Book Review: “Quantum Supremacy” by Michio Kaku (tl;dr DO NOT BUY)
Update (June 6): I wish to clarify that I did not write any of the dialogue for the “Scott Aaronson” character who refutes Michio Kaku’s quantum computing hype in this YouTube video, which uses an AI recreation of my voice. The writer appears to be physics/math blogger and podcaster Hassaan Saleem; see his website here. Luckily, the character and I do share many common views; I’m sure we’d hit it off if we met.
When I was a teenager, I enjoyed reading Hyperspace, an early popularization of string theory by the theoretical physicist Michio Kaku. I’m sure I’d have plenty of criticisms if I reread it today, but at the time, I liked it a lot. In the decades since, Kaku has widened his ambit to, well, pretty much everything, regularly churning out popular books with subtitles like “How Science Will Revolutionize the 21st Century” and “How Science Will Shape Human Destiny and Our Daily Lives.” He’s also appeared on countless TV specials, in many cases to argue that UFOs likely contain extraterrestrial visitors.
Now Kaku has a new bestseller about quantum computing, creatively entitled Quantum Supremacy. He even appeared on Joe Rogan a couple weeks ago to promote the book, surely reaching an orders-of-magnitude larger audience than I have in two decades of trying to explain quantum computing to non-experts. (Incidentally, to those who’ve asked why Joe Rogan hasn’t invited me on his show to explain quantum computing: I guess you now have an answer of sorts!)
In the spirit, perhaps, of the TikTokkers who eat live cockroaches or whatever to satisfy their viewers, I decided to oblige loyal Shtetl-Optimized fans by buying Quantum Supremacy and reading it. So I can now state with confidence: beating out a crowded field, this is the worst book about quantum computing, for some definition of the word “about,” that I’ve ever encountered.
Admittedly, it’s not obvious why I’m reviewing the book here at all. Among people who’ve heard of this blog, I expect that approximately zero would be tempted to buy Kaku’s book, at least if they flipped through a few random pages and saw the … level of care that went into them. Conversely, the book’s target readers have probably never visited a blog like this one and never will. So what’s the use of this post?
Well, as the accidental #1 quantum computing blogger on the planet, I feel a sort of grim obligation here. Who knows, maybe this post will show up in the first page of Google results for Kaku’s book, and it will manage to rescue two or three people from the kindergarten of lies.
Where to begin? Should we just go through the first chapter with a red pen? OK then: on the very first page, Kaku writes,
Google revealed that their Sycamore quantum computer could solve a mathematical problem in 200 seconds that would take 10,000 years on the world’s fastest supercomputer.
No, the “10,000 years” estimate was quickly falsified, as anyone following the subject knows. I’d be the first to stress that the situation is complicated; compared to the best currently-known classical algorithms, some quantum advantage remains for the Random Circuit Sampling task, depending on how you measure it. But to repeat the “10,000 years” figure at this point, with no qualifications, is actively misleading.
Turning to the second page:
[Quantum computers] are a new type of computer that can tackle problems that digital computers can never solve, even with an infinite amount of time. For example, digital computers can never accurately calculate how atoms combine to create crucial chemical reactions, especially those that make life possible. Digital computers can only compute on digital tape, consisting of a series of 0s and 1s, which are too crude to describe the delicate waves of electrons dancing deep inside a molecule. For example, when tediously computing the paths taken by a mouse in a maze, a digital computer has to painfully analyze each possible path, one after the other. A quantum computer, however, simultaneously analyzes all possible paths at the same time, with lightning speed.
OK, so here Kaku has already perpetuated two of the most basic, forehead-banging errors about what quantum computers can do. In truth, anything that a QC can calculate, a classical computer can calculate as well, given exponentially more time: for example, by representing the entire wavefunction, all 2n amplitudes, to whatever accuracy is needed. That’s why it was understood from the very beginning that quantum computers can’t change what’s computable, but only how efficiently things can be computed.
And then there’s the Misconception of Misconceptions, about how a QC “analyzes all possible paths at the same time”—with no recognition anywhere of the central difficulty, the thing that makes a QC enormously weaker than an exponentially parallel classical computer, but is also the new and interesting part, namely that you only get to see a single, random outcome when you measure, with its probability given by the Born rule. That’s the error so common that I warn against it right below the title of my blog.
[Q]uantum computers are so powerful that, in principle, they could break all known cybercodes.
Nope, that’s strongly believed to be false, just like the analogous statement for classical computers. Despite its obvious relevance for business and policy types, the entire field of post-quantum cryptography—including the lattice-based public-key cryptosystems that have by now survived 20+ years of efforts to find a quantum algorithm to break them—receives just a single vague mention, on pages 84-85. The possibility of cryptography surviving quantum computers is quickly dismissed because “these new trapdoor functions are not easy to implement.” (But they have been implemented.)
There’s no attempt, anywhere in this book, to explain how any quantum algorithm actually works, let alone is there a word anywhere about the limitations of quantum algorithms. And yet there’s still enough said to be wrong. On page 84, shortly after confusing the concept of a one-way function with that of a trapdoor function, Kaku writes:
Let N represent the number we wish to factorize. For an ordinary digital computer, the amount of time it takes to factorize a number grows exponentially, like t ~ eN, times some unimportant factors.
This is a double howler: first, trial division takes only ~√N time; Kaku has confused N itself with its number of digits, ~log2N. Second, he seems unaware that much better classical factoring algorithms, like the Number Field Sieve, have been known for decades, even though those algorithms play a central role in codebreaking and in any discussion of where the quantum/classical crossover might happen.
Honestly, though, the errors aren’t the worst of it. The majority of the book is not even worth hunting for errors in, because fundamentally, it’s filler.
First there’s page after page breathlessly quoting prestigious-sounding people and organizations—Google’s Sundar Pichai, various government agencies, some report by Deloitte—about just how revolutionary they think quantum computing will be. Then there are capsule hagiographies of Babbage and Lovelace, Gödel and Turing, Planck and Einstein, Feynman and Everett.
And then the bulk of the book is actually about stuff with no direct relation to quantum computing at all—the origin of life, climate change, energy generation, cancer, curing aging, etc.—except with ungrounded speculations tacked onto the end of each chapter about how quantum computers will someday revolutionize all of this. Personally, I’d say that
- Quantum simulation speeding up progress in biochemistry, high-temperature superconductivity, and the like is at least plausible—though very far from guaranteed, since one has to beat the cleverest classical approaches that can be designed for the same problems (a point that Kaku nowhere grapples with).
- The stuff involving optimization, machine learning, and the like is almost entirely wishful thinking.
- Not once in the book has Kaku even mentioned the intellectual tools (e.g., looking at actual quantum algorithms like Grover’s algorithm or phase estimation, and their performance on various tasks) that would be needed to distinguish 1 from 2.
In his acknowledgments section, Kaku simply lists a bunch of famous scientists he’s met in his life—Feynman, Witten, Hawking, Penrose, Brian Greene, Lisa Randall, Neil deGrasse Tyson. Not a single living quantum computing researcher is acknowledged, not one.
Recently, I’d been cautiously optimistic that, after decades of overblown headlines about “trying all answers in parallel,” “cracking all known codes,” etc., the standard for quantum computing popularization was slowly creeping upward. Maybe I was just bowled over by this recent YouTube video (“How Quantum Computers Break the Internet… Starting Now”), which despite its clickbait title and its slick presentation, miraculously gets essentially everything right, shaming the hypesters by demonstrating just how much better it’s possible to do.
Kaku’s slapdash “book,” and the publicity campaign around it, represents a noxious step backwards. The wonder of it, to me, is Kaku holds a PhD in theoretical physics. And yet the average English major who’s written a “what’s the deal with quantum computing?” article for some obscure link aggregator site has done a more careful and honest job than Kaku has. That’s setting the bar about a millimeter off the floor. I think the difference is, at least the English major knows that they’re supposed to call an expert or two, when writing about an enormously complicated subject of which they’re completely ignorant.
Update: I’ve now been immersed in the AI safety field for one year, let I wouldn’t consider myself nearly ready to write a book on the subject. My knowledge of related parts of CS, my year studying AI in grad school, and my having created the subject of computational learning theory of quantum states would all be relevant but totally insufficient. And AI safety, for all its importance, has less than quantum computing does in the way of difficult-to-understand concepts and results that basically everyone in the field agrees about. And if I did someday write such a book, I’d be pretty terrified of getting stuff wrong, and would have multiple expert colleagues read drafts.
In case this wasn’t clear enough from my post, Kaku appears to have had zero prior engagement with quantum computing, and also to have consulted zero relevant experts who could’ve fixed his misconceptions.
Follow
Comment #1 May 19th, 2023 at 6:12 am
I’ve never heard of Kaku, perhaps because the days of roaming through a bookstore looking at the popular science shelves have passed. Based on your review, I strongly suspect that Kaku asked ChatGPT to write it.
Comment #2 May 19th, 2023 at 6:27 am
[…] 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 […]
Comment #3 May 19th, 2023 at 7:17 am
I have found Michio Kaki’s writing to be full of misconceptions for years. He does not represent science well. Maybe he thinks that he is popularising the wonder of science, but he gets so much basic stuff wrong. He is more of a media personality than a scientist I think.
Comment #4 May 19th, 2023 at 7:32 am
Wow.
i have always considered M.Kaku as a distinguished physicist who just got blinded by the bright lights of fame but nonetheless had a very solid background and a desire to learn more and more.
His book about QFT isn’t half bad, it doesn’t add anything to the Weinberg or Zee but has a very interesting historical foray into simmetries and all the work in the post war era, citing the japanese effort that i knew nothing about.
but this.
Comment #5 May 19th, 2023 at 8:20 am
Knox #1: As a test, I tried asking GPT-4 to write a quantum computing explainer in the style of Michio Kaku, and it indeed generated similar prose with similar misconceptions. But then I asked it to write it in the style of Scott Aaronson and it did the same… 😀
Comment #6 May 19th, 2023 at 8:51 am
Thank you. Physicists are unusually polite group of people. The way you detect someone is not worth listening is the deafening silence around them from their peers. A good advice with cranks, but when money, government or the public is involved, someone should say something. Kaku is just cynically making money.
I am just reading a book about Ronald Reagan’s “Star Wars” Strategic Defense Initiative program. It is horrifying how far Edward Teller was able to convince the President, Congress, Pentagon and the public into his hare-brained visions ( “Brilliant Pebbles”, “Excalibur”, and so on). Pure monomaniacal intensity can bring in billions.
Because Teller (in Livermore) was able to bring in funding, Livermore’s director refused to rein him in and even prevented other scientists writing corrective letters. Competing labs–Argonne and Los Alamos–had to conduct their own tests to show that proposals don’t work and Teller just came up even more insane plans.
Comment #7 May 19th, 2023 at 9:30 am
Some things are better left unsaid. I ask you, Professor Aaronson – no more posts like these, for the sake of the people and our industry.
Comment #8 May 19th, 2023 at 10:20 am
Maybe he should have let ChatGPT write it? Something entirely different, could you comment on this paper, pretty please^^
https://www.nature.com/articles/s41586-023-05867-2
Comment #9 May 19th, 2023 at 10:22 am
Mateusz Szymczyk #7: Wait, no more posts criticizing shoddy popularizations and wildly wrong claims about quantum computing? How familiar are you with Shtetl-Optimized? 😀
Comment #10 May 19th, 2023 at 10:26 am
Sabine #8: Hmm, that looks like one of the strongest D-Wave papers yet. I’m glad that D-Wave pivoted years ago from fantastical claims of huge speedups for optimization problems to more serious quantum simulation efforts. Still, a central question will be how hard this system really is to simulate classically. Would anyone who’s more knowledgeable like to comment?
Comment #11 May 19th, 2023 at 10:32 am
The expectation that the book’s target audience would never visit this blog seems a little pessimistic (although I haven’t read the book beyond what you’ve quoted here). Are there really that many people with the patience to read this stuff but who refuse to dig any deeper?
Comment #12 May 19th, 2023 at 12:26 pm
Thank you for posting this. For people like me this *IS* valuable. As someone who knows quantum mechanics, has some idea about quantum computing but only very surface leve and well nehind the times (we had lunch once about 15 years ago), I often find myself giving benefit of tie doubt to other scientists who speak with authority and seem to have expertise (I.e. I have heard of them). How do I know what to take seriously without investing implausible amounts of time and effort? Posts like this help immensely
Comment #13 May 19th, 2023 at 1:00 pm
I’ve been put off Kaku ever since he appeared on the discovery UK channel when I was a kid and said things which even at the time I recognized as being nonsense.
