaker Heights High School, which is another, you know, some of these good public, suburban public schools in the United States still do produce a good education. And you went from there to Princeton, but you already, you went and majored in physics. Let's go back. What got you interested in physics? Were your parents, either your parent, what do your parents do?

What did they do? My parents were both trained as lawyers. They both had law degrees. In my mom's case, that was very unusual. She was the only woman in her law school class. In fact, her law degree is from Western Reserve, now Case Western Reserve, where you will, you know, at one time on the faculty. Yeah.

She grew up in Cleveland, and her father was an attorney. And she was, I think she might have, you know, joined his practice, except World War II intervened. My dad was a bit of a prodigy. He graduated from law school at age 20. Wow. He went to the University of Chicago. He grew up in Chicago.

And it was possible then to concurrently get a bachelor's degree and a law degree. Wow. So 1932, he's a 20-year-old lawyer, except he couldn't take the bar exam because he was a minor. He had to wait until he was 21 for that under at least the rules that held in those days. Now, he practiced law for nearly 10 years, but then when the war came, he was 4F because of a medical condition. So he couldn't serve in the military, but he did work for the federal government. And that's where he met my mom. He went to Baltimore for what was then called the Federal Security Agency.

which is the forerunner of what became health education and welfare. Anyway, their main legal agenda was to flesh out the Social Security program, which was still kind of new at the time, which there were many, you know, legal issues. And anyway, they worked in that same law office, and that's where they met. Wow. And then after the war, they came back to Chicago where my dad was from. Oh, okay. And they moved to the suburbs, Highland Park before my brothers and I were born.

And that's where I was raised. Presumably moved because of the education or the opportunities for a place to bring up kids. Yeah. Okay. Wow. Wow. Your father's not still alive, is he? My father is not alive. He was born in 1911, so it would be quite a feet if he were. He did live to be 90. Oh, that's good. My mom just died. year she was 1921 but she she made it to 100 at least but uh no i was going to say i would have thanked him for social security's been good to me lately um and uh so now so they were both

well at they were both trained at you know in college as lawyers uh you did they where did you get interested physics well i think it was the space program for my generation And that was very transformational. I became fascinated when I was in second grade with human spaceflight, which was sort of the biggest, you know, news of the day. I can remember hearing about Yuri Gagarin being the first human in orbit. And then, you know, there was the Mercury program with Alan Shepard and Gus Grism and John Glenn and so on. Each one of those flights was a huge national event.

where, you know, in those days, we had three television networks and they'd all stop regular programming to follow these space flights. And so I became very curious about rockets and space and that drew me to the public library where I started reading books about things like that. So I think that was the earliest thing that really got me started. I actually, when I was a fourth grader, when I was 10 years old, I read a book which turned out to be very influential. And I didn't understand until many years later the origin of this book. It was called the World of Science. And it contained chapters on various scientific disciplines. And the one of the ones, which I found especially interesting, was about theoretical physics.

and it doesn't mention the name of any scientists, but it does mention various scientific developments and achievements. And considering this book was actually published in 1958, it's a rather amazing thing that it contained a detailed and accurate discussion of the discovery of parity violation in the weak interactions. Wow. Physics is different in a mirror than it is in real life. And I thought that was the most amazing thing.

It is. But what I found out years later is that all of the scientists who were interviewed to provide the content of this book, the author was Jane Warner Watson and her husband was a Caltech faculty member. All of the information came from Caltech faculty. And the chapter on theoretical physics, the main sources were Richard Feynman and Murray Galman. Really? Wow. Wow, that's kind of poetic, isn't it? I thought so. Yeah. So this got me excited about theoretical physics, and then, you know, 20 years later, I joined the Caltech faculty. Yeah.

And Murray Dillman and Richard Feynman became my colleagues. Wow, that must have been, wow. Did you know about that before you joined the Caltech faculty? Did you know about this? It was only after. Well, actually, let me think about that. Here's how I figured it out. In this book on theoretical physics, it starts out with a story, which you may recognize, about a boy pulling a little red wagon with a ball in the back. And when he would pull the wagon forward, the ball would roll to the back.

And when he'd stop pulling, the ball would roll to the front. And then, as the story was told, he went to his father and said, why does that happen? And his father said, well, that's called inertia. but nobody knows why. And then years later, when Christopher Sykes did an interview with Feynman, which became a TV show on BBC Horizon, which was broadcast in the U.S. on Nova, Feynman tells this exact same story. And when I first heard that interview, I thought, Feynman stole the story from that golden book that I read when I was a little kid. Then that was what stirred me to look at the book after many years.

It was still on my shelf. And there are front notes in very tiny print which say, you know, who was interviewed for each one of the chapters. And not only that, there is a photo of two unidentified physicists in front of a blackboard making diagrams. And that was Feynman and Gellner. Wow. I would now recognize years later. But, of course, when I was 10 years old, I had no idea who they were. Wow. That is an amazing story. Yeah, I know the story, obviously, but that's, well, now, did you read, so that book was instrumentally in grade four. For me, it was a book about Galileo in grade five or six that was really influential. But it was, did you? I read a book. I don't know. I wonder if you ever read this one. We are pretty much contemporary. Yeah, we're with year apart. There was a book for children by Isaac Asimov, which was called Breakthroughs in Science. It had a chapter about Galileo and about, about,

Newton and about, you know. Yeah, all those people. Yeah. And it was a fantastic book for a kid my age, you know, like fourth or fifth grade. And yeah, the biographies of all these great scientists and the things they accomplished. And that was also very fascinating to me at the time. So did you read widely? Your parents did they encourage you to read?

Did you, was it reading about scientists? I mean, in addition to the space program, for me, it was reading about scientists and then books by scientists, like Asimov, George Gamow, also a little bit and that really had a big influence. And then Feynman later on. But I'll tell you, I had another passion besides science

and that was baseball. I think I read more about baseball than about anything else from about, you know, the age of six till 14. And I was very interested in statistics. I was going to say baseball is a great place.

Baseball is a very quantitative. sport. I collected baseball cards. And the way I did it was the way one used to do it in those days. I take a nickel down to the stationary store and buy a pack, which had five cards and a piece of very stale bubble gum. I did too. Hope when you open the pack that you, you know, see one of your favorite players. And I bought a lot of gum and collected a lot of cards, which I still have. And on the back of these cards, there were the, you know, career statistics of, um,

whoever the player was on the front. And for some reason, I memorized many of these. Sure. And one thing that's interesting is back in the early 1960s when I was really into baseball, people didn't really understand baseball very well. You know, there was kind of a revolution in the 1980s in baseball analysis where people had their amazing, insight that in order to win, you have to score runs and stop the other team scoring runs.

And so if you want to quantify the valuable player, you should figure out how many runs they help score or prevent it. And the statistics are much more in line with an actual player value now after that revolution than they were back in the 1960s. That great book and then movie Moneyball, I think it was about the money ball. Money balls are really good book. My hero who actually was influenced on Billy Bean, the

Moneyball, was Bill James who was involved in the 1980s and understanding what's really important for scoring runs and preventing runs. But you know, it may have changed baseball, but you know, it's funny because

I grew up in Canada, but I, baseball was the sport I also liked growing up. You know, the thing I liked about baseball was one of many things besides liking baseball was that it was statistics were used more than in any other area and as a kid it's a great way to teach kids statistics because learning mathematics often for many people even college even often they wonder why they're learning what they're learning and if they can see applications then it makes it much more relevant and and and certainly you and to understand what a batting average of 300 is you had to know a little bit about statistics at least

And, you know, actually when I was chair of the physics department, a case, we wanted to move our physics class in electromagnetism down. It was taught a little bit later after the mathematicians taught vector calculus. And we moved it at the same time, and the mathematicians were really concerned. But the students, as happened to me when I learned it, found it much better because the real place you understand, vector calculus when you apply it in things like electromagnetism. Otherwise, you'd sort of don't understand why you're doing it. And so real world applications of math are a real good way to learn math. And I think baseball, therefore, should be used more often in schools as an example. Well, it was very inspiring for me when it came to understanding statistics and how to

compute them myself, you know, which was very empowering. Incidentally, there's a lot of interesting physics in baseball, too, but I was less attuned to that in those days. My late colleague at Yale, Bob Adair, wrote a book called The Physics of Baseball. I read the book, The Physics of Baseball. And he became because, as you probably know, the president of Yale became a commissioner of baseball, Bart Jiamatti, the actor's father, who unfortunately died early. But he appointed Adair as official. Before that, he was head of the National League, and he appointed Adair of physicist to the National League.

I think he had that official position, whatever that was. He was well qualified for that. Yeah, yeah. Anyway, so baseball, science, but you read more baseball than you read science books. By the way, did you ever read the character of physical law? That had a big impact on me, Feynman's book, when you were younger. Not so much later. Oh, okay.

Well, in any case, you went off to, did your parents want you, did they want you to be a lawyer, or do they not want you to be a lawyer because they were lawyers? Or did they care? They didn't. Well, you know, my dad left the law. He decided the law was boring. And he became a businessman. He worked in marketing for a company called Allied Radio for over 20 years.

And my mom did not, you know, return to law after her child was born. She was a very amazing woman. though, and she was very active in many ways. She had natural leadership qualities, and so she became the president of everything, you know, the PTA and the League of Women Voters and a local philanthropic organization. And, you know, she was volunteering for everything. She has a very different personality than I do. She, or she did. She was very outgoing and extroverted. I'm quite introverted. Yeah, you are. It's funny. interesting to me because you are introverted, but you're a wonderful lecturer and teacher.

It's a nice combination, you know, it's really interesting. Well, it seems to work for me. Yeah, yeah, no, I think it, I think because I suspect, I don't know, because I'm the opposite, obviously, in a ways. I'm not an introvert. My wife keeps reminding me, but, but, maybe because you're introverted, you put more energy into preparing things so that you're well prepared to speak about them. Well, I, I have a constant fear of, humiliation if I'm not prepared. Yeah.

I remember that. I actually remember that. Well, anyway, the answer to your question is, no, they didn't care whether I was a lawyer or anything else. They just encouraged me to do what I found interesting. Neither one had a scientific background at all or any special knowledge of science. And did you decide before we went to Princeton that you were going to study physics? When did you decide you want to study physics? As a matter of fact, I became more interested in math than in physics when I was in high school.

