We’re hearing this week from two very different parts of the string theory community that quantum supremacy (quantum computers doing better than classical computers) is the answer to the challenges the subject has faced.
New Scientist has an article Quantum computers could simulate a black hole in the next decade which tells us that “Understanding the interactions between quantum physics and gravity within a black hole is one of the thorniest problems in physics, but quantum computers could soon offer an answer.” The article is about this preprint from Juan Maldacena which discusses numerical simulations in a version of the BFSS matrix model, a 1996 proposal for a definition of M-theory that never worked out. Maldacena points to this recent Monte-Carlo calculation, which claims to get results consistent with expectations from duality with supergravity.
Maldacena’s proposal is basically for a variant of the wormhole publicity stunt: he argues that if you have a large enough quantum computer, you can do a better calculation than the recent Monte-Carlo. In principle you could look for quasi-normal modes in the data, and then you would have created not a wormhole but a black hole and be doing “quantum gravity in the laboratory”
seeing these quasinormal modes from a quantum simulation of the quantum system under discussion, would be a convincing evidence that we have created something that behaves as a black hole in the laboratory.
This isn’t a publicity stunt like the wormhole one, because the only publicity I’ve seen is a New Scientist article, and this is just a proposal, not actually executed. Maldacena estimates that to reproduce the recent Monte-Carlo calculation you’d need 7000 or so logical qubits, which the New Scientist reporter explains would be something like one million physical qubits. So, there’s no danger Quanta magazine will be producing videos about the creation of a black hole in a Google lab any time soon.
Maldacena has been chosen to give the presentation tomorrow at the SLAC P5 Town Hall about a vision for the future of fundamental theory, no idea whether creating black holes in the lab using quantum computers will be part of it.
At the other extreme of respectability and influence in the physics community, Michio Kaku has a new book out, Quantum Supremacy. I took a quick look yesterday at a copy at the bookstore. I’ll leave it to others to discuss the bulk of the book, which seems to be about how “There is not a single problem humanity faces that couldn’t be addressed by quantum computing.” The last few pages are about string theory, beginning with the usual bogus pro-string theory arguments, working up to the ending of the book: “So quantum computers may hold the key to creation itself” (i.e. they will “solve all the equations of string theory”). His argument for the relevance of quantum computers to string theory is that they will calculate paths in the landscape:
One day, it might be possible to put string theory on to a quantum computer to select out the correct path. Perhaps many of the paths found in the landscape are unstable and quickly decay, leaving only the correct solution . Perhaps our universe emerges as the only stable one.
This is justified by a bizarre paragraph about lattice gauge theory, which explains that since we can’t solve QCD analytically, here’s what theorists do:
One solves the equations for one tiny cube, uses that to solve the equations for the next neighboring cube, and repeats the same process for all that follow. In this way, eventually the computer solves for all the neighboring cubes, one after the other.
This pretty conclusively shows that the explanation for the Kaku phenomenon is simply that he has no idea what he is talking about.
Update: Michio Kaku was on a very recent Joe Rogan Experience, getting a huge audience for his explanations of quantum computing. Some commentary here.
Update: The reviews of the book have been pretty uniformly very enthusiastic, with the reviewers evidently incapable of distinguishing sense from nonsense. A depressing example is at Science magazine. Why choose as reviewers of a book on quantum computing two people who know nothing about the subject? Is it because Science couldn’t find anyone who does know about quantum computing willing to read the book and write about it?
Update: Scott Aaronson has read the book and confirms that it’s every bit as awful as it seems. For a different look at out-of-control quantum computing hype, see here
Update: The one thing keeping my spirits up while reading the almost uniformly glowing reviews of this piece of junk has been the thought that “at least the New York Times is kind of doing the right thing”: not reviewing the book. Just noticed they do have a review up:
That mind-blowing future is the focus of the final five or so hours of the audiobook, which explores the real-world impacts quantum computing could have: altering our immune systems to avoid cancer and Alzheimer’s, increasing crop yields, ending world hunger. As Kaku puts it, “the familiar laws of common sense are routinely violated at the atomic level”; but his lucid prose and thought process make abundant sense of this technological turning point.
