Jack Krupansky, a long-time observer of the quantum computing industry, is interviewed by Yuval Boger. Jack rates the quantum progress in 2023, revisits his Christmas wish for 48 fully-connected qubits, discusses whether quantum computing is an industry or a sector, provides advice for those entering quantum, discusses his definition of a quantum ChatGPT moment, and much more.
Full Transcript
Yuval Boger: Hello, Jack. Thank you for joining me today.
Jack Krupansky: Thanks, Yuval. Glad to be here, and I’m looking forward to whatever questions you have to throw my way.
Yuval: So, who are you, and what do you do?
Jack: Okay, I’m Jack Krupansky. I hope I don’t need to spell that. I’m a former software developer. I’m semi-retired. I’ve worked in a lot of different companies, a lot of different kind of areas of computing. I’ve worked for computer companies. I’ve worked for startups. I’ve worked for a little bit of applications, but I’m more of a technologist than an applications guy. So, in the case of quantum computers, I don’t have an application where I need to figure out how to use this computer for my application. I’m interested in three things with technology. I’m interested in what its capabilities are. What can it do? What are its limits? What maybe can’t it do? And what are all the issues, all the things that kind of make it hard? Like with quantum computers, they can do a lot, but it’s hard to do much at all. So I’m trying to figure out, spending last, it’s been over six years now. Trying to figure out where are quantum computers really, what can they and really what can’t they do? What can we really expect from them?
Yuval: About a year ago, I read one of your missives, and you said, “Oh, I’m finishing up my exploration of quantum computing, and I’m going to wind down.” And here we are a year later, and I don’t know how many missives later. How come you’re still here? How come you’re still interested in quantum?
Jack: Well, that’s a great question, of course. Well, first, I go back to, say, five or six years ago when it was just a year from when I started. And it’s like, it just always felt that, gee, if another year or two go by, and quantum will finally be there. And a year ago, it was like, well, I’ve been saying this for every year, for every two years saying that we’re almost there. And it just got to the point where it just started to feel like we weren’t there. But there’s still a lot of, I was going to say noise flow, but news flow. And people doing advances, research advances, people putting out products. And so it always seemed like there was, it’s like a carrot dangled in front of me. It’s like, just stay with it a little longer, and it’ll be there a little bit longer.
So right now, where am I? Well, IBM had their big dog and pony show, and at the end of December, a few other announcements came out since then. There were a bunch of announcements in the fall. I’m almost in the mode of I want to just see what happens over the next few months, next three months or so, and just kind of judge from that whether we have, we have a lot of activity, but not necessarily a lot of advances. You know, it’s sometimes, somebody puts out a new computer and it’s got, oh, it’s got 20 qubits. It’s like, well, weren’t we there like five years ago? But it’s somehow there’s 20 qubits that are a little bit different. So why am I still here? I mean, it’s just, it just feels like there’s a little bit more and I’m semi-retired, like I said, so I really don’t have anything better to do. It’s not like I have a job that needs the computer today. And if it doesn’t have it today, then there’s no way to look at it. There’s just something else. There’s always something new brewing. And I am really interested in the research and there’s, regardless of what’s available for practical uses, there’s always a lot of interesting research going on.
Yuval: So in your mind lots of motion, little progress.
Jack: Sorry, where for me progress is that we’re getting really close to a practical quantum computer. So we’re making progress like we’re making the early steps in a long journey, we’re making that kind of progress, but we’re not making the progress where we’re near the end of the journey. That’s what I’m looking for.
Yuval: If you were to start your quantum explorations again, I think you said you started them six years ago. Where would you start? Or I get a lot of questions from people who say, I want to get into quantum. How do I get started in a meaningful way? What would your answer be?
Jack: Well, I don’t actually… well, two things, I don’t have a good answer because I just came up with this just recently, that gee, resetting or restarting my quantum journey and saying from scratch, I knew nothing. What can I learn? I think one of the first things I do is say, go back to how I got started, seeing accounts in the media, general media or the tech media and saying, oh, quantum computers can do this and that. And I look at it and say, well, what can I tell from this report? And if I see some keywords, let me do a Google search on these keywords and see what pops up in Google. And just like, how would you do a journey of exploration when you know nothing? Like if you came to the new world with Columbus, it’s like there’s no map. What do you do? How do you go and search? But also look at some of the how-to guys, ask the question, well, now since we have chat GPT, I can use Bing chat. I can say I want to learn about quantum computing. Where should I start? And it’ll give me a much more interesting answer than a Google search would. I mean, actually, the Google search might come up. I’ll compare the two. The Google search might actually have the first item be the right actual place to start. But with a chat GPT like Bing chat, it’s going to give me a little bit more of a little deeper sense of, you know, that there might be more than one place to start, for example.