Okay. I had learned about Goodell's theorem. Oh, yeah. And the idea that there are things that are true that can't be proven. I thought this was the deepest insight the human mind had ever achieved and that I was going to be a mathematician, studying logic and set theory and so on. and I had this view. I'm not exactly sure what it was based on,

that if you were going to do math, you had to go to Princeton. Maybe it was because Einstein had been there. Yeah, yeah. I assume that's the reason. But, yeah, so I went to Princeton intending to be a math major, and then I had to talk my way in as a freshman to a graduate course on set theory and logic,

and, you know, which I did well. But I started to realize when I was a freshman at Princeton, I think I had really known all along that I just am not cut out to be a mathematician. I don't have the ability, you know, mathematicians. And they have, you know, extraordinary powers of thought in their own avenue of discourse, which I really just could not match. And around that same time, though, I was getting more and more interested in physics. I had some inspiring teachers. The first one was Val Fitch, who won the Nobel Prize for discovering the CP violation and the neutral Kion system. He taught electricity and magnetism, and he was great.

And then I took a full year course from John Wheeler. Wow, that would be inspiring. And, you know, it's funny because at the time, I thought he was ancient, impossibly old. can anyone be that old? He had worked with Bell's bore for goodness sake. But in fact, he was 61 then. You and I both know that's not really so old. But at any rate, he was

inspiring, very idiosyncratic. Yeah, sure. But he deepened my interest in physics. So I didn't have any doubt at that point that I should study physics.

That's, there's boy, there's so many interesting things about that. I want to parse a little bit. One was the mathematics thing. It's interesting because I did a degree in math and a separate degree in math and physics. And the funny thing was that you're absolutely right. I realized I wasn't a mathematician as you did. The difference for me was when I was doing physics, I could tell what was next, what was going to happen next, where I was going In math, I did very well in, but I really never got a sense of what was on the horizon. And I was just sort of doing it.

And that was immediate for me, made it clear to me that I wasn't really cut out to be a mathematician. The interesting thing for me was that there were people, there were other people in my math classes who were better mathematicians than me. And they had to take a physics class as part of their course. And I thought it should be, I just figured physics has applied math at that time and I thought it would be trivial for them. But I was surprised at how many mathematicians had trouble in physics classes, which is really a surprise to me. Yeah, my experience is very similar. And I don't understand why the mathematicians don't, you know, beat the pants off the physicists when they come to advancing physics. But for some reason, they don't.

Yeah, well, you know, and I think it's, you know, you don't, well, anyway, it's a demand. for different levels of truth, I guess, more understanding, as a number of people have talked about. And I was, and John Wheeler, I was influential,

me as an undergraduate occurred to me. I may have actually seen you. If you, did undergraduates go to physics department picnics? At Princeton? Yeah. Well, I don't remember the picnics.

I do remember Wheeler having us over to his house for tea. Yeah, well, that would be nice. I'm not sure. I probably did go to the picnic. if there was a picnic, but I don't remember that. I just remember going down at one point. It's when I was trying to encourage him to come speak.

I was the president of the Canadian Undergraduate of Physics Association. I got him to come up and speak. And I went down to Princeton to try and convince him. It happened to be a physics department picnic. And I remember I still have a picture that day. And were you successful in? Yeah, yeah, yeah, yeah. And he came up and he, you know, what he was like, just wonderful and charming.

He brought, I think he brought me the gravitation book, or I owned it, the Big Bible. and he wrote what is what it still remains the longest inscription i've ever had in any book it was about two pages because i don't think he's so typical he was lovely but anyway i was probably too introverted to ask him to sign my copy of grapies yeah i can tell you a lot of uh funny stories about weiler maybe you'll appreciate this one um It's the first day of class. We're sophomores. John Wheeler is going to teach us classical mechanics.

The book is Oldstein. Yeah. Classics, which I dipped into before the first class. And I thought, oh, this is going to be great. We're going to learn about, you know, the principle of least action. The Euler-Legrange equation, I was excited. And so Wheeler comes in, you know, he's impeccably dressed in suit and tie. And he had very, he's very skillful with the chalk.

And he, the first thing he did is he made two dots on the board and labeled one of them A, the other one, B. And he said, all right, there's an electron. And it's going to go from A to B. But how does it know what path to take? There's no way to know. So it takes all the paths. Oh my God. adds them all together with an E to the I S, and that's how it knows where to go.

He was trying to explain to us that this principle of lead action really comes from quantum mechanics, and it was the insight of his most successful student, Richard Feynman, that we can think about quantum mechanics that way and how classical mechanics arises. And To Wheeler was very important that the students who were about to learn classical mechanics understand that it comes from something more fundamental. Well, that's fascinating. Most of us teach classical mechanics. And in fact, it was not appreciated by everybody in the class.

I thought it was wonderful, but it didn't help you to do the problems. Yeah. Write down the Lagrangian for a bunch of masses and springs. Well, you know, again, this may be, well, for the physicists who are listening, for me, there was another example of utility. I remember taking classical account out of Goldstein. and well actually first one was a slightly lower level and Goldstein I taught myself

but but and I just kept thinking well what's the point of all this you know the Lagrangian blah blah blah and like many things was only when I got to quantum mechanics that I appreciated the utility of the class of the things I'd learned in classical mechanics because that's where you really use it Hamiltonian Lagrangian I really I mean they seemed

abstract and useless or not useless but but pedantic to me as a as a when I was taking classical mechanics. It was only later. So, you know, some people have, I don't know if you ever taught it. I think, I don't know if Feynman ever tried to teach at Caltech. I've known one place where they tried to teach classical mechanics or quantum mechanics before classical mechanics, using Feynman's book, by the way. It was an abject failure, but it would be an interesting thing to try.

I have taught classical mechanics and I enjoy teaching it. I probably because I don't mind being pedantic. I'll tell you, I had a role model in teaching classical mechanics because when I was at Princeton, the next year after Wheeler's class, I took a more advanced class in classical mechanics. And it was one of the best lecture courses I ever attended. And guess who the instructor was? You're not going to be able to guess it. It was Alan Gooth.

Was Alan Gooth? Wow. He was at that time an instructor at Princeton. He just finished his PhD at MIT, right? He worked with Francis Lowe. And he just come to Princeton, and he worked very, very hard on that class. He told me years later, which I can easily believe. He works very hard. He was perfect.

And so I stole a lot of the things that I do when teaching classical mechanics from Alan Gooth. This is now even more poetic. God, this whole, your whole history is going to be poetry in so many ways. This is great. Because I want to move on, because you, in my opinion, so you have. have, you, oh, Alan Gooth that, but I think Alan Gooth owes you, as we'll talk about. Because you, you, you completed your undergraduate work, and then you went to Harvard as a graduate student, which is where, again, where we first met. And you achieved one of the mere impossible things,

which was to be a student of Steve Weinberg's, and actually get a BFD, very few, probably, as we all Luke, because if Steve was interested in something, he did it, and he generally could do it faster than any graduate student. And so it was hard to, unless you found your own problem, in some ways, it was probably hard to keep up. I mean, did, I have a story about that too. Yeah, good. I want to hear it. Yeah. I worked up my courage to go to Steve when I was a graduate student, and, you know, it took me a while to build up to it, and asked him if he could suggest a problem. That was 1977. and he had just read a paper by two physicists named Roberto Pichet and Helen Quinn, I think, you know, this paper.

And he said to me, you know, it might be interesting to work out the phenomenology of this model that Pichet and Quinn have just constructed. You know, what experimentally detectical consequences you did? And I thought, oh, well, that sounds interesting. graduate students, well, not any, but many graduate students would do, I then obsessively read everything I could find about the phenomenology of the Higgs sector. Yeah. And while I was doing so, of course, Steve was discovering what we now call the axion. He actually called it the Higlet.

Yeah. So some weeks later, he announced he was going to give a seminar and he had solved the problem he had suggested and discovered that there was a very interesting consequence of that model so that in a sense it could be ruled out, but then it turned out the model could survive in a different form. But anyway, you know, at the time, I felt a little resentful.

I thought, boy, Steve gave me this problem and then he did it himself. Of course, he probably had no recollection he'd even mentioned it to me. I just happened to walk into his office at the moment he had, you know, read the paper and was thinking about it. And, you know, when it comes to working with Steve,

it's kind of like he said, Steve would talk to me, but he wanted to talk to me about what he was working on and get me to, you know, tell him things that would be useful for what he was like a vacuum cleaner. Yeah. And that's how I first impressed him that I had studied. You mentioned, you mentioned a tuft. Yeah. earlier. He was one of my heroes, and I had studied his work in much detail as a graduate student, and Steve had not. And Steve got interested in something that we call instantons, which are an interesting quantum phenomenon in particle physics. It was one of the things that was exciting in the late 70s. And I knew all about instantons, and Steve didn't know anything about them. So whenever we talk, he would be pumping me for information about instantons.

you know it's you know i was a i was a graduate student at mity as you know at time and i i was having trouble finding as well being happy with a supervisor and at one point the great thing about harvard mit i did all my classes at harvard i almost did no class at mit because at the time harvard had a much more in my opinion at the time more powerful physics department at least in the years I was interested in. And so I took all my courses from Steve. And I remembered a very similar thing. At the time, I was very mathematical, and I was working on the geometry of gauge, called the geometry of gauge theories at the time. And that's what I was focusing on with my supervisor at the time at MIT. And I learned all about something called fiber bundles. And Steve was getting

very interested through instantons, in fact. And so I'd done well in this class. And at some point I brought up fiber bundles or something. And it was amazing to me because, again, it was like a vacuum leader. I remember Steve Weinberg phoning me at home at 11 o'clock at night to ask me a question. And as graduate student, I felt like, wow, I can't believe it. It's so wonderful. And the interesting thing, I wanted to work. So I asked Steve if he would be my supervisor because I thought, you know, I wasn't really getting ahead at MIT. And Steve gave me the greatest. He was very honest. He said, I have a problem that I know I would, it would be a good PhD problem. It was. have to do with what's we now called Cairo-Lagrangians

and sort of understanding the phenomenology of these things. And it would have been a great thesis. He was fascinated by it. He said, I know it won't interest you because it's not mathematical. But I'll be willing to supervise you in that problem. He said, however, I have an obligation to Harvard graduate students. And if any Harvard graduate student asked me that I like asked me for a problem, I'll give them exactly the same problem and I'll say I never gave it to you.