Just shoot me…
I’m a bit surprised that these folks haven’t pivoted from “quantum computers will save HEP” to “large language models will save HEP.”
On a different topic, is there an accessible Google sheet detailing the planned schedule for this run of the LHC, like you linked to last year? That was great for us watching at home.
Estimates that I have seen say it’ll require about a million physical qubits for 110 to 150 or so logical qubits, though that depends on error rate and algorithm. I strongly doubt that you can do 7000 logical with only 1 million physical ones, it’s almost certainly considerably more. At the very least I’d like to see a source for the NS estimate.
Besides that, I think it would be good if journalists writing about this stuff would just ask people making such claims “and then what?”. I mean, let’s assume for a moment we actually manage to build a 7000 logical qubit quantum computer and actually simulate Maldacena’s whatever model on it. And then what? They’ll write papers and press releases. And then what? What will we learn from it? What will we do with it?
Well, nothing. We’ll learn that you can run a certain code on a certain computer. Will it tell us what’s going on in black holes? Of course not because you still have no idea if that code you ran on your computer actually describes a black hole. What will we do with it? There’s nothing you can do with it.
I know that many people in the field dismiss my output-orientation as unscientific but I would like to argue that this is just a tactic they have developed to shield themselves from criticism. There’s a LOT of public money going into quantum technology and people deserve an honest answer to the question what it’s supposed to do for them.
PS: I just recently made a video about what we could do with quantum computers, in case someone is interested.
@Sabine: Your comment is really much too negative about numerical/quantum simulations in general.
This experiment certainly could possibly give a negative answer to the question: is the BFSS matrix model the same as one formulation of type IIA string theory. Negative answers can be valuable.
And if you think that a positive answer on a numerical experiment can never definitively tell you that this is what is going on in the physics world, then why did the Nobel committee give a prize to the discovery of the Hulse–Taylor pulsar system, where numerical calculations confirmed that it was losing energy because of gravitational waves. (For all we know, there might a different mechanism that just happens to have the same orbital decay parameters.)
Would Maldacena’s proposed experiment tell us much? Probably not … looking at the abstracts of the papers Peter links to above, it seems to be testing whether two theories, neither of which comes close to describing the real world, predict the same result. Would this tell us anything about real-world black holes? It’s not clear.
@Peter Shor
You don’t need numerical calculations for the Hulse-Taylor pulsar, it’s just quadrupole radiation described in textbooks, see e.g. eq. (3) in
https://arxiv.org/pdf/1011.0718.pdf
But the broader point of needing numerical GR for discovery is valid. In the case of the discovery of gravitational waves by LIGO they needed waveform templates from numerical simulations.
I thought, as Seth Lloyd reportedly proved in the 1990’s, Feynman’s conjecture that quantum computers could do quantum simulations no classical computer could do is a notion that still holds up despite the huge advances in classical computing. Whether building a useful universal quantum simulator is technically feasible I don’t know, but seems like a worthwhile objective to pour some cash in, to support some very down-to-earth basic science in e.g. chemistry and condensed matter.
One rather hopes this highly speculative AdS/CFT-inspired “tabletop black hole” stuff doesn’t end up killing interest in the entire field once the backlash reaches a critical level.
David Metzler,
The “machine learning will solve the problems of string theory” thing has been going on for a while, see for example
https://news.northeastern.edu/2019/08/05/how-theoretical-physicist-james-halverson-is-using-data-science-and-machine-learning-to-study-string-theory/
https://iaifi.org/
https://lims.ac.uk/string-data-2022/
For the 2023 LHC schedule, see
https://edms.cern.ch/ui/file/2785501/1.2/2023-LHC-schedule_V1.2.pdf
LMMI,
The danger to the field of quantum computing is from people like Michio Kaku, getting huge audiences for absurd claims at Joe Rogan. There’s already a lot of discussion of a possible coming “quantum winter”, with the hype hitting the wall of the reality of slow progress.
