A quantum podcast

A few months ago I sat down with Craig Cannon of Y Combinator for a discussion about quantum technology and other things. A lightly edited version was published this week on the Y Combinator blog. The video is also on YouTube:

If you’re in a hurry, or can’t stand the sound of my voice, you might prefer to read the transcript, which is appended below. Only by watching the video, however, can you follow the waving of my hands.

I grabbed the transcript from the Y Combinator blog post, so you can read it there if you prefer, but I’ve corrected some of the typos. (There are a few references to questions and comments that were edited out, but that shouldn’t cause too much confusion.)

Here we go:

Craig Cannon [00:00:00] – Hey, how’s it going? This is Craig Cannon, and you’re listening to Y Combinator’s Podcast. Today’s episode is with John Preskill. John’s a theoretical physicist and the Richard P. Feynman Professor of Theoretical Physics at Caltech. He once won a bet with Stephen Hawking and he writes that it made him briefly almost famous. Basically, what happened is John and Kip Thorne bet that singularities could exist outside of black holes. After six years, Hawking conceded. He said that they were possible in very special, “non-generic conditions.” I’ll link up some more details to that in the description. In this episode, we cover what John’s been focusing on for years, which is quantum information, quantum computing, and quantum error correction. Alright, here we go. What was the revelation that made scientists and physicists think that a quantum computer could exist?

John Preskill [00:00:54] – It’s not obvious. A lot of people thought it couldn’t. The idea that a quantum computer would be powerful was emphasized over 30 years ago by Richard Feynman, the Caltech physicist. It was interesting how he came to that realization. Feynman was interested in computation his whole life. He had been involved during the war in Los Alamos. He was the head of the computation group. He was the guy who fixed the little mechanical calculators, and he had a whole crew of people who were calculating, and he figured out how to flow the work from one computer to another. All that kind of stuff. As computing technology started to evolve, he followed that. In the 1970s, a particle physicist like Feynman, that’s my background too, got really interested in using computers to study the properties of elementary particles like the quarks inside a nucleus, you know? We know a proton isn’t really a fundamental object. It’s got little beans rattling around inside, but they’re quantum beans. Gell-Mann, who’s good at names, called them quarks.

John Preskill [00:02:17] – Now we’ve had a theory since the 1970s of how quarks behave, and so in principle, you know everything about the theory, you can compute everything, but you can’t because it’s just too hard. People started to simulate that physics with digital computers in the ’70s, and there were some things that they could successfully compute, and some things they couldn’t because it was just too hard. The resources required, the memory, the time were out of reach. Feynman, in the early ’80s said nature is quantum mechanical damn it, so if you want a simulation of nature, it should be quantum mechanical. You should use a quantum system to behave like another quantum system. At the time, he called it a universal quantum simulator.

John Preskill [00:03:02] – Now we call it a quantum computer. The idea caught on about 10 years later when Peter Shor made the suggestion that we could solve problems which don’t seem to have anything to do with physics, which are really things about numbers like finding the prime factors of a big integer. That caused a lot of excitement,  in part because the implications for cryptography are a big disturbing. But then physicists — good physicists — started to consider, can we really build this thing? Some concluded and argued fairly cogently that no, you couldn’t because of this difficulty that it’s so hard to isolate systems from the environment well enough for them to behave quantumly. It took a few years for that to sort out at the theoretical level. In the mid ’90s we developed a theory called quantum error correction. It’s about how to encode the quantum state that you’d like to protect in such a clever way that even if there are some interactions with the environment that you can’t control, it still stays robust.

John Preskill [00:04:17] – At first, that was just kind of a theorist’s fantasy — it was a little too far ahead of the technology. But 20 years later, the technology is catching up, and now this idea of quantum error correction has become something you can do in the lab.

Craig Cannon [00:04:31] – How does quantum error correction work? I’ve seen a bunch of diagrams, so maybe this is difficult to explain, but how would you explain it?

John Preskill [00:04:39] – Well, I would explain it this way. I don’t think I’ve said the word entanglement yet, have I?

Craig Cannon [00:04:43] – Well, I have been checking off all the Bingo words yet.

John Preskill [00:04:45] – Okay, so let’s talk about entanglement because it’s part of the answer to your question, which I’m still not done answering, what is quantum physics? What do we mean by entanglement? It’s really the characteristic way, maybe the most important way that we know in which quantum is different from ordinary stuff, from classical. Now what does it mean, entanglement? It means that you can have a physical system which has many parts, which have interacted with one another, so it’s in kind of a complex correlated state of all those parts, and when you look at the parts one at a time it doesn’t tell you anything about the state of the whole thing. The whole thing’s in some definite state — there’s information stored in it — and now you’d like to access that information … Let me be a little more concrete. Suppose it’s a book.

John Preskill [00:05:40] – Okay? It’s a book, it’s 100 pages long. If it’s an ordinary book, 100 people could each take a page, and read it, they know what’s on that page, and then they could get together and talk, and now they’d know everything that’s in the book, right? But if it’s a quantum book written in qubits where these pages are very highly entangled, there’s still a lot of information in the book, but you can’t read it the way I just described. You can look at the pages one at a time, but a single page when you look at it just gives you random gibberish. It doesn’t reveal anything about the content of the book. Why is that? There’s information in the book, but it’s not stored in the individual pages. It’s encoded almost entirely in how those pages are correlated with one another. That’s what we mean by quantum entanglement: Information stored in those correlations which you can’t see when you look at the parts one at a time. You asked about quantum error correction?

John Preskill [00:06:39] – What’s the basic idea? It’s to take advantage of that property of entanglement. Because let’s say you have a system of many particles. The environment is kind of kicking them around, it’s interacting with them. You can’t really completely turn off those interactions no matter how hard you try, but suppose we’ve encoded the information in entanglement. So, say, if you look at one atom, it’s not telling you anything about the information you’re trying to protect. The environment isn’t learning anything when it looks at the atoms one at a time.

John Preskill [00:07:15] – This is kind of the key thing — that what makes quantum information so fragile is that when you look at it, you disturb it. This ordinary water bottle isn’t like that. Let’s say we knew it was either here or here, and we didn’t know. I would look at it, I’d find out it’s here. I was ignorant of where it was to start with, and now I know. With a quantum system, when you look at it, you really change the state. There’s no way to avoid that. So if the environment is looking at it in the sense that information is leaking out to the environment, that’s going to mess it up. We have to encode the information so the environment, so to speak, can’t find out anything about what the information is, and that’s the idea of quantum error correction. If we encode it in entanglement, the environment is looking at the parts one at a time, but it doesn’t find out what the protected information is.

Craig Cannon [00:08:06] – In other words, it’s kind of measuring probability the whole way along, right?

John Preskill [00:08:12] – I’m not sure what you mean by that.

Craig Cannon [00:08:15] – Is it Grover’s algorithm that was as quantum bits roll through, go through gates– The probability is determined of what information’s being passed through? What’s being computed?

John Preskill [00:08:30] – Grover’s algorithm is a way of sort of doing an exhaustive search through many possibilities. Let’s say I’m trying to solve some problem like a famous one is the traveling salesman problem. I’ve told you what the distances are between all the pairs of cities, and now I want to find the shortest route I can that visits them all. That’s a really hard problem. It’s still hard for a quantum computer, but not quite as hard because there’s a way of solving it, which is to try all the different routes, and measure how long they are, and then find the one that’s shortest, and you’ve solved the problem. The reason it’s so hard to solve is there’s such a vast number of possible routes. Now what Grover’s algorithm does is it speeds up that exhaustive search.

John Preskill [00:09:29] – In practice, it’s not that big a deal. What it means is that if you had the same processing speed, you can handle about twice as many cities before the problem becomes too hard to solve, as you could if you were using a classical processor. As far as what’s quantum about Grover, it takes advantage of the property in quantum physics that probabilities … tell me if I’m getting too inside baseball …

Craig Cannon [00:10:03] – No, no, this is perfect.

John Preskill [00:10:05] – That probabilities are the squares of amplitudes. This is interference. Again, this is another part of the answer. Well, we can spend the whole hour answering the question, what is quantum physics? Another essential part of it is what we call interference, and this is really crucial for understanding how quantum computing works. That is that probabilities add. If you know the probability of one alternative, and you know the probability of another, then you can add those together and find the probability that one or the other occurred. It’s not like that in quantum physics. The famous example is the double slit interference experiment. I’m sending electrons, let’s say — it could be basketballs, but it’s an easier experiment to do with electrons —

John Preskill [00:11:02] – at a screen, and there are two holes in the screen. You can try to detect the electron on the other side of the screen, and when you do that experiment many times, you can plot a graph showing where the electron was detected in each run, or make a histogram of all the different outcomes. And the graph wiggles, okay? If you could say there’s some probability of going through the first hole, and some probability of going through the second, and each time you detected it, it went through either one or the other, there’d be no wiggles in that graph. It’s the interference that makes it wiggle. The essence of the interference is that nobody can tell you whether it went through the first slit or the second slit. The question is sort of inadmissible. This interference then occurs when we can add up these different alternatives in a way which is different from what we’re used to. It’s not right to say that the electron was detected at this point because it had some probability of going through the first hole, and some probability of going through the second

John Preskill [00:12:23] – and we add those probabilities up. That doesn’t give the right answer. The different alternatives can interfere. This is really important for quantum computing because what we’re trying to do is enhance the probability or the time it takes to find the solution to a problem, and this interference can work to our advantage. We want to have, when we’re doing our search, we want to have a higher chance of getting the right answer, and a lower chance of getting the wrong answer. If the different wrong answers can interfere, they can cancel one another out, and that enhances the probability of getting the right answer. Sorry it’s such a long-winded answer, but this is how Grover’s algorithm works.

John Preskill [00:13:17] – It can speed up exhaustive search by taking advantage of that interference phenomenon.

Craig Cannon [00:13:20] – Well this is kind of one of the underlying questions among many of the questions from Twitter. You’ve hit our record for most questions asked. Basically, many people are wondering what quantum computers really will do if and when it becomes a reality that they outperform classical computers. What are they going to be really good at?

John Preskill [00:13:44] – Well, you know what? I’m not really sure. If you look at the history of technology, it would be hubris to expect me to know. It’s a whole different way of dealing with information. Quantum information is not just … a quantum computer is not just a faster way of computing. It deals with information in a completely new way because of this interference phenomenon, because of entanglement that we’ve talked about. We have limited vision when it comes to predicting decades out what the impact will be of an entirely new way of doing things. Information processing, in particular. I mean you know this well. We go back to the 1960s, and people are starting to put a few transistors on a chip. Where is that going to lead? Nobody knew.

Craig Cannon [00:14:44] – Even early days of the internet.

John Preskill [00:14:45] – Yeah, good example.

Craig Cannon [00:14:46] – Even the first browser. No one really knew what anyone was going to do with it. It makes total sense.

John Preskill [00:14:52] – For good or ill. Yeah. But we have some ideas, you know? I think … why are we confident there will be some transformative effect on society? Of the things we know about, and I emphasize again, probably the most important ones are things we haven’t thought of when it comes to applications of quantum computing, the ones which will affect everyday life, I think, are better methods for understanding and inventing new materials, new chemical compounds. Things like that can be really important. If you find a better way of capturing carbon by designing a better catalyst, or you can design pharmaceuticals that have new effects, materials that have unusual properties. These are quantum physics problems because those properties of the molecule or the material really have to do with the underlying quantum behavior of the particles, and we don’t have a good way for solving such problems or predicting that behavior using ordinary digital computers. That’s what a quantum computer is good at. It’s good — but maybe not the only thing it’s good at — one thing it should certainly be good at is telling us quantitatively how quantum systems behave. In the two contexts I just mentioned, there’s little question that there will be practical impact of that.

Craig Cannon [00:16:37] – It’s not just doing the traveling salesman problem through the table of elements for why it can find these compounds.