So I said, under those circumstances, it was too much of a gamble and I didn't work with Steve, but at least he was honest about it. And it was true. In retrospect, it was a,

it would have been, it was a very fruitful area, obviously, though. And as you pointed out, Steve did not have many graduate students. At Harvard at that time, the graduate students in particle theory

flocked to Sidney Coleman and Howard George High. Howard, because he was such a, a supportive mentor and Sydney because he knew everything and could answer off the cuff, any question you could come up with. But, and I intended to work with Sydney, actually. But then I realized that the things that Steve was interested in resonated very well with what I was interested in.

And I don't know if you had this interest at that time in the late 70s, but what I found particularly promising and fascinating in the late 70s was the connection. between particle physics and cosmology. Which is exactly. Which he had pioneered. Actually, it was around that time that his book, first few minutes, had just come out a little bit earlier.

And I remember reading it and getting to sign it. And yeah, that was what got me interested in cosmology too. And indeed, I was going to say, one of the ways to, well, I alluded to earlier, one of the ways to survive as a graduate student of Steve's was to have a problem that you could build on that wasn't the focus of his attention. And I don't know if your thesis, sorry, go on.

I was going to say that book, the book, the first three minutes, was quite unusual because it was a book intended for a popular audience, but it had a big influence on physicists. Yeah, it really is. A physicist who came from a particle theory background learned about cosmology and in particular the early universe from that book. and that launched some scientific careers, including Alan Gooths. I think he was very much influenced by that point.

Yeah, it was really a powerful book. And although he claimed it was written for a smart lawyer, as you may know at the beginning of the book. Yeah, I remember that. Of course, he must have had Louise in mind. Yeah, yeah. Very smart lawyer. Anyway, you were asking about my thesis problem,

and that's also an interesting story, because it combined two different things I was interested in the time. Exciting developments in the late 70s in particle theory. One was, as you mentioned, the fiber bundles, the ideas from topology, which were becoming increasingly relevant to particle physics, and the other, the potential to learn about the very early universe, and testing our ideas about particle physics by studying the very early universe. And in the case of the topological ideas,

I learned a lot about that from Sidney Coleman and from reading his papers. And in the case of the connection between particle physics and cosmology, I learned a lot about that by reading things that Steve had written. But at that time, at least, there weren't many people who knew about both things. And so I was able to arrive at a question that, connected the two, and in particular the possibility that what we call topological defects, which our magnetic monopoles could be created in the early universe. And I was interested in how many of those particles would be produced in the early universe and how many would still be left

over today. And I told Steve about this interest, and he was very dismissive. He really didn't seem to think it was a good question. He didn't know anything at that time about these magnetic monopoles or about topological ideas in particle physics. And he thought, you know, it seemed very speculative and not well-founded. You know, he liked, of course, he liked the things that he understood. Yeah. You know, he liked arguments that were based on sort of the, you know, the most general bedrock

principles. Yeah. And this involves speculation. about grand unified theories and he didn't really understand the topology. So it was a little discouraging that my advisor was not enthusiastic about what I was doing, but there were others who were, including Howard Georgia and Sidney Coleman, were supportive. Also, Bert Halperin, who has a background in particle physics,

but knew a lot about these topological things. And also postdocs who were at Harvard at the time, who were brilliant scientists, especially Ed Witten. and Michael Peskin. So I had a community of people who were supportive, but it wasn't really coming from Steve. Well, that's probably lucky for you

because if he had been, he would have scooped you on it, I think. I mean, he was a Steve roller. That's true. Yeah, and I have to say, I dragged my feet because I, you know, I kept thinking I didn't understand it well enough,

so it took me quite a while. Well, you know, I... Go on. Here's something that seems amazing from the perspective of, you know, the young people who do theoretical physics today. I had no papers

after four years of graduate school. I wrote my first paper, the one about magnetic monopoles in the early universe at the end of my fourth year, submitted it to physical review letters and it was promptly rejected. Oh, really?

As, you know, not being of sufficiently broad interest, you know, that story. And it was very discouraging because, you know, I was, I thought I was going to finish the following year. I'd be applying for postdocs in the fall. I had no publications that at least appeared in journals. So I remember Roberta, my wife, to cheer me up.

We went out and bought a colored television set, which actually did cheer me up. Up until then, we just had this little black and white set that carried around the apartment. Baseball is much better in color. Oh, absolutely. And then I resubmitted the paper, and it was accepted. But, you know, I was applying for postdocs with just that one paper,

which at the time hadn't even been published, and somehow I did okay. Yeah, it's, it is, it's, it's, it is interesting. Well, it helps to have people, especially good people who know you and, and, and, you know, they're famous. Yeah, yeah, I mean, they're famous stories, the famous story about Dirac. I'll tell you that later, but you probably know the story. But that paper, which you did write, I was going to say, I didn't know if that was your thesis per se. But that... It was not, actually.

Okay, I bet it wasn't. Yeah, I didn't think it was. But that paper that you wrote on monopoles, which is these predictions of things that in some way, in most of these, in many theories of the universe must appear, caused a problem in particle physics. And I think that was probably, I think it's fair to say, the most single influential paper that convinced people there was something that particle physics had something to say about the early universe and needed to be thought about, and the early universe needed to be thought about seriously, not just speculatively and back of the envelope. the paper you produced produced ultimately a problem which for many of us, I remember when Alan Gooth came out with inflation. Yeah, I had never heard of the flatness problem or the horizon problem, which are part of the things which we now think of inflation as solving. The really significant implication of inflation, at least for us in the Boston community, as I

remember was that it would solve the problem that you'd present it as a generic problem in cosmology. And I think it has to be understood as having been profoundly important in that regard. You know, the context was interesting because until a few years before I wrote that paper, it would have seemed ridiculous and we wouldn't have really known how to get started to talk about the first 10 to the minus 35 seconds after the Big Bang. Yeah. But one of the things that made it possible to do that was the discovery of asymptotic freedom, which came in 1973.

Until then, you know, when the universe was a hundredth of a second old or, you know, the things were so hot that physicists had no idea what was going on. But then with asymptotic freedom, we understood that you could go to much, much higher temperatures and therefore earlier times and still have theories that were. predictive and made sense. And the other important context was the idea of grand unification. Sure. That the interactions that we knew about could be descended from some more unified theory.

And that unification would have consequences. The one that got a lot of attention from experimentalists at first was that the proton would be unstable and you could look for the decay of the proton and people did. But another one of those predictions was the magnetic monopoles. And so this was a situation where we wanted to explore physics at fantastically high energies, and we couldn't do that with accelerators, but we could use the early universe as the accelerator and look for the, you know, the vestige of that very early period and learn something about fundamental physics.

Which then, I should say, by the way, personally sponsored my own, prompted my own conversion, sort of movement out of mathematical physics to my thesis was, on the early universe and ways to try and resolve problems that having to do with the monopole problem but also the problem with entropy. So yeah, these things really got, they were in the air and the thought that the hubris, the hoodspa that the particle physics community had at that time, having just produced a standard model and then gran unification, it looked like literally the grand synthesis was in the air in the late 70s and early. the 80s and everyone expected when these machines are built, they discover proton decay and

even, and as you know, around that time, Blas Cabrera discovered what I guess is the only monopole in our universe because he never saw another. And in principle, I suppose, inflation would suggest there might be one monopole in our universe, so he might have been lucky. Blas would have had to been very, very, very tough. Yeah, yeah. In any case, that, So that was, I think it's really important. And from there, I mean, your work, you know, having no, we work, we've worked together too, but I mean, having been close at Harvard, the work continued to be many, a number of

ideas related to particle physics and cosmology, axon physics, and, and then related to black holes. I, I, we, we, we, we've, we wrote a paper, which, as you know, I'd got. and interested in in a topic that was relevant to quantum effects in black holes, something that our friend and colleague, Frank Wilczek and I worked on something called discrete hair, like holes. And there was an aspect of it that was nagging at me. It was based on something Sidney Coleman had said. And the typical of Sydney, he sort of threw off a remark. And it required, and I thought, well, this is really a subject that should be explored.

Lord, which is, you know, you can't have a, it turns out in a closed universe, you can't have an electric charge. But in these weird kind of models we'd made, those same arguments wouldn't tell us you couldn't have these things we call discrete charges. And so I talked to Sidney about it. And I guess you and I talked to you, I think it might, I just looked at that paper and we thanked the Aspen. So it may have been an Aspen that we started to talk about, but I figured who would be the person to flesh out this problem if Sydney wouldn't, you know, know, be interested. And the fascinating question was, for me at the time, what was fascinating was whether this could be a way that black holes could store information quantum mechanically

to solve the information paradox problem in black holes, where the information paradox problem, which I've talked about a number of people, but for listeners, I'll just remind them that objects fall into a black hole in a principle. Once they've fallen in, all information about what fell in is gone, except for the mass, the total mass, the black hole and its charge and it's spin. And then if the black hole evaporates by hawking radiation and the thermal radiation, all that information goes away. And there's a property of quantum mechanics called Unitarianity that says that shouldn't happen.

And so maybe, so it, you know, that's why black holes become so fascinating because there are, they're an area where general relativity confronts quantum mechanics, the two forefront fields of fundamental fields of physics. And so at the time I thought, well, this is an interesting potential area of interest, which is maybe whether you could get quantum information from these, this quantum hair. And we talked about it. And I was kind of fascinated because you began to, I don't know if it had any influence of you, but when I look at your work, you began to think more about black hole information around that time. So I don't know if those discussions ever had any, if that, or you'd been thinking about it.

it beforehand or not. I meant to ask you that. I was starting to think about it. And part of what sparked my interest or accentuated it was the paper you wrote with Frank Wilczak. Actually, that's a nice paper we wrote together. There are a lot of interesting things. It was nice because of you. Yeah, there were a lot of interesting things. And one of the things that we discuss is related to an interest that, you know, I continue to pursue, which doesn't have to do specifically with black holes, but with objects called aneons. Yeah. Articles that obey unusual statistics. And in particular, non-abillion anions, which are the particularly exotic form of eneons.

And some of the mathematics we did in that paper was related to anyons, and I continued to pursue that. We should mention that really, in a way, the hero of this whole discussion about black hole information with Stephen Hawking. Until 1974, one could just say, okay, information goes into a black hole and never comes out again and fine. we could say it's locked inside the black hole behind the event horizon, but when Hawking discovered that in fact, because of quantum effects, black holes evaporate and can eventually disappear, this caused a very profound tension between the idea that information is not destroyed

in quantum mechanics. And on the other hand, what comes out of a black hole doesn't seem to be related to what falls into it. And you know, we're still struggling with that. there's been a lot of progress on that question. And it's a problem now, which is, you know, nearly 50 years old. Yeah, and still not solved, in my opinion. Still not completely solved. I'm glad you agree.