Yuval: Since you brought up GPT, do you see quantum as intersecting significantly with AI? Or do you think there’s just a ploy to get more money from VCs that were so excited about AI, and maybe they have something left for quantum?
Jack: Well, I’ve been, you know, at the beginning, six years ago, I knew nothing. And I said, well, okay, people are saying quantum computers are going to be good for AI. I mean, Google’s whole effort from the beginning and still today is billed as quantum AI. I mean, they’re the quantum AI group. They’re linked together, and they won’t separate the two. But my feeling now that I’ve passed through all this time is through all these different stages of my journey, is that, first of all, a lot of this AI needs, like LLMs, the large language models, need big data. Well, quantum computers don’t do big data well. They don’t do it at all, whether they could eventually, that’s a separate discussion. But if you can’t do the big data, then you can’t do an LLM. So then, what can you do with AI? And the real point of quantum computer, the way I look at it, is it’s small data with a large solution space. You feed a little bit of data in, like a number you want to factor, and then the computer is going to process a lot of different possibilities that you don’t have to feed in. They’re just looking in a large solution space. So the question is, for AI, are there small data problems like that in AI, where there’s a small amount of input that can be processed over a large solution space but without requiring big data? And so where we are right now is, I don’t see any problems like that in AI, where you can say, I mean, you could do a little thing where you say, here’s an image that’s got like 25 pixels. Compare that against this other image with 25 pixels. But that’s a little data problem, not a high-intensity compute problem. So that’s where I think we are right now.
Yuval: But wouldn’t a lot of people say that quantum could be used to generate data that could be used to teach AI? So synthetic data, more test cases, true randomness, sort of generative quantum adversarial networks?
Jack: Well, there’s two little issues there. One is, you said randomness and random numbers. Well, that’s certainly, regardless of what I have to say about quantum applications, quantum computers use another application, and what applications can be used in. Generating true random numbers is the one single thing that quantum computers can do that classical computers can’t do it all, and they can do it well today. So we don’t have to wait two years or ten years to generate true random numbers. Now, there’s some caveats on that. But classical computers can generate pseudo random numbers, which for most applications are good enough. So that’s kind of where that is on it.
As far as generation, as long as it’s a small data generation, you’re not going to generate a 50 gigabyte video movie from a quantum computer. Although at some point, maybe you could, because the integration of sensors, one of the ideas I have is that if you could integrate sensors with a quantum computer where underneath some of the qubits is an actual quantum sensor, then you could be processing quantum sense data directly in quantum without having to go through a conversion from this quantum sensor to classical data and then trying to figure out how to get the classical data into a quantum computer. Well, do the flip side of that as well. In what ways could we have the qubits actually generate data right there at that qubit level? Because now, even if you’re on a slow quantum computer, you can still be generating thousands of images a second, as long as you don’t have to feed them out as classical data. So there’s an option there. But to me, that’s a research project, and I haven’t heard anybody even talking about it, but it’s there.
Yuval: We’re recording this close to New Year. In your opinion, was 2023 a good year for quantum? An average year, a bad year?
Jack: Well, I’ve tried to think about that, and it’s difficult. So, I mean, I don’t want to be too negative, but I’m unable to be too positive either. So maybe I would say it’s like an average year. Like if I look at the last six years, I mean, when IBM went from five qubits to 17 qubits, that was a huge leap. When they got to 27-bit Falcon, I mean, even today, I’m not so sure that the 127-bit Eagle gives most people much benefit over a 27-bit Falcon. So when we get to Falcon, that’s a really big deal. But since then, I don’t think we’ve given people a lot more. Simulators may be a little bit better, but maybe not a lot better. I think if I had to say, pick one thing I wanted to see in 2024, other than my 48-bit quantum computer, that would be that we want to see much better simulation so that we can give people a tool that actually helps them debug their programs. Because that’s the problem with a quantum computer ultimately is if you use real hardware, you don’t know what happened with the quantum state in the middle. We just saw the bits at the end. That’s a real difficulty, and you can’t really debug your code, and you can’t simulate the whole thing if you’re more than 50 qubits.