John Preskill [00:16:49] – No. If it were, that wouldn’t be very efficient.

Craig Cannon [00:16:52] – Exactly.

John Preskill [00:16:53] – Yeah. No, it’s much trickier than that. Like I said, the exhaustive search, though conceptually it’s really interesting that quantum can speed it up because of interference, from a practical point of view it may not be that big a deal. It means that, well like I said, in the same amount of time you can solve an instance which is twice as big of the problem. What we really get excited about are the so-called exponential speed ups. That was why Shor’s algorithm was exciting in 1994, because factoring large numbers was a problem that had been studied by smart people for a long time, and on that basis, the fact that there weren’t any fast ways of solving it was pretty good evidence it’s a hard problem. Actually, we don’t know how to prove that from first principles. Maybe somebody will come along one day and figure out how to solve factoring very fast on a digital computer. It doesn’t seem very likely because people have been trying for so long to solve problems like that, and it’s just intractable with ordinary computers. You could say the same thing about these quantum physics problems. Maybe some brilliant graduate student is going to drop a paper on the arXiv tomorrow which will say, “Here, I solved quantum chemistry, and I can do it on a digital computer.” But we don’t think that’s very likely because we’ve been working pretty hard on these problems for decades and they seem to be really hard. Those cases, like these number theoretic problems,

John Preskill [00:18:40] – which have cryptological implications, and tasks for simulating the behavior of quantum systems, we’re pretty sure those are hard problems classically, and we’re pretty sure quantum computers … I mean we have algorithms that have been proposed, but which we can’t really run currently because our quantum computers aren’t big enough on the scale that’s needed to solve problems people really care about.

Craig Cannon [00:19:09] – Maybe we should jump to one of the questions from Twitter which is related to that. Travis Scholten (@Travis_Sch) asked, what are the most problem pressings in physics, let’s say specifically around quantum computers that you think substantial progress ought to be made in to move the field forward?

John Preskill [00:19:27] – I know Travis. He was an undergrad here. How you doing, Travis? The problems that we need to solve to make quantum computing closer to realization at the level that would solve problems people care about? Well, let’s go over where we are now.

Craig Cannon [00:19:50] – Yeah, definitely.

John Preskill [00:19:51] – People have been working on quantum hardware for 20 years, working hard, and there are a number of different approaches to building the hardware, and nobody really knows which is going to be the best. I think we’re far from collapsing to one approach which everybody agrees has the best long-term prospects for scalability. And so it’s important that a lot of different types of hardware are being pursued. We can come back to what some of the different approaches are later. Where are we now? We think in a couple of years we’ll have devices with about 50 qubits to 100, and we’ll be able to control them pretty well. That’s an interesting range because even though it’s only 50 to 100 qubits, doesn’t sound like that big a deal, but that’s already too many to simulate with a digital computer, even with the most powerful supercomputers today. From that point of view, these relatively small, near-term quantum computers which we’ll be fooling around with over the next five years or so, are doing something that’s kind of super-classical.

John Preskill [00:21:14] – At least, we don’t know how to do exactly the same things with ordinary computers. Now that doesn’t mean they’ll be able to do anything that’s practically important, but we’re going to try. We’re going to try, and there are ideas about things we’ll try out, including baby versions of these problems in chemistry, and materials, and ways of speeding up optimization problems. Nobody knows how well those things are going to work at these small scales. Part of the reason is not just the number of qubits is small, but they’re also not perfect. We can perform elementary operations on pairs of qubits, which we call quantum gates like the gates in ordinary logic. But they have an error rate a little bit below an error every 100 gates. If you have a circuit with 1000 qubits, that’s a lot of noise.

Craig Cannon [00:22:18] – Exactly. Does for instance, 100-qubit quantum computer really mean 100-qubit quantum computer or do you need a certain amount of backup going on?

John Preskill [00:22:29] – In the near term, we’re going to be trying out, and probably we have the best hopes for, kind of hybrid classical-quantum methods with some kind of classical feedback. You try to do something on the quantum computer, you make a measurement that gives you some information, then you change the way you did it a little bit, and try to converge on some better answer. That’s one possible way of addressing optimization that might be faster on a quantum computer. But I just wanted to emphasize that the number of qubits isn’t the only metric. How good they are, and in particular, the reliability of the gates, how well we can perform them … that’s equally important. Anyway, coming back to Travis’ question, there are lots of things that we’d like to be able to do better. But just having much better qubits would be huge, right? If you … more or less, with the technology we have now, you can have a gate error rate of a few parts in 1,000, you know? If you can improve that by orders of magnitude, then obviously, you could run bigger circuits. That would be very enabling.

John Preskill [00:23:58] – Even if you stick with 100 qubits just by having a circuit with more depth, more layers of gates, that increases the range of what you could do. That’s always going to be important. Because, I mean look at how crappy that is. A gate error rate, even if it’s one part in 1,000, that’s pretty lousy compared to if you look at where–

Craig Cannon [00:24:21] – Your phone has a billion transistors in it. Something like that, and 0%–

John Preskill [00:24:27] – You don’t worry about the … it’s gotten to the point where there is some error protection built in at the hardware level in a processor, because I mean, we’re doing these crazy things like going down from the 11 nanometer scale for features on a chip.

Craig Cannon [00:24:45] – How are folks trying to deal with interference right now?

John Preskill [00:24:50] – You mean, what types of devices? Yeah, so that’s interesting too because there are a range of different ways to do it. I mentioned that we could store information, we could make a qubit out of a single atom, for example. That’s one approach. You have to control a whole bunch of atoms and get them to interact with one another. One way of doing that is with what we call trapped ions. That means the atoms have electrical charges. That’s a good thing because then you could control them with electric fields. You could hold them in a trap, and you can isolate them, like I said, in a very high vacuum so they’re not interacting too much with other things in the laboratory, including stray electric and magnetic fields. But that’s not enough because you got to get them to talk to one another. You got to get them to interact. We have this set of desiderata, which are kind of in tension with one another. On the one hand, we want to isolate the qubits very well. On the other hand, we want to control them from the outside and get them to do what we want them to do, and eventually, we want to read them out. You have to be able to read out the result of the computation. But the key thing is the control. You could have two of those qubits in your device interact with one another in a specified way, and to do that very accurately you have to have some kind of bus that gets the two to talk to one another.

John Preskill [00:26:23] – The way they do that in an ion trap is pretty interesting. It’s by using lasers and controlling how the ions vibrate in the trap, and with a laser, kind of excite, wiggles of the ion, and then by determining whether the ions are wiggling or not, you can go address another ion, and that way you can do a two-qubit interaction. You can do that pretty well. Another way is really completely different. What I just described was encoding information at the one atom level. But another way is to use superconductivity — circuits in which electric current flows without any dissipation. In that case, you have a lot of freedom to sort of engineer the circuits to behave in a quantum way. There are many nuances there, but the key thing is that you can encode information now in a system that might involve the collective motion of billions of electrons, and yet you can control it as though it were a single atom. I mean, here’s one oversimplified way of thinking about it.

John Preskill [00:27:42] – Suppose you have a little loop of wire, and there’s current flowing in the loop. It’s a superconducting wire so it just keeps flowing. Normally, there’d be resistance, which would dissipate that as heat, but not for the superconducting circuit, which of course, has to be kept very cold so it stays superconducting. But you can imagine in this little loop that the current is either circulating clockwise or counterclockwise. That’s a way of encoding information. It could also be both at once, and that’s what makes it a qubit.

Craig Cannon [00:28:14] – Right.

John Preskill [00:28:15] – And so in that case, even though it involves lots of particles, the magic is that you can control that system extremely well. I mentioned individual electrons. That’s another approach. Put the qubit in the spin of a single electron.

Craig Cannon [00:28:32] – You also mentioned better qubits. What did you mean by that?

John Preskill [00:28:35] – Well, what I really care about is how well I can do the gates. There’s a whole other approach, which is motivated by the desire to have much, much better control over the quantum information than we do in those systems that I mentioned so far, superconducting circuits and trapped ions. That’s actually what Microsoft is pushing very hard. We call it topological quantum computing. Topological is a word physicists and mathematicians love. It means, well, we’ll come back to what it means. Anyway, let me just tell you what they’re trying to do. They’re trying to make a much, much better qubit, which they can control much, much better using a completely different hardware approach.

Craig Cannon [00:29:30] – Okay.

John Preskill [00:29:32] – It’s very ambitious because at this point, it’s not even clear they have a single qubit, but if that approach is successful, and it’s making progress, we will see a validated qubit of this type soon. Maybe next year. Nobody really knows where it goes from there, but suppose it’s the case that you could do a two-qubit gate with an error rate of one in a million instead of one in 1,000. That would be huge. Now, scaling all these technologies up, is really challenging from a number of perspectives, including just the control engineering.

Craig Cannon [00:30:17] – How are they doing it or attempting to do it?

John Preskill [00:30:21] – You know, you could ask, where did all this progress come from over 20 years, or so? For example, with the superconducting circuits, a sort of crucial measure is what we call the coherence time of the qubit, which roughly speaking, means how much it interacts with the outside world. The longer the coherence time, the better. The rate of what we call decoherence is essentially how much it’s getting buffeted around by outside influences. For the superconducting circuits, those coherence times have increased about a factor of 10 every three years, going back 15 years or so.

Craig Cannon [00:31:06] – Wow.

John Preskill [00:31:07] – Now, it won’t necessarily go on like that indefinitely, but in order to achieve that type of progress, better materials, better fabrication, better control. The way you control these things is with microwave circuitry. Not that different from the kind of things that are going on in communication devices. All those things are important, but going forward, the control is really the critical thing. Coherence times are already getting pretty long, I mean having them longer is certainly good. But the key thing is to get two qubits to interact just the way you want them to. Even if there is, now I keep saying the key thing is the environment, it’s not the only key thing, right? Because you have some qubit, like if you think about that electron spin, one way of saying it is I said it can be both up and down at the same time. Well, there’s a simpler way of saying that. It might not point either up or down. It might point some other way. But there really are a continuum of ways it could point. That’s not like a bit. See, it’s much easier to stabilize a bit because it’s got two states.

John Preskill [00:32:31] – But if it can kind of wander around in the space of possible configurations for a qubit, that makes it much harder to control. People have gotten better at that, a lot better at that in the last few years.

Craig Cannon [00:32:44] – Interesting. Joshua Harmon asked, what engineering strategy for quantum computers do you think has the most promise?

John Preskill [00:32:53] – Yeah, so I mentioned some of these different approaches, and I guess I’ll interpret the question as, which one is the winning horse? I know better than to answer that question! They’re all interesting. For the near term, the most advanced are superconducting circuits and trapped ions, which is why I mentioned those first. I think that will remain true over the next five to 10 years. Other technologies have the potential — like these topologically protected qubits — to surpass those, but it’s not going to happen real soon. I kind of like superconducting circuits because there’s so much phase space of things you can do with them. Of ways you can engineer and configure them, and imagine scaling them up.

John Preskill [00:33:54] – They have the advantage of being faster. The cycle time, time to do a gate, is faster than with the trapped ions. Just the basic physics of the interactions is different. In the long term, those electron spins could catapult ahead of these other things. That’s something that you can naturally do in silicon, and it’s potentially easy to integrate with silicon technology. Right now, the qubits and gates aren’t as good as the other technologies, but that can change. I mean, from a theorist’s perspective, this topological approach is very appealing. We can imagine it takes off maybe 10 years from now and it becomes the leader. I think it’s important to emphasize we don’t really know what’s going to scale the best.

Craig Cannon [00:34:50] – Right. And are there multiple attempts being made around programming quantum computers?