There's been a lot of heat and a little bit of light in the process. I actually think there's been a lot of light. But nevertheless, not completely solved. Yeah, yeah. But it's interesting. The reason, it's a nice segue, because I think it's important when I think of, when I think of who might be these things might be useful for, one of them is young students.

And you hit one point when you talked about what you did as a graduate student. And it really relates to my feeling of being an early graduate student. One of the problems of being a graduate student initially, and I often try and encourage my students to get away from this, is the first thing you want to do when you work in any problem is understand everything. and it's a natural tendency when you're starting out saying before I bark at this problem

I have to read absolutely everything and I have to understand absolutely everything and it takes a long time before you realize you actually just have to understand something and you can't and you keep getting delayed by learning more and more and you really want to hit something concrete and I know you talked about that

in your early period with Weinberg and I think it is a real it's probably the biggest change that a student needs to make from being an undergraduate to being a research graduate student is to realize is to begin to focus is to say I I can't learn everything. It's I have to understand. I have to actually do something. So that's what you can't, but I have to admit I tried. Yeah, yeah. I really want to do. No, me too. No, I know what it's like and it's very seductive. But the other thing that that your history and now I want to move to

interestingly after an hour and 20 minutes to the topic I wanted to eventually get to. Time flies, huh? When you're having fun, I hope that you feel it that way, which is ultimately quantum information theory, which obviously become one of the world's leaders and you direct an institute at Caltechian. But I think it's another interesting thing is, I assume that your interest in information theory in general initiated with black holes.

And, yeah, and it's, It's a nice thing to know for students that things you learn are often useful in ways you never expected them to be. And it's, you know, we ran up, we created a program when I was chair at Case on physics entrepreneurship, which the dean of the business school said was an oxymoron, but it isn't because, because scientists have to learn that. Sometimes the problem you think you're solving, you have a problem and you try and solve it, but you really try and end up solving another problem. And the tools you've learned are never wasted in that way. And sometimes you don't completely solve the problem, but you partially solve it. And that also is a valuable contribution.

And somebody can build on that and take it farther. Absolutely. And maybe that's the thing. Maybe that's why mathematicians don't like to be physicists because physicists won't mind partially solving problems. Mathematicians, I think, probably have a harder time with partially solving something. Partial proof. Well, you know, if it's not a theorem,

then it doesn't count. Yeah. That's a very high standard. Yeah, exactly. And a higher one than we have. But around that time, right around the time we were thinking about a little bit after we were all thinking

about black hole information and the world of quantum computing changed. A world which of course had been created as you described in your beautiful many review articles. And I guess I talked about even in my book about Feynman.

Well, we'll talk about Feynman's introduction in this field, but for most of us, for me, and I suspect you, I first learned about quantum computation when I learned about this result by a guy named Peter Shore. Is that what, I mean, that's when the, Peter Shore had a proof that these things called quantum computers could do something that at that time was protecting the world's, and still is, the world's banking system,

which was how to factor a large number into its product of primes, which a classical computer can take longer than the age of the universe to do if the primes are, if the number is big enough. And that was a key, that was the way to encrypt information. And Peter Shore showed that a quantum computer could do that potentially in a human lifetime or less. And that was like a lightning bolt that reverberated throughout the community. So for me, that's why I first heard about quantum computers.

I don't know if it's where you did. Well, it's not where I first heard about them, but where I first began to take the idea seriously and get deeply interested in it. There's several things prime me to get so deeply interested. One is what you mentioned. I was interested in information because I wanted to understand how information can escape from a black hole. And being the type of person who does these things, I thought, well, I should learn everything about information. and in particular about quantum information, which was not a, you know, a deeply developed subject at the time and had only a few practitioners. But there were ideas like quantum cryptography, quantum teleportation that I learned about because I thought maybe that would help me to understand black holes better.

And then another thing was happening around that time, as I'm sure you remember, for my generation, which is also, or your generation of particle physicists, our great hope was to discover the physics beyond the standard model, which had been established a little bit too early for you and me to contribute to erecting that core theory. Beyond the standard model physics is where we would get our chance to understand new facts and principles about fundamental physics, and where we thought that,

data was going to come from was something called the superconducting supercollider, which was already under construction in Texas. And that project was canceled in 1993 for complicated reasons, but at any rate it was canceled. And many of us, including me, realized at that time that while our opportunity to have data about physics beyond the standard model has been pushed further into the future, we anticipated, as turned out to be the case, that there would. be a machine at CERN that would not quite as well as the SSC would have. Yeah. Explore some of that physics. But in the meantime, what were we supposed to do? And so I was in a mood to learn about new things and maybe explore different possibilities.

But I heard about Shore's algorithm in May of 1994. He had announced the discovery the previous month. And I was immediately fascinated because the whole idea that the difference between a hard problem and an easy problem to solve between problems will never be able to solve in the age of the universe and problems we can solve very efficiently on future machines, that that difference between hard and easy pivotably essentially depends on the fact that it's a quantum world instead of a classical world. What an amazing idea. Yeah. I mean, I've read from you, the quote for Feynman really basically knocks that home. But go on. I'm sorry.

Yeah. So, you know, I was still thinking about black halls. And it was kind of transitional time in our group because I had students working on all kinds of things. You know, black holes and particle physics and aneons. And some of them got interested in quantum computing. and some of them did not. But those who did really dove into the subject. And, you know, when you're in the midst of what turns out to be kind of career transition,

you're not so aware that it's happening because things seem more adiabatic at that time. But it did turn out to be a major change in direction in my scientific direction. And the Shores algorithm was the key to spurring that. Well, it was good for you. I was happy to see that. And it was good for the world because you did it. It is, you know, being able to change, I think it did affect the whole generation. I moved more towards astrophysics for the same reason because it was clear to me that terrestrily were. I was trying to look for ways we could use the universe in other ways to try and constrain particle physics. But it was a similar motivation and we weren't going to get the results on the ground. And then I watched you from a distance because you already moved to Caltech.

and begin to take that. And it's funny because I, you know, I guess I had an inner, we'd work together. We wrote that paper in 1990. And I saw, when I saw that field and saw you be new, I thought, this is, this is just what John is going to eat up. This because it's exactly the type of thinking that I knew you carried out. And I was therefore not surprised. And I was really, really happy that you were doing it.

It just seemed to me a field that was made for you. and it has been, and you've really helped lead the developments in that field, and that's what I want to spend the next, you know, four or six hours on, no, anyway, but next little while talking about, because quantum computing is a subject one people hear a lot about, and there's always stuff in the press about it, because it's one of these things that promises a lot. It's not quite like fusion, which is always 25 years in the future. But nevertheless, there's great, promise, but there are great challenges.

And so I want to talk about both. And I want to spend a little time going into what quantum computing is and where you think it's going, if that's okay with you. The Feynman quote that you always say basically is something. I think it's, I was going to look it up. I have it on my screen here somewhere. But basically that nature isn't classical, damn it. And if you want to simulate nature, then you've got to do it quantum mechanically,

more or less. You may have the quote memorized in your mind because you've used it a lot. Do you have it? Well, that was pretty close. Nature isn't classical damage. So if you want to make a simulation of nature, you better make it quantum mechanical. And it's a wonderful

problem because it doesn't look easy. That's right. And by golly, it's a wonderful problem because it doesn't look so. By golly, it's weird for him. I don't know if he really said by golly. I know. It doesn't sound like him. Yeah, it certainly doesn't sound like him. Well, he was right about it not being easy. Yeah, it was right about a lot of it. And so, and Feynman had first thought about

this problem a long time ago. And I, you know, we could talk about the history a little bit. Maybe, well, the, the thing I've gotten out of, you know, I've obviously followed quantum computing, although I haven't written anything on it, but, and I'd always thought of one aspect of quantum computing. And I guess I got to appreciate a second aspect by reading more of your review articles over the last few weeks. for me, the power of quantum computing was always the fact that quantum mechanics, because of the fact that quantum objects are doing many things at the same time, can be doing many calculations at the same time, whereas classical systems can do one. And we'll get into the mechanics of how that happens.

But that's sort of my basic gut. When people ask me of quantum mechanics, I basically say a bit is one or zero, but a a qubit can be in many different states at the same time, and therefore, as it evolves, it can be doing many different, effectively calculations at the same time. And so using quantum computers in a way to improve the calculational ability. But the other aspect, which I guess really I've come to appreciate from you, is that quantum systems themselves cannot really be ever practically explored fully using

classical computers as a matter of principle, not just as a matter of practicality, and that there are therefore systems that you really need a quantum computer to understand if you understand the physics of those systems. So have I encapsulated the two major, briefly in my own way, the two major strengths of quantum computers? Or maybe let me have you do it. I didn't hear anything that I disagreed with. Okay.

Let you elaborate, because you'll do it much. more beautifully. Well, let's dive in a little bit to what you said about quantum system being many things at once. Yeah. Well,

one way of saying why we think quantum computing is powerful is essentially the way you put it a moment ago, which is we don't know how to efficiently simulate what a quantum computer does with an ordinary classical computer. And that's not for lack of trying because physicists and chemists have been trying for many decades to come up with better ways for computing how quantum systems with many particles behave,

like a molecule that has many electrons. We know the equations that describe that system. we can write down those equations with very high confidence, but they're just too hard to solve. And the root of that difficulty is that these quantum systems, particularly ones with many particles, kind of speak a different language than the language that we know and understand or that our classical computers understand.

And in particular, they have the capacity to become very highly, entangled. And what that word means is that they have very complex correlations among the particles which can't be easily captured in terms of classical data. So if I have a system of just say 100 cubits, the quantum analog of bits, which we call qubits, in some highly entangled states, if these qubits have been strongly interacting with one another for a while, if I wanted to write down a complete description of that system in terms of ordinary bits, it's completely infeasible to do so. It would require, in fact, more bits than the number of atoms in the visible universe.

So there's this kind of extravagant complexity in a many-cubit or many-particle quantum system that we can only get a little glimpse of. Because while the quantum world has this great complexity, our ability to interact, with that quantum world is quite limited. What we can do is prepare a simple initial state of our qubits and we can measure them. But if there are 100 cubits and we measure them, all we get is 100 bits of information, which I could easily write down on a piece of paper. So where's all this enormous complexity? Well, it has to do with how the system can evolve from some initial state to some final state. This goes back to what I was saying about Wheeler, actually.