So anyway, maybe you want to ask that, see if I follow up with that same question, because I may have gotten too far away from the real question there. You’re saying, was it an average year or a better year? So I feel kind of average because we didn’t have the gigantic leaps that we saw in the past. And now we’re struggling. We want to see, like, okay, so maybe I’ll call it a less than average year because I thought at the beginning of the year from a year ago, IBM was almost promising us that we’d see three nines of qubit fidelity with Heron, and it came up short of that. And 18 months ago, they had promised us that the 27-bit processors were going to hit. They said they saw in the lab that the quantum volume hit 256, and then a couple of months later hit 512. And they promised us, I mean, Jay made a comment on LinkedIn I saw to me that they thought they were going to hit 1024 by the end of the year, which is last Summit last November. And even now where we are now, the latest dashboard shows that Falcon is only sitting at 128. We never saw the 1024, Heron is listed as 512, but IBM has kind of given us mixed messages as to whether or not they’re going to continue pushing quantum volume. They’ve kind of hidden it now. It’s still there. Anyway, so that leads me back to the original question of kind of, I guess I feel like it’s less than I hoped for a year. It has a lot of advances, but I just wished there was more progress.
Yuval: One of the things we have seen this year is people starting to build what they consider factories to build quantum computers. Do you think that’s a sign that the computer engineers are taking over from the physicists? And is that a good thing in your mind?
Jack: You mean like, for example, IonQ? They’re building a factory in the Seattle area, or something? OK, I remember seeing that. And again, it’s probably a mixed message thing. I think it is a good sign, but I’m not sure how well they’ll do and I’m not sure if it’ll change the net result. See, I think the computers, quantum computers aren’t ready to leave the lab. So I would rather leave them in the lab and focus on how to make science advances in the quantum computer rather than trying to figure out how to manufacture them as cheaply as possible. I don’t think a cheaper IonQ machine or if IBM’s big behemoth, the IBM Quantum System 2, which is like ten times bigger than I thought it was going to be. If you manufacture that ten times more efficiently, I don’t think it would it’s not going to help the people who are really struggling with algorithms. And I think that’s where the main issue is, people want to do bigger algorithms with more data and less noise. And manufacturing the computers faster or cheaper isn’t going to improve qubit quality.
Yuval: You mentioned quantum volume and that seems to be like a little bit of a two-dimensional benchmark, because once you’re starting to have qubits that are better and better, then the number of qubits could be much smaller than the number of steps. So you have to move to more of a rectangular benchmark than a square one. Do you still see quantum volume as something that’s going to be useful going forward?
Jack: Another mixed message issue. At the beginning, I was not enthusiastic about quantum volume at all. It’s ridiculous because they give you this number, and the very first thing you have to do is math. You do the logarithm base two of the number to find the number of qubits. That’s all it’s really telling you is how many qubits you can use in an algorithm and expect to get a reasonable result. So if you had a quantum volume of 128, logarithm base two, number of bits is seven. So if you use seven qubits, you can pretty much count on getting a reasonable result. The second thing it did was the whole point of having that two to the exponential in there that I took the logarithm of was take seven, two to the seventh power that allows you to do have 128 quantum states. And that’s the computing power or what I call quantum parallelism of that configuration. So if I have 20 qubits now, two to the 20th is a million, that means I can essentially evaluate with a Grover’s algorithm or whatever, a million possibilities all at the same time. And so it was a useful metric in that sense for those two senses. It has a definite end date or end size. I mean, once you get to you can do 50 qubits with high quality, you can no longer simulate that algorithm. And the essence of how they define the benchmark requires simulation. It requires you to do a classical simulation. So if you were 50 or 55 qubits and they work just fine, you wouldn’t be able to simulate the 55 qubits because it’s too many quantum states to fit on even a large computer.