John Preskill [00:34:55] – Yeah. I mean, some of these companies– That are working on quantum technology now, which includes well-known big players like IBM, and Google, and Microsoft and Intel, but also a lot of startups now. They are trying to encompass the full stack, so they’re interested in the hardware, and the fabrication, and the control technology. But also, the software, the applications, the user interface. All those things are certainly going to be important eventually.

Craig Cannon [00:35:38] – Yeah, they’re pushing it almost to like an AWS layer. Where you interact with your quantum computer in a server farm and you don’t even touch it.

John Preskill [00:35:49] – That’s how it will be in the near term. You’re not going to have, most of us won’t, have a quantum computer sitting on your desktop, or in your pocket. Maybe someday. In the near term, it’ll be in the Cloud, and you’ll be able to run applications on it by some kind of web interface. Ideally, that should be designed so the user doesn’t have to know anything about quantum physics in order to program or use it, and I think that’s part of what some of these companies are moving toward.

Craig Cannon [00:36:24] – Do you think it will get to the level where it’s in your pocket? How do you deal with that when you’re below one kelvin?

John Preskill [00:36:32] – Well, if it’s in your pocket, it probably won’t be one kelvin.

Craig Cannon [00:36:35] – Yeah, probably not.

John Preskill [00:36:38] – What do you do? Well, there’s one approach, as an example, which I guess I mentioned in passing before, where maybe it doesn’t have to be at such low temperature, and that’s nuclear spins. Because they’re very weakly interacting with the outside world, you can have quantum information in a nuclear spin, which — I’m not saying that it would be undisturbed for years, but seconds, which is pretty good. And you can imagine that getting significantly longer. Someday you might have a little quantum smart card in your pocket. The nice thing about that particular technology is you could do it at room temperature. Still have long coherence times. If you go to the ATM and you’re worried that there’s a rogue bank that’s going to steal your information, one solution to that problem — I’m not saying there aren’t other solutions — is to have a quantum card where the bank will be able to authenticate it without being able to forge it.

Craig Cannon [00:37:54] – We should talk about the security element. Kevin Su asked what risk would quantum computers pose to current encryption schemes? So public key, and what changes should people be thinking about if quantum computers come in the next five years, 10 years?

John Preskill [00:38:12] – Yeah. Quantum computers threaten those systems that are in widespread use. Whenever you’re using a web browser and you see that little padlock and you’re at an HTTPS site, you’re using a public key cryptosystem to protect your privacy. Those cryptosystems rely for their security on the presumed hardness of computational problems. That is, it’s possible to crack them, but it’s just too hard. RSA, which is one of the ones that’s widely used … as typically practiced today, to break it you’d have to do something like factor a number which is over 2000 bits long, 2048. That’s too hard to do now. But that’s what quantum computers will be good at. Another one that’s widely used is called elliptic curve cryptography. Doesn’t really matter exactly what it is.

John Preskill [00:39:24] – But the point is that it’s also vulnerable to quantum attack, so we’re going to have to protect our privacy in different ways when quantum computers are prevalent.

Craig Cannon [00:39:37] – What are the attempts being made right now?

John Preskill [00:39:39] – There are two main classes of attempts. One is just to come up with a cryptographic protocol not so different conceptually from what’s done now, but based on a problem that’s hard for quantum computers.

Craig Cannon [00:39:59] – There you go.

John Preskill [00:40:02] – It turns out that what has sort of become the standard way doesn’t have that feature, and there are alternatives that people are working on. We speak of post-quantum cryptography, meaning the protocols that we’ll have to use when we’re worried that our adversaries have quantum computers. I don’t think there’s any proposed cryptosystem — although there’s a long list of them by now which people think are candidates for being quantum resistant, for being unbreakable, or hard to break by quantum computers. I don’t think there’s any one that the world has sufficient confidence in now that’s really hard for a quantum adversary that we’re all going to switch over. But it’s certainly time to be thinking about it. When people worry about their privacy, of course different users have different standards, but the US Government sometimes says they would like a system to stay secure for 50 years. They’d like to be able to use it for 20, roughly speaking, and then have the intercepted traffic be protected for another 30 after that. I don’t think, though I could be wrong, that we’re likely to have quantum computers that can break those public key cryptosystems in 10 years, but in 50 years seems not unlikely,

John Preskill [00:41:33] – and so we should really be worrying about it. The other one is actually using quantum communication for privacy. In other words, if you and I could send qubits to one another instead of bits, it opens up new possibilities. The way to think about these public key schemes — or one way — that we’re using now, is I want you to send me a private message, and I can send you a lockbox. It has a padlock on it, but I keep the key, okay? But you can close up the box and send it to me. But I’m the only one with the key. The key thing is that if you have the padlock you can’t reverse engineer the key. Of course, it’s a digital box and key, but that’s the idea of public key. The idea of what we call quantum key distribution, which is a particular type of quantum cryptography, is that I can actually send you the key, or you can send me your key, but why can’t any eavesdropper then listen in and know the key? Well it’s because it’s quantum, and remember, it has that property that if you look at it, you disturb it.

John Preskill [00:42:59] – So if you collect information about my key, or if the adversary does, that will cause some change in the key, and there are ways in which we can check whether what you received is really what I sent. And if it turns out it’s not, or it has too many errors in it, then we’ll be suspicious that there was an adversary who tampered with it, and then we won’t use that key. Because we haven’t used it yet — we’re just trying to establish the key. We do the test to see whether an adversary interfered. If it passes the test, then we can use the key. And if it fails the test, we throw that key away and we try again. That’s how quantum cryptography works, but it requires a much different infrastructure than what we’re using now. We have to be able to send qubits … well, it’s not completely different because you can do it with photons. Of course, that’s how we communicate through optical fiber now — we’re sending photons. It’s a little trickier sending quantum information through an optical fiber, because of that issue that interactions with the environment can disturb it. But nowadays, you can send quantum information through an optical fiber over tens of kilometers with a low enough error rate so it’s useful for communication.

Craig Cannon [00:44:22] – Wow.

John Preskill [00:44:23] – Of course, we’d like to be able to scale that up to global distances.

Craig Cannon [00:44:26] – Sure.

John Preskill [00:44:27] – And there are big challenges in that. But anyway, so that’s another approach to the future of privacy that people are interested in.

Craig Cannon [00:44:35] – Does that necessitate quantum computers on both ends?

John Preskill [00:44:38] – Yes, but not huge ones. The reason … well, yes and no. At the scale of tens of kilometers, no. You can do that now. There are prototype systems that are in existence. But if you really want to scale it up —  in other words, to send things longer distance — then you have to bring this quantum error correction idea into the game.

John Preskill [00:45:10] – Because at least with our current photonics technology, there’s no way I can send a single photon from here to China without there being a very high probability that it gets lost in the fiber somewhere. We have to have what we call quantum repeaters, which can boost the signal. But it’s not like the usual type of repeater that we have in communication networks now. The usual type is you measure the signal, and then you resend it. That won’t work for quantum because as soon as you measure it you’re going to mess it up. You have to find a way of boosting it without knowing what it is. Of course, it’s important that it works that way because otherwise, the adversary could just intercept it and resend it. And so it will require some quantum processing to get that quantum error correction in the quantum repeater to work. But it’s a much more modest scale quantum processor than we would need to solve hard problems.

Craig Cannon [00:46:14] – Okay. Gotcha. What are the other things you’re both excited about, and worried about for potential business opportunities? Snehan, I’m mispronouncing names all the times, Snehan Kekre asks, budding entrepreneurs, what should they be thinking about in the context of quantum computing?

John Preskill [00:46:37] – There’s more to quantum technology than computing. Something which has good potential to have an impact in the relatively near future is improved sensing. Quantum systems, partly because of that property that I keep emphasizing that they can’t be perfectly isolated from the outside, they’re good at sensing things. Sometimes, you want to detect it when something in the outside world messes around with your qubit. Again, using this technology of nuclear spins, which I mentioned you can do at room temperature potentially, you can make a pretty good sensor, and it can potentially achieve higher sensitivity and spatial resolution, look at things on shorter distance scales than other existing sensing technology. One of the things people are excited about are the biological and medical implications of that.

John Preskill [00:47:53] – If you can monitor the behavior of molecular machines, probe biological systems at the molecular level using very powerful sensors, that would surely have a lot of applications. One interesting question you can ask is, can you use these quantum error correction ideas to make those sensors even more powerful? That’s another area of current basic research, where you could see significant potential economic impact.

Craig Cannon [00:48:29] – Interesting. In terms of your research right now, what are you working on that you find both interesting and incredibly difficult?

John Preskill [00:48:40] – Everything I work on–

Craig Cannon [00:48:41] – 100%.

John Preskill [00:48:42] – Is both interesting and incredibly difficult. Well, let me change direction a little from what we’ve been talking about so far. Well, I’m going to tell you a little bit about me.

Craig Cannon [00:48:58] – Sure.

John Preskill [00:49:00] – I didn’t start out interested in information in my career. I’m a physicist. I was trained as an elementary particle theorist, studying the fundamental interactions and the elementary particles. That drew me into an interest in gravitation because one thing that we still have a very poor understanding of is how gravity fits together with the other fundamental interactions. The way physicists usually say it is we don’t have a quantum theory of gravity, at least not one that we think is complete and satisfactory. I’ve been interested in that question for many decades, and then got sidetracked because I got excited about quantum computing. But you know what? I’ve always looked at quantum information not just as a technology. I’m a physicist, I’m not an engineer. I’m not trying to build a better computer, necessarily, though that’s very exciting, and worth doing, and if my work can contribute to that, that’s very pleasing. I see quantum information as a new frontier in the exploration of the physical sciences. Sometimes I call it the entanglement frontier. Physicists, we like to talk about frontiers, and stuff. Short distance frontier. That’s what we’re doing at CERN in the Large Hadron Collider, trying to discern new properties of matter at distances which are shorter than we’ve ever been able to explore before.

John Preskill [00:50:57] – There’s a long distance frontier in cosmology. We’re trying to look deeper into the universe and understand its structure and behavior at earlier times. Those are both very exciting frontiers. This entanglement frontier is increasingly going to be at the forefront of basic physics research in the 21st century. By entanglement frontier, I just mean scaling up quantum systems to larger and larger complexity where it becomes harder and harder to simulate those systems with our existing digital tools. That means we can’t very well anticipate the types of behavior that we’re going to see. That’s a great opportunity for new discovery, and that’s part of what’s going to be exciting even in the relatively near term. When we have 100 qubits … there are some things that we can do to understand the behavior of the dynamics of a highly complex system of 100 qubits that we’ve never been able to experimentally probe before. That’s going to be very interesting. But what we’re starting to see now is that these quantum information ideas are connecting to these fundamental questions about gravitation, and how to think about it quantumly. And it turns out, as is true for most of the broader implications of quantum physics, the key thing is entanglement.

John Preskill [00:52:36] – We can think of the microscopic structure of spacetime, the geometry of where we live. Geometry just means who’s close to who else. If we’re in the auditorium, and I’m in the first row and you’re in the fourth row, the geometry is how close we are to one another. Of course, that’s very fundamental in both space and time. How far apart are we in space? How far apart are we in time? Is geometry really a fundamental thing, or is it something that’s kind of emergent from some even more fundamental concept? It seems increasingly likely that it’s really an emergent property.

John Preskill [00:53:29] – That there’s something deeper than geometry. What is it? We think it’s quantum entanglement. That you can think of the geometry as arising from quantum correlations among parts of a system. That’s really what defines who’s close to who. We’re trying to explore that idea more deeply, and one of the things that comes in is the idea of quantum error correction. Remember the whole idea of quantum error correction was that we could make a quantum system behave the way we want it to because it’s well-protected against the damaging effects of noise. It seems like quantum error correction is part of the deep secret of how spacetime geometry works. It has a kind of intrinsic robustness coming from these ideas of quantum error correction that makes space meaningful, so that it doesn’t just evaporate when you tap on it. If you wanted to, you could think of the spacetime, the space that you’re in and the space that I’m in, as parts of a system that are entangled with one another.