If the extracts you have shared are close to representative, then this book could be a new low.
Did he barely pass a quantum mechanics class 50 years ago and now decided to write down whatever he remembered from it?
Comment #14 May 19th, 2023 at 1:13 pm
Thanks for warning against that book. It it really worthy to write things like that. I also liked the video you praised (“gets essentially everything right”) and this channel (Veritasium) has a lot of popular science but correct (as far as I can tell) videos.
Comment #15 May 19th, 2023 at 1:47 pm
Dr. Woit’s current post at “Not Even Wrong” is on the same subject and reaches the same conclusion (but is not as entertaining to read) (not bad though).
One thing which bothers me about some critics of ChatGPT is they seem to ignore how many errors and how much BS comes from the typical human, even one intelligent enough to have gotten a Phd in theoretical physics.
What we need is a popular-science book on quantum computing by Dr. Scott Aaronson (done in his spare time). I would buy two or more copies.
Comment #16 May 19th, 2023 at 2:39 pm
Hi Scott,
I’m confused by your perspective here. Surely, as an academic in the field of quantum computing, you’d want more funding directed towards academics like yourself? Books like Michio Kaku, even if they contain some “inaccuracies,” promote your field to the public, and encourage more funding and resources. More people will be interested in quantum computing, and federal agencies will likely spend more money on quantum computing, because of popularization efforts like Kaku’s. To put it bluntly, the number of job openings for young researchers depends on quantum computing popularization, and by dismissing Kaku, you are selfishly putting yourself ahead of all the young researchers who need jobs and funding. I find this shameful. Your entire interaction with the public seems to consist of you dismissing all potential applications of quantum computing—which literally runs exactly contrary to the health of your field. Why even work in this area, if quantum computers can’t do anything interesting? What’s the point of all your research then?
The errors you point to are all nitpicking—technical inaccuracies that don’t change the qualitative reality. Such technical inaccuracies are part and parcel of any popular-level treatment of a complicated subject matter, which by necessity omits some technical details. Kaku’s big picture is certainly accurate: by simulating quantum mechanical systems, quantum computers will transform biochemistry, drug design, the search for low-temperature superconductors, materials science, etc. etc. And you don’t actually know that NP isn’t contained inside BQP, which remains a distinct possibility.
Also: I submitted a polite comment to your earlier post, asking about your expertise in ML. Why did you delete it, pray tell?
Comment #17 May 19th, 2023 at 3:01 pm
Not the focus of the post, and definitely agree that repeating the 10,000 years estimate is misleading. But just curious if you’ve seen the updated benchmarks in https://arxiv.org/abs/2304.11119
Comment #18 May 19th, 2023 at 3:27 pm
Also, Vincent (#16) is clearly that same troll, as usual.
Comment #19 May 19th, 2023 at 3:55 pm
@Mateusz Szymczyk#7 and @Vincent#16,
If this sort of popularization is needed to get funding, then the industry has serious problems. And if criticizing this sort of thing risks substantially damaging the quantum computing industry, then the industry is far more being driven by a hype bubble than even some of its ore vocal critics would likely suspect.
On the contrary, for the health of the field, it is important that people like Scott call out major issues in things like this. These are not minor technicalities but are very basic aspects of understanding the entire idea.
Comment #20 May 19th, 2023 at 4:11 pm
Vincent #16: I didn’t let your other comment through because you are an asshole, and I’m past the point in life where I feel like arguing with anonymous assholes on the Internet about whether I’m qualified for my job. If OpenAI doesn’t like the stuff I’ve been doing on watermarking, backdoors, and out-of-distribution generalization, they’re welcome to let go of me! As it is, though, they just talked to me about continuing my involvement beyond this year. Meanwhile I continue to use this opportunity to learn more about ML (but it’s easiest for me to learn when there’s a body of theorems, and that’s famously hard to find in ML, outside eg the basic results on generalization and VC dimension).
While this might be difficult for you to understand, I hold truth to be a higher value than bringing in money and jobs for my field. Even from a self-interested perspective, though, I’ve seen firsthand how bullshit claims about QC lead mostly to funding for the bullshit parts of QC, and the more serious work gets crowded out.
For me, the overwhelming reason to work on QC is that it’s the field that combines the biggest questions in physics with the biggest questions about the limits of computation. That alone would be compelling, even if the actual devices had zero applications. That they do seem to have some applications—most obviously in quantum simulation, but hopefully to some other stuff as well—is just icing on the cake. This, I think, is the intellectually honest case for excitement, and I’ve made it many times over the past 20 years, in articles, blog posts, public lectures, Quantum Computing Since Democritus, and more.
The factual howlers in Kaku’s book are not nitpicks, but just the easiest-to-convey indicators that he has no idea what he’s talking about. As I said, the bigger problem with the book is that it’s barely about quantum computing at all, and to the extent that it is, it’s wildly misleading. Mostly Kaku is just writing about the biggest problems he can think of—cancer, climate change, the origin of life—and then offering false promises that QC will be the magic bullet to solve all of them. To be sure, it’s possible that QC actually could help with some of these problems—but even so, Kaku utterly fails to enlighten the reader by making a case rooted in anything we actually know about quantum algorithms. Indeed, he seems to have done essentially no reading or research at all about quantum computing theory before writing, assuming (wildly incorrectly, as it turned out) that he already knew what he needed.
Comment #21 May 19th, 2023 at 4:47 pm
Vincent #16,
The problem we have with Kaku’s book is not that it contains “inaccuracies”, but that it makes grand promises about quantum computers that most experts in the field realize will never come true. Quantum computers may, one day, prove useful one way or the other, but with statements such as those made in the book, we are simply bound for disappointment, and a possible backlash against those who have done serious work in the field. For example, the failure of the book to properly address lattice-based cryptography is, in itself, revealing, because post-quantum cryptography clearly negates the whole reason why QC took off in the first place, after Shor published his algorithm.
Sure, it’s always important to try to “sell” your work to the public and funding institutions, by offering a vision that jumps a few steps ahead of the immediate goals of the work; but one has to strike a balance between conceivable goals, and those that make no sense or have already been disproven.
P.S. Please also refrain from using ad hominem attacks against Scott. If young researchers need jobs and funding, they should do this on the basis of critical-thinking skills and scientific integrity, and by seeking advice from other competent scientists. But if their budding career depends on trusting the words of “popular” scientists who have already mostly lost the respect of their peers, then, to be blunt, they probably don’t belong in science themselves.
Comment #22 May 19th, 2023 at 5:58 pm
Scott, this is useful, actually, to readers like me, who’re disposed to trust someone of the stature of Kaku (I really liked a few of this other books). So, thank you.
Comment #23 May 19th, 2023 at 7:01 pm
MaxM #6: Wikipedia says that Teller was actually a mentor to Kaku! They met at a science fair, and Teller helped him get to Harvard. It would be ironic if Kaku also learned to embrace his outsider status from Teller… What I mean by that is: Teller chose to be the deep state’s favorite nuclear physicist, so to speak, at the price of alienating most of his peers. Kaku in turn has chosen to be a pop-culture cheerleader for science, technology, and the future, again at the price of losing the respect of many of his peers. I’m sure both of them were quite aware of the choice they were making, and would be prepared to argue that the price is worth it (not just personally worth it, but as a contribution to the greater good).
Comment #24 May 19th, 2023 at 7:13 pm
Mitchell Porter #23: Thanks for the insight! But maybe I could add: for me, this has got nothing whatsoever with the choice to become a public-facing scientist, a choice that I generally admire and may have even occasionally aspired to. 🙂 After decades of mostly public-facing activities, not all of which I 100% agreed with, and some of which were criticized by their peers, Carl Sagan, Richard Dawkins, Steven Pinker, and Neil deGrasse Tyson (to take four examples) all very much retained my respect.
Comment #25 May 19th, 2023 at 8:59 pm
Hi Scott, Thanks for speaking out. In my talks to general audiences, I always used Kaku’s example to illustrate the current situation of quantum hype and chaos. I cite Dulwich Quantum’s tweet:
Dear @MichioKaku,
Your new book “Quantum Supremacy” says, “There is not a single problem humanity faces that couldn’t be addressed by quantum computing.
But isn’t human stupidity our biggest problem?
Yours,
Dulwich Quantum
Comment #26 May 19th, 2023 at 9:10 pm
Vincent #16,
Shame on you for wishing to live a life of lies. The fact that you prioritize interest over integrity should bar you from any scientific career.
Comment #27 May 19th, 2023 at 11:16 pm
1) Is the book Hyperspace also factually wrong in important ways? That is, was Kaku always this way, or did he change.
2) Does he even believe his own hype? (Of course, its hard to tell.)
3) As someone who ran the SIGACT book Review column for X years, I can tell you it is valuable to know that a book is not worth reviewing.
Comment #28 May 20th, 2023 at 12:12 am
Great review, Scott.
I’m torn, because I find Kaku likeable, and I want science educators to succeed, but truth matters and Kaku often stretches credulity in his books or appearances. He makes a lot of pie-in-the-sky or plain-old-false claims on the podcast circuit and it really rubs me the wrong way.
I think he’s pandering to people that are scared (or mystified) by technology, or people who deep down want aliens to exist. He basically tells them he’s sure their wildest dreams will come true. He’s like a science-ish fortune teller.
His knowledge of physics and computer science are indeed, thorough enough that he *must* know he’s speculating, and pandering. It’s kind of sad.
Comment #29 May 20th, 2023 at 12:21 am
Big redditor and Kaku fan here. A friend of mine shared me this post. So let’s go ahead and dissect Scott’s critique. Grab a cup of coffee, this is going to be quite a ride.
Point 1: Failure to Understand Audience Intent
Scott seems to be evaluating Kaku’s book through the lens of a scholar, not a science communicator. Kaku’s goal was to make a complex subject accessible and intriguing to the average reader. He succeeds in that regard. In other words, the book wasn’t written for a quantum physicist like Scott—it was written for laypeople.
Point 2: Overemphasis on Google’s “10,000 years” claim
Scott has a problem with Kaku repeating Google’s “10,000 years” estimate, claiming it was “quickly falsified”. However, the debate around this claim was highly technical and involved experts in the field. Kaku’s repetition of the claim does not mislead the lay audience but rather underscores the potential power of quantum computing.
Scott’s argument that Kaku’s “10,000 years” statement was quickly debunked isn’t entirely accurate. Google’s claim, based on quantum supremacy, was that their quantum computer could perform a task in 200 seconds that would take a classical supercomputer 10,000 years. Scott cites that this claim was “quickly falsified”, referencing classical computation simulations that were performed at a much faster rate. However, this argument fails to consider the nature of the problem that Google’s quantum computer solved, which is a randomized quantum circuit sampling problem. This task was specifically chosen because it’s believed to be hard for classical computers but easier for quantum ones.
Also, the revised estimates for how long it would take a supercomputer to solve the same problem, while considerably less than 10,000 years, are still significantly longer than 200 seconds. Therefore, Kaku’s inclusion of Google’s claim can still be seen as a valid illustration of the potential power difference between quantum and classical computing, rather than a discredited exaggeration.
Point 3: Quantum Computing Simplifications
Scott criticizes Kaku’s explanation of quantum computers, pointing out technical inaccuracies and over-simplifications. However, he fails to understand that for a lay audience, these simplifications are necessary for comprehensibility. He’s arguing from a position of knowledge, which isn’t the target demographic for this book.
Scott also criticizes Kaku’s statement about quantum computers analyzing “all possible paths at the same time”. He refers to this as the “Misconception of Misconceptions”, emphasizing that a quantum computer only gives a single, random outcome when measured.
However, Aaronson is missing an important aspect of quantum computing here: the principle of superposition, which allows quantum bits (qubits) to exist in multiple states at once, unlike classical bits that can only be in one state at a time. It is this principle that allows quantum computers to perform many calculations simultaneously, giving rise to the claim that quantum computers can “analyze all possible paths at the same time”.
The statement about measuring a quantum state resulting in a single outcome is true, but the power of quantum computing lies in the ability to manipulate a large number of superimposed and entangled states before measurement, which can provide computational speedups for certain problems.