Yeah. Suppose you're looking at an electron and you see yesterday that it's at some point in space A and you'd like to predict where it's going to be today. Is it going to be at point B today? And quantum mechanics modifies our notion of probability. The best we can do is make some statement about the probability that the electron, which was at A yesterday, is at B today. And how do we compute that? Well, while we're not watching the electron,

we don't know what it was doing. So it could have been anywhere. So we have to consider all the possible paths that the electron takes from A to B and there are rules about how to assign numbers to all those paths, which are called amplitudes. And then we have to add together all those amplitudes to, and then we square it to find that probability. But where these amplitudes are very different from probabilities is that they're not, they can be negative. They don't have to be zero or a positive number. In fact, they can even be complex numbers. So while it doesn't make sense to say the probability it's going to rain today is minus 50%, these amplitudes can be positive or negative. They're different from probabilities, and that means they can cancel out. So when you

add together a lot of amplitudes, the positive ones can cancel against the negative ones. And when that happens, that means that the probability you're going to see the electron at B is very small, but they can also add up to give a big number. And that means the probability you're going to see the electron in B is actually fairly large. And so what does that have to do with quantum computing? Well, what we can do in a quantum computer is we can put in an initial state to our computation, and then the quantum computer processes that state somehow. And we're not watching it while that's happening. But then at the end of the computation, we measure all the qubits, and we just get a bit string out, which, you know, is not a very complex thing. It's just a short list of bits.

But to determine the probability of the different possible bit strings we might see when we measure, we have to consider all the possible paths the computation could have taken from the initial state to the final state. And there are an enormous number of possible paths, and we can't possibly add them all up with our classical computer, there are just too many. And when we do so, it might be that the amplitudes give a large positive number, and that's an outcome of the computation, which is likely to occur, but it might give almost zero, and that's an outcome which is unlikely to occur. And even though with a classical computer, it's impossibly hard to add up all those amplitudes, the quantum computer does it effortlessly, just by following

the rules of quantum mechanics, which nature tells us, are the fundamental rules for how a system of cupid should behave. And so the art of quantum computing, and it's a big challenge, is to figure out how to get those amplitudes to add up to a large number for the answers we want and to give something close to zero for the answers we don't want. And we figured out how to do that in a few cases, like the example you mentioned, Shores algorithm for vectoring large integers. And there are a lot of cases for which we really don't know how to do it, and it might not even be possible. But Feynman's interest was in the applications of quantum computing to understanding quantum systems that chemists and physicists are interested. And actually, we used to talk about that in the 1980s when I was at Caltech. I arrived in 1983, and he died in 1988. We overlap for like four and a half years.

And we never talked about quantum computing per se, but we did talk about computation. He was very interested in quantum chromodynamics. Yeah. The theory of how nuclear particles behave. That's another case where we know the equations. We know with very high confidence what the correct equations are. But if I want to describe, say, two protons colliding with one another at very high energy and predict what's going to come out, we don't know how to solve those equations to predict that.

There are some things that we can compute, but we can compute the outcome of collision between two particles of energy, just like we can do with the Large Hadron Collider at CERN. And the reason is that, although we know the equations, they're just too darn hard to solve, and that's because there are too many amplitudes to add up. And so Phelan was very interested in doing that kind of computation, and I think that was actually quite important.

important for arousing his interest in the concept of a quantum computer. Okay. Let's see. When did he write his, I should remember this, I've written the book, but when did he write the first about, well, the first time he talked about was lots of physics down below or whatever. I love that piece. That was, that was, that was really early. But he was thinking really early about, at least about the physical.

He was interested in computation his whole life. Yeah. back at least to Los Alamos, where he was the head of computation group. Yeah. That lecture in 1959 is quite remarkable. Yeah. That he speculates about computers and which information is stored in individual atoms, which we can, in fact, do today, quantum computers.

But it wasn't until 1981 when he gave a talk at a conference at MIT, May of 1981, one, which was transcribed and became a paper called simulating physics with computers. Yeah. In which he discussed the idea that, first of all, quantum computing is, well, that quantum systems are very hard to simulate on ordinary computers. Well, of course, people who try to do that already knew why. But then he suggests that, as that a quote you mentioned indicates, that if we want to do it, we should have a quantum system be our computer to, in effect, simulate another quantum system. And that was the idea of a quantum computer.

And he deeply appreciated it already in 1981 that a quantum computer would be capable of solving some problems that would be just too hard to solve with an ordinary computer, including ones which are very interesting to chemists and materials, scientists, and so on. I mean, in a sense, there's something called the strong church-chering thesis, which codifies that in some ways to basically say that that classical computers cannot do what quantum computers can do when it comes to quantum systems. That's a, well, that's a very, I think it's a very interesting connection between computer science and physical. and I really think the foundations of computer science are very much foundations of physics because computing machine is undergoing some physical process. The original Church Turing thesis, which dates back to the 1930s,

was that a certain mathematical model, what we now called Turing machine, could capture any computation which could be carried out in nature by any. physical process. The extended church touring thesis says something stronger. It says that what can be computed efficiently with a Turing machine coincides exactly with the things that we can compute efficiently with any physically realizable device. Efficiently has to do with how many steps in the computation we need, how that scales with the size of the input to the problem. And we think the extended Church-Turring thesis as initially formulated is wrong. Wrong.

Because it didn't take into account quantum physics. So the modern version that has replaced it is the thesis that anything that can be solved efficiently in nature with any conceivable computing machine can be simulated efficiently by a quantum computer. and we can define mathematically what you mean by a quantum computer and we don't know for sure whether that's correct it's a statement about physics it's not just a mathematical statement but there is evidence indicating that it is correct

and that's very exciting because it means that anything that can occur in nature in principle because we're only interested in things that can happen efficiently in nature because the things that take a zillion years we don't really care about You know, those will always be things that we can simulate with quantum computers.

Although, if it turns out that the quantum version of the extended church touring thesis is incorrect, that's really exciting too, because it means that we haven't yet captured fully with our current concept of a quantum computer, what nature is capable of computationally. Yeah, no, the statement that it's a fascinating statement saying that a quantum computer can basically, in principle, and we'll get to principle versus practice. practice, answer any, any, any, compute any physical process. But implicit in that is a statement, though, that a classical commuter cannot. I think that's an important thing. I mean, there's the positive aspect, but there's the negative aspect, which says that it is not true that a classical computer can simulate any physical process because the world is quantum mechanical damage. Yeah, but to be honest, we can't prove that as a mathematical statement, though we have good reason to think it's true. the best reason probably just that people have tried very hard to come up with ways of

simulating complex quantum systems using classical computers. And nevertheless, the best algorithms that we have are very inefficient that requires a time on the classical computer, which grows exponentially with the size of the quantum problem. Well, okay. Now, I want to step back and parse some of the, because you gave a great summary of this, but I want to parse this a little more carefully. I want to take people, to a little bit of the nuts and bolts of quantum computers to understand why some of the statements you made and I made are work or are true.

And to understand that, I will, we need to describe the difference between a qubit and a bit, and although people can hear that a lot over the internet, I will ask you to basically give the difference now, if you wouldn't mind. Okay. Well, actually, can I just give a few examples? Yeah, that's even, that's fine with me.

Examples are always good. Yeah. So, well, I think the concept of a bit is familiar to most people. A number, which is either zero one or zero or one, a switch in a chip with a transistor, which can be said either on or off. That's the concept of a bit. an example of a qubit is what businesses call a spin. And so how is a spin or a qubit different from an ordinary bit?

Well, one way of saying what the difference is is that there's just one way to look at a bit, no matter how you look at it. It's either going to be definitely a zero or definitely one. but in the case of a qubit, we have different complementary ways of looking at it. And so one way of describing that in kind of a geometrical language,

which is why we use the word spin, is you can think of the spin as pointing in some direction in space, and it might be up or down along the vertical axis. But there's another way we can measure it, which we can look on a horizontal axis and asks whether it points left or right. And if we just know what happens when we look on the vertical axis

to see whether it points up or down, we don't know everything about the qubit. It has a richer structure because we have these alternative ways of looking at it. And one consequence of that is that the correlations among qubits are richer and more complex. then correlations among bits. Because if I have two qubits, I can ask, okay, if I look at both of them along the vertical axis, you know, are they pointing in the same direction or opposite directions? But there's a different question you can ask, which is, what if you look at both of them along some horizontal direction? Will they both be pointing left or, no, one be pointing left and one

to be pointing right. And because of this richer structure, as you increase the number of qubits, the correlations become more and more complex and harder and harder to encode in terms of ordinary classical information in terms of bits. And that's really the secret of the qubit, I would say, that it has. There are different possible ways of looking at it. And that makes the correlations very different. Let me elaborate a little bit. on that because just so we get everyone clear. The, the strange, the question is, you know, how to encapsulate the strange weirdness of quantum mechanics.

And as you pointed out, if you, if you measure a spin, you know, a particle spin and you measure whether it's spinning up or down. And then that doesn't tell you anything. if you make a measurement of whether of the spin in the in the horizontal direction, you'll find out that it might be spinning left or right. And you might say, okay, well, now I know what it's doing. But then when you go back, having made that second measurement, and after the horizontal, you say, okay, but I knew it was spinning up at the beginning.

When you go back and find out it may not be spinning up. And that process of these, as we say, non-communitive, but the process of saying, well, when I know, just because I know what the particle is doing, I measured it here, I really can't say it was doing anything specific in the X direction. And part of the proof is when I measure in the X direction, I suddenly find out I can't say anything specific about what was doing in the Z direction anymore. And that's really strange. That's part of the part of the long chain of arguments that suggests that this uncertainty in quantum mechanics

This doesn't come from just not knowing enough, not having made enough measurements, but there's something intrinsic. You can't, there's no classical way in which you could picture what just happened in that, by saying that the particle was in some definite state before you made the measurement. Yeah, that's a very good point. And it highlights something that we should emphasize, which is a difference between qubits and ordinary bits. one can look at a bit and ascertain whether it's a zero or one without disturbing the bit in any way. So it might be a switch on a transistor, and I can shine light on it and see whether the switch is open or closed, and that's not going to change it from open to close or vice versa.

Cubits are more delicate. If we acquire information about the state of the qubit, if we observe it, then that will typically, disturb it in some uncontrollable and unpredictable way. And that's part of the challenge of quantum computing. It's part of the wonder.