John Preskill [00:54:45] – What would happen if we broke that entanglement and your part of space became disentangled from my part? Well what we think that would mean is that there’d be no way to connect us anymore. There wouldn’t be any path through space that starts over here with me and ends with you. It’d become broken apart into two pieces. It’s really the entanglement which holds space together, which keeps it from falling apart into little pieces. We’re trying to get a deeper grasp of what that means.

Craig Cannon [00:55:19] – How do you make any progress on that? That seems like the most unbelievably difficult problem to work on.

John Preskill [00:55:26] – It’s difficult because, well for a number of reasons, but in particular, because it’s hard to get guidance from experiment, which is how physics historically–

Craig Cannon [00:55:38] – All science.

John Preskill [00:55:38] – Has advanced.

Craig Cannon [00:55:39] – Yeah.

John Preskill [00:55:41] – Although it was fun a moment ago to talk about what would happen if we disentangled your part of space from mine, I don’t know how to do that in the lab right now. Of course, part of the reason is we have the audacity to think we can figure these things out just by thinking about them. Maybe that’s not true. Nobody knows, right? We should try. Solving these problems is a great challenge, and it may be that the apes that evolved on Earth don’t have the capacity to understand things like the quantum structure of spacetime. But maybe we do, so we should try. Now in the longer term, and maybe not such a long term, maybe we can get some guidance from experiment. In particular, what we’re going to be doing with quantum computers and the other quantum technologies that are becoming increasingly sophisticated in the next couple of decades, is we’ll be able to control very well highly entangled complex quantum systems. That should mean that in a laboratory, on a tabletop, I can sort of make my own little toy space time …

John Preskill [00:57:02] – with an emergent geometry arising from the properties of that entanglement, and I think that’ll teach us lessons because systems like that are the types of system that, because they’re so highly entangled, digital computers can’t simulate them. It seems like only quantum computers are potentially up to the task. So that won’t be quite the same as disentangling your side of the room from mine, in real life. But we’d be able to do it in a laboratory setting using model systems, which I think would help us to understand the basic principles better.

Craig Cannon [00:57:39] – Wild. Yeah, desktop space time seems pretty cool, if you could figure it out.

John Preskill [00:57:43] – Yeah, it’s pretty fundamental. We didn’t really talk about what people sometimes, we did implicitly, but not in so many words. We didn’t talk about what people sometimes call quantum non-locality. It’s another way of describing quantum entanglement, actually. There’s this notion of Bell’s theorem that when you look at the correlations among the parts of a quantum system, that they’re different from any possible classical correlations. Some things that you read give you the impression that you can use that to instantaneously send information over long distances. It is true that if we have two qubits, electron spins, say, and they’re entangled with one another, then what’s kind of remarkable is that I can measure my qubit to see along some axis whether it’s up or down, and you can measure yours, and we will get perfectly correlated results. When I see up, you’ll see up, say, and when I see down, you’ll see down. And sometimes, people make it sound like that’s remarkable. That’s not remarkable in itself. Somebody could’ve flipped a pair of coins, you know,

John Preskill [00:59:17] – so that they came up both heads or both tails, and given one to you and one –

Craig Cannon [00:59:20] – Split them apart.

John Preskill [00:59:20] – to me.

Craig Cannon [00:59:21] – Yeah.

John Preskill [00:59:22] – And gone a light year apart, and then we both …  hey, mine’s heads. Mine’s heads too!

Craig Cannon [00:59:24] – And then they call it quantum teleportation on YouTube.

John Preskill [00:59:28] – Yeah. Of course, what’s really important about entanglement that makes it different from just those coins is that there’s more than one way of looking at a qubit. We have what we call complementary ways of measuring it, so you can ask whether it’s up or down along this axis or along that axis. There’s nothing like that for the coins. There’s just one way to look at it. What’s cool about entanglement is that we’ll get perfectly correlated results if we both measure in the same way, but there’s more than one possible way that we could measure. What sometimes gets said, or the impression people get, is that that means that when I do something to my qubit, it instantaneously affects your qubit, even if we’re on different sides of the galaxy. But that’s not what entanglement does. It just means they’re correlated in a certain way.

John Preskill [01:00:30] – When you look at yours, if we have maximally entangled qubits, you just see a random bit. It could be a zero or a one, each occurring with probability 1/2. That’s going to be true no matter what I did to my qubit, and so you can’t tell what I did by just looking at it. It’s only that if we compared notes later we can see how they’re correlated, and that correlation holds for either one of these two complementary ways in which we could both measure. It’s that fact that we have these complementary ways to measure that makes it impossible for a classical system to reproduce those same correlations. So that’s one misconception that’s pretty widespread. Another one is this about quantum computing, which is in trying to explain why quantum computers are powerful, people will sometimes say, well, it’s because you can superpose –I used that word before, you can add together many different possibilities. That means that, whereas an ordinary computer would just do a computation once, acting on a superposition a quantum computer can do a vast number of computations all at once.

John Preskill [01:01:54] – There’s a certain sense in which that’s mathematically true if you interpret it right, but it’s very misleading. Because in the end, you’re going to have to make some measurement to read out the result. When you read it out, there’s a limited amount of information you can get. You’re not going to be able to read out the results of some huge number of computations in a single shot measurement. Really the key thing that makes it work is this idea of interference, which we discussed briefly when you asked about Grover’s algorithm. The art of a quantum algorithm is to make sure that the wrong answers interfere and cancel one another out, so the right answer is enhanced. That’s not automatic. It requires that the quantum algorithm be designed in just the right way.

Craig Cannon [01:02:50] – Right. The diagrams I’ve seen online at least, involve usually you’re squaring the output as it goes along, and then essentially, that flips the correct answer to the positive, and the others are in the negative position. Is that accurate?

John Preskill [01:03:08] – I wouldn’t have said it the way you did– Because you can’t really measure it as you go along. Once you measure it, the magic of superposition is going to be lost.

John Preskill [01:03:19] – It means that now there’s some definite outcome or state. To take advantage of this interference phenomenon, you need to delay the measurement. Remember when we were talking about the double slit and I said, if you actually see these wiggles in the probability of detection, which is the signal of interference, that means that there’s no way anybody could know whether the electron went through hole one or hole two? It’s the same way with quantum computing. If you think of the computation as being a superposition of different possible computations, it wouldn’t work — there wouldn’t be a speed up — if you could know which of those paths the computation followed. It’s important that you don’t know. And so you have to sum up all the different computations, and that’s how the interference phenomenon comes into play.

Craig Cannon [01:04:17] – To take a little sidetrack, you mentioned Feynman before. And before we started recording you mentioned working with him. I know I’m in the Feynman fan club, for sure. What was that experience like?

John Preskill [01:04:32] – We never really collaborated. I mean, we didn’t write a paper together, or anything like that. We overlapped for five years at Caltech. I arrived here in 1983. He died in 1988. We had offices on the same corridor, and we talked pretty often because we were both interested in the fundamental interactions, and in particular, what we call quantum chromodynamics. It’s our theory of how nuclear matter behaves, how quarks interact, what holds the proton together, those kinds of things. One big question is what does hold the proton together? Why don’t the quarks just fall apart? That was an example of a problem that both he and I were very interested in, and which we talked about sometimes. Now, this was pretty late in his career. When I think about it now, when I arrived at Caltech, that was 1983, Feynman was born in 1918, so he was 65. I’m 64 now, so maybe he wasn’t so old, right? But at the time, he seemed pretty ancient to me. Since I was 30.

John Preskill [01:05:58] – Those who interacted with Dick Feynman when he was really at his intellectual peak in the ’40s, and ’50s, and ’60s, probably saw even more extraordinary intellectual feats than I witnessed interacting with the 65 year old Feynman. He just loved physics, you know? He just thought everything was so much fun. He loved talking about it. He wasn’t as good a listener as a talker, but actually – well that’s a little unfair, isn’t it? It was kind of funny because Feynman, he always wanted to think things through for himself, sort of from first principles, rather than rely on the guidance from experts who have thought about these things before. Well that’s fine. You should try to understand things as deeply as you can on your own, and sort of reconstruct the knowledge from the ground up. That’s very enabling, and gives you new insights. But he was a little too dismissive, in my view, of what the other guys knew. But I could slip it in because I didn’t tell him, “Dick, you should read this paper by Polyakov” — well maybe I did, but he wouldn’t have even heard that  — because he solved that problem that you’re talking about.

John Preskill [01:07:39] – But I knew what Polyakov had said about it, so I would say, “Oh well, look, why don’t we look at it this way?” And so he thought I was, that I was having all these insights, but the truth was the big difference between Feynman and me in the mid 1980s was I was reading literature, and he wasn’t.

Craig Cannon [01:08:00] – That’s funny.

John Preskill [01:08:01] – Probably, if he had been, he would’ve been well served, but that wasn’t the way he liked to work on things. He wanted to find his own approach. Of course, that had worked out pretty well for him throughout his career.

Craig Cannon [01:08:15] – What other qualities did you notice about him when he was roaming the corridors?

John Preskill [01:08:21] – He’d always be drumming. So you would know he was around because he’d actually be walking down the hallway drumming on the wall.

Craig Cannon [01:08:27] – Wait, with his hands, or with sticks, or–

John Preskill [01:08:29] – No, hands. He’d just be tapping.

Craig Cannon [01:08:32] – Just a bongo thing.

John Preskill [01:08:33] – Yeah. That was one thing. He loved to tell stories. You’ve probably read the books that Ralph Leighton put together based on the stories Feynman told. Ralph did an amazing job, of capturing Feynman’s personality in writing those stories down because I’d heard a lot of them. I’m sure he told the same stories to many people many times, because he loved telling stories. But the book really captures his voice pretty well.

John Preskill [01:09:12] – If you had heard him tell some of these stories, and then you read the way Ralph Leighton transcribed them, you can hear Feynman talking. At the time that I knew him, one of the experiences that he went through was he was on the Challenger commission after the space shuttle blew up. He was in Washington a lot of the time, but he’d come back from time to time, and he would sort of sit back and relax in our seminar room and start bringing us up to date on all the weird things that were happening on the Challenger commission. That was pretty fun.

Craig Cannon [01:09:56] – That’s really cool.

John Preskill [01:09:56] – A lot of that got captured in the second volume. I guess it’s the one called, What Do You Care What Other People Think? There’s a chapter about him telling stories about the Challenger commission. He was interested in everything. It wasn’t just physics. He was very interested in biology. He was interested in computation. I remember how excited he was when he got his first IBM PC. Probably not long after I got to Caltech. Yeah, it was what they called the AT. We thought it was a pretty sexy machine. I had one, too. He couldn’t wait to start programming it in BASIC.

Craig Cannon [01:10:50] – Very cool.

John Preskill [01:10:51] – Because that was so much fun.

Craig Cannon [01:10:52] – There was a question that I was kind of curious to your answer. Tika asks about essentially, teaching about quantum computers. They say, many kids in grade 10 can code. Some can play with machine learning tools without knowing the math. Can quantum computing become as simple and/or accessible?

John Preskill [01:11:17] – Maybe so. At some level, when people say quantum mechanics is counterintuitive, it’s hard for us to grasp, it’s so foreign to our experience, that’s true. The way things behave at the microscopic scale are, like we discussed earlier, really different from the way ordinary stuff behaves. But it’s a question of familiarity. What I wouldn’t be surprised by is that if you go out a few decades, kids who are 10 years old are going to be playing quantum games. That’s an application area that doesn’t get discussed very much, but there could be a real market there because people love games. Quantum games are different, and the strategies are different, and what you have to do to win is different. If you play the game enough, you start to get the hang of it.