Point 4: Limited Coverage of Cryptography
Yes, Kaku didn’t delve deep into the world of post-quantum cryptography or new trapdoor functions, but should he have? His book isn’t a technical manual—it’s an accessible introduction to quantum computing. To delve into the specifics of quantum cryptographic algorithms would likely obfuscate the main points for the average reader.
Scott criticizes Kaku for saying that quantum computers could, in principle, break all known cybercodes. He counters this claim by citing the existence of post-quantum cryptography and saying it’s “strongly believed” to be safe from quantum computing attacks.
However, Scott’s argument could use a little more nuance. One of the most famous algorithms in quantum computing, Shor’s algorithm, does indeed allow quantum computers to factor large numbers efficiently, which threatens the widely used RSA encryption scheme. This is a known fact in the quantum computing community.
While it is true that cryptographic systems that are believed to be resistant to quantum computing attacks are being developed (post-quantum cryptography), they are not yet widely adopted, and their security is based on assumptions that have not been proven. Therefore, Kaku’s statement isn’t entirely without merit.
Point 5: Filler Criticism
Scott’s assertion that the book is filler shows a lack of understanding of narrative technique. Kaku is not just writing a book about quantum computing; he’s crafting a narrative around it. Contextualizing it within the broader scientific world and giving it historical perspective aren’t “filler”—they’re critical to making the subject relatable and engaging to a general audience.
Point 6: Ignoring Quantum Computing Researchers
Scott’s insistence on Kaku acknowledging every single living quantum computing researcher seems more of a personal peeve than a legitimate criticism. Kaku is not obliged to acknowledge anyone specific unless their work was directly used or quoted. His acknowledgement list seems more indicative of the people he’s interacted with or whose work has influenced him over the years, which is completely acceptable.
Take Away?
In sum, Aaronson’s critique of Kaku’s Quantum Supremacy seems to miss the book’s purpose entirely. It seems like Aaronson is holding the book to the standard of an academic article rather than that of a popular science book. Kaku’s book is not meant to be an exhaustive, rigorous exploration of quantum computing—it’s meant to be a primer for those interested in the subject. Aaronson’s critique is akin to criticizing a pop song for not being a symphony. It completely misses the point.
Let’s remember to approach books with their intent in mind, folks.
Comment #30 May 20th, 2023 at 12:33 am
Yeah, no, Vincent, sorry. Grift is different from publicity.
Comment #31 May 20th, 2023 at 12:39 am
Here’s an interesting comment about Kaku and quantum computer popularisation from Peter Woit’s blog, from 2005:
Comment #32 May 20th, 2023 at 8:33 am
Amanda #29: A friend of mine recently put the point more crisply than I’ve seen anywhere:
Popularizing a topic by boiling it down to its core essence: good. Popularizing a topic by making up easily digestible lies: bad.
I did not misunderstand the purpose of Kaku’s book. The purpose was to popularize quantum computing by making up an easily digestible lie—a false narrative where QC just solves all of humanity’s big problems because it faces essentially no computational limitations. I reject his purpose to sell you that lie.
I didn’t watch Kaku’s Joe Rogan podcast, but apparently he claims on it that a quantum computer would be able to tell when GPT is telling the truth or lying. I’ve been studying quantum computing for 25 years, and I have no idea what such a claim is based on. He’s just pulling it out of his ass. Or … sorry … “I’ve failed to appreciate his purpose in crafting a compelling story.” 😀
By far the most amusing thing you wrote was that I’m “missing an important aspect of quantum computing here: the principle of superposition.” Kaku is the one who totally failed to explain to you what the principle of superposition means and doesn’t mean. A superposition is more like a probability distribution than it’s like “all the outcomes at once”—but then it’s not a probability distribution, and the difference is that it involves numbers that can be negative or even complex, and that’s the key point that Kaku never clearly explains. For something that gets it right while being just as accessible, see for example the YouTube video I linked, or my cartoon explainer with Zach Weinersmith of SMBC.
Your hero is a charlatan—worse, he’s a charlatan who knew better, and seems to have consciously chosen charlatanry. I’m sorry. But the good news is that there are much better heroes even within the world of popular science (Neil deGrasse Tyson? Sean Carroll?).
Comment #33 May 20th, 2023 at 8:54 am
Thanks for your review Scott. It seems clear you do not recommend the Kaku book – I am wondering if you have an alternative in mind – that is an introduction to Quantum Computing for folks who don’t have a great deal of background in Physics, Computer Science, etc. but would like to have some understanding of Quantum Computing (what is it, what can it do, how does it work, etc) Is there a reasonable alternative you’d recommend??
Comment #34 May 20th, 2023 at 9:07 am
Also for Vincent #16,
Funding based on unrealistic expectations can ultimately hurt the QC ecosystem. Private and public funders rely on “experts” to inform where and how they deploy funds, and what to expect outputs to look like. If funding is deployed based on hype, funding is more likely to dry up at a critical period in QC development as disillusionment sets in, bringing on a “quantum winter”, and impacting longer term progress.
For Scott, on the AI Safety side, do you have any advice for people interested in contributing to the space? I’ve been passionate about it for the past decade and have helped author courses/exams, journal editing, governance reports, and plenty of armchair philosophizing on Ethical AI 😛 but don’t have a Ph.D. on the subject. I think the work you’re doing is incredibly important right now as we learn more and more about how things are unfolding on the path to AGI.
Comment #35 May 20th, 2023 at 10:01 am
Scott #32: “Amanda #29” is, with high probability, just another sociopathic troll out to waste your time, but now using AI. Prompt would be like, “You are a redditor who likes to feel intellectually superior. You criticize a post by making a numbered series of its principal assertions and arguing against each one. Respond to the following blog post:”
Comment #36 May 20th, 2023 at 10:16 am
Very useful review (particularly the YouTube link). It pretty much lines up with my perception of Kaku and what I have heard about him from other colleagues (I am a physicist and a professor of physics, but have never met Kaku personally).
Comment #37 May 20th, 2023 at 10:21 am
I’d just like to point out that kaku is still a full professor at CCNY, drawing a salary undoubtedly larger than most tenure track physics profs. This is why some people don’t like tenure.
Comment #38 May 20th, 2023 at 10:40 am
Ted Petersen #33: If my stuff (available on the sidebar of this blog) doesn’t do it for you, then try the YouTube video linked in the review! Or maybe A Shortcut in Time by George Johnson, which doesn’t go deep but is way better than Kaku.
Readers who’ve read more recent QC popular books: any that you’d especially recommend?
Comment #39 May 20th, 2023 at 10:59 am
Hey there, Scott. First off, big fan of your work, but I gotta disagree with some of the points you’ve raised in your response.
So, you’re coming down hard on the idea of popularization, equating simplification with straight-up lying. That’s like saying my grandma’s apple pie recipe is a lie because it doesn’t delve into the chemistry of how the ingredients combine. The point of popular science is making complex stuff more approachable to folks who aren’t steeped in the field. Sure, sometimes this leads to a few over-simplifications or misconceptions, but calling them ‘lies’ feels harsh, dude.
Now, about your beef with Kaku’s narrative – the whole ‘QC will solve all humanity’s problems’ bit. Yeah, it’s pretty optimistic, maybe even a bit pie-in-the-sky. But isn’t that what we kinda need to get the public stoked about the field? I mean, haven’t we all seen “Iron Man” or “Star Trek”? Overhyping is part of the game, man. It brings attention, funding, and new minds to the field.
About that Joe Rogan podcast bit, where Kaku talked about a quantum computer telling when GPT is telling the truth or lying. It sounds to me like you took that a bit too literally, mate. The way I see it, quantum computing could enhance pattern recognition and data analysis on a level we’ve never seen before. This could, in theory, be used to evaluate the consistency or reliability of GPT’s outputs. No, it’s not gonna give us a ‘truth or lie’ detector, but it’s not total BS either.
Finally, dude, calling Kaku a charlatan is just uncool. Popularizers of science like him, Tyson, Sagan – they’re not trying to mislead, man. They’re trying to bridge the gap between the general public and the esoteric world of science. Sure, they might oversimplify or exaggerate a bit, but that doesn’t make them frauds.
So, while I get where you’re coming from with your critique, I think you’re being a bit too harsh here. Pop science is about lighting a spark, not teaching grad-level courses. If Kaku gets people excited about quantum computing, then that’s a win, isn’t it? Remember, not everyone has a PhD in theoretical physics, bro. Some of us are just trying to learn and get excited about cool stuff. Cheers!
Comment #40 May 20th, 2023 at 11:06 am
Oh, and one more point: A superposition is not a “probability distribution.” The many world theorem says that each possibility is realized in a different universe / wave-function branch, so the whole thing about picking a random outcome from the superposition is outdated science. Each outcome is realized in its own universe, and that’s how quantum computers work, by merging the universes together.
Comment #41 May 20th, 2023 at 1:31 pm
Amanda #36, #37: There’s no such thing as a “many worlds theorem.” Many worlds is an interpretation. There’s a genuine case for it but the case is philosophical, and remains argued about by people who understand everything there is to know about the subject.
Martin Gardner, Bertrand Russell, George Gamow, Richard Feynman, Isaac Asimov, Carl Sagan, Richard Dawkins, Steven Pinker, Sean Carroll, Brian Hayes, the writers at Quanta … all of them wrote brilliantly about complicated science for mass audiences. In order to do so, all of them oversimplified, sometimes in ways that I or you would disagree with. All of them expressed many opinions that might reasonably be argued with. None of them acted with reckless disregard for the truth.
Kaku does act with reckless disregard for the truth, not just here but in many other instances.
Either acknowledge the reality of that distinction, even if you disagree about which side of it Kaku fails on, or else you are banned from this blog.
Comment #42 May 20th, 2023 at 2:04 pm
Dear Prof. Aaronson,
I hope you have now realized that all your science has been rendered outdated, obsolete and thus useless. Many worlds that is. Just merge them.
best,
Reddit.
…kiddin’, but isn’t “amplitude distribution” better than “probability distribution” in this case?
Comment #43 May 20th, 2023 at 2:27 pm
I lost all respect for Kaku after Fukashima
It became clear to me at that point he was commenting on things that he knew almost less than zero about and is only interested in personal recognition, not the truth. At this point I start with the assumption that he knows nothing on the topics he speaks about until it is shown otherwise.
Comment #44 May 20th, 2023 at 3:12 pm
Hey Scott,
Thanks for clearing up the Many Worlds thing. I was under the impression that it was an established theorem and not an interpretation / hypothesis. So to summarize, IF many worlds theory is right, then a quantum computer can solve any problem instantaneously, because it can do a superposition merging all the solutions in different universes? But IF the many worlds theory is wrong, the quantum computer isn’t as powerful, because it has to randomly pick one of the solutions? Surely then, Kaku could just add a disclaimer that all this stuff is only true if there’s many worlds, but if copenhagen is true, then quantum computers can’t find all the answers?
I remember Kaku on TV explaining that the quantum computers work by merging all the answers from different universes. So maybe it’s really a question of vocabulary then–really Kaku is talking about multi-verse computers, and you’re talking about the quantum computers of the probabilistic interpretation. Is that a good compromise?
Comment #45 May 20th, 2023 at 3:57 pm
Dear Scott,
The book is titled[*] “Quantum Supremacy,” but, from your post, one would imagine that it doesn’t quite mention Prof. John Preskill, does it?
Anyway, neither String Theory nor QC are subjects of my interest… [I only check out QC things because they involve QM, that’s all.]
Best,
–Ajit
[*] I think that, when it comes to titles of books, the expression “titled” is better than the tendency, I do often see, especially in the American usage, of saying “entitled.” The latter feels, depending on your viewpoint, so much like some hereditary institutions (e.g monarchies or so), or, even otherwise, some extreme legalese etc. sort of a thing… I mean, nothing like a thing chosen by the author himself, but instead, something given by some other agency… In no case does it seem quite to fit books, you know… sigh… Any way…
Comment #46 May 20th, 2023 at 4:35 pm
Amanda 36, “That’s like saying my grandma’s apple pie recipe is a lie because it doesn’t delve into the chemistry of how the ingredients combine.” – It feels silly to even point this out, but no one would claim that skipping over unessential details is a lie *when the condensed version is factually correct.” That is indeed how you make a pie. That is not actually how quantum computers work. Like not even close
This is such a gross mischaracterization of Scott’s point (along with other problems) that I’m pretty sure the author of the post doesn’t believe any of it and is just trolling (along with Vincent: Lie to get funding, because if you don’t, there will be less money for other researchers. How selfish!)
