When it comes to quantum computing, I regard the choice made by many colleagues and friends to pursue this direction of research as entirely (3000%) justified, and it has led to beautiful mathematics and science. (I hope that my own choice to pursue the skeptical direction was justified as well, though this is a more difficult call.)I do not know who is right about that factual question, but I am included to believe that Google did not really prove it, if Kalai is till unconvinced seven years later.For the record, I expect that scalable quantum computation — and even important milestones toward this goal — are inherently impossible. ...
One very clear and important point of disagreement between Scott and me concerns the following simple factual question:
Is it currently possible to produce (without classical-computing interference) samples of size 500K for depth-14 random circuits with 20 qubits and an XEB fidelity above 0.2? (This is something Google claimed in 2019.)
I tend to think that the answer to this question is negative; Scott strongly believes that this and much more have already been achieved experimentally. Thus these are sharp factual disagreements, and I hope they will be resolved within the next few years.
Aaronson is also wound up about some Middle East war issues. I understand why he is so pro-Israel, but I do not understand why he is so anti-Trump. At any rate, it is out of my expertise.
Kalai writes in such an excruciatingly detailed (even maths-like) manner, and yet so voluminously, that it becomes impossible to locate the right document where he said something of topical relevance.to you. That's precisely what happened to me once again, right now.
ReplyDeleteIn connection