John Preskill [01:12:26] – I don’t see any reason why kids who have not necessarily deeply studied physics can’t get a pretty good feel for how quantum mechanics works. You know, the way ordinary physics works, maybe it’s not so intuitive. Newton’s laws … Aristotle couldn’t get it right. He thought you had to keep pushing on something to get it to keep moving. That wasn’t right. Galileo was able to roll balls down a ramp, and things like that, and see he didn’t have to keep pushing to keep it moving. He could see that it was uniformly accelerated in a gravitational field. Newton took that to a much more general and powerful level. You fool around with stuff, and you get the hang of it. And I think quantum stuff can be like that. We’ll experience it in a different way, but when we have quantum computers, in a way, that opens the opportunity for trying things out and seeing what happens.

John Preskill [01:13:50] – After you’ve played the game enough, you start to anticipate. And actually, it’s an important point about the applications. One of the questions you asked me at the beginning was what are we able to do with quantum computers? And I said, I don’t know. So how are we going to discover new applications? It might just be, at least in part, by fooling around. A lot of classical algorithms that people use on today’s computers were discovered, or that they were powerful was discovered, by experimenting. By trying it. I don’t know … what’s an example of that? Well, the simplex method that we use in linear programming. I don’t think there was a mathematical proof that it was fast at first, but people did experiments, and they said, hey, this is pretty fast.

Craig Cannon [01:14:53] – Well, you’re seeing it a lot now in machine learning.

John Preskill [01:14:57] – Yeah, well that’s a good example.

Craig Cannon [01:14:58] – You test it out a million times over when you’re running simulations, and it turns out, that’s what works. Following the thread of education, and maybe your political interest, given it’s the year that it is, do you have thoughts on how you would adjust or change STEM education?

John Preskill [01:15:23] – Well, no particularly original thoughts. But I do think that STEM education … we shouldn’t think of it as we’re going to need this technical workforce, and so we better train them. The key thing is we want the general population to be able to reason effectively, and to recognize when an argument is phony and when it’s authentic. To think about, well how can I check whether what I just read on Facebook is really true? And I see that as part of the goal of STEM education. When you’re teaching kids in school how to understand the world by doing experiments, by looking at the evidence, by reasoning from the evidence, this is something that we apply in everyday life, too. I don’t know exactly how to implement this–

John Preskill [01:16:36] – But I think we should have that perspective that we’re trying to educate a public, which is going to eventually make critical decisions about our democracy, and they should understand how to tell when something is true or not. That’s a hard thing to do in general, but you know what I mean. That there are some things that, if you’re a person with some — I mean it doesn’t necessarily have to be technical — but if you’re used to evaluating evidence and making a judgment based on that evidence about whether it’s a good argument or not, you can apply that to all the things you hear and read, and make better judgments.

Craig Cannon [01:17:23] – What about on the policy side? Let’s see, JJ Francis asked that, if you or any of your colleagues would ever consider running for office. Curious about science policy in the US.

John Preskill [01:17:38] – Well, it would be good if we had more scientifically trained people in government. Very few members of Congress. I know of one, Bill Foster’s a physicist in Illinois. He was a particle physicist, and he worked at Fermilab, and now he’s in Congress, and very interested in the science and educational policy aspects of government. Rush Holt was a congressman from New Jersey who had a background in physics. He retired from the House a couple of years ago, but he was in Congress for something like 18 years, and he had a positive influence, because he had a voice that people respected when it came to science policy. Having more people like that would help. Now, another thing, it doesn’t have to be elective office.

Craig Cannon [01:18:39] – Right.

John Preskill [01:18:42] – There are a lot of technically trained people in government, many of them making their careers in agencies that deal with technical issues. Department of Defense, of course, there are a lot of technical issues. In the Obama Administration we had two successive secretaries of energy who were very, very good physicists. Steve Chu was Nobel Prize winning physicist. Then Ernie Moniz, who’s a real authority on nuclear energy and weapons. That kind of expertise makes a difference in government.

John Preskill [01:19:24] – Now the Secretary of Energy is Rick Perry. It’s a different background.

Craig Cannon [01:19:28] – Yeah, you could say that. Just kind of historical reference, what policies did they put in place that you really felt their hand as a physicist move forward?

John Preskill [01:19:44] – You mean in particular–

Craig Cannon [01:19:45] – I’m talking the Obama Administration.

John Preskill [01:19:49] – Well, I think the Department of Energy, DOE, tried to facilitate technical innovation by seeding new technologies, by supporting startup companies that were trying to do things that would improve battery technology, and solar power, and things like that, which could benefit future generations. They had an impact by doing that. You don’t have to be a Nobel Prize winning physicist to think that’s a good idea. That the administration felt that was a priority made a difference, and appointing a physicist at Department of Energy was, if nothing else, highly symbolic of how important those things are.

Craig Cannon [01:20:52] – On the quantum side, someone asked Vikas Karad, he asked where the Quantum Valley might be. Do you have thoughts, as in Silicon Valley for quantum computing?

John Preskill [01:21:06] – Well… I don’t know, but you look at what’s happening the last couple of years, there have been a number of quantum startups. A notable number of them are in the Bay Area. Why so? Well, that’s where the tech industry is concentrated and where the people who are interested in financing innovative technical startups are concentrated. If you are an entrepreneur interested in starting a company, and you’re concerned about how to fundraise for it, it kind of makes sense to locate in that area. Now, that’s what’s sort of happening now, and may not continue, of course. It might not be like that indefinitely. Nothing lasts forever, but I would say… That’s the place, Silicon Valley is likely to be Quantum Valley, the way things are right now.

Craig Cannon [01:22:10] – Well then what about the physicists who might be listening to this? If they’re thinking about starting a company, do you have advice for them?

John Preskill [01:22:22] – Just speaking very generally, if you’re putting a team together… Different people have different expertise. Take quantum computing as an example, like we were saying earlier, some of the big players and the startups, they want to do everything. They want to build the hardware, figure out better ways to fabricate it. Better control, better software, better applications. Nobody can be an expert on all those things. Of course, you’ll hire a software person to write your software, and microwave engineer to figure out your control, and of course that’s the right thing to do. But I think in that arena, and it probably applies to other entrepreneurial activity relating to physics, being able to communicate across those boundaries is very valuable, and you can see it in quantum computing now. That if the man or woman who’s involved in the software has that background, but there’s not a big communication barrier talking to the people who are doing the control engineering, that can be very helpful. It makes sense to give some preference to the people who maybe are comfortable doing so, or have the background that stretches across more than one of those areas of expertise. That can be very enabling in a technology arena like quantum computing today, where we’re trying to do really, really hard stuff, and you don’t know whether you’ll succeed, and you want to give it your best go by seeing the connections between those different things.

Craig Cannon [01:24:28] – Would you advise someone then to maybe teach or try and explain it to, I don’t know their young cousins? Because Feynman maybe recognizes the king of communicating physics, at least for a certain period of time. How would you advise someone to get better at it so they can be more effective?

John Preskill [01:24:50] – Practice. There are different aspects of that. This isn’t what you meant at all, but I’ll say it anyway, because what you asked brought it to mind. If you teach, you learn. We have this odd model in the research university that a professor like me is supposed to do research and teach. Why don’t we hire teachers and researchers? Why do we have the same people doing both? Well, part of the reason for me is most of what I know, what I’ve learned since my own school education ended, is knowledge I acquired by trying to teach it. To keep our intellect rejuvenated, we have to have that experience of trying to teach new things that we didn’t know that well before to other people. That deepens your knowledge. Just thinking about how you convey it makes you ask questions that you might not think to ask otherwise, and you say “Hey, I don’t know the answer to that.” Then you have to try to figure it out. So I think that applies at varying levels to any situation in which a scientist, or somebody with a technical background, is trying to communicate.

John Preskill [01:26:21] – By thinking about how to get it across to other people, we can get new insights, you know? We can look at it in a different way. It’s not a waste of time. Aside from the benefits of actually successfully communicating, we benefit from it in this other way. But other than that… Have fun with it, you know? Don’t look at it as a burden, or some kind of task you have to do along with all the other things you’re doing. It should be a pleasure. When it’s successful, it’s very gratifying. If you put a lot of thought into how to communicate something and you think people are getting it, that’s one of the ways that somebody in my line of work can get a lot of satisfaction.

Craig Cannon [01:27:23] – If now were to be your opportunity to teach a lot of people about physics, and you could just point someone to things, who would you advise someone to be? They want to learn more about quantum computing, they want to learn about physics. What should they be reading? What YouTube channel should they follow? What should they pay attention to?

John Preskill [01:27:44] – Well one communicator who I have great admiration for is Leonard Susskind, who’s at Stanford. You mentioned Feynman as the great communicator, and that’s fair, but in terms of style and personality of physicists who are currently active, I think Lenny Susskind is the most similar to Feynman of anyone I can think of. He’s a no bullshit kind of guy. He wants to give you the straight stuff. He doesn’t want to water it down for you. But he’s very gifted when it comes to making analogies and creating the illusion that you’re understanding what he’s saying. He has … if you just go to YouTube and search Leonard Susskind you’ll see lectures that he’s given at Stanford where they have some kind of extension school for people who are not Stanford students, people in the community. A lot of them in the tech community because it’s Stanford, and he’s giving courses. Yeah, and on quite sophisticated topics, but also on more basic topics, and he’s in the process of turning those into books. I’m not sure how many of those have appeared, but he has a series called The Theoretical Minimum

John Preskill [01:29:19] – which is supposed to be the gentle introduction to different topics like classical physics, quantum physics, and so on. He’s pretty special I think in his ability to do that.

Craig Cannon [01:29:32] – I need to subscribe. Actually, here’s a question then. In the things you’ve relearned while teaching over the past, I guess it’s 35 years now.

John Preskill [01:29:46] – Shit, is that right?

Craig Cannon [01:29:47] – Something like that.

John Preskill [01:29:48] – That’s true. Yeah.

Craig Cannon [01:29:51] – What were the big thing, what were the revelations?

John Preskill [01:29:55] – That’s how I learned quantum computing, for one thing. I was not at all knowledgeable about information science. That wasn’t my training. Back when I was in school, physicists didn’t learn much about things like information theory, computer science, complexity theory. One of the great things about quantum computing is its interdisciplinary character, that it brings these different things into contact, which traditionally had not been part of the common curriculum of any community of scholars. I decided 20 years ago that I should teach a quantum information class at Caltech, and I worked very hard on it that year. Not that I’m an expert, or anything, but I learned a lot about information theory, and things like channel capacity, and computational complexity — how we classify the hardness of problems — and algorithms. Things like that, which I didn’t really know very well. I had sort of a passing familiarity with some of those things from reading some of the quantum computing literature. That’s no substitute for teaching a class because then you really have to synthesize it and figure out your way of presenting it. Most of the notes are typed up and you can still get to them on my website.That was pretty transformative for me … and it was easier then, 20 years ago, I guess than it is now because it was such a new topic.

John Preskill [01:31:49] – But I really felt I was kind of close enough to the cutting edge on most of those topics by the time I’d finished the class that I wasn’t intimidated by another paper I’d read or a new thing I’d hear about those things. That was probably the one case where it really made a difference in my foundation of knowledge which enabled me to do things. But I had the same experience in particle physics. When I was a student, I read a lot. I was very broadly interested in physics. But when the first time, I was still at Harvard at the time –later I taught a similar course here — I’m in my late 20s, I’m just a year or two out of graduate school, and I decide to teach a very comprehensive class on elementary particles … in particular, quantum chromodynamics, the theory of nuclear forces like we talked about before. It just really expanded my knowledge to have that experience of teaching that class. I still draw on that. I can still remember that experience and I think I get ideas that I might not otherwise have because I went through that.

Craig Cannon [01:33:23] – I want to get involved now. I want to go back to school, or maybe teach a class. I don’t know.