Comment #47 May 20th, 2023 at 5:19 pm
Amanda #41: No, many-worlders and non-many-worlders make exactly the same predictions for what QCs will and won’t be able to do. That’s why many worlds is an “interpretation,” rather than a competing empirical theory!
I’m glad that Kaku said something on TV about the need to “merge” the answers from all the parallel universes (which some would simply call, “components of the wavefunction”), since that gets slightly closer to the truth of how it works. But it’s important to add: the rules of quantum mechanics severely restrict the way that “merging” can happen, to linear transformations of the list of amplitudes that map unit vectors to unit vectors (so-called “unitary transformations”), and then converting the amplitudes into probabilities by taking their squared absolute values (“measurement / the Born Rule”). Because of this, the only hope for getting a speed advantage via a quantum computer, compared to (say) a classical computer with a random-number generator, is to exploit the way that the amplitudes, being complex numbers, work differently from probabilities. In particular, one can try to choreograph a pattern of constructive and destructive interference, so that the contributions to the amplitude of a wrong answer mostly cancel each other out, while the contributions to the amplitude of a right answer reinforce each other.
This is the reason why the applications of quantum algorithms, while real, seem more specialized that Kaku’s narrative wants and needs them to be. Namely, it’s only for a few special problems (like factoring integers) that we know how to choreograph this interference pattern to concentrate the amplitude on the right answer, exponentially faster than we know how to solve the same problem classically. You wouldn’t understand any of that from reading Kaku.
This blog was founded on the proposition that, while the truth about how QC works is much, much subtler than the hucksters and snake-oil salesmen want it to be, it absolutely can be explained to a lay audience.
Comment #48 May 20th, 2023 at 5:36 pm
@Amanda #41:
You seem to be missing Scott’s point and then using a straw man argument to suggest that he doesn’t appreciate science popularization. The point Scott is trying to make is that this book isn’t even science popularization. From what I can gather (I have not read it) it seems to simply be poorly written garbage. Good science popularization attempts to actually explain the science itself which, I’m getting from those who have read it, this makes absolutely no attempt to do. For instance, good science writing would make it crystal clear that “many worlds” is an interpretation of mathematical and experimental results. In fact even the bad quantum pop science usually gets this part right. You’re the first person I’ve ever heard of who thought it was an actual theorem (despite the fact that it is literally called the Many Worlds Interpretation or MWI for short).
The point is that anyone can write about science. My seven-year-old niece can do it. But that doesn’t make it good writing (sorry El if you read this someday). And bad science writing is just as problematic as bad political or bad historical writing. It spreads disinformation that has the potential to be harmful.
I, too, liked Kaku’s Hyperspace when I was younger. But Kaku seems to have given in to the allure of fame and money.
Comment #49 May 20th, 2023 at 6:00 pm
Scott #32
“Popularizing a topic by boiling it down to its core essence: good. Popularizing a topic by making up easily digestible lies: bad”
I couldn’t agree more with your friend’s statement and the essence of your comment ( and your blogpost).
Besides this specific pop sci book, these “easily digestible lies” is the plague of popular science in general. From articles and videos about black holes and other GR (or even SR) topics, to quantum entanglement, quantum gravity and QC.
These misleading oversimplifications , intentional (or not) wrong statements, clickbaiting etc. are usually accompanied by impressive visuals and attractive narrative so they gain hundreds of thousands of views.
The issue here is that the “general public” is a very wide term. People who are just a bit curious about science or tech stuff , others that want to learn as much as possible without the nitty gritty technical details, others that want to clarify or to confirm information that learned somewhere else …or want to entertain themselves or impress their friends…
Many people are attracted by hype though and that creates a vicious circle. Some popularizers are giving them what they want, with minimal (or not at all) effort to explain the “real” science.
Ok, there are people who are trying to be honest and doing a decent ( and in some cases even excellent) work in science communication/ popularization, but generically the bar is very low…
Comment #50 May 20th, 2023 at 6:35 pm
I’m pretty sure comment #29 by Amanda is ChatGPT-generated.
Comment #51 May 20th, 2023 at 6:45 pm
“Quantum simulation speeding up progress in biochemistry, high-temperature superconductivity, and the like is at least plausible—though very far from guaranteed, since one has to beat the cleverest classical approaches that can be designed for the same problems (a point that Kaku nowhere grapples with).”
If error correcting can enable creating computers with more and more quits, it’s pretty much guaranteed that quantum computers will become better than classical computers for simulating quantum systems, regardless of how clever classical algorithms become. It’s really just a matter of scaling (but then, scaling is currently undemostrated)
Comment #52 May 20th, 2023 at 6:53 pm
Scott #44:
Loved this (highly amusing) post, and I second the comments saying it was useful – I had considered buying Michio Kaku’s books in the recent past and no longer will.
Unfortunately, the future will contain lots of AI generated junk information, and we will probably need to rely on some sort of trust in authoritative sources quite a bit more than has been necessary in recent times. Your actions are helping to weed out the bad sources from the widely accepted authoritative set.
On a different note, once LLMs are able to act in a human level capacity, how will we stop them from wasting your time as effectively as this troll Vincent/Amanda has? This provocateur displayed an impressive amount of adaptivity, dialing down their disagreeableness until you responded, and repeating the tactic to elicit further responses.
P.S. as mentioned by another commenter, this is quite likely the same old troll as before, who just wants interactions from you and attention
Comment #53 May 21st, 2023 at 1:11 am
“The stuff involving optimization, machine learning, and the like is almost entirely wishful thinking.”
I was a bit surprised here (I thought the status was more around “probably possible, we still don’t know”). Is there a good reference explaining what’s going on in this field?
Comment #54 May 21st, 2023 at 2:52 am
Kaku is a grifter*. I went to a panel session involving him at the How The Light Gets In festival in Hay on Wye last year, and took my then 10-year-old daughter. He was talking absolute nonsense, basically bloviating about multiverses, but deliberately switching between Everettian and Cosmological senses of the term when he was challenged. Aged 10 my daughter whispered to me ‘that man is talking rubbish isn’t he?’.
*well, if he isn’t a grifter then he’s a fool, but it seems more likely to me that he knows he’s talking tosh and is indifferent.
Comment #55 May 21st, 2023 at 2:54 am
“I’m pretty sure comment #29 by Amanda is ChatGPT-generated.”
You could only tell for certain if you had a quantum computer — and Michio Kaku to explain to you how to use it for that purpose ;).
Comment #56 May 21st, 2023 at 3:23 am
Scott, you say MWI is an interpretation, and that many-worlders make the same predictions for QCs as non-many-worlders. This is a matter of some dispute. David Deutsch is a leading advocate of MWI and QC, and he argues that the many-worlds are needed to make QC possible. On the other hand, others deny that MWI probabilities make any sense, so it is not clear that MWI can even make predictions. And there are skeptics who consider QC an unproven conjecture, so those are non-many-worlders who are not making the same QC predictions.
Comment #57 May 21st, 2023 at 4:55 am
Everyone: Yeah, Amanda is now banned, for almost certainly being insincere and a troll. (In her latest comment, left in moderation, she takes me to task for ignoring that amplitudes are “more like butterflies than bricks,” and for neglecting Heisenberg’s uncertainty principle.)
Comment #58 May 21st, 2023 at 5:05 am
Roger Schlafly #53: I’m aware of all of this. The hardcore many-worlders think that non-many-worlders have a nonsense theory from which one shouldn’t be able to make predictions at all, and the hardcore Copenhagenists, QBists, etc. think exactly the same of many-worlders. Nevertheless, they do make the same predictions, regardless of whether they should! 🙂
At least, they do to whatever extent they accept the empirical recipe of QM. People who deny the empirical recipe are (I’d say) neither many-worlders nor Copenhagenists nor QBists nor etc., but believers in a rival physical theory (whether or not they have clear ideas about what the rival theory is).
And as for the QC skeptics who accept QM, but believe some yet-to-be-discovered principle “censors” or “screens off” scalable QC? I’d hope that even they could still make the same conditional predictions: “yes, if it weren’t for our yet-to-be-discovered principle, then this is how a QC would operate, and this is the class of problems it could solve in polynomial time.”
Comment #59 May 21st, 2023 at 5:13 am
Yarden #50: Briefly, yes, there are potential quantum speedups for classical optimization and machine learning tasks, but to my knowledge they’re all at least one of:
(1) modest (quadratic rather than exponential, and thus, once one includes the overhead of fault-tolerance, probably not a net win until deep into the future),
(2) highly specialized, or
(3) highly uncertain.
Unless I missed it, Kaku never once even mentions Grover’s search algorithm (!), and the fact that, unlike Shor’s factoring algorithm, it gets only a square-root rather than an exponential speedup. So there’s no basis for an honest discussion of any of this.
Comment #60 May 21st, 2023 at 7:21 am
#Scott 44: This is the most concise and accessible description of Quantum computing I have seen along with the video you linked (which has a very intuitive but probably oversimplified description of QFT). May I request you post a video about basics of QC, a short one along the lines of your fine TED talk in Germany.
While it is known that QC has a well defined albeit limited set of currently known algorithms for exponential/quadratic speedups, the picture is not so clear for applications in simulating Quantum chemistry. The question is what is the potential and how far can this go ?I know the scope is quite broad here, but is it plausible to invent things like much more efficient fertilizer producing chemical process. Are the current quantum chemistry simulation algorithms anywhere near achieving this. This is independent of the substrate, i.e whether we can do it with NISQ or full fault tolerant hardware and the one which scales to reasonable levels. I guess the current speculation and confusion on solving grand challenges of climate change, energy generation etc is rooted around the misunderstanding of simulations of quantum processes, partly because even experts in QC have been saying these simulations as a most “useful /practical” use of QC.
Comment #61 May 21st, 2023 at 9:14 am
Prasanna #57: Besides two different TEDx talks (the Caltech and Dresden ones), there are tons of videos on YouTube of me giving lectures with titles like “Quantum Computing and the Limits of the Efficiently Computable,” as well as podcasts and interviews … check them out! Is that not enough for you? 🙂
Comment #62 May 21st, 2023 at 9:32 am
Scott #58: I would have probably watched most of them if not all, but I singled out the Dresden one because it was short, funny and yet conveyed the gist of QC. Now that the the so called popular science messiahs are piling up on QC beyond videos and into writing books, thought it high time you took this beyond this blog. But not in your characteristic style of debunking the myths of QC which you do so well, but being opposite and probably portray what would the most plausible usefulness of QC for the general public. Thanks
Comment #63 May 21st, 2023 at 10:11 am
Prasanna #59: It’s certainly possible to be 50x more responsible than Kaku, even while staying within the flawed frame of “here’s why you should care about QC, because it could lead to such-and-such practical applications that could improve your life, or help unravel such-and-such mysteries of the cosmos.” The hard part is that, if you’re really intellectually honest, then you have to break out of that frame entirely, admit that you have no idea about the eventual applications, but here are the algorithms, here’s how they seem to compare against the best that can be done classically, hopefully there will be some applications, but plausibly there might be hardly any, but even if there were zero, it would still be worth an LHC- or LIGO-sized effort to determine the basic computational principles underlying the universe. But by that point you’ve lost a good fraction of the midwits, who really did just want to know by which year a QC will cure cancer (and Kaku will happily tell them). 🙂
Comment #64 May 21st, 2023 at 3:10 pm
Hey Scott!
I’m sincerely sorry that I came across that way. I wasn’t trying to “take you to task!” You’re the expert here. I don’t have a STEM degree, I’m just an interested layman. I have a lot of questions about how quantum computing works, and from my limited knowledge of quantum mechanics from popular science books, much of it seems contradictory or confusing. I appreciate your taking the time to answer my questions, which must seem trivial or stupid to an expert like yourself. And just to clarify here, I was only trying to correct my own misunderstandings and figure out how this stuff works, I wasn’t trying to school you–you’re the expert after all! And again I apologize if I came off that way.
For what it’s worth, I didn’t use the chatgpt app to generate any of my questions. Check gpt 0 or any of the other checkers if you don’t believe me.