Oh, so but you asked about a second dimension. So I don’t like quantum volume per se, has some utility. I don’t like two-qubit gate error rate per se, but it has some utility. The Q-CTRL people, they like a variety of benchmarks, and I agree with that. But I think these two, each of these benchmarks gives you a point and we need a bunch of points to get a big good picture of what your computer can do. One of the benchmark points that I want to see and Q-CTRL has on their list is the quantum Fourier transform and the quantum phase estimation, because that’s telling you a lot about how much you can do in parallel. And it requires connectivity, you know, like full connectivity. If you don’t have full connectivity, it’s not going to work very well. So, a variety of benchmarks. I don’t think we’re ready for true application benchmarks like machine learning because we don’t have the real applications yet. And I don’t I don’t like variational methods, so I don’t want a benchmark that’s benchmarking variational methods. It doesn’t tell me anything useful. But I think having these algorithmic building block benchmarks like quantum Fourier transform, I think that’s useful.
Yuval: Some are asking whether quantum computing is an industry or a sector. One, does it matter? And two, what criteria would you use to evaluate the answer?
Jack: OK, I was going to ask that question or if I hadn’t already. I think of an industry as like a big thing that’s it’s also kind of stodgy, and it’s stagnant. It’s stable, like the car industry. You know, it’s a clear thing. It’s obvious and doesn’t really change. Although you have electric cars that you know, that’s a kind of a change of batteries now to worry about charging stations, etc. So quantum computing. Well, as I said earlier, it’s still captive to the physicists. OK, so I don’t know if I would want to take anything that’s physics and say it’s an industry. I mean, because physics should be a science, not an industry. So it just kind of feels weird. When quantum computers get turned over to the engineers, the computer engineers, and the computer software engineers and operating system designers, then we’ll have quantum computing as an industry. Maybe at that stage it might be an industry, but should it be a separate industry? But I don’t think you want it to be because I don’t think you want when we get to real classical, real quantum computers doing practical work, I think we’re then going to start looking at wanting to integrate classical and quantum much more. Not just that there’s a hybrid, but that it’s all in one industry. OK, so we already have a computer industry. I don’t think we want this other industry and the classical computer industry to be separate. So I think that’s where I think of the bottom line here is I think of quantum computing as ultimately a sector of the computing industry, even though it’s currently captive to the physics scientists.
Yuval: A year ago, you published your wishes for Christmas, and I think one of them was 48 fully connected, good enough qubits. And then, this Harvard group, together with QuEra and MIT and NIST and UMD published 48 fully connected, logical qubits. Not perhaps good enough yet, but how do you feel about that publication?
Jack: Well, actually, I did some posts on the Harvard experiment and paper. Do I have to give a long list of people who participated? Harvard slash QuEra. Actually, simple question for you. Is it QuEra or Q-era? Which do you prefer?
Yuval: I say QuEra, but it’s up to you.
Jack: Well, because you had a podcast with the guy at Harvard who’s the PI for the project, and it sounded as if he was saying Q-era.
Yuval: But anyway, either way is fine.
Jack: All right. So, repeat the question again just if I get the words right.
Yuval: You mentioned 48 logical, 48 fully connected, good enough qubits a year ago. The Harvard paper shows something that’s not perhaps too far from that. How do you feel about that?
Jack: OK, so let me start the question over there that’s fresh with that. So I had written a bunch of comments about reading the paper and reading on that project and reading the press release and all sorts of things and some initial comments. And I pretty much stick by them. But then it occurred to me after the fact is like, well something that should have stood out right at the beginning, but I just ignored was 48 qubits. Well, that’s what I was asking for Christmas a year ago: 48 qubits. In fact, I did do a Christmas wish list this year for next year, and that’s still 48 qubits. OK, so how does it relate to the Harvard paper? Well, the 48 qubits are there. The fully connected is there. One of my open questions is still what is the effective two qubit gate error rate? And I don’t know what the answer is. That’s one of the things that’s kind of missing. It’s a missing, not missing link, but it’s missing piece of data. I have like 10 different 10 or 12 different questions that I have with a project that kind of will need to go into at some point to get the answers. So I don’t know how perfect they are. And from reading Scott Aronson, it sounded like they aren’t doing full error correction, the full path. So we’re not looking at a full algorithm yet, really. So it’s still too early. So it’s too early to actually deliver as a Christmas present that a child could use. Put it that way.
Yuval: You mentioned GPT earlier. And to me, in the evolution of AI, there were two really important moments. One was the AlphaGo moment where AI beat the world champion in Go and the other one, of course, was the chat GPT where it became ubiquitous and lots of people saw the potential of AI. What do you see as the chat GPT moment of quantum and how soon would you estimate that would come?