John Preskill [01:33:27] – Well, what’s stopping you?

Craig Cannon [01:33:29] – Nothing. Alright, thanks John.

John Preskill [01:33:32] – Okay, thank you Craig.

Rock-paper-scissors, granite-clock-idea

I have a soft spot for lamassu. Ten-foot-tall statues of these winged bull-men guarded the entrances to ancient Assyrian palaces. Show me lamassu, or apkallu—human-shaped winged deities—or other reliefs from the Neo-Assyrian capital of Nineveh, and you’ll have trouble showing me the door.

Assyrian art fills a gallery in London’s British Museum. Lamassu flank the gallery’s entrance. Carvings fill the interior: depictions of soldiers attacking, captives trudging, and kings hunting lions. The artwork’s vastness, its endurance, and the contact with a three-thousand-year-old civilization floor me. I tore myself away as the museum closed one Sunday night.


I visited the British Museum the night before visiting Jonathan Oppenheim’s research group at University College London (UCL). Jonathan combines quantum information theory with thermodynamics. He and others co-invented thermodynamic resource theories (TRTs), which Quantum Frontiers regulars will know of. TRTs are quantum-information-theoretic models for systems that exchange energy with their environments.

Energy is conjugate to time: Hamiltonians, mathematical objects that represent energy, represent also translations through time. We measure time with clocks. Little wonder that one can study quantum clocks using a model for energy exchanges.

Mischa Woods, Ralph Silva, and Jonathan used a resource theory to design an autonomous quantum clock. “Autonomous” means that the clock contains all the parts it needs to operate, needs no periodic winding-up, etc. When might we want an autonomous clock? When building quantum devices that operate independently of classical engineers. Or when performing a quantum computation: Computers must perform logical gates at specific times.


Wolfgang Pauli and others studied quantum clocks, the authors recall. How, Pauli asked, would an ideal clock look? Its Hamiltonian, \hat{H}_{\rm C}, would have eigenstates | E \rangle. The labels E denote possible amounts of energy.

The Hamiltonian would be conjugate to a “time operator” \hat{t}. Let | \theta \rangle denote an eigenstate of \hat{t}. This “time state” would equal an even superposition over the | E \rangle’s. The clock would occupy the state | \theta \rangle at time t_\theta.

Imagine measuring the clock, to learn the time, or controlling another system with the clock. The interaction would disturb the clock, changing the clock’s state. The disturbance wouldn’t mar the clock’s timekeeping, if the clock were ideal. What would enable an ideal clock to withstand the disturbances? The ability to have any amount of energy: E must stretch from - \infty to \infty. Such clocks can’t exist.

Approximations to them can. Mischa, Ralph, and Jonathan designed a finite-size clock, then characterized how accurately the clock mimics the ideal. (Experts: The clock corresponds to a Hilbert space of finite dimensionality d. The clock begins in a Gaussian state that peaks at one time state | \theta \rangle. The finite-width Gaussian offers more stability than a clock state.)

Disturbances degrade our ability to distinguish instants by measuring the clock. Imagine gazing at a kitchen clock through blurry lenses: You couldn’t distinguish 6:00 from 5:59 or 6:01. Disturbances also hinder the clock’s ability to implement processes, such as gates in a computation, at desired instants.

Mischa & co. quantified these degradations. The errors made by the clock, they found, decay inverse-exponentially with the clock’s size: Grow the clock a little, and the errors shrink a lot.


Time has degraded the lamassu, but only a little. You can distinguish feathers in their wings and strands in their beards. People portray such artifacts as having “withstood the flow of time,” or “evaded,” or “resisted.” Such portrayals have never appealed to me. I prefer to think of the lamassu as surviving not because they clash with time, but because they harmonize with it. The prospect of harmonizing with time—whatever that means—has enticed me throughout my life. The prospect partially underlies my research into time—perhaps childishly, foolishly—I recognize if I remove my blurry lenses before gazing in the mirror.

The creation of lasting works, like lamassu, has enticed me throughout my life. I’ve scrapbooked, archived, and recorded, and tended memories as though they were Great-Grandma’s cookbook. Ancient civilizations began alluring me at age six, partially due to artifacts’ longevity. No wonder I study the second law of thermodynamics.

Yet doing theoretical physics makes no sense from another perspective. The ancient Egyptians sculpted granite, when they could afford it. Gudea, king of the ancient city-state of Lagash, immortalized himself in diorite. I fashion ideas, which lack substance. Imagine playing, rather than rock-paper-scissors, granite-diorite-idea. The idea wouldn’t stand a chance.

Would it? Because an idea lacks substance, it can manifest in many forms. Plato’s cave allegory has manifested as a story, as classroom lectures, on handwritten pages, on word processors and websites, in cartloads of novels, in the film The Matrix, in one of the four most memorable advertisements I received from colleges as a high-school junior, and elsewhere. Plato’s allegory has survived since about the fourth century BCE. King Ashurbanipal’s lion-hunt reliefs have survived for only about 200 years longer.

The lion-hunt reliefs—and lamassu—exude a grandness, a majesty that’s attracted me as their longevity has. The nature of time and the perfect clock have as much grandness. Leaving the British Museum’s Assyrian gallery at 6 PM one Sunday, I couldn’t have asked for a more fitting location, 24 hours later, than in a theoretical-physics conversation.


With thanks to Jonathan, to Álvaro Martín-Alhambra, and to Mischa for their hospitality at UCL; to Ada Cohen for the “Art history of ancient Egypt and the ancient Near East” course for which I’d been hankering for years; to my brother, for transmitting the ancient-civilizations bug; and to my parents, who fed the infection with museum visits.

Click here for a follow-up to the quantum-clock paper.


The word dominates chapter one of Richard Holmes’s book The Age of WonderHolmes writes biographies of Romantic-Era writers: Mary Wollstonecraft, Percy Shelley, and Samuel Taylor Coleridge populate his bibliography. They have cameos in Age. But their scientific counterparts star.

“Their natural-philosopher” counterparts, I should say. The word “scientist” emerged as the Romantic Era closed. Romanticism, a literary and artistic movement, flourished between the 1700s and the 1800s. Romantics championed self-expression, individuality, and emotion over convention and artificiality. Romantics wondered at, and drew inspiration from, the natural world. So, Holmes argues, did Romantic-Era natural philosophers. They explored, searched, and innovated with Wollstonecraft’s, Shelley’s, and Coleridge’s zest.

Age of Wonder

Holmes depicts Wilhelm and Caroline Herschel, a German brother and sister, discovering the planet Uranus. Humphry Davy, an amateur poet from Penzance, inventing a lamp that saved miners’ lives. Michael Faraday, a working-class Londoner, inspired by Davy’s chemistry lectures.

Joseph Banks in paradise.

So Holmes entitled chapter one.

Banks studied natural history as a young English gentleman during the 1760s. He then sailed around the world, a botanist on exploratory expeditions. The second expedition brought Banks aboard the HMS Endeavor. Captain James Cook steered the ship to Brazil, Tahiti, Australia, and New Zealand. Banks brought a few colleagues onboard. They studied the native flora, fauna, skies, and tribes.

Banks, with fellow botanist Daniel Solander, accumulated over 30,000 plant samples. Artist Sydney Parkinson drew the plants during the voyage. Parkinson’s drawings underlay 743 copper engravings that Banks commissioned upon returning to England. Banks planned to publish the engravings as the book Florilegium. He never succeeded. Two institutions executed Banks’s plan more than 200 years later.

Banks’s Florilegium crowns an exhibition at the University of California at Santa Barbara (UCSB). UCSB’s Special Research Collections will host “Botanical Illustrations and Scientific Discovery—Joseph Banks and the Exploration of the South Pacific, 1768–1771” until May 2018. The exhibition features maps of Banks’s journeys, biographical sketches of Banks and Cook, contemporary art inspired by the engravings, and the Florilegium.

online poster

The exhibition spotlights “plants that have subsequently become important ornamental plants on the UCSB campus, throughout Santa Barbara, and beyond.” One sees, roaming Santa Barbara, slivers of Banks’s paradise.

2 bouganvilleas

In Santa Barbara resides the Kavli Institute for Theoretical Physics (KITP). The KITP is hosting a program about the physics of quantum information (QI). QI scientists are congregating from across the world. Everyone visits for a few weeks or months, meeting some participants and missing others (those who have left or will arrive later). Participants attend and present tutorials, explore beyond their areas of expertise, and initiate research collaborations.

A conference capstoned the program, one week this October. Several speakers had founded subfields of physics: quantum error correction (how to fix errors that dog quantum computers), quantum computational complexity (how quickly quantum computers can solve hard problems), topological quantum computation, AdS/CFT (a parallel between certain gravitational systems and certain quantum systems), and more. Swaths of science exist because of these thinkers.


One evening that week, I visited the Joseph Banks exhibition.

Joseph Banks in paradise.

I’d thought that, by “paradise,” Holmes had meant “physical attractions”: lush flowers, vibrant colors, fresh fish, and warm sand. Another meaning occurred to me, after the conference talks, as I stood before a glass case in the library.

Joseph Banks, disembarking from the Endeavour, didn’t disembark onto just an island. He disembarked onto terra incognita. Never had he or his colleagues seen the blossoms, seed pods, or sprouts before him. Swaths of science awaited. What could the natural philosopher have craved more?

QI scientists of a certain age reminisce about the 1990s, the cowboy days of QI. When impactful theorems, protocols, and experiments abounded. When they dangled, like ripe fruit, just above your head. All you had to do was look up, reach out, and prove a pineapple.


Typical 1990s quantum-information scientist

That generation left mine few simple theorems to prove. But QI hasn’t suffered extinction. Its frontiers have advanced into other fields of science. Researchers are gaining insight into thermodynamics, quantum gravity, condensed matter, and chemistry from QI. The KITP conference highlighted connections with quantum gravity.

…in paradise.

What could a natural philosopher crave more?


Artwork commissioned by the UCSB library: “Sprawling Neobiotic Chimera (After Banks’ Florilegium),” by Rose Briccetti

Most KITP talks are recorded and released online. You can access talks from the conference here. My talk, about quantum chaos and thermalization, appears here. 

With gratitude to the KITP, and to the program organizers and the conference organizers, for the opportunity to participate. 

Standing back at Stanford

T-shirt 1

This T-shirt came to mind last September. I was standing in front of a large silver-colored table littered with wires, cylinders, and tubes. Greg Bentsen was pointing at components and explaining their functions. He works in Monika Schleier-Smith’s lab, as a PhD student, at Stanford.

Monika’s group manipulates rubidium atoms. A few thousand atoms sit in one of the cylinders. That cylinder contains another cylinder, an optical cavity, that contains the atoms. A mirror caps each of the cavity’s ends. Light in the cavity bounces off the mirrors.

Light bounces off your bathroom mirror similarly. But we can describe your bathroom’s light accurately with Maxwellian electrodynamics, a theory developed during the 1800s. We describe the cavity’s light with quantum electrodynamics (QED). Hence we call the lab’s set-up cavity QED.

The light interacts with the atoms, entangling with them. The entanglement imprints information about the atoms on the light. Suppose that light escaped from the cavity. Greg and friends could measure the light, then infer about the atoms’ quantum state.

A little light leaks through the mirrors, though most light bounces off. From leaked light, you can infer about the ensemble of atoms. You can’t infer about individual atoms. For example, consider an atom’s electrons. Each electron has a quantum property called a spin. We sometimes imagine the spin as an arrow that points upward or downward. Together, the electrons’ spins form the atom’s joint spin. You can tell, from leaked light, whether one atom’s spin points upward. But you can’t tell which atom’s spin points upward. You can’t see the atoms for the ensemble.