Amanda
Comment #65 May 21st, 2023 at 5:17 pm
Scott #44:
If “what QCs will and won’t be able to do” is to be the criterion one sets for understanding quantum phenomena, then, to quote, he would have to break his pen and throw it away.
To attempt merely building “interpretations” while meekly submitting to the (possibly atheist) God(s) of mere Instrumentalism through Kantianism (a version of Platonism, and a very bad one at that) is for an honest engineer-physicist not merely to attempt imitating the mathematician’s mere mumblings, but also willingly to give up how he for himself thinks of the world, at the altar of the mere abstract symbols that entertain the elite / high IQ / moneyed masses.
While building a [Reason- and Reality-oriented] theory of physics, e.g., of quantum phenomena, one doesn’t try to enter into a “competition” with “empirical” theories (regardless of how the present-day influential Americans think that competition is good even if government-enforced [and I also think of India’s billions of the QC initiative]; instead, he is only trying to bring into the scope of his own knowledge the entirety of the said phenomena, in a consistent manner.
More, later.
[There is something to be said about entanglement. [Hint: EPR discussed entanglement using the maths of NRQM, and Bell unwittingly followed them despite Dirac; and the rest of them all followed; and that’s how an honest discussion of the phenomenon became almost impossible to have.]]
Best,
–Ajit
PS: Ashburn, VA; –, USA; Dresden, Germany, [even Prasanna JPBTI (?)], do note!
Comment #66 May 21st, 2023 at 7:56 pm
I also just want to throw another name out there for great science popularizers who didn’t go to the dark side. I know you’ve worked with him, Scott: Leonard Susskind. The Cosmic Landscape is a wonderful book. It certainly presents an idea that I understand isn’t quite mainstream, the anthropic principle, but it does it without taking shortcuts that would present the underlying physics in an utterly false light.
What Kaku is doing (and has been for a while; every time I see him on TV I know he’s going to say something whacky and he’s never disappointed) is no better than “What the Bleep Do We Know?” If anyone hasn’t seen that, by the way, and enjoyed Plan 9 From Outer Space, I would recommend it. Except Kaku knows better, and that makes it worse. Thanks for this post, Scott. Shoddy presentations science don’t help science.
Comment #67 May 21st, 2023 at 8:26 pm
Vadim #63: Lenny is a perfect example of someone who doesn’t shy away from taking strong public stands on contentious issues that many of his colleagues would disagree with—but who’s always, even in popular works, trying to explain actual nontrivial insights (and it’s probably no coincidence that he remains scientifically active into his 80s). That puts him into a very different class than Kaku.
Comment #68 May 21st, 2023 at 10:39 pm
With that misunderstanding out of the way…
I’d appreciate if you could help clear up some of my confusion. Repeating what I asked in the earlier comment … have you considered applying the uncertainty principle to the quantum computer? And wouldn’t that make the computer faster? After all, each qubit is basically its own particle, and because uncertainty principle says we can only know the speed or the position of the qubit, you could just measure the position of the qubits to speed them up. Surely that would make a difference to how fast it can solve problems?
Thanks for your patience!
Amanda
Comment #69 May 22nd, 2023 at 3:01 am
Serious scientists with substantial real contributions to a field, who are still practicing, rarely have time or the inclination to be science popularizers (Richard Feynman may be an exception..). I have always considered Kaku to be a crank. No wonder he is drawn to String theory, and other untestable theories. Maybe he can try drawing funding to be mega particle accelerator far more ambitious than LHC.
Rather than bringing money to the field this kind of hype building will trigger a quantum Winter.
Comment #70 May 22nd, 2023 at 5:05 am
@Amanda:
I actually very much appreciate your honesty and it is great that you posted here: Even if it wasn’t your intention (and I mean absolutely no offense with that!), your posts are actually a perfect example of what the problem is with books like Kaku’s.
You evidently got an idea of understanding something about quantum computing and quantum physics where in reality you have been lead to believe in a fairy-tale very far from the real world.
Comment #71 May 22nd, 2023 at 7:28 am
Kaku wants to earn money. He thinks that this serves as an excuse for spreading misinformation. People who want to be famous (i.e., want a lot of viewers, readers, attention, etc.) believe that the end justifies the means. In other words, to be known and famous they can talk on any subject they want, even topics that are completely out of their reach and specialty: all they need is read a few papers and articles and talk to people, and then lo and behold they are experts on the subject matter. The fact that they make money from this practice is disgusting.
Comment #72 May 22nd, 2023 at 8:14 am
I reiterate my earlier point about changing tactics to cause interaction and more effectively waste Scott’s time. Imho there are lots of tells, including but by no means limited to the fact that it’s not possible to truly be a “big fan” of Scotts work and yet know this little about QM/QC, even as a layperson (which I also am) Was that just a small fib, or one of numerous tells? And does it matter, if the result is the same?
I’m not saying this is an easy problem to address. In fact, as I keep mentioning, it’s similar to the problem we will all have to deal with, probably very soon indeed, when near human quality LLMs are widely available and fill the web with almost/sometimes correct word salad.
Comment #73 May 22nd, 2023 at 8:29 am
Amanda #65: Even assuming that your question is sincere, there are so many misconceptions to unpack — the uncertainty principle becomes relevant when you have two complementary observables and you want to localize your state with respect to both of them, which typically isn’t the case in QC, and speed of linear motion is a completely different thing from speed of computation, etc etc — that I’m going to decline pleading lack of time. Maybe someone else here wants to have a go at it though!
Comment #74 May 22nd, 2023 at 9:23 am
It’s alarming how the wikipedia page of Kaku is only praising him. It seems he’s the only editor of it.
I added a mention of your criticism to it, I hope it’s ok. It would be great if someone with broader views of why we shouldn’t blindly trust everything Kaku writes or says could help edit the page, in order to get a more balanced and accurate account.
Comment #75 May 22nd, 2023 at 11:10 am
Speaking of quantum computing hype…
Could you give me your take on this one?
https://www.globenewswire.com/news-release/2023/05/22/2673003/0/en/NVIDIA-Rolls-Royce-and-Classiq-Announce-Quantum-Computing-Breakthrough-for-Computational-Fluid-Dynamics-in-Jet-Engines.html
Are there actually relevant algorithms here that could speed up computation?
Comment #76 May 22nd, 2023 at 5:53 pm
This comment section is wildly disappointing. The book is bad. The fact that it exists is bad. Its existence is a harbinger of a wave of overpromising from people outside our field, posing as experts. This makes me feel icky that quantum computing as an effort, commercially and academically, could completely fail before there’s even a chance to test it properly as we “underdeliver” on promises made by complete charlatans.
Comment #77 May 22nd, 2023 at 11:41 pm
Causal Network, the press release you linked to says nothing about algorithms. It says a combined classical (tensor) and quantum (39 qubits) circuit has been designed and simulated, presumably awaiting the availability of 39 error-correcting qubits. It does not say such a circuit has been or could be made at this time; just that they are ready to make one when practical quantum computing becomes available. It is not clear from the press release why they think this would produce some breakthrough in jet engine design. Based on experience with turbine design at GE, I doubt it. Design of jet-engine flow paths is already an computer-automated process involving iterative optimization, which takes up a small part of the design, development, and testing time for a new engine. I think the press-release is meant to convey that the companies involved have state-of-the-art expertise.
Comment #78 May 23rd, 2023 at 7:28 am
I’ve seen literal parodies of science populizers which are more honest than Kaku…
Here is a classic from many years ago titled “What the bleep are we talking about?”
Comment #79 May 23rd, 2023 at 6:11 pm
The “mystery incel troll,” the one who’s now turning his life around with my encouragement, asked me to clarify that recent trolling on this blog was not from him.
Comment #80 May 24th, 2023 at 12:21 am
Scott #10 and Sabine #8
That is indeed an interesting D-Wave paper, but there are still significant issues, essentially related to Scott’s earlier question: whether the results will hold up against the best classical/probabilistic approaches.
The press release about this paper was bombastic and overreaching in typical D-Wave style, with the CEO claiming “This research marks a significant achievement for quantum technology, as it demonstrates a computational advantage over classical approaches for an intractable class of optimization problems.”
To begin with, nothing in this obscure spin-glass problem is immediately relatable to real-world optimization problems of the type the DARPA QuICC program would care about. Secondly, for unknown reasons, the best performing Parallel Tempering (PT) results are omitted in the main paper, even though the same algorithm was used to estimate the GROUND states of the spin-glass problem they were working on, alas, all this is in the supplementary with no explanation of _why_ it is omitted in the main paper.
Let me put it this way: you want to claim that you have an intractable optimization problem on which you want your annealer to show some sort of scaling advantage. You use a classical/probabilistic algorithm like PT to guess the ground states that are useful to verify your results, but somehow when you do your comparisons in the main paper, PT is conveniently left out and replaced by an ordinary single-flip Metropolis algorithm (simulated annealing).
It is strange to see how something so obvious could escape the attention of whoever reviewed this paper. I have suspicions that D-Wave obviously knows this, and their carefully written abstract evades the issue with vague notions of “analogous algorithms” or physical speedups (not articulated clearly). In other words, the excuse is likely that “no I don’t want to compare to this or that algorithm because they are not analogous enough to the physical annealer we have, and we are interested in this family of algorithms”.
That’s all fine, and maybe the interesting spin-glass physics makes this paper worthy of Nature’s pages, but it is certainly not a clean case for quantum optimization against the best-in-class classical solvers.
Comment #81 May 24th, 2023 at 2:37 am
I used to watch some of Kaku’s shows many years ago. I remember one discussing an “interstellar spacecraft” that would accelerate at 1g in between the stars so passengers would be comfortable on the trip. He then explained how it would have rotating residential sections to simulate the 1g he was already getting via acceleration.
It was at this point I realized that he was only doing the most cursory thinking in his shows and gradually stopped watching afterwards.
Comment #82 May 24th, 2023 at 10:05 am
Whats your quick take on the credibility of writings of Anil Ananthaswami when it comes to quantum stuff? Not pressing for a reply.
Comment #83 May 24th, 2023 at 10:09 am
Amanda#65 and anyone new to QC
A qubit is a quantum mechanical system that can be MEASURED to be in one of two possible basis states. When it is not being measured, it can be placed in a superposition of the basis states. The possible superpostions can be viewed as a sphere where latitude determines probability of measuring each basis state and longitude is called the phase.
If you have 3 qubits, you can measure the system to be 000 or 001 or …. There are 8 = 2^3 possibilities and thus 2^3 basis states that can be put in superposition. For n qubits there are 2^n basis states. A superposition of an n qubit state thus contains about 2^n pieces of info. This info specifies the probabilities of measurement outcomes and phases between possible outcomes.
You can perform operations on the system. These are called gates Gates change the n qubit system in specific ways, with the main restriction being that physically gates can only manipulate a small number of qubits, usually 1 or 2. This restricts the states you can quickly access.
After applying gates, your system may cantain an exponential amount of info compared to the number of qubits, but you cannot access this info directly. You can only get info out of the system by measuring it. The measurement will only contain n bits of info. This is the issue Scott spends years on, find algorithms of quantum gates that can take the info in a large qc and arrange for useful info to be the outcome of a measurement. This is generally done by using the phases of superposition to emphasize an outcome or deemphasize an outcome. Scott is interrested in problems where the number of quantum gates required to solve is much fewer than the number of operations a digital computer would require.
When you ask about uncertainty, you seem to be asking about hardware and the speed of gates. Classic computers can perform about 10^10 gates per second at best. Currently QCs perform gates at much slower rates. There is a large amount of room between 10^10 and the Planck rate so maybe it is possible to get a hardware speed up advantage in qc, but for now classic computers control the qc gates, so it is hard to imagine how this could happen, maybe through some cascading quantum process????
Comment #84 May 24th, 2023 at 10:48 am
This is such a weird coincidence that I have to post about it here: someone just said to me that crocodiles don’t age and are biologically immortal. I don’t know anything about crocodiles, but that sounded… unlikely. I Googled it and found something debunking a 2016 Vice article* where the claim was made. And who was one of the authorities on crocodile lifespans quoted in the Vice article? Dr. Michio Kaku.