Yeah, we can do it. And that's part of what makes me so hard to do. And an important point that you made also, which I would like to reiterate, is that there's the way probabilities arise when we talk about ordinary bits and when we talk about qubits

is very different. Like if I have an ordinary coin, which is a bit, I might flip it, and it lands on the table, and I cover it up. And now we know it's either heads or tails, but we don't know which. And so we might say, well, it has probability one half

of being heads and probability one half being tails, because I don't have any a prior information about which is true. But it really is either heads or tails. It's just that we don't know. Whereas with quantum computers or with qubits, when we observe them, it's not that the qubit already is determined to be pointing up or pointing down along the vertical axis.

The probability is really intrinsic. Even if we have the most complete description that nature will allow us to have of that cubit, we are still powerless to predict whether we'll see it pointing up or down along the vertical axis. exactly whereas in some sense if you with the head or tail if you had all the information you could possibly have about the motion and speed with which you you flipped it and the air resistance and everything else you might be able to calculate exactly the probability probably enters because of our ignorance not because of some fundamental exactly the other thing the other aspect i want to elaborate on is actually one i guess i learned from you

from reading at least learned how to emphasize from reading some of you some of you your work. It's a really important difference between, and this is really relevant and has to do with this phenomenon of entanglement, the fact that a quantum system has correlations between many different parts of it that are just absent classically. You can think of, if atoms were billiard balls, you could think of them as each billiard ball doing its thing, but you can't separate an atomic system of a bunch of molecules that may look like billiard balls into a separate set of billiard balls. There are correlations between them, which are quantum mechanical and intrinsic. And the way that that manifests itself, which is probably also another way of thinking

about why classical computers can't mimic or understand quantum system so much and why quantum systems are so much richer, it's a statement that if I have, and again, I learned this from you, that if you have a system, a large system made up of a lot of entangled particles, if you I know the state of that whole system. If it's classical, if I know the state of the whole system, then I know the state of every one of the particles that every object within it.

If I know a whole bunch of bits, if I know the state, whether it's one zero, one zero, zero, zero, then I know where each bit is in. But knowing the state of an entangled, a complete description of that entangled state does not allow you, even in principle, know what the separate components are to describe the separate components. So you don't have enough information when you understand the whole state to understand the state of every single particle

within that system. And that's a really important difference. And I think you emphasize that. And that leads to the intrinsic, of course, complexity of quantum systems and also the fact that clearly there can be much more information stored in such systems than you can access just by measuring the whole system itself. Well, it's a bit debatable whether it's there or if we can access it. Yeah. In some sense, it is,

because we don't know how to simulate what's going on before we observe it in an efficient way using just a classical machine. Yeah, but as you point out, so this gives some sort of heuristic understanding of the power. Once again, these systems are doing many things at the same time, And if you can manipulate them appropriately, you can effectively perform many calculations at the same time, which you could never do classically. But you point out, you can only do this.

These quantum mechanical correlations are so strange to us. Why don't we understand quantum mechanics and why did Feynman hope that a quantum computer would teach in quantum mechanics? Because we're classical because these weird correlations just vanish in the world we live in. We don't see them. Particle billiard balls behave like billiard balls and taking a cue ball and doing something to it doesn't affect the eight ball over the other end of the table. And so it's so strange because those quantum mechanical correlations vanish. And yet the whole point of quantum computers is to ensure a system which is macroscopic in some sense, but which the correlations don't vanish. And that means it's really difficult. And there's a quote from one of my former colleagues when I taught at Yale,

Sergei Roche was at Yale at the time I was there, and now back in France, he won the Nobel Prize for the work he did in measuring atomic systems of quantum mechanics. I think search and Hiroshi and Ramon said that, you know, these quantum algorithms, these quantum computers are a computer scientist's dream and an experimentalist nightmare. Maybe that's still true, although, you know, Horosch was not alone when he said that in that article with Ramon. It was, maybe 1996. So it was a couple of years after Schor's algorithm where there was a lot of interest in quantum computing that had been ignited by Peter Schor's discovery and interest shared by a growing theoretical community and also many experimentalists.

But there was skepticism, which was reasonable skepticism. about whether quantum computers could ever really be built and operated to solve really hard problems. The essential difficulty being what you were highlighting, and what Hirosh was one of the world's great experts on, what we call decoherence. This comes back to the observation we spoke of a few minutes ago, that you can't observe a qubit or a quantum system. without disturbing it in some uncontrollable way. And in the case of a quantum computer,

even if we're not looking at the content of a quantum memory ourselves, it's always interacting with the environment. And in some sense, the environment is observing the system. Information about the state of the system is leaking to the environment. Now, that can happen for a classical computer too, but it's fine. It might be that the environment, is affected differently by a bit, which is a zero, and then a bit, which is a one,

but it's still a zero or a one. But it's the case of a qubit when information about the state of the qubit leaks to the outside, the cubit is damaged. So in some sense, if we want a quantum computer to really operate without errors, we have to keep it perfectly isolated from the outside world, and that's extremely difficult, if not impossible.

So what was very important, which followed after, Shore's initial discovery of his algorithm just by a couple of years is the idea we call quantum error correction, which is a way of protecting a complex quantum system from damage. And the key is to encode the information that we want to protect in a sufficiently clever way that it's very hard for the environment to find out what the state of the quantum memory is. and so in effect it becomes very well isolated from the outside so we can manipulate it and do a reliable computation.

And that's how we expect eventually sufficiently large-scale quantum computers will operate and give answers to very hard problems that we can solve classically. They will make use of this idea of quantum error correction. Yeah, then quantum error correction is very important. I have to say as the eternal, I wasn't going to say cynic, I'm skeptic.

I remember I was skeptical. I didn't think the gravitational wave detector would ever work. And when I first heard about quantum computers, I said, well, that's nice. But it's the problems are so immense that, I mean, people talk about, and you know, back early on, people were saying, oh, we'd have quantum computers who'd do this or that quickly. And I was skeptical for this clear problem that quantum mechanical systems are beautiful for precisely the reasons that we never see them as being quantum mechanical because they, because they. Did you get did you did you lose me there for a second? Yeah, yeah, I did, but you're back. Okay, you're back. I saw you the whole time, but I could see that you were.

But the fact that quantum mechanical systems are are beautiful for the same reasons that we never see them, that this coherence is a very special property of quantum mechanical systems that are isolated from the environment. And isolating them from the environment, I always thought would just be impractical enough. And of course, as usual, I underestimated experimentalists and theorists

for being able to think of ways to get around this problem of decoherence and isolating for the environment. I want to talk about that a little bit. For a second, I want to go back, though, and ask you, I was trying to think about how, when I was reading your work about your statements about, you know, but the key fact

of a quantum mechanical system is that measuring its whole state doesn't give you the state of the part of each of the objects. And I was trying to think of a simple example. And I, and I, tell me if this simple example works. It's the simplest example, have entanglement in general.

If I, if I prepare a system of two particles with their spins opposite, so we say that they basically cancel out and the total spin of the system is zero. We can define that system completely. by saying it's a spin zero system. But then, you know, and these particles can separate,

and it's still a spin zero system if it doesn't interact the environment. But having said that, that doesn't tell us anything about the spin directions of the individual particles. So knowing that it's in a total spin zero system, which is all you need to know to describe that quantum mechanical system, doesn't tell you anything about what you're going to measure when you measure the spin directions of the individual particles. Is that a reasonable example of the simplest example, can think of that effect that you were talking about? Yeah, that's the prototypical example, the two-cubit state. But the thing I would emphasize is that in the case of qubits, you know, we have more than one way of observing them.

And what that spin singlet state that you described has the interesting property that if we look at the two spins along, say, the vertical axis, they'll be correlated in a certain way, namely they'll be opposite. but the same is true if we look along a different axis, like a horizontal axis. And that's something very different from correlations in a classical system where we had just one way of looking at them. It could be that somebody decided to prepare two coins where one is heads and one is tails and gave one of those coins to me and another one to you. And then when we uncover the coins, we would say, hey, we have opposite. You have heads. I have tails. But that's, first of all, because we didn't have the most complete description.

If we really had, we would have known who had heads and who had tails. But also, we just have that one way of looking at things. That's what makes the entanglement of qubits a lot more interesting than the correlations of bits. Yeah, absolutely. If I look at my heads, I may know in advance you have tails, even if you're an Alpha Centauri, but there's no communication between us. But, you know, in the qubit, that's fine. I may know one thing, but if you make a different measurement,

I won't know anything about that measurement of that particular cubit. If you decide to measure it not in the vertical direction, but the horizontal direction, nothing about my measurement will be able to tell you anything about the one you made. It's a bit subtle, and it often causes confusion because if this is not a mechanism for instantaneous communication between Prince Edward Island, where you are, and a distant galaxy,

it's just that the correlations have a characteristic structure that is different from correlations among bits. Yeah, the way I think of it, you know, I get asked this question a lot about quantum about quantum teleportation and why it isn't given instantaneous, why it doesn't violate Einstein or anything. And my answer is usually that we're just thinking about the system wrong. I mean, we're thinking about the system classically. We're thinking about it as if they're two separate objects, but they're not two separate objects. They're the same object. They're part of the same object. and you're making two measurements of the same object.

And so it's just, we're just thinking, as always, the quantum paradoxes occur because we're thinking classically when our old friend and in some case mentor, Sidney Coleman would have said, we've got to, it's quantum mechanics in your face. We've got to think about it quantum mechanically. And the real thing is people get hung up on the classical interpretation of quantum mechanics, and as he would say, that's backwards. The world is quantum mechanical. you shouldn't try and talk about how it interpreted in terms of this classical

cluge that happens to be something we're used to. We really should talk about the interpretation of classical mechanics in terms of the underlying theory. And so much of the weirdness of quantum mechanics, when expressed classically, is just a property of the fact that we're expressing it badly. That's right. In the case of two entangled qubits, our classical reasoning will just lead us to incorrect conclusions because that's not the way nature works. It's quantum, not classical.