Monika’s team can. They’ve cut a hole in their cylinder. Light escapes the cavity through the hole. The light from the hole’s left-hand edge carries information about the leftmost atom, and so on. The team develops a photograph of the line of atoms. Imagine holding a photograph of a line of people. You can point to one person, and say, “Aha! She’s the xkcd fan.” Similarly, Greg and friends can point to one atom in their photograph and say, “Aha! That atom has an upward-pointing spin.” Monika’s team is developing single-site imaging.


Aha! She’s the xkcd fan.

Monika’s team plans to image atoms in such detail, they won’t need for light to leak through the mirrors. Light leakage creates problems, including by entangling the atoms with the world outside the cavity. Suppose you had to diminish the amount of light that leaks from a rubidium cavity. How should you proceed?

Tell the mirrors,

T-shirt 2

You should lengthen the cavity. Why? Imagine a photon, a particle of light, in the cavity. It zooms down the cavity’s length, hits a mirror, bounces off, retreats up the cavity’s length, hits the other mirror, and bounces off. The photon repeats this process until a mirror hit fails to generate a bounce. The mirror transmits the photon to the exterior; the photon leaks out. How can you reduce leaks? By preventing photons from hitting mirrors so often, by forcing the photons to zoom longer, by lengthening the cavity, by shifting the mirrors outward.

So Greg hinted, beside that silver-colored table in Monika’s lab. The hint struck a chord: I recognized the impulse to

T-shirt 3

The impulse had led me to Stanford.

Weeks earlier, I’d written my first paper about quantum chaos and information scrambling. I’d sat and read and calculated and read and sat and emailed and written. I needed to stand up, leave my cavity, and image my work from other perspectives.

Stanford physicists had written quantum-chaos papers I admired. So I visited, presented about my work, and talked. Patrick Hayden introduced me to a result that might help me apply my result to another problem. His group helped me simplify a mathematical expression. Monika reflected that a measurement scheme I’d proposed sounded not unreasonable for cavity QED.

And Greg led me to recognize the principle behind my visit: Sometimes, you have to

T-shirt 4

to move forward.

With gratitude to Greg, Monika, Patrick, and the rest of Monika’s and Patrick’s groups for their time, consideration, explanations, and feedback. With thanks to Patrick and Stanford’s Institute for Theoretical Physics for their hospitality.

Entropy Avengers

As you already know if you read my rare (but highly refined!) blog samples, I have spent a big chunk of my professorial career teaching statistical mechanics. And if you teach statistical mechanics, there is pretty much one thing you obsess about: entropy.

So you can imagine my joy of finally seeing a fully anti-entropic superhero appearing on my facebook account (physics enthusiasts out there – the project is seeking support on Kickstarter):

Apart from the plug for Assa Auerbach’s project (which, for full disclosure, I have just supported), I would like to use this as an excuse to share my lessons about entropy. With the same level of seriousness. Here they are, in order of increasing entropy.

1. Cost of entropy. Entropy is always marketed as a very palpable thing. Disorder. In class, however, it is calculated via an enumeration of the ‘microscopic states of the system’. For an atomic gas I know how to calculate the entropy (throw me at the blackboard in the middle of the night, no problem. Bosons or Fermions – anytime!) But how can the concept be applied to our practical existence? I have a proposal:

Quantify entropy by the cost (in $’s) of cleaning up the mess!

Examples can be found at all scales. For anything household-related, we should use the H_k constant. H_k=$25/hour for my housekeeper. You break a glass – it takes about 10 minutes to clean. That puts the entropy of the wreckage at $4.17. Having a birthday party takes about 2 hours to clean up: $50 entropy.

Another insight which my combined experience as professor and parent has produced:

2. Conjecture: Babies are maximally efficient topological entropy machines. If you raised a 1 year-old you know exactly what I mean. You can at least guess why maximum efficiency. But why topological? A baby sauntering through the house leaves a string of destruction behind itself. The baby is a mess-creation string-operator! If you start lagging behind, doom will emerge – hence the maximum efficiency. By the way, the only strategy viable is to undo the damage as it happens. But this blog post is about entropy, not about parenting.

In fact, this allows us to establish a conversion of entropy measured in k_B units, to its, clearly more natural, measure in dollar units. A baby eats about 1000kCal/day=4200kJ/day. To fully deal with the consequences, we need a housekeeper to visit about once a week. 4200kJ/day times 7 days=29400 kJoules. These are consumed at T=300K. So an entropy of S=Q/T~100J/K, which is also S~6 \times 10^{24} (Q/k_B T) in dimensionless units, converts to S~$120, which is the cost of our weekly housekeeper visit. This gives a value of $ 10^{-23} per entropy of a two-level system. Quite a reasonable bang for the buck, don’t you think?

3. My conjecture (2) fails. The second law of thermodynamics is an inequality. Entropy \geq Q/T. Why does the conjecture fail? Babies are not ‘maximal’. Consider presidents. Consider the mess that the government can make. It is at the scale of trillions per year. $ 10^{12}. Using the rigorous conversion rule established above, this corresponds to 10^{35} two-level systems. Which happens to quite precisely match the combined number of electrons present in the human bodies of all our military personnel. But the mess, however, is created by very few individuals.

Given the large amounts of taxpayer money we dish out to deal with entropy in the world, Auerbach’s book is bound to make a big impact. In fact, maybe Max the demon would one day be nominated for the national medal of freedom, or at least be inducted into the National Academy of Sciences.

Glass beads and weak-measurement schemes

Richard Feynman fiddled with electronics in a home laboratory, growing up. I fiddled with arts and crafts.1 I glued popsicle sticks, painted plaques, braided yarn, and designed greeting cards. Of the supplies in my family’s crafts box, I adored the beads most. Of the beads, I favored the glass ones.

I would pour them on the carpet, some weekend afternoons. I’d inherited a hodgepodge: The beads’ sizes, colors, shapes, trimmings, and craftsmanship varied. No property divided the beads into families whose members looked like they belonged together. But divide the beads I tried. I might classify them by color, then subdivide classes by shape. The color and shape groupings precluded me from grouping by size. But, by loosening my original classification and combining members from two classes, I might incorporate trimmings into the categorization. I’d push my classification scheme as far as I could. Then, I’d rake the beads together and reorganize them according to different principles.

Why have I pursued theoretical physics? many people ask. I have many answers. They include “Because I adored organizing craft supplies at age eight.” I craft and organize ideas.


I’ve blogged about the out-of-time-ordered correlator (OTOC), a signature of how quantum information spreads throughout a many-particle system. Experimentalists want to measure the OTOC, to learn how information spreads. But measuring the OTOC requires tight control over many quantum particles.

I proposed a scheme for measuring the OTOC, with help from Chapman University physicist Justin Dressel. The scheme involves weak measurements. Weak measurements barely disturb the systems measured. (Most measurements of quantum systems disturb the measured systems. So intuited Werner Heisenberg when formulating his uncertainty principle.)

I had little hope for the weak-measurement scheme’s practicality. Consider the stereotypical experimentalist’s response to a stereotypical experimental proposal by a theorist: Oh, sure, we can implement that—in thirty years. Maybe. If the pace of technological development doubles. I expected to file the weak-measurement proposal in the “unfeasible” category.

But experimentalists started collaring me. The scheme sounds reasonable, they said. How many trials would one have to perform? Did the proposal require ancillas, helper systems used to control the measured system? Must each ancilla influence the whole measured system, or could an ancilla interact with just one particle? How did this proposal compare with alternatives?

I met with a cavity-QED2 experimentalist and a cold-atoms expert. I talked with postdocs over skype, with heads of labs at Caltech, with grad students in Taiwan, and with John Preskill in his office. I questioned an NMR3 experimentalist over lunch and fielded superconducting-qubit4 questions in the sunshine. I read papers, reread papers, and powwowed with Justin.

I wouldn’t have managed half so well without Justin and without Brian Swingle. Brian and coauthors proposed the first OTOC-measurement scheme. He reached out after finding my first OTOC paper.

According to that paper, the OTOC is a moment of a quasiprobability.5 How does that quasiprobability look, we wondered? How does it behave? What properties does it have? Our answers appear in a paper we released with Justin this month. We calculate the quasiprobability in two examples, prove properties of the quasiprobability, and argue that the OTOC motivates generalizations of quasiprobability theory. We also enhance the weak-measurement scheme and analyze it.

Amidst that analysis, in a 10 x 6 table, we classify glass beads.


We inventoried our experimental conversations and distilled them. We culled measurement-scheme features analogous to bead size, color, and shape. Each property labels a row in the table. Each measurement scheme labels a column. Each scheme has, I learned, gold flecks and dents, hues and mottling, an angle at which it catches the light.

I’ve kept most of the glass beads that fascinated me at age eight. Some of the beads have dispersed to necklaces, picture frames, and eyeglass leashes. I moved the remnants, a few years ago, to a compartmentalized box. Doesn’t it resemble the table?


That’s why I work at the IQIM.


1I fiddled in a home laboratory, too, in a garage. But I lived across the street from that garage. I lived two rooms from an arts-and-crafts box.

2Cavity QED consists of light interacting with atoms in a box.

3Lots of nuclei manipulated with magnetic fields. “NMR” stands for “nuclear magnetic resonance.” MRI machines, used to scan brains, rely on NMR.

4Superconducting circuits are tiny, cold quantum circuits.

5A quasiprobability resembles a probability but behaves more oddly: Probabilities range between zero and one; quasiprobabilities can dip below zero. Think of a moment as like an average.

With thanks to all who questioned me; to all who answered questions of mine; to my wonderful coauthors; and to my parents, who stocked the crafts box.

The entangled fabric of space

We live in the information revolution. We translate everything into vast sequences of ones and zeroes. From our personal email to our work documents, from our heart rates to our credit rates, from our preferred movies to our movie preferences, all things information are represented using this minimal {0,1} alphabet which our digital helpers “understand” and process. Many of us physicists are now taking this information revolution at heart and embracing the “It from qubit” motto. Our dream: to understand space, time and gravity as emergent features in a world made of information – quantum information.

Over the past two years, I have been obsessively trying to understand this profound perspective more rigorously. Recently, John Preskill and I have taken a further step in this direction in the recent paper: quantum code properties from holographic geometries. In it, we make progress in interpreting features of the holographic approach to quantum gravity in the terms of quantum information constructs. 

In this post I would like to present some context for this work through analogies which hopefully help intuitively convey the general ideas. While still containing some technical content, this post is not likely to satisfy those readers seeking a precise in-depth presentation. To you I can only recommend the masterfully delivered lecture notes on gravity and entanglement by Mark Van Raamsdonk.  

Entanglement as a cat’s cradle


A cat’s cradle serves as a crude metaphor for quantum mechanical entanglement. The full image provides a complete description of the string and how it is laced in a stable configuration around the two hands. However, this lacing does not describe a stable configuration of half the string on one hand. The string would become disentangled and fall if we were to suddenly remove one of the hands or cut through the middle.

Of all the concepts needed to explain emergent spacetime, maybe the most difficult is that of quantum entanglement. While the word seems to convey some kind of string wound up in a complicated way, it is actually a quality which may describe information in quantum mechanical systems. In particular, it applies to a system for which we have a complete description as a whole, but are only capable of describing certain statistical properties of its parts. In other words, our knowledge of the whole loses predictive power when we are only concerned with the parts. Entanglement entropy is a measure of information which quantifies this.

While our metaphor for entanglement is quite crude, it will serve the purpose of this post. Namely, to illustrate one of the driving premises for the holographic approach to quantum gravity, that the very structure of spacetime is emergent and built up from entanglement entropy.

Knit and crochet your way into the manifolds

But let us bring back our metaphors and try to convey the content of this identification. For this, we resort to the unlikely worlds of knitting and crochet. Indeed, by a planned combination of individual loops and stitches, these traditional crafts are capable of approximating any kind of surface (2D Riemannian surface would be the technical term).