Jack: Okay. So, yeah, first, I don’t feel that we’ve hit a chat GPT moment. So just to get that on the table. Anytime soon, probably not. So let me take you back to two things kind of classically. I’ve actually written something I saw, which I titled something to the effect, words to the effect. When will quantum computing hit its ENIAC moment? So this, I wrote this before chat GPT even came up. And so the ENIAC moment is when the ENIAC computer was produced and it could actually do calculations that people could sort of understand. It was calculating artillery firing tables. Not that people understand what that really means, unless you’re in the Israeli military, then you probably know that kind of thing. But it was something that people could say, “Oh, okay, I can guess what that is.” And so it was the first time we had a computer doing something really kind of interesting that people could relate to. Although there was another moment, the 1952 presidential election, where they used CBS News, I believe, and they had, who was the famous broadcaster [Charles Collingwood]. Anyway, they used the whirlwind computer, which is MIT [actually, it was a Remington Rand UNIVAC – Edward R. Murrow did a CBS story about the MIT Whirlwind computer in 1951], to forecast the winner of the outcome of the election. So that was something that people could really relate to. But if you do ground states of weird molecules or many body system simulations, these are things, almost anything you can do in physics, people aren’t going to relate to at all
So the question then is like my ENIAC moment, you’ve got sufficient hardware, you’ve got sufficient capability, quality, and capability to do algorithms, and you’ve identified some problem that people can relate to, like chat GPT. You just type in anything and it does something interesting. It may be hallucination, but it does something interesting with very little work. And so we’re way far away from being able to do anything with little work. But my definition of the ENIAC moment, I have three things. There’s the ENIAC moment, there is something in the middle called the configurable package quantum solutions. And then the third thing is the FORTRAN moment. And the FORTRAN moment would be the time when a lot of average IT organizations could use quantum computers. But with the ENIAC moment, I’m saying it doesn’t have to be average people. These could be the super-elite. But at least the super-elite are able to produce something that actually does something that looks practical. So that’s the thing I’m looking for. It’s still out there. It’s just a pie in the sky off in the future. I don’t even have a timeline anymore.
If I went back to what I wrote, I may have said maybe by now we’d see something. Are we going to see something next year? Well, it’s possible. I wouldn’t hold my breath in two years. I mean, I think it comes back to that qubit quality. We need to get past this three nines barrier that even IBM failed to do with Heron. We need to get up to three and a quarter, three and a half. We don’t have to go all the way to four, but maybe we’ll get to three and a half. We’ll have enough qubit quality where somebody could write what I call a 40 qubit algorithm that actually does something. It may simulate the simplest module, but at least it actually does the whole thing instead of doing this kind of smoke and ears kind of simulation as we do today with quantum, where we know the answer. So we only ask it to solve problems where we know the answer. Whereas like chat GPT, you could ask it a question about the origin of the universe or the destination of humanity, and it’ll come up with something.
Yuval: There’s a famous 1976 movie “Network”, and then the character Howard Beale, a fictional broadcaster, stands and says, “I’m mad as hell, and I can’t take it anymore.” Do you feel you’re the Howard Beale of quantum?
Jack: Actually, here in DC, there is a think tank where they use a regular movie theater, and they just rent out of room, a theater, and they show old science movies and geopolitical movies and have a discussion afterward. And one of the movies they showed was Network, and people got to talk about that. So I had never seen it when it originally came out. I wasn’t in the era when I was watching a lot of movies like that. But watching it, so the scene was, he does it on the air, and then people go to their windows. We have all these people in the apartment buildings in New York City going to their windows, leaning out the window, saying, “I’m mad as hell, I’m not going to take it anymore.” But the answer to your simple question is, no, I’m not quite at that stage yet. But maybe it’s coming up, but I’m being philosophical, and I don’t have skin in the game where it matters to me one way or the other. So it’s like my original question, how I got started in this journey was, gee, people are starting to do stuff with quantum computers. How real are they? How realistic a problem can you do? And I’m just asking the question, and then, well, I’m curious what the answer is. My life is dependent on what the answer is. Because I think I would say about classical computers, sure, there are limits there, and Moore’s Law is slowing down, blah, blah, blah, but there’s still plenty of runway ahead. And as even IBM, Jay Gambetta, has acknowledged that as IBM does larger algorithms on their Eagle with 100 qubits or whatever, that’s feeding back into classical, and the classical people are then upping their game. So if quantum is, it does nothing else, if at a minimum, it’s just offering an incentive for the classical people to get their act together, then that’s a big win. So anyway, I’m not ready to jump up and down and pull my hair out yet.