Things like the article in New Scienctist about Maldacena’s proposal for creating black holes don’t get a lot of wider public attention. The danger of this is not to quantum computing but to fundamental theoretical physics. There’s a lot of money and attention out there for anyone who claims to have an idea for how to do quantum gravity using machine learning, quantum computers, quantum information theory, etc. If prominent theorists change their research programs to chase that money and attention, it could move a field already in trouble because of string theory/susy into a complete intellectual collapse dominated by pursuit of nonsense.
Peter Shor,
I think (hope) that you have missed the point of Sabine’s comment: using a numerical model of a theory to test its predictions against some actual experimental measurements, as in the case of Hulse-Taylor or a huge number of other cases (the detection of gravitational waves, for instance) is an entirely different thing than ‘testing’ a theory for which no experimental evidence exists using a numerical simulation.
Low Math, Meekly Interacting,
I am not sure what Feynman thought quantum computers could do, but they gain you no formal power over classical machines: any problem which can be solved with a quantum computer can be solved with a classical computer, and vice versa. What they do gain is an improvement in time complexity for some problems. That in practice makes some problems soluble which would not be soluble on a classical machine because they have some awful time complexity.
How anyone can claim that “there is not a single problem humanity faces that couldn’t be addressed by quantum computing” is entirely mysterious to me.
Out of interest, why is there no public effort by physicists to counter the relentless nonsense coming from Kaku?
I know that there is a LOT of hype and inaccurate information in pop Sci books and science magazines, but Kaku seems to be responsible for a disproportionately large share of the garbage!
Maybe a joint letter in Scientific American? I don’t suppose it would help much. Kaku will still churn out the books, TV documentaries and be invited for interviews as the expert of anything and everything. My heart sinks everything I hear Kaku has taken an interest in a new topic.
“How anyone can claim that ‘there is not a single problem humanity faces that couldn’t be addressed by quantum computing’ is entirely mysterious to me.”
He does say *addressed*, not *solved* ;).
Honest question about the things Michio Kaku said (maybe dumb question… but still, honest): how much is fair to expect from a venue like Joe Rogan’s podcast? I wouldn’t expect accuracy or even factuality of people having casual conversation in a media not famous for high level talks. My point is: isn’t this, by itself, a proof that these kind of theories will eventually get to the level of “were the piramids built by aliens”? The kind of thing that just will not go away because it provides entertainment. There is lot of people like me out there: people who knows nothing about math or physics (and never will, by the way, I already have lots of thing to read just from my job), but enjoy watching edutainments like scishow, pbs spacetime, etc. I think most of these people knows there is something sketchy or at least controversial about string theory, even if they can’t quite explain what is the matter. I mean… people are watching Sabine’s channel too.
Mike,
That’s a good question, one I’ve asked people over the years. On the string theory front, Kaku has been writing misleading popular books on the topic for over 35 years. Most string theorists agree, at least privately, that what he writes and says has become increasingly inaccurate and increasingly embarrassing. They generally however make the argument: “dealing with this is not my responsibility, there’s loads of misleading stuff out there about science. And, maybe it will have a positive effect, getting young people and the public interested in this kind of science, going on to read better books.” A cynical point of view on this would be that you can’t expect people to go out of their way to challenge something untrue which is promoting their own interests. When it gets to the point of damaging their interests, they might do something and we’re not there yet.
As for his nonsense about quantum computing, there’s so much promotional bullshit going on that it’s understandable that anyone wanting to do something about this has no particular reason to start with Kaku.
No Math, Random Reader,
The problem is Kaku, not the Rogan podcast.
Kaku says something wise in this scientific debate: https://www.youtube.com/watch?v=mY5V5jqdX9U.
Trying to summarize:
Strings become mainstream because a generation of smart theorists risked their careers attracted by string theory. Now string theorists control theory groups and this can only change if somebody proposes a better alternative. Not by un-constructive criticism.
Alessandro Strumia
Of course, “better alternatives” are themselves risks to the careers of those wedded to comatose ideas. And they take great pains to ensure that they are buffered against the slings and arrows of outrageous fortune by indoctrinating the next generation with this same professional angst.