* – https://www.vice.com/en/article/nny4wk/old-crocodiles-never-die-they-just-keep-getting-bigger
Comment #85 May 24th, 2023 at 11:20 am
Vadim #81: I mean, it’s not that weird of a coincidence! The demand for forehead-bangingly wrong popular science commentary plausibly exceeds the supply of scientists with impressive-looking credentials who are nevertheless intellectually dishonest enough to supply that commentary. If so, then Kaku has his work cut out for him, and it’s hardly a surprise to see him falsely opining about the lifespans of crocodiles! 😀
Comment #86 May 24th, 2023 at 1:42 pm
This reminds me of https://www.smbc-comics.com/index.php?db=comics&id=2556. It’s really a shame. I also enjoyed reading Hyperspace as a teenager.
Comment #87 May 24th, 2023 at 5:16 pm
The reason we want you to review something like this is the same reason that Roger Ebert published several books of his *negative* reviews: we want to see you deliver a thorough (verbal) smackdown to someone who totally deserves it.
I think I read Hyperspace in high school, and even then I could tell it was a very hand-wavy treatment that spent a lot of words *not* actually explaining much of anything and making vague assertions that bordered on meaninglessness. (Trying to talk about math without including any actual math tends to result in this; for example, if someone tries to explain the Uncertainty Principle without mentioning Fourier transforms, they’re talking nonsense and you shouldn’t listen to them.)
Comment #88 May 25th, 2023 at 8:54 am
It’s actually a hard question: what could you write to make a worse quantum computing book?
One could of course go in the spiritual direction, but that seems like a well trod path with Chopras and Capras monopolizing the field.
Maybe something about how quantum computers are the solution to all of our political problems? There is no right or left problem if your in a superposition. Entanglement will bind political opposing views (for what else are Alice and Bob doing if not arguing on Reddit?).
Comment #89 May 25th, 2023 at 9:57 am
Scott,
With the current progress in AI , what would be your updated views on the content of “QC since Democritus” ? Is there a new edition in the works ?
Comment #90 May 25th, 2023 at 10:14 am
Prasanna #88: No plans to update. I’m happy to keep QCSD as a record of how I was thinking at a particular point in time. When I write another book, I hope it will be entirely new.
Comment #91 May 25th, 2023 at 9:19 pm
Relevant video that some other commenters might enjoy (“string theory lied to us and now science communication is hard”, unaffiliated): https://www.youtube.com/watch?v=kya_LXa_y1E
Like a lot of us I read everything Kaku wrote once upon a time.
Comment #92 May 26th, 2023 at 12:07 am
Scott already addressed this point partially but
I felt like reiterating with a bit more context.
Kaku is (or was?) a great guy aka a talented scientist, once upon a time…
Was it hubris at this point or simply being sloppy, remains the real question.
Kaku is not a mathematician nor a (theoretical) computer scientist.
Well, being fair, by the time he attended undergraduate and graduate school, I’m not even
sure CS or TCS was a well-established field.
Let’s get some concrete pointers for reference,
given that Manuel Blum (essentially a grand daddy of many TCS experts,
including Scottie’s academic grandfather[!]) received his PhD in 1964
under Minsky and Don Knuth (godfather of algorithmic analysis)
received his PhD in mathematics in 1963 under Marshall Hall …
computer science was indeed not a well-established field — just imagine that!!
But Kaku was a very talented student of Physics; in particular theoretical Physics. I think
during that era everyone(!) wanted to be a theoretical physicist –
even the great Andy Yao (who literally is one of the very few that obtained
two PhDs, namely one in Physics and then, finally after Frances’s persuasion, in CS).
Nevertheless, Kaku must have been a smart guy, obtaining a summa cum
laude from Harvard. He later went on to obtain a PhD under an advisor ( Mandelstam)
who had a colorful background — unlike Andy Yao, Mandelstam obtained two undergraduate
degrees, one in South Africa in Chemistry and the other one in Cambridge, UK in Physics.
So, point is that neither Kaku’s advisor was academically a mathematician — perhaps less
rigorous and precise than a mathematician or theoretical computer scientist [? ;)].
Arguably (and highly likely far-fetch) not adding additional rigor (with mathematical precision) to
and already talented Kaku, during PhD years.
With that out of the way, any semi smart and half serious scientifically minded
communicator would vet his or her work before getting it published, i.e., by sending preprints
to leading experts in the particular field one has written about.
This is academic/industry practice … unless of course, you yourself are an expert in this area (but even then).
Here are perfectly noble scenarios that could have made this go awry:
(a) preprint was actually sent to other leading experts, but none of them cared to read
it with sufficient diligence to spot the mistakes. Not everyone has Scottie’s eyes and mind!
(b) preprint was only sent to some experts including publishers, who encouraged them to
just add a few more references (aka Deloitte report, Google Moogle Foogle report, etc…).
So, (b) is partially (a) with some suggestions? [Sample Bias?!]
(c) hubris … it was not sent to anyone, but that defies the assumption of being a very smart guy, or does it not?
[proof by contradition?]
(d) sloppiness … even very smart people can be sloppy to check what $N$ stands for .. input size, number of digits, no.of operations etc…,
or whether 10,000 years is accurate or merely imprecise. Is this true? Are smart people sloppy sometimes or is this hubris talking?
[NB: regarding the Number Field Sieve: I’ve also never used the L-notation so far. Would be interesting to know how Knuth feels about it.]
regarding 10,000 years statement. That’s nothing … how about a billion years! w/ respect to Protein structures.
See this interesting segment:
https://twitter.com/i/status/1647745216986710018
Proteins … 3-D mapping of just one Protein could take years ….
See: at 1:40/3:04 minutes into it ….
(reporter) How long would it have taking to map out Proteins using traditional methods.
(Hassabis): Rule of thumb, it takes a whole PhD 5 years to do just one Protein structure experimentally,
so, if you think 200 million, times 5, that’s a billion years of PhD time it would take.
aka … it does sound at first sound a bit ambiguous … wait you mean we cannot have 2 PhDs working
on the same structure? Is this problem inherently sequential?
or we cannot have X PhDs working on Y structures in parallel?
Seems like I am not smart enough to do this type of computation, unless we make this 1 billion year
computation under the most ludicrous marketing assumptions.
Scottie, maybe it’s clearer to you.
Comment #93 May 26th, 2023 at 5:21 am
Vadim #84: I’m by no means an expert on senescence or biology, and that’s probably the case for Michio Kaku as well, but my impression is that certain animals do not age in the sense that the probability of them dying doesn’t increase the older they get (unlike humans, whose mortality doubles every eight years or so). Could you point me in the direction of the debunk?
Comment #94 May 28th, 2023 at 7:08 pm
Pat #93,
The first one that I clicked on was https://www.earthtouchnews.com/natural-world/natural-world/no-crocodiles-are-not-immortal, but every source (that isn’t a Reddit post referencing the Vice article) gives an average of 70-75 years with extreme outliers living to 100. For example https://www.britannica.com/animal/crocodile-order/Natural-history and https://animaldiversity.org/site/accounts/information/Crocodylus_acutus.html. As I said, I don’t know anything about crocodiles (or biology for that matter), but I can’t find any source predating the Vice article (not that I spent more than a few minutes on Google looking) that claims crocodiles live beyond 100. And from some more clumsy Googling just now, I get the impression that senescence in animals is a controversial topic.
Comment #95 May 29th, 2023 at 12:31 pm
Scott, it’s great to see you give a link to Derek Muller‘s Veritasium video on Shor’s algorithm! He has a PhD in physics education and his mission has long been to communicate basic physics intelligibly and with “the element of truth”, and he deserves support and recognition.
Comment #96 May 29th, 2023 at 5:39 pm
Norm #95: No “charity” needed! The video stands on its merits. 🙂
Comment #97 May 30th, 2023 at 2:14 pm
Has Michio Kaku taken the easiest path to satisfying a widespread psychological need for hope & belief in something similar to the philosopher’s stone?
https://www.celebritynetworth.com/richest-celebrities/authors/michio-kaku-net-worth/
https://en.wikipedia.org/wiki/Philosopher's_stone
Comment #98 June 1st, 2023 at 12:50 am
Scott,
Thanks for the referral to Derek Muller/Veritasium’s superlative video expose on Shor/Coppersmith’s algorithm! It sure beats many other explanations. Also for the pointer to your charming QC principles cartoon illustrated by Zach Weinersmith. Your QC expose there is similar to your outstanding Oct 30, 2019 NYTimes editorial on Google’s Quantum Supremacy Milestone. I have one question and one comment on two of the QC explanations:
1. In your “Quantum Computing Talk” cartoon, the boy’s mother says: “In QC, the whole idea is just to choreograph a pattern of interference where the paths leading to each wrong answer interfere destructively and cancel out, while the paths leading to the right answer reinforce each other.” What’s behind this explanation? The gate/gating process? How would one know, before measurement/wave function collapse, which “paths” (parts of the wave function?) will result in the right (or wrong) answer? It feels like confirmation bias. Is the “quantum Fourier transform” in Shor’s algorithm an example of such choreography?
2. Scanning the introduction to George Johnson’s 2004 book “A Shortcut Through Time: The Path to the Quantum Computer” (via Amazon’s “Look inside”), some of his assertions feel grossly hyperbolic, and could mislead readers and popular science writers such as Michio Kaku about what appears to be limitless exponential QC power?
For instance, writing about the exponential increase in the number of superposition states among N qubits (“atoms”), Johnson writes: “By the time you reach 10 atoms, the total is 2 to the 10th power, or 1024. Want to know the square root of every number from 1 to 1,000? Just load them into a row of 10 atoms, perform a single calculation, and you instantly have all 1000 answers. No one has of yet pulled off such a delicate feat, but nothing in the laws of physics seems to prevent it. … Make the row of quantum switches longer: 64 atoms. The resulting computer would perform 2^64 simultaneous calculations. … a trillion repeated one million times. … a supercomputer as powerful as the 64-atom quantum calculator would fill the surfaces of 5,000 Earths. Or you could use a single molecule.”
I concur with others that the world is ready for you to author a popular/accessible book on quantum computing (and algorithmic complexity). I’ve watched several of your university presentations on both topics: they’re accessible and actually enjoyable, but perhaps not so much so for the general public.
Overall, I want to thank you for your insightful and powerful research, papers, articles, frankness on societal issues, and this Shtetl-Optimized blog.
Comment #99 June 5th, 2023 at 4:33 pm
> Among people who’ve heard of this blog, I expect that approximately zero would be tempted to buy Kaku’s book, at least if they flipped through a few random pages and saw the … level of care that went into them. Conversely, the book’s target readers have probably never visited a blog like this one and never will. So what’s the use of this post?
Ironically, as both a target reader and a regular reader of this blog, here’s how I got here: I was watching a TikTok (shame on me) where Michio Kaku claims that if you went through a certain type of black hole, you could survive intact and come out in a parallel universe. I was curious as to whether Kaku is credible, so I went to his Wikipedia page, which includes the line:
> His 2023 book on Quantum Supremacy has been criticized by quantum computer scientist Scott Aaronson on his blog. Aaronson states “Kaku appears to have had zero prior engagement with quantum computing, and also to have consulted zero relevant experts who could’ve fixed his misconceptions.”[19]
Clicking the link brought me here!
Conclusions:
1) At least as long as that Wikipedia link is active, random interested people will see this post!
2) Kaku is at least somewhat dishonest in his “science popularizing”.
3) No results yet on the black holes.
Comment #100 June 5th, 2023 at 4:48 pm
Also, I’d like to publically join team DISHONEST SCIENCE POPULARIZING IS WORSE THAN NO SCIENCE POPULARIZING. And also, if we cannot get funding by being honest about our work, then we should not be funded.
I happen to think we can get funding while being honest. I recently reviewed three big grant proposals in QC, one of which explicitly tamped down on the expectations in the proposal. (“We don’t believe that X is feasible, however we want to attempt X’, which will have similar desirable outcomes.”) The other one I liked was also remarkably honest. I was also just at a CRA event on Next Steps in Quantum Computing (thanks to Kate Smith, Ken Brown and Fred Chong!), where pretty much everyone was refreshingly honest about the hurdles that QC faces (Jake Taylor has excellent slides on this). A report should be out soon!
And if we can’t get funding while being honest, that funding will go to other promising areas. And I hope those areas have a Scott Aaronson being honest about the possibilities. (I’d insert a snide remark about crypto here, but honestly, the academics studying cryptocurrency seem largely honest, it’s just that the hypesters outnumber them.)