Yuval: There’s still a good number of quantum modalities where people are trying to build qubits this way or that way and so on. I know you’ve been spending a good bit of time looking at IBM. Are you more optimistic about some modalities than others?
Jack: Well, the first simple answer is, unfortunately today, not really, but there had been some pluses, and I had been optimistic, for example, initially I was optimistic about silicon spin qubits with Intel. Intel kind of stumbled with Transmons initially. They had a 49-qubit machine they were building. I don’t know if they actually built it or not. And then they switched horses in midstream and bid on, at the farm, on silicon spin qubits, but they haven’t done enough. They’ve done some, had some, a lot of activity, and they did have a 12-qubit machine last year, but it’s just too lackluster so far. So there are other people in that area, but I’m not seeing the kind of level of activity that, well, for me, would be progress. So that’s one. There was another one with fluxonium qubits. I won’t name the names. I don’t want to embarrass anybody, except I embarrassed Intel just there. They need to up their game, you know, accelerate or something. Either get in full force or go big or go home, you know, that kind of mentality. So fluxonium qubits are very similar to the Transmon, but have some qualities where they were promising an order of magnitude improvement qubit quality. Well, we haven’t really seen that yet. And then the last I heard from them, the company, one company, the main company was pursuing that, was that they were going to, they signed a deal to go whole hog into quantum error correction, which to me was like the kiss of death for them. It’s like they’re going to be distracted by quantum error correction and not focus on qubit quality. Because the idea is, well, once you get two nines or three nines of qubit quality, you don’t have to bother anymore because quantum error correction will do the rest for you, which I don’t quite believe in. But that’s a very enticing, very alluring come-on for people who are struggling to make their qubits faster. I would adopt faster, but the higher quality. And I would rather see people in the lab making qubits of higher quality than I would focusing on quantum error correction, which ultimately is still, well, with IBM, they’re saying 10 years, actually nine years. It was 10 years to get to 2,000 logical qubits, but it was only nine years to get to 200 logical qubits. So maybe in eight years, they could have something that’s actually useful. But that’s a long time to wait. And it’s like, why not make better qubits now?
Yuval: As we get closer to the end of our conversation, I know you live in DC, do you worry about the geopolitical aspects of quantum, you know, the sort of space race between different countries or different continents as it relates to quantum, about the quantum initiatives of the U.S. and others?
Jack: Well, actually, just a separate kind of just two issues there with Washington. There’s the international aspects, the competitive stuff you were just getting to. And then there’s the domestic. Okay. Even if we just ignore the rest of the world, which in general, I think the United States can safely do. We need, we want to have our industry be the best it can be. And so I’m in favor of some national efforts, but I’m more of a private sector guy than a public sector guy. I think where the government could help most is with the government, the applications that the government needs. I mean, that’s how ENIAC came about. You know, they were doing firing tables, not for the general public, but for the military. And once they actually built the ENIAC, they didn’t have it on their roadmap, but they were just, they did an enhancement to the original machine, and then they were able to use it for some of the calculations for the hydrogen bomb. Okay. So again, that’s a government application.
So unfortunately, one of the biggest government applications is cracking foreign intelligence messages. So that’s not exactly the kind of, so it’s a mixed bag with the government. But then again, that’s a lot of this domestic or foreign policy, not foreign policy, but national security, which is how it looks outside, but it’s focused on the inside. So, back to your original question about the international competition, we have China. And besides China, what do you have? You have Europe. Well, we’re not trying to compete with Europe per se, but we are. I mean, it’s a friendly competition kind of thing. Whereas with China, we’re worried about national security implications. And what about Russia? Is Russia doing anything? I mean, as far as we can tell, I mean, I’m sure they’re doing something, but nobody ever talks about that. India is doing a fair amount. We don’t talk about that. And then you have, of course, Australia. Well, is Australia international or is it, you know, we have AUKUS, the United States, UK and Australia. So we’re almost like three countries that are almost semi-joined together. So there’s cooperation and competition too. So initially with a race, I don’t personally worry about a race or see a race. I think it’s more of a media way to get attention from the media. And maybe it does raise some money, help people raise some money. But every time someone says, look what China did, they just did a paper on boson sampling or whatever, which is kind of useless to me. But some people think that’s important.