Alessandro Strumia,
Given the present state of the HEP theory community in general, its ability to recognize and respond appropriately to a good new idea is not so clear. One thing that is very clear is that Michio Kaku is a convincing example of a theorist with zero chance of even understanding a new idea (his discussion of lattice gauge theory shows he’s incapable of understanding even the most basic aspects of current ideas about QFT).
Alessandro Strumia,
The “only game in town” argument has not been valid for quite some time now. There actually are various alternative proposals out there, some arguably better than others, but for some reason they fail to attract attention as promising research directions. As a consequence, very few people work on those proposals, and therefore progress is slow.
And in the meantime string theorists are also making no progress, but they represent the vast majority within hep-th and gr-qc, and are not interested in abandoning their old community for some underdeveloped new idea. It’s a viscious circle — lack of interest by the community produces lack of manpower, which produces lack of development, which produces lack of interest by the community…
That’s the actual problem with better alternatives to ST.
Best, 🙂
Marko
“And, maybe it will have a positive effect, getting young people and the public interested in this kind of science, going on to read better books.”
Let me say, as someone who was once on a track to become a physicist partly due to the influence of popular science, that one problem with Kaku-style hype is it gives people an unrealistic idea of what science is about and what it can do. He paints a picture of a magical, science-fictional world where the human intellect and its technology have god-like powers, avoiding the messy details and limiting realities. They’re alluring stories for the uninitiated, but when you start to look under the hood, understand the science in more detail and see that it’s not really like Kaku says, it’s easy to become disillusioned, to realize that you’ve been taken in by propaganda, and to lose interest in the field. That’s kinda what happened to me anyway.
Unfortunately, in America we tend to reward the hucksters, the entertainers, the PR people and the imaginers of fantasy worlds much more than those who deal in less glamorous reality. A capitalist entertainment-centric society has little incentive to reign in the Michio Kakus of the world, and every incentive to promote them. Science should be the complete opposite of this ethos, but it has clearly been corrupted by the larger society.
Someone here almost coined a very useful term for describing the slow progress of fundamental physics: “viscous circle”.
John,
Feel free to coin the term! 🙂
I meant to write “vicious circle”, but made a typo, which can then be reinterpreted as a typo of “viscous circle” instead, with a curiously similar meaning to the former…
Best, 🙂
Marko
Peter : something OT. I am surprised you haven’t mentioned anything about passing away of Stanley Deser one of the doyens of GR and also supergravity, who has been working on GR for past 70 years starting from the days he overlapped with Einstein at IAS.
Shantanu,
Nothing about Deser just because I never met him and know little about his work. Part of the problem is that I’ve always shared Sidney Coleman’s view on supergravity.
I can point to places that have more about Deser, the NYT at
https://www.nytimes.com/2023/05/08/science/stanley-deser-dead.html
and an interview at
https://heritageproject.caltech.edu/interviews-updates/stanley-deser
The Arnowitt-Deser-Misner formulation of general relativity in terms of initial data on a spacelike slice is a big deal.
While we’re on the subject of Stanley Deser, I’ll mention that World Scientific is providing free access to his autobiography, _Forks in the Road_, until July 31 of this year:
https://www.worldscientific.com/doi/epdf/10.1142/12205
Scott Aaronson has taken one for the team: https://scottaaronson.blog/
Shockingly he did not love the book!
It gets worse. Neil deGrasse Tyson interviews Michio Kaku:
https://www.youtube.com/watch?v=IfvoBlDUrPU
Mark Hillery,
What an embarrassing bunch of bullshit. What is wrong with Tyson?
Interview with Kaku about the book by Michael Shermer
https://youtube.com/watch?v=0lsO-dSR-7s
Giovanni Ronchi,
That’s really depressing. Shermer just swallows all Kaku’s nonsense whole. Amazing that someone whose whole thing is about the importance of skepticism, just believes every ridiculous and untrue thing Kaku says because he is supposedly a scientist.
The June 4 New York Times Book Review led its Audiobooks column with “Quantum Supremacy.”