Okay, well, now I want to get to the, okay, so we've outlined the problems. I want to talk about what's really happening. I mean, and so, you know, 20 years ago or not more than 20 years ago, yeah, boy, a lot more than 20 years ago, short did his stuff. And people, you know, and people have been talking about quantum computers for a long time. And then people say, yeah, but all we have is, You know, I think we have 50 cubits, 75, I don't know what the number, what the record is now. I know the one that achieved the term that I think you created, right, called quantum supremacy,

which is the point where a quantum computer can do something in a finite time that a classical computer would do in an either longer than the age of the universe or an unfeasnably long time. When that happens, when there's a calculation that quantum computer can do, that it has achieved quantum supremacy. see. That was announced, I think, a year or two ago, right, like Google or, and it involved a 53-cubit system. Am I right? Yeah, that was in 2019. 20, already three years ago, four years ago now soon. Wow. Yeah, we're 2023 now. But so there are challenges, but you can do things. So I would like you to talk about the systems that are being used to, two examples. maybe ions and maybe synoductive, the systems that are being used to try and in capture and physically become quantum computers.

And also the techniques that are being used to try and recognize that you need to do error correction, that quantum computers don't give answers. They give answers that because of errors, you have to repeat the calculation many times and see where the answer is tending rather than what the answer is. And so could you go over those those areas? Because those are the real logistical, practical aspects of what are making quantum computers useful or not useful, and the ones which will eventually determine whether they ever become all that is promised. These things take time, don't they?

Yeah, they do. You mentioned gravitational wave detection. I guess I have a skeptical nature also. And, you know, when I first came to Caltech in 1983, they were already constructing the 40-meter prototype, which was 1-100th scale of LIGO. And the idea that we'd actually be able to build the big thing and make it work, it seemed, you know, quite a reach.

A lot of smart people worked very hard and made it happen. In the case of quantum computing, Look, it's 40 years now since Feynman first suggested the idea. 26 years, well, a little longer than that. Almost 28 years and shows algorithm. Yeah. And so where are we? We're just getting to the point where quantum computers are arguably capable of doing interesting things.

What are the different hardware approaches, I think, was one of your questions. Well, one is to use individual atoms as. cubits. You know, it's interesting that around the time that Shores algorithm was discovered, it happened that for rather different reasons with different motivation. Physicists were developing the tools to manipulate single quantum systems like individual atoms. Serge Hiroshi, who you mentioned was one of the early heroes.

And that was very timely. In fact, people had developed trapped ion technology where we trap charged atoms with electromagnetic fields by the mid-1990s for the purpose of making better clocks. It turned out a lot of the technology that you need to make the world's best clock is also very relevant to quantum computing because what they had learned to do was to manipulate individual atoms with lasers. and it sounds like it would be very hard to see a little atom, but it's actually not so hard. If the qubit could be encoded in either the atom being in its lowest energy state, its ground state, or some excited state, and if you shine a laser on the atom with the right frequency,

it will either absorb and reemmit the light, so it will fluoresce and glow, and you can see a little spot of light, or it might not interact with the light at all and stay dark. and that's a way of reading out whether it's a zero or one. Of course, we need to get the qubits to interact with one another. That's the hardest part of any quantum computing technology, and we need to do that in a very well-controlled way.

And in the case of ions in a trap, you can take advantage of the fact that they vibrate, that the trap is like a potential that confines the motion of the ions, and they rock back and forth, and you can excite the vibrations of the ion in the trap in a way that allows the state of one ion to be manipulated, conditioned on the state of another ion. And that's sort of the fundamental operation in a quantum computer. It enables you to entangle two atoms. And you put together many of those two-cubit entangling gates, and you're doing a quantum computation.

reading out the ions at the end the way I said. Now, the current state of the art is, you know, there are ion traps with, say, 32 ions. They can do pretty good gates, but not really great gates. In a device with many ions, typically every time you do one of these two cubit entangling gates, you make an error about one time out of 100. Yeah. Now, that's sort of the state of the art, though, for other technologies as well. another competing technology is to use superconducting electrical circuits, circuits at very low temperature,

which conduct electricity without resistance. And in that case, the detailed physical setting is different, but it's kind of an artificial atom. It's kind of amazing, actually, because this superconducting circuit involves the collective motion of billions of pairs of electrons, but we learned how to make it behave as though it were a single atom and to manipulate it. In that case, not with visible light with lasers like we do with the ions, but with microwave light, a much, you know, lower frequency type of electromagnetic radiation. But again, we can, we can manipulate those superconducting cubits and we can entangle them. And the devices now are up to over 100 cubits. But again, the problem is that those entangling gates just aren't good enough. They have a probability

of error of about 1% every time you do a gate. And so that's a limitation on how large a computation you can do. If you try to do too many gates, you'll just get random junk. Yeah, I want to, I want to, there's a number you quoted there, but I want to, before I get there, I wanted to ask, so you talk about how you can entangle the atoms in a trap. How do you entangle the superconducting circuits? Well, interesting question. in the case of the superconducting qubits, the

key to the technology is what we call a joseous injunction. It's a kind of quantum mechanical device in a superconducting circuit. And what it crucially does is it makes the system behave non-linearly,

which means it doesn't depend, but it doesn't behave just like a light that doesn't interact with other light, but it makes potential for interactions. And these superconducting devices, people call them transmons, are coupled to a microwave resonator. This is kind of the analog of the atom being coupled to its vibrations in the trap. And so I can have two of these transmons, these superconducting devices with a microwave resonator coupling them.

and that makes it possible for quantum information to be transmitted back and forth between them. Or actually, I could say it more simply than that. In the case of the ions, the key thing is the ions are charged, so they interact with one another. Just the coulom repulsion of the ions is the key thing. And that means the different vibrational modes

will rise to normal modes that we can manipulate. And in the case of the superconducting circuits, well, they're circuits. And so, you know, you can put in ductors and capacitors. There's the electric fields can couple or the magnetic fields can couple. And that allows two transomones to talk to one another. And but you hit the key point, which is the gates, the things that basically do the entangling in ways that you determine in advance so that you can do the computation you want to have happen. the difference between classical gates and quantum mechanical gates, again, pretty clear to state. If you have a one or a zero, you know, you can have a gate that takes one to a zero or zero to one.

But in quantum mechanics, you have these things called unitary transformations, but basically you have a continuum, which is part of the reason that the quantum mechanical system is so much richer. And if you have a continuum, then if you're doing an experimental thing where you want to, let's say, of a continuum as an angle and you want to turn it by 27 degrees, well, you're going to turn it by maybe 26 or 28 degrees because all experiments have an intrinsic uncertainty. And if you can only do it to 1%, and you keep multiplying that error, as you pointed out, if you want to, if you had 100, a hundred gates, you'd have to have something like 100,000 cubits if you had a 1%, a 0.1% error, which is 10 times better than you can even try and achieve now.

You'd have 100,000 cubits before you could get a reliable, before you could overcome that error. That's a number you quotes, so I'm assuming it's right. When I spoke of 100,000 cubits, I was probably talking about using quantum error correction, which is not the story. But having more cubits if we don't do quantum error correction just makes things worse, right?

Yeah, yeah. It means more things that can fail. So you say you have an error of 1%, but let's talk about that quantum error correction because that seems to me to be the great hope. I mean, that's what, if anything, is going to overcome my original skepticism about whether these things can work.

I'm still convinced, you know, you can't keep these things coherent for a long time or maybe you can, but it's going to be hard. But if you can overcome some of these problems by error correction, then maybe my concerns are not so great. So why don't you talk for a few minutes about quantum error correction? And maybe... Well, the idea of quantum error correction is, as I noted earlier,

if we want to manipulate quantum information accurately, we have to prevent it from interacting with the environment. And our hardware isn't perfectly isolated from the environment. So what we do is we make use of entanglement. We encode the information that we want to protect in a highly entangled state. And what that means is, and you mentioned this earlier, if we look at the parts of the system one at a time, we don't see the hidden information.

We don't see the encoded information. But that's how the environment typically interacts with the system in a way that's spatially local. And we spread out the quantum information in the form of entanglement involving many qubits, and that makes it possible to protect it. And we've learned how to efficiently manipulate or process information that's encoded in that very entangled way, and that's how quantum error correction will work in the long run. But it's expensive because if our gates have 1% is just a little too high, but let's say, and this will probably happen reasonably soon,

the hardware improves to the point where the probability of error per gate is about one in the thigh, instead of one in 100, then in principle quantum air correction will work, but we'll need a lot of extra qubits to have enough redundancy to protect the information well. And that's probably where that 100,000 number came from. If I wanted to run Shores algorithm to factor a number which is cryptographically relevant, like to break codes that people use today to protect their privacy, that would require a few thousand protected qubits and the number of physical qubits that we would need is in the tens of millions if the error rate per gate is 10 to the minus three. So that's a long way from where we are now because where now is more at the level of 100 cubits and we're probably going to want to have millions and that's going to take some time. In the meantime, though, the quantum computing technology, I think, is already signed.

scientifically interesting because it does give us an opportunity to study the behavior of many highly entangled quantum particles in a way that's never been experimentally accessible before. And I anticipate that we'll be learning things about quantum dynamics from experiments with quantum computers and quantum simulators in the next few years. And arguably, that's starting to happen already. So for me, as a physicist, who's interested in understanding nature better, I think quantum computing has reached a very interesting stage. In terms of economic impact, applications of broad interest in the business community,

those are probably still considerably further off because this might be wrong, but as far as we can currently tell, we'll probably need quantum error correction for that. And that's a big leap from the current state of the technology. We'll get there. But it's going to take time. This is perfect because it segues to the last thing I want to talk about.

But what you're talking about, I don't know if you invented this term, NISQ, which basically says these systems are workable enough now to do interesting things. What does NISQ stand for? Right. I pronounce it NISC as though the Q were okay. Yeah, yeah. It's an acronym. It stands for noisy intermediate scale.

quantum. And what intermediate scale means is that we now have devices with, say, of order 100 cubits or 50 to 100 cubits, which are of a scale that we can by brute force simulate with an ordinary computer exactly what the quantum device is doing. It's just too hard. But noisy reminds us that these are not error-corrected devices, and the noise is a limitation on their computational power, a limitation on how many gates we can do. and still read out a useful result. So in this experiment that you mentioned that was announced with some fanfare by Google in late 2019, they had a 53-cubit device, and they had a gate error rate around 1%.

And they performed computations with hundreds of intergates. My dog isn't a huge fan of air gates. Go on. What's your dog's name? Well, we have two of them. Levi is my dog. And then my inherited my mother's dog, her name is Tasha. And they, once one starts to bark the other, can't help it. Yes, Jewish dog. A Jewish dog, Levi, exactly. Yeah, that's right. Okay, good. Yeah, so in that case, because there were hundreds of entangling gates, when they, when they read out the result in the end, about one time in 500, they were able to get a valid result. And 499 times out of 500, they just got random junk. So they had to repeat the computation millions of times.