So I can see both sides on that, but I’m not leaning towards the, you know, we’re in a race. It’s not like the Oppenheimer movie trailer, where we’re in a race, you know, the race against time in the Germans, blah, blah, blah. I don’t see that we’re in that. We don’t have any. But that’s actually negative, too, because if we were in such a race, then it’s clear like we have to get to a certain point in two years, not nine or ten years, in two years, which is what they did with the Apollo project. Or in the case of the Apollo program, sorry, that was the Manhattan project. But in the case of the space program with the Apollo program and Mercury in Gemini, you know, we had a whole decade, but that was a huge project for a decade. But we’re not even in that kind of situation. There’s no like urgency to … ten years, and we’re doing logical qubits doing what? We don’t even have a sense of what it is we’re going to be doing with these 2,000 logical qubits, other than maybe trying to crack Shor’s algorithm, which I don’t, which I think will ultimately fail anyway. But that’s a different story.
Yuval: I think you’ve listened to a good number of episodes of this podcast. So first, thank you very much for being a listener. I know you send me corrections on the transcript from time to time. So, you know, this question is coming. If you could have dinner with one of the quantum greats, dead or alive, who would that be?
Jack: Well, I was just trying to think of a good answer yesterday. I mean, well, first of all, as I’ve already indicated that the last time I said Feynman, which is the obvious answer. I still want to have dinner with Feynman like, well, do it maybe once a week, have dinner with Feynman just as a normal kind of thing, because there’s so many different things you could really get into. But I don’t want to say him again. It’s just it’s just kind of too pat an answer. And I was trying to figure out who I want to select. And I came up with a totally different kind of answer is I think I’d like to have dinner with like on a regular weekly basis with a bunch of young kids who were at different levels of starting, some who were just getting in, and they just have a taste and they just have they have a vision, maybe what the future is. And then some PhDs who have been doing the heavy research, and then they know what it can do and who’s going to be the next Feynman. So I want to have dinner with the next Feynman. How’s that for an answer?
So I think you’d have to talk to the kids, postdoctoral researchers, some of the some of the professors. I mean, I worry that we’ve been doing quantum now for so long that the leading lights, so to speak, are past their prime. I mean, they’ve done … I don’t want to name names, but they’ve done their best work and I don’t see them really adding a lot of value at this stage. And I’d like to see these postdocs in these in all states. I mean, people who are in the undergrad, the masters level and PhD level postdoc, the people who are really going to actually do the real breakthroughs that are actually going to make quantum computers practical. Even if you believe in quantum error correction, IBM keeps changing horses. You know, I mean, they at one point when I started when I wrote a paper on it, I read the IBM papers and they’re talking about 57 or 65 qubits for a logical qubit. But then a while later, they’re talking about the standard Google mantra of a thousand qubits, you know, and now recently they’ve with LDPC, I think it’s called a low-density parity check, something like that. Parity code, I know which it is. They’re talking about maybe 144 or 72 qubits. And it’s like, OK, that’s this year now, what maybe next year. And they’re not going to have this stuff for nine or 10 years, how many times might they change it in the middle there that, you know, we don’t really know where it’s going. And that’s kind of why I would prefer to have better qubits. And I think the guys working in the lab, the young Turks who need to put their mark on the world, they need to do something other than just extend the work of their predecessors. These tired old dinosaurs who have done their best work. You know, they need to be like you had people, people who Oppenheimer and his cohort, they mentored for. OK, sorry they were their proteges, their mentors, their mentors who passed their prime and didn’t really add any value to quantum mechanics. You know, OK, so they had to like they’re the ones who started it out. So that’s why I want to have dinner with the young people who are just really sinking their teeth into this and are really gung ho and who are going to be the next Feynman and the next Preskill and other guys. The next Chris Monroe and the next Dave Wineland.
Yuval: I did have an opportunity to chat with Professor Preskill at a trade show not too long ago and ask him this question. And he says, no, I had plenty of dinners with Feynman. But then he said, actually, I would love to have another one just to see how he feels about the progress in this industry. Thank you so much for joining me today.
Jack: OK. Yeah, it was great, Yuval, and I’ll maybe see you again next year. We’ll see if I’m still around.