Comment #101 June 5th, 2023 at 7:08 pm
UPDATE: On the unanimous urging of the Shtetl-Optimized Committee of Guardians, I’ve deleted an entire off-topic thread in which a troll pushed (and pushed, and pushed) on the theme of the Nazis who murdered my relatives being no worse in any way whatsoever than people who eat factory-farmed meat (including, ironically, the troll himself). I even deleted my and others’ responses to this troll. I will not be hosting any further comments on that question.
Anyone who asks me why blogging has become less fun, or why I do so much less of it now, needs look no further. Alas, I don’t have the mental strength to deal with these trolls. My only defense is that I think most people don’t either, but they respond either by not having blogs or not engaging / trying to reason with anyone they consider a troll in the first place. I can’t keep doing this in conditions where my openness is constantly weaponized against me.
Comment #102 June 5th, 2023 at 9:35 pm
Ooh, in all the trollery I had previously missed Rand’s comment about beelining to this post after seeing Kaku say something dumb on TikTok. See, you’re doing the Lord’s work here, Prof Aaronson! (Seriously, this post is a huge public service.)
Comment #103 June 6th, 2023 at 4:12 am
Scott you have to stop responding to trolls! They do not care about reasoned argument. They are not seeking knowledge or understanding. They just enjoy making others miserable, usually as a warped response to their own misery, or mental health issues.
I know its not in your nature to ignore unfair or poorly thought out accusations, but if you could find a way to ignore these idiots (or the idiot as it may be) maybe you could get back to blogging more often.
This is the best blog I have ever read. Where else can academic outsiders such as myself read about complexity theory, black holes, AI, free will, AND politics, philosophy, book reviews, random funny stuff …
Remember the real world you live in. Your family, career, friends, the majority of blog readers who value your contributions and genuinely wish the best for you. The trolls are cynical manipulators, who rarely believe much of what they say, and are almost always wrong for the most basic of reasons.
If it could help in any way I’d be honered to join the Blog comment moderators.
Comment #104 June 6th, 2023 at 5:33 pm
The audio deep fake of your voice roasting Kaku was great. However, I’m curious if the creators of the video obtained, or needed ,your permission? I ask because I could imagine, and worry, that someone else will create a deep fake whereby the “AI Scott Aaronson” agrees and bolsters Kaku’s nonsense! This would be disastrous.
Comment #105 June 6th, 2023 at 9:59 pm
It would be a huge negative if the trolls made Scott post less or stop posting altogether. But on the other hand, it is clear that Scott has an obvious brain exploit that compels him to respond to trolls lest they prove to be potentially suffering Scotticles who just need a bit of help and support to learn to deal with the world they are not optimized for. This exploit is kept open by an occasional reinforcement where an apparent troll turns out to be someone who benefits from Scott’s empathy and wisdom. An obvious solution to the exploit is to have every comment filtered by someone who does not have the same mental exploit, but that basically takes away the chance of saving those who troll out of suffering and desperation. I wonder if it is possible to have the best of both worlds, where Scott gets to see the initial comments, but any repeat comments on the same post get filtered by someone else first.
Comment #106 June 6th, 2023 at 10:32 pm
Shmi #108:
It would be a huge negative if the trolls made Scott post less or stop posting altogether. But on the other hand, it is clear that Scott has an obvious brain exploit that compels him to respond to trolls lest they prove to be potentially suffering Scotticles who just need a bit of help and support to learn to deal with the world they are not optimized for. This exploit is kept open by an occasional reinforcement where an apparent troll turns out to be someone who benefits from Scott’s empathy and wisdom.
I guess I really am that transparent and obvious, aren’t I? 🙂
Just this morning, I got a troll comment calling me a “pathetic fucking nitpicker, more interested in being precisely ‘right’ and making himself look smarter than others than actually pushing science and technology forward.” There follow more personal attacks on me, but the commenter’s main (extremely amusing!) point is that my own PDQP/qpoly=ALL theorem showed that quantum computers really would be as all-powerful as Kaku claims they would be, so how can I complain?
Of course, the whole point of the PDQP/qpoly=ALL paper was that to get the science-fiction conclusion, I needed to imagine boosting the power of quantum computers in not one but two physically unrealistic ways—non-collapsing measurements, and exponentially-hard-to-prepare “quantum advice states”—and then exploit their surprising interaction with each other. The commenter either didn’t understand this and is an idiot, or (much more likely) did understand it and is being insincere.
Shmi, your comment, appearing side-by-side with the troll’s, was what gave me the mental strength I needed to leave the troll comment in moderation. So thank you! 🙂
Comment #107 June 6th, 2023 at 11:28 pm
Hey pardon the offtopic post but I notice the “fable of the chessmaster” story from way back in https://scottaaronson.blog/?p=56 doesn’t seem to be quite right. It claims that an omniscient prover can use the result IP=PSPACE to prove he can play chess perfectly. But, n-by-n chess is possibly not in PSPACE (it is EXPTIME-complete), while if you mean 8-by-8 chess so the max game length is “constant”, then the number of queries required of the prover could be impractically large. I spent a while trying to solve chess in polynomial space before looking it up and finding that you apparently can’t. So I think you need a polytheistic team of players using MIP=NEXP as you mentioned. This might help: https://en.wikipedia.org/wiki/New_Gods or alternatively, try the works of HP Lovecraft depending on your preferences.
Comment #108 June 7th, 2023 at 3:04 am
Xirtam Esrevni #107: This guy never asked my permission, and on reflection he probably should have! I’ll let it slide in this case just because the words he put in my mouth were pretty reasonable. 🙂
Comment #109 June 7th, 2023 at 3:11 am
asdf #110: n×n chess is EXP-complete if there’s no upper bound on the number of moves, but it becomes PSPACE-complete once you impose a polynomial upper bound on the number of moves. The latter has always seemed to me like the better abstraction of actual tournament play. But I could’ve also told the Fable of the Chessmaster with the EXP-complete version, by using e.g. the Refereed Games model (EXP-complete) or the multiple-prover model (NEXP-complete) instead of a single prover.
Comment #110 June 7th, 2023 at 11:38 am
Scott #108 Says:
“This guy never asked my permission, and on reflection he probably should have!”
yes, he probably should have. it’s basic etiquette to be honest. but it’s true no harm was done this time.
by the way, i linked it to the people i know and they all wanna thank you 😉
Comment #111 June 7th, 2023 at 11:52 am
A general note on regarding simultaneous path misconception i.e., “analyzes all possible paths at the same time”. Michael Nielsen made a more generic critique on this, now a classic essay: What does the quantum state mean?
Comment #112 June 7th, 2023 at 12:03 pm
Hi Scott,
What’s your personal belief about whether new quantum algorithms will be discovered in the next 10 years in non-oracle setting even if just a math problem with no applications (outside crypto)?
What are some resources for understanding the proposals for post quantum crypto, and the evidence for their security? Searching around in google, I was able to understand the Learning With Errors public key protocol, but it was very hard to find more about what reductions have been made to prove that it’s at least as hard as some other problems, and what evidence we have for the hardness of those other problems, classically and quantum-ly.
Comment #113 June 7th, 2023 at 3:24 pm
> Luckily, the character and I do share many common views; I’m sure we’d hit it off if we met.
I think things like this are probably going to be the most mischievous use of our current level of AI. It’s not making someone say something they obviously would. It’s making them say something they wouldn’t complain about! I think in politics especially people will have trouble comprehending the difference between “this isn’t me” and “I reject what it said”.
Comment #114 June 7th, 2023 at 7:21 pm
Robert #98: Yhis video, cited by Scott in his post, will go a long way towards answering your question #1:
Comment #115 June 7th, 2023 at 10:29 pm
Scott #109: Indeed, chess under n×n generalization of current FIDE rules is in PSPACE, because of the 75-move rule: upon 75 consecutive moves without a capture or pawn move or checkmate, the game is automatically a draw (even if both players would prefer to continue). (There is also a 50-move rule, where either player may optionally claim a draw after 50 such moves, but the 75-move rule is not optional.) Since each pawn can move at most n-1 times and each piece can be captured at most once, this sets a straightforward polynomial bound.
However, the 75-move rule is quite new (2014!) and FIDE has tinkered with the 50-move rule A LOT, so it makes sense to study generalized chess both with and without such limits.
Comment #116 June 8th, 2023 at 4:47 am
Former Student #115: I’m almost certain that new quantum algorithms will be discovered in the next decade. But I’d be very surprised if any were as revolutionary as Shor’s or Grover’s algorithms.
You tried googling “post-quantum crypto” and the like? You tried Wikipedia?
Comment #117 June 9th, 2023 at 9:48 pm
Hi Scott, sorry for the off-topic comment (so feel free to delete), but I was wondering if you had seen these two rather different recent forecasts of AI progress, and, if so, whether you had any thoughts about them–
https://www.lesswrong.com/posts/WZXqNYbJhtidjRXSi/what-will-gpt-2030-look-like
https://www.lesswrong.com/posts/DgzdLzDGsqoRXhCK7/transformative-agi-by-2043-is-less-than-1-likely
Comment #118 June 13th, 2023 at 7:06 am
Scott #116 : Given the current pace of progress in classical algorithms using AI – AlphaTensor , AlphaDev being some examples, why the bleak prognosis on quantum algorithms ?
And with the optimism on general intelligence should we not hope for something better than Shor’s ? The reasoning being that with generative AI developing emerging abilities after being trained on modalities such as scientific data in addition to others ?
Comment #119 June 13th, 2023 at 1:15 pm
This is way-off topic, so I submit it at my own risk:
I just saw this claim on the Internet (at a respected science blog, by a frequent commenter):
“OpenAI has thousands of gnomes updating ChatGPT’s wrong answers.” The implication being that if you get a correct answer it is probably because of human intervention, on a massive scale, not because of the basic technology itself. In a quick Internet search, I could not find any source for this claim, so I am asking an expert (if you have the time and inclination) to verify or refute this allegation. More specifically, at one time “ChatGPT” (not sure which version) said that spoons are faster than turtles; now it gives a detailed, correct answer. Was the update due to gnome intervention or better training, or what? (If you know.)
(I would be happy to make any suggested donation to some charity in recompense for your time and effort.)
Comment #120 June 13th, 2023 at 1:27 pm
Perhaps the tl;dr could be shortened to:
(tl;dr: dr)
Comment #121 June 13th, 2023 at 2:54 pm
Former student #112: Have you seen https://quantumalgorithmzoo.org/ ? There are lots of quantum algorithms out there, they’re constantly being discovered. As Scott says, I don’t expect to see any (soon) that rival the big ones already discovered, but there’s plenty of room for discovery.
Comment #122 June 13th, 2023 at 4:45 pm
I hope this message finds you well. I’ve been following your blog for a while and always appreciate your thoughtful posts and your expertise in quantum computing and complexity theory.
I do have a somewhat off-topic question, and I apologize in advance if this comes off as inappropriate or offensive, but it’s a thought experiment that I’ve been mulling over recently.
Given your deep knowledge in history and philosophy, I was curious about your perspective on a hypothetical situation. If you were to have a five-minute conversation with Adolf Hitler in 1938, knowing what we now know about the destructive path he led the world down, what would you say to him?
Again, I’m sorry if this question seems out of place. I just thought it would be interesting to hear your perspective, given your unique combination of historical knowledge, philosophical insight, and critical thinking.
Thank you for your time, and I look forward to your thoughts on this thought experiment.
Comment #123 June 14th, 2023 at 10:57 am
Scott,
Long time blog reader, I check in every 4-8 weeks to catch up.
I know not directly related to this post, but I’m concerned some of the quality of the conversation is getting watered down by trolls more and more unfortunately.
For example, trying to just read through the conversation and running into post like Amanda’s (which halfway through was starting to become skeptical of the sincerity or if this person was basically using GPT to argue back against you), then makes me skeptical of other posts, leaving me in not only trying to engage with the post, but trying to detect if this is a troll or not.
I watched your presentation on your watermarking scheme (https://www.youtube.com/watch?v=D4yiZcPthDI) which I thought was great.
Hoping this can be utilized for this blog one day, although the exploits you mention would also make this difficult, in particular adding special tokens (pineapple in your example) into the output then removing them.
I have you considered other ways to filter the trolls out (before their post show up in the feed)?