And then they were able to extract a statistically useful signal. So that's kind of the state of the art now. But the point is, but the point is if you can do it, if it takes a few seconds or minutes, you can repeat it thousands or millions of times. And that's okay if the classical calculation would take thousands or millions of years. You don't mind repeating a quantum calculation a thousand times or a million times if you're doing many of them per second, I guess. Yeah, well, that's right. Right. And in fact, to do that, do millions of repetitions did take only a few minutes.

An interesting thing that happened is that the classical team was inspired by this experiment and other related ones to come up with better methods for simulating what computer was. Those methods are better now than they were in 2019. So the gap between how long it would take for a classical computer to imitate what the quantum computer is, what the quantum computer does for this particular experiment has narrowed a lot. But the essential point is that as you increase the number of cubits, the difficulty of doing that classical simulation, the number of steps that it takes grows exponentially with the number of cubits. So if the gap is still not so wide for 53 cubits,

if we go up to say 70 cubits, the quantum computer will really be left in the, well, far ahead of the classical. Yeah, that's the thing I was going to get. It's a really important is that if you have 53 cubits. qubits and it takes a minute time, 100 cubits, it's not going to just double things. It's, it's exponential. And so, yeah, if you're, if they're close to each other now, you know, that's the great hope of quantum computers is you just, because it improves exponentially, as you get a few more, you'll be away from the domain where classical computers could compete. But it is an interesting fact.

You're absolutely right. Within a few days, and that does bring, I can't help, but when I think of the object lessons that comes from, from our talk, we've mentioned a few for graduate students. But the other thing that is interesting, I've found in my career as a scientist. In my own career, I've written papers that I could have written 10 years earlier, but until the experiment was done,

I guess I never took it seriously enough to think about the details. I mean, and it's amazing how actually doing something inspires people to think about the implications a lot more carefully than they would have. You know, Gunnuckin experiments work to some extent, but really, having the data is an inspiration for theorists that you might not imagine it would be.

You know, so sometimes people say the theorists take their ideas too seriously, but when I was a student, Steve Weinberg told me he thought the opposite was true, that the theorists don't take their ideas seriously enough? Because it's just so hard to believe that the scribbles we make on a piece of paper are really going to correspond to the way nature behaves on the mental life. It is intimidating, it's intimidating and almost frightening to think when you're doing something that nature may actually behave that way. It is a really, it's exhilarating if it's right, but it's terrifying.

It's hard to believe you're absolutely true. And we don't. And yes, on my own career, and I think all of us, you know, we, oh, well. And, you know, often I've written papers saying, well, you know, I remember there were, or haven't written papers saying, well, this would be. interesting, but they'll never be able to do this. So I won't write down the paper. Well, that's a mistake because experimentalists are actually hardworking and can do

amazing things. So speaking of that, and we're at the point where now we've got sort of this, we're able to do useful stuff. And you talk about, it's true. And I think it's really important to point out that, you know, I've been at meetings with recently a meeting of crypto people talking about they need new quantum error correction codes that, you know, will preserve the world's banking systems. The world's banking systems are not threatened immediately. And unlikely in the near future, would you agree? Well, yes and no.

I think it's true, which I think is what you're suggesting, that we will not, most likely, have quantum computers that are capable of breaking the crypto systems that we currently use in the next, say, 20 years. But on the other hand, it's urgent to replace the cryptosystems, which will eventually become vulnerable to attack by quantum computers with new crypto systems, which we think are resistant to quantum attacks, because first of all, it takes a long time to implement that new key infrastructure. And secondly,

you want a crypto system that you're using to keep information secure for some time after you begin using it. You know, it's always possible for an adversary to capture traffic, which can't be encrypted

right away, but when technology is more advanced, later on, we'll be able to encrypt it. So you have to ask yourself, not just how long will it be before people can run things like Shores algorithm, but also how long will it take to implement alternatives and how long do we want to protect the information? I don't think anybody can promise you that we won't have quantum computers that break the crypto system we're using today in 25 years.

So, you know, it's time to start worrying about it. It's trying to start worrying about it. One should also add that quantum mechanics is double-edged sword. It brings not just the threat, but solutions as well. Quantum mechanics, by the very same kinds of entanglement, allows one to actually know whether messages that have been sent have been eavesdropped on. And so one may use similar technology that one uses to build quantum computers to build technologies to ensure that one's protected from eavesdropping or disturbances.

So it's a nice kind of complementarity there in quantum mechanics. It's another manifestation of what we said earlier, about quantum information having the feature that you can't observe it without disturbing it in some detectable way. And that's the fundamental physics principle that makes quantum cryptography, well, in principle, in principle. In principle. In principle. So in the future, to come to the end now, the two things.

One is scalability. And, I mean, you've talked about these technologies. and I've been involved at various times in proposing experiments to look for things like dark matter, which we're fine when you have gram-sized detectors, but if you need 10-ton detectors, it's a different story. Are the technologies that you've discussed scalable, or will one need new technologies in order to scale up to go from 50 to 5,000 qubits? Well, it certainly needs some new technologies. I think we're not at the stage, where, you know, we can hand it over to the engineers,

and I think we still have a lot of basic science to do. I think as of right now, there are really two fundamental questions about the future of quantum computing, both of which I regard as largely open. One is, how are we going to use these things? What will be the most important applications for powerful quantum computing technology when we have it? And the other is how are we going to scale up from a relatively small?

small quantum computers we have now to much bigger ones, which are capable of solving very hard problems. And we don't know for sure what the answer is to either one of those questions. Okay, great. I mean, not knowing, as I've, my whole new book, my most recent book is about the importance of not knowing, at least recognizing that one doesn't know because it gives one an invitation to learn. The, the, and I don't like asking people to make predictions, because people always ask me to make predictions and I say I don't, not unless they're 10 trillion years in the future that I'm happy to predict. But so rather than saying we're asking you where quantum computing will be 25 or 50 years, because who the heck knows, let me ask you, what's,

what excites you the most for the next 10 or 20 years? What, what, what, what are your, what are you, what do you expect and what do you hope for? Well, one thing I expect on a shorter time scale than you mentioned, is that we will see significant progress towards realizing quantum error correction, not at the scale that we'll eventually need, but, you know, up to now, we haven't reached the milestone of showing that if we use a quantum error correcting code, we can make a computation much more reliable and continue to do so as we scale up to larger and larger codes, theoretically, but we believe that's the case. But we'd really like to see that demonstrated in hardware because the theory is based on certain assumptions about the noise, which we won't really

be able to validate until we try it in different hardware platforms. I think we'll see a lot of progress on that and on a five-year time scale. As far as the applications to physics are concerned, I think we're going to learn things on a scale of five to ten years about how quantum chaos works about what happens when a quantum system has many strongly interacting particles. You know, there was kind of a revolution in classical physics back in the 60s and 70s when people started to simulate using their conventional computers, the behavior of classical nonlinear dynamical systems. And that led to a lot of insights into the types of chaos that can arise. We know relatively little about chaos in a quantum setting because we can't simulate those systems with

our conventional computers, and that's part of what our quantum devices will be capable of. And even relatively noisy ones are going to teach us interesting things. In the longer term, I think we can anticipate applications of quantum computing to the problems that Feynman originally had in mind to understanding chemistry and materials more deeply. We have to keep in mind that the classical algorithms are not so bad, even though they don't scale well. And they'll continue to improve. In fact, they have improved a lot in the last 10 years. So exactly when we'll see quantum computers surpassing our best conventional computers,

running the best algorithms for problems in chemistry and materials. Well, we don't really know that. It will happen eventually. And I think that will be one of the ways in which quantum computing will eventually have a big practical impact on the world. If you had to ask one question, a quantum computer and get an answer, what question would it be? Well, you know, this reminds me of something that you asked me when you were writing, you

know, the sequel to the physics of Star Trek. Yeah. I was supposed to ask not a quantum computer, but some all-knowing Oracle. a question and I asked it is physics and environmental science? In other words, are the laws of physics really determined or are there a roll of the dice? Anyway, so I don't think since then have you asked me a similarly profound question? Well, the answer that comes to mind is ever since I first got into the subject in the 1990, I have been interested in what quantum computing might teach us about quantum gravity.

And we touched on that briefly before. But one way in which ideas from quantum information have had a significant impact on our understanding of fundamental physics is the people who do quantum gravity for a living, the people with string theory backgrounds and so on, They use a rather different language now than they did 10 years ago. If you go to a conference, you'll hear people talking about quantum error correction and quantum complexity and so on in the context of quantum gravity. And part of what's driven that is the realization that we can think of space itself as an emergent property in which the underlying mechanism is quantum entanglement. In a sense, what's holding space together is quantum entanglement. And what I would like to see when I'm still around to enjoy it is insights into how space can emerge from a highly entangled system coming from simulations run on quantum computers.

Great. Well, you know, I was going to ask you a more leading question. I was hoping you'd go there. So I'm glad that's where you went. But in a way, this talk and our discussion has been full of poetry because you're really a poet, even though you don't know it. the poetry of many aspects of your life. And it's kind of interesting. You mentioned that question I asked you a long time ago,

but whether physics is an environmental science. In some sense, since string theory or since quantum gravity may, one of the things quantum gravity may tell us is that physics is an environmental science, in the sense it may tell us that the universe we live in is an accident of our circumstances. It may answer that same question that, you want it to know way back then. So I think it's kind of poetic in a way that maybe in the long run, when we learn about whether space is an emergent property,

we may learn about whether our space, whether it can emerge in many different ways. And that would be interesting. And so I certainly hope you're around to keep asking profound questions and keep pushing an important field. And also, I hope you continue to always try, your graduate predilection

of understanding everything because it helps to understand something in your case and I really do appreciate you're taking this incredibly generous time to talk to me

about your time as a physicist and about this incredibly new field which where a lot is said and one often has to parse it carefully to see what's accurate and it's nice to go to the horse's mouth not saying that you're a horse but nevertheless

I appreciate it. It's been a lot of fun, Lawrence. I hope you enjoyed today's conversation. This podcast is produced by the Origins Project Foundation, a non-profit organization whose goal is to enrich your perspective of your place in the cosmos by providing access to the people who are driving the future of society in the 21st century and to the ideas that are changing our understanding of ourselves and our world. To learn more, please visit OriginsprojectFoundation.org