Here I have presented some examples with uniform curvature R: flat in green, positive curvature (ball) in yellow and negative curvature (coral reef) in purple. While actual practitioners may be more interested in getting the shape right on hats and socks for loved ones, for us the point is that if we take a step back, these objects built of simple loops, hooks and stitches could end up looking a lot like the smooth surfaces that a physicist might like to use to describe 2D space. This is cute, but can we push this metaphor even further?

Well, first of all, although the pictures above are only representing 2D surfaces, we can expect that a similar approach should allow approximating 3D and even higher dimensional objects (again the technical term is Riemannian manifolds). It would just make things much harder to present in a picture. These woolen structures are, in fact, quite reminiscent of tensor networks, a modern mathematical construct widely used in the field of quantum information. There too, we combine basic building blocks (tensors) through simple operations (tensor index contraction) to build a more complex composite object. In the tensor network world, the structure of the network (how its nodes are connected to other nodes) generically defines the entanglement structure of the resulting object.


This regular tensor network layout was used to describe hyperbolic space which is similar to the purple crochet. However, they apriori look quite dissimilar due to the use of the Poincaré disk model where tensors further from the center look smaller. Another difference is that the high degree of regularity is achieved at the expense of having very few tensors per curvature radius (as compared to its purple crochet cousin). However, planarity and regularity don’t seem to be essential so the crochet probably provides a better intuitive picture.

Roughly speaking, tensor networks are ingenious ways of encoding (quantum) inputs into (quantum) outputs. In particular, if you enter some input at the boundary of your tensor network, the tensors do the work of processing that information throughout the network so that if you ask for an output at any one of the nodes in the bulk of the tensor network, you get the right encoded answer. In other words, the information we input into the tensor network begins its journey at the dangling edges found at the boundary of the network and travels through the bulk edges by exploiting them as information bridges between the nodes of the network.

In the figure representing the cat’s cradle, these dangling input edges can be though of as the fingers holding the wool. Now, if we partition these edges into two disjoint sets (say, the fingers on the left hand and the fingers on the right hand, respectively), there will be some amount of entanglement between them. How much? In general, we cannot say, but under certain assumptions we find that it is proportional to the minimum cut through the network. Imagine you had an incredible number of fingers holding your wool structure. Now separate these fingers arbitrarily into two subsets L and R (we may call them left hand and right hand, although there is nothing right or left handy about them). By pulling left hand and right hand apart, the wool might stretch until at some point it breaks. How many threads will break? Well, the question is analogous to the entanglement one. We might expect, however, that a minimal number of threads break such that each hand can go its own way. This is what we call the minimal cut. In tensor networks, entanglement entropy is always bounded above by such a minimal cut and it has been confirmed that under certain conditions entanglement also reaches, or approximates, this bound. In this respect, our wool analogy seems to be working out.


Holography, in the context of black holes, was sparked by a profound observation of Jacob Bekenstein and Stephen Hawking, which identified the surface area of a black hole horizon (in Planck units) with its entropy, or information content:BHentropyF1

S_{BH} = \frac{k A_{BH}}{4\ell_p^2} .

Here, S_{BH} is the entropy associated to the black hole, A_{BH} is its horizon area, \ell_p is the Planck length and k is Boltzmann’s constant.
Why is this equation such a big deal? Well, there are many reasons, but let me emphasize one. For theoretical physicists, it is common to get rid of physical units by relating them through universal constants. For example, the theory of special relativity allows us to identify units of distance with units of time through the equation d=ct using the speed of light c. General relativity further allows us to identify mass and energy through the famous E=mc^2. By considering the Bekenstein-Hawking entropy, units of area are being swept away altogether! They are being identified with dimensionless units of information (one square meter is roughly 1.4\times10^{69} bits according to the Bousso bound).

Initially, the identification of area and information was proposed to reconcile black holes with the laws of thermodynamics. However, this has turned out to be the main hint leading to the holographic principle, wherein states that describe a certain volume of space in a theory of quantum gravity can also be thought of as being represented at the lower dimensional boundary of the given volume. This idea, put forth by Gerard ‘t Hooft, was later given a more precise interpretation by Leonard Susskind and subsequently by Juan Maldacena through the celebrated AdS/CFT correspondence. I will not dwell in the details of the AdS/CFT correspondence as I am not an expert myself. However, this correspondence gave S. Ryu and T. Takayanagi  (RT) a setting to vastly generalize the identification of area as an information quantity. They proposed identifying the area of minimal surfaces on the bulk (remember the minimal cut?) with entanglement entropy in the boundary theory.

Roughly speaking, if we were to split the boundary into two regions, left L and right R it should be possible to also partition the bulk in a way that each piece of the bulk has either L or R in its boundary. Ryu and Takayanagi proposed that the area of the smallest surface \chi_R=\chi_L which splits the bulk in this way would be proportional to the entanglement entropy between the two parts

S_L = S_R = \frac{|\chi_L|}{4G} =\frac{|\chi_R|}{4G}.

It turns out that some quantum field theory states admit such a geometric interpretation. Many high energy theory colleagues have ideas about when this is possible and what are the necessary conditions. By far the best studied setting for this holographic duality is AdS/CFT, where Ryu and Takayanagi first checked their proposal. Here, the entanglement features of  the lowest energy state of a conformal field theory are matched to surfaces in a hyperbolic space (like the purple crochet and the tensor network presented). However, other geometries have been shown to match the RT prediction with respect to the entanglement properties of different states. The key point here is that the boundary states do not have any geometry per se. They just manifest different amounts of entanglement when partitioned in different ways.


The holographic program suggests that bulk geometry emerges from the entanglement properties of the boundary state. Spacetime materializes from the information structure of the boundary instead of being a fundamental structure as in general relativity. Am I saying that we should strip everything physical, including space in favor of ones and zeros? Well, first of all, it is not just me who is pushing this approach. Secondly, no one is claiming that we should start making all our physical reasoning in terms of ones and zeros.

Let me give an example. We know that the sea is composed mostly of water molecules. The observation of waves that travel, superpose and break can be labeled as an emergent phenomenon. However, to a surfer, a wave is much more real than the water molecules composing it and the fact that it is emergent is of no practical consequence when trying to predict where a wave will break. A proficient physicist, armed with tools from statistical mechanics (there are more than 10^{25} molecules per liter), could probably derive a macroscopic model for waves from the microscopic theory of particles. In the process of learning what the surfer already understood, he would identify elements of the  microscopic theory which become irrelevant for such questions. Such details could be whether the sea has an odd or even number of molecules or the presence of a few fish.

In the case of holography, each square meter corresponds to 1.4\times10^{69} bits of entanglement. We don’t even have words to describe anything close to this outrageously large exponent which leaves plenty of room for emergence. Even taking all the information on the internet – estimated at 10^{22} bits (10 zettabits) – we can’t even match the area equivalent of the smallest known particle. The fact that there are so many orders of magnitude makes it difficult to extrapolate our understanding of the geometric domain to the information domain and vice versa. This is precisely the realm where techniques such as those from statistical mechanics successfully get rid of irrelevant details.

High energy theorists and people with a background in general relativity tend to picture things in a continuum language. For example, part of their daily butter are Riemannian or Lorentzian manifolds which are respectively used to describe space and spacetime. In contrast, most of information theory is usually applied to deal with discrete elements such as bits, elementary circuit gates, etc. Nevertheless, I believe it is fruitful to straddle this cultural divide to the benefit of both parties. In a way, the convergence we are seeking is analogous to the one achieved by the kinetic theory of gases, which allowed the unification of thermodynamics with classical mechanics.

So what did we do?

The remarkable success of the geometric RT prediction to different bulk geometries such as the BTZ black holes and the generality of the entanglement result for its random tensor network cousins emboldened us to take the RT prescription beyond its usual domain of application. We considered applying it to arbitrary Riemannian manifolds that are space-like and that can be approximated by a smoothly knit fabric.

Furthermore, we went on to consider the implications that such assumptions would have when the corresponding geometries are interpreted as error-correcting codes. In fact, our work elaborates on the perspective of A. Almheiri, X. Dong and D. Harlow (ADH) where quantum error-correcting code properties of AdS/CFT were laid out; it is hard to overemphasize the influence of this work. Our work considers general geometries and identifies properties a code associated to a specific holographic geometry should satisfy.

In the cat cradle/fabric metaphor for holography, the fingers at the boundary constitute the boundary theory without gravity and the resulted fabric represents a bulk geometry in the corresponding bulk gravitational theory. Bulk observables may be represented in different ways on the boundary, but not arbitrarily. This raises the question of which parts of the bulk correspond to which parts of the boundary. In general, there is not a one to one mapping. However, if we partition the boundary in two parts L and R, we expect to be able to split the bulk into two corresponding regions  {\mathcal E}[L]  and  {\mathcal E}[R]. This is the content of the entanglement wedge hypothesis, which is our other main assumption.  In our metaphor, one could imagine that we pull the left fingers up and the right fingers down (taking care not to get hurt). At some point, the fabric breaks through \chi_R into two pieces. In the setting we are concerned with, these pieces maintain part of the original structure, which tells us which bulk information was available in one piece of the boundary and which part was available in the other.

Although we do not produce new explicit examples of such codes, we worked our way towards developing a language which translates between the holographic/geometric perspective and the coding theory perspective. We specifically build upon the language of operator algebra quantum error correction (OAQEC) which allows individually focusing on different parts of the logical message. In doing so we identified several coding theoretic bounds and quantities, some of which we found to be applicable beyond the context of holography. A particularly noteworthy one is a strengthening of the quantum Singleton bound, which defines a trade-off between how much logical information can be packed in a code, how much physical space is used for encoding this information and how well-protected the information is from erasures.

One of the central observations of ADH highlights how quantum codes have properties from both classical error-correcting codes and secret sharing schemes. On the one hand, logical encoded information should be protected from loss of small parts of the carrier, a property quantified by the code distance. On the other hand, the logical encoded information should not become accessible until a sufficiently large part of the carrier is available to us. This is quantified by the threshold of a corresponding secret sharing scheme. We call this quantity price as it identifies how much of the carrier we would need before someone could reconstruct the message faithfully. In general, it is hard to balance these two competing requirements; a statement which can be made rigorous. This kind of complementarity has long been recognized in quantum cryptography. However, we found that according to holographic predictions, codes admitting a geometric interpretation achieve a remarkable optimality in the trade-off between these features.

Our exploration of alternative geometries is rewarded by the following guidelines


In uberholography, bulk observables are accessible in a Cantor type fractal shaped subregion of the boundary. This is illustrated on the Poincare disc presentation of negatively curved bulk.

  • Hyperbolic geometries predict a fixed polynomial scaling for code distance. This is illustrated by a feature we call uberholography. We use this name because there is an excess of holography wherein bulk observables can be represented on intricate subsets of the boundary which have fractal dimension even smaller than the boundary itself.
  • Hyperbolic geometries suggest the possibility of decoding procedures which are local on the boundary geometry. This property may be connected to the locality of the corresponding boundary field theory.
  • Flat and positive curvature geometries may lead to codes with better parameters in terms of distance and rates (ratio of logical information to physical information). A hemisphere reaches optimum parameters, saturating coding bounds.


    Seven iterations of a ternary Cantor set (dark line) on the unit interval. Each iteration is obtained by punching holes from the previous one and the set obtained in the limit is a fractal.

Current day quantum computers are far from the number of qubits required to invoke an emergent geometry. Nevertheless, it is exhilarating to take a step back and consider how the properties of the codes, and information in general, may be interpreted geometrically. On the other hand, I find that the quantum code language we adapt to the context of holography might eventually serve as a useful tool in distinguishing which boundary features are relevant or irrelevant for the emergent properties of the holographic dual. Ours is but one contribution in a very active field. However, the one thing I am certain about is that these are exciting times to be doing physics.