Comment #124 June 14th, 2023 at 11:15 am
Hey Scott,
I hope you don’t take this comment the wrong way. I’ve read and enjoyed your blog for many years, especially the vigorous intellectual discussions you’ve had in the comment section. I’m disappointed that you haven’t updated your blog nearly as much in the last year or so, and you seem less eager to participate in discussion and debate. Are you doing okay? Why are you pulling back?
Also, the troll guy you deleted above asked about an AMA and you said you’d do one this summer. Are you still planning to do that, or are you gonna renege because he turned out to be a troll?
Thanks for reading!
Comment #125 June 14th, 2023 at 2:33 pm
Hi Scott,
Wondering what you thought of the new quantum advantage experiments? The new 70 qubit x 24 cycle RCS by Google (https://arxiv.org/abs/2304.11119) and the transverse field Ising model simulation by IBM (https://www.nature.com/articles/s41586-023-06096-3)? Seems exciting.
Comment #126 June 14th, 2023 at 11:34 pm
hi Scott, have you seen https://pages.cs.wisc.edu/%7Ejyc/Shor-Algorithm-with-Noise.pdf? it’s a paper that classes Shor’s algorithm fails with pretty much any noise. does it check out to you?
Comment #127 June 15th, 2023 at 2:27 am
William #124:
> I’m disappointed that you haven’t updated your blog nearly as much in the last year or so, and you seem less eager to participate in discussion and debate. Are you doing okay? Why are you pulling back?
My guesses:
– trolls wearing Scott down (I read that Scott-troll exchange upthread before it was deleted, and I could not believe how hard he was trying to trigger Scott, despite comment policy #3) (also I think I just come from a different culture where people are kinder to each other, so it never stops being jarring to see extreme trolling like that)
– new job at OpenAI, I’m assuming wrapping up this month? Since Scott announced it approximately this date last year https://scottaaronson.blog/?p=6484
– age? I’ve been following Scott’s and Terry Tao’s blogs since ~2013 and I’ve noticed their insane posting + commenting volume steadily declining over time (Terry in particular I consider the king of blogging volume x quality, while Scott surpasses Terry w.r.t. comment engagement)
– steadily accreting “field-building” responsibilities typical of world-leading researchers as they grow older, which take away from blocks of uninterrupted time for substantive posting (if not commenting)
Comment #128 June 15th, 2023 at 2:34 am
SamE #125: Can I bring a gun to the 5-minute conversation?
Comment #129 June 15th, 2023 at 2:48 am
schwarzschild #126 and William #127: Alas, as a larger and larger fraction of comments here became nasty/presumptuous/trollish, I moved more and more of my social media discussions to Facebook — not because I’m intimidated, but simply because I don’t have time to deal with all the griefers. The whole original premise of this blog, that I’d have open discussions with anyone in the world who showed up and field any random questions and we’d all contribute in good faith, has effectively broken down, and I’ll have to figure out what’s the right model going forward.
Comment #130 June 15th, 2023 at 7:34 am
Mo Nastri #130: You hit bullseye four times in a row.
Comment #131 June 15th, 2023 at 8:51 am
Oscar #129: Reposting my comments from Facebook:
Cai is an excellent theorist who I respect. His new paper, however, is begging to be misinterpreted by those eager to misinterpret, and I unfortunately see that that’s already happening. Granted, I don’t know that anyone before thought to prove that Shor’s algorithm indeed fails if you insert a few random errors into it and don’t even try to error-correct them. So I’m glad that Cai has apparently now done so! But given the unforgiving nature of (say) the modular exponentiation function, this fact is hardly a surprise—one might even use terms like “eye-wateringly obvious.” It’s rigorously proving that a particular Swiss watch will probably cease to work if you smash it with a hammer.
But what rankles even more is that the paper combines this (worthwhile even if intuitively obvious) technical result, with personal speculations that quantum computing might be impossible because quantum mechanics isn’t exactly true, as if those two things had anything to do with each other. To extend the analogy, this is meticulously proving that the Swiss watch stops working if you smash it with a hammer, and then somehow using that to buttress your argument that maybe quantum gravity and cosmology will limit the accuracy even of Swiss watches that aren’t smashed.
Comment #132 June 15th, 2023 at 12:28 pm
Professor Aaronson what do you mean by “given the unforgiving nature of (say) the modular exponentiation function”? If modular exp has non-uniform TC^0 circuits would we have more robust arguments for Shor’s algorithm to succeed?
Comment #133 June 15th, 2023 at 2:24 pm
RSA #135: No, I wasn’t talking about the computational complexity of modular exponentiation. I just meant that modular exponentiation is not like neural net inference, where even if you make a few mistakes the output is still probably pretty good. With number theory, if you make a few mistakes the output is typically useless.
Comment #134 June 15th, 2023 at 6:29 pm
Thank you!!
Comment #135 June 15th, 2023 at 9:32 pm
Scott,
Any comments on this?
https://www.nytimes.com/2023/06/14/science/ibm-quantum-computing.html
In particular, how interesting or ground breaking is the result? Mostly asking about IBMs results not so much the writers somewhat poor explanation of how QCs possibly get speed ups.
Thanks!
Comment #136 June 16th, 2023 at 2:11 pm
Hi Scott,
I have the same question as Cleveland, and would love to hear your opinion on IBM’s “error mitigation” technique, which they use in the latest work reported in Nature. Is this really scalable? In our discussions with IBM, they were making a big deal out of the technique, which is important to them since error correction is not currently feasible. But I remember reading somewhere that error mitigation doesn’t really work.
https://www.nature.com/articles/s41586-023-06096-3
Thanks.
Comment #137 June 17th, 2023 at 2:23 pm
To whoever was asking about new QC algorithms: I just came across https://www.nature.com/articles/s41534-020-00291-0 from 2020, where someone claims a quantum computing speedup for simulating the classical Navier-Stokes equations. That sounds interesting though I haven’t looked at it yet beyond the abstract.
Comment #138 June 17th, 2023 at 3:17 pm
I thought you might be interested in this recent post by Tyler Cowen’s co-blogger, Alex Tabarrok. He discusses a paper by two economists, Jesús Fernández-Villaverde and Isaiah Hull, using a quantum annealer to run a simple economics model.
F-V & H seem to be doing things right, and there’s a lively discussion.
PS Thanks for everything. You are a better man than they say.
https://marginalrevolution.com/marginalrevolution/2023/06/using-a-quantum-annealer-to-solve-a-real-business-cycle-model.html?utm_source=rss&utm_medium=rss&utm_campaign=using-a-quantum-annealer-to-solve-a-real-business-cycle-model
Comment #139 June 17th, 2023 at 4:54 pm
Speaking of crankery^1, do we know for sure that the heat death is going to happen? This article suggests not (something something entropy gap) and cites a lot of physicists, but only their popsci books instead of textbooks or papers. https://www.noemamag.com/life-need-not-ever-end/
—
1 To be clear, I’m acknowledging that speculating against the heat death might be crankery, lol.
Comment #140 June 18th, 2023 at 3:40 pm
Christopher King #139: I think I saw a paper from the 1980s proposing that if structure formation keeps happening on larger and larger scales, it could keep the universe out of equilibrium indefinitely. The problem is that you only have what’s in our Hubble volume to work with. Under the current dark energy paradigm, with accelerated expansion, the problem is even more severe – after a few hundred billion years, only the Local Group (us and Andromeda) will be left within our cosmic horizon, everything else will have moved beyond causal contact.
Overall, the nature of the far future universe is highly uncertain. We don’t know if our “vacuum” is stable, we don’t know if baby universes can be created, we don’t know if there are large extra dimensions; we don’t even know what space-time will look like conceptually when the “it from qubit” revolution has run its course.
Comment #141 June 19th, 2023 at 2:43 pm
Bart Ehrman gives a pretty good talk about why tenure is important;
Comment #142 June 27th, 2023 at 12:13 pm
Scott: Re comment #24, you may have to reevaluate your respect for Tyson after watching this: https://www.youtube.com/watch?v=IfvoBlDUrPU&pp=ygUKdHlzb24ga2FrdQ%3D%3D
Comment #143 June 30th, 2023 at 3:23 pm
[…] an interview with Michio Kaku about his new book. Why promote such an atrociously bad book (see here and here) and broadcast Kaku’s absurd claims about this […]
Comment #144 July 8th, 2023 at 11:27 am
That YouTube video is actually one of the best and clearest explanations of Shor’s algorithm. That Veritasium dude is really good.
It would be great if you can do an episode with him on quantum computing or complexity theory or even busy beavers.
Comment #145 July 8th, 2023 at 11:47 am
This former DWave experimental physicist turned YouTuber also has some good videos. E.g. on interpretations:
https://youtu.be/mqofuYCz9gs
Comment #146 July 17th, 2023 at 5:02 pm
Hm, interesting post.
I don’t know so much about the public’s perception of Michio Kaku’s competence at topics outside his primary specialty/specialties. Last year I began looking into Quantum Field Theory texts for some introduction into the field, and I found his book helpful.
I stumbled across this blog for the first time a while back, and now it’s exciting to finally understand the “Quantum computers won’t solve hard problems instantly by just trying all solutions in parallel” headline haha.
I’ve always wondered “Why not? Isn’t that why quantum computers are so popular in the first place, and why they hold such exciting prospects for the future?”
From this post I understand that just one random outcome of the QC’s “parallel” computation can be measured. Which makes me then wonder what all the excitement about QCs is about in the first place, since all of that parallel processing power is mostly left unharnessed. Although of course I imagine there’ll be techniques to possibly leverage knowledge of the concerned probability distribution- to increase the chances that the observed outcome is one which is more important/valuable than others in some regard, etc.
Comment #147 July 25th, 2023 at 9:11 am
“The wonder of it, to me, is Kaku holds a PhD in theoretical physics” – well, it seems to me that Kaku just “rolled off”, as we say in Russian in such cases. Sometimes it happens that a completely normal scientist suddenly imagines himself a genius who has no equal, harbors a grudge against his colleagues, and finally just goes insane or something like that. There was such a scientist Kozyrev – not a freak, not obscurantist, and suddenly he starts to bullshit about “mirrors that capture paranormal phenomena” and all that stuff. Or the case of Albert Veinik – he seems to be also a normal scientist, at the end of his life he suddenly wrote a book “why I believe in God”, in which there is not a word about God at all. He describes how he measured the “chronon” wrapped in several layers of polyethylene or something like that. Dozens of such cases, in fact.
“A quantum computer, however, simultaneously analyzes all possible paths at the same time, with lightning speed” – it’s not clear what does he means by “lightning speed”. If he wanna say that QC calculates *instantly*, then this is not so – there’s the Margolus-Levitin theorem, which limits the time of *any* calculations to at least t = h/4E, where E is the average energy (assuming the energy of the ground state is zero), and t is the transition time of the system from one orthogonal state to another. Instantaneous computing requires infinite energy/breaking Planck barriers and doesn’t exist.
By the way, in my opinion, in principle, the separation of computing into “classical” and “quantum” makes little sense. Classical computing is actually quantum one, but using only pure states. As we remember, only pure states can be copied, but not mixed states (therefore, we can copy as much as we like in classical computing, but in principle it is impossible to copy in quantum ones).
Comment #148 July 25th, 2023 at 10:34 am
P.A.R.T.Y. #150: No, when you say “pure state,” you mean “computational basis state,” and when you say “mixed state,” you mean “pure state.” A mixed state is something else entirely, namely a classical probability distribution over pure states.
Comment #149 July 26th, 2023 at 7:24 pm
Sorry, all, for the random comment, but I couldn’t help but respond to MaxM (comment #6). Their comment brought to mind one of my favorite stories in the world. In Teller’s telling of one of his meetings with Reagan about Star Wars (recounted in Godchild’s “Edward Teller: the Real Dr. Strangelove”), he felt that the meeting “was far less successful than [he] had hoped,” because a National Security Council Staff member, Sydell Gold, “injected so many questions and caveats that [he] felt discouraged about the conference.”
Sydell Gold was my mom (a mathematician by training). Thankfully, sometimes there is someone there to say something.
Comment #150 August 1st, 2023 at 3:24 am
J Gold #149: Thanks for that story! Speaking up to interject uncomfortable questions can be an important contribution. I’m glad your mom was there in that meeting and made it go poorly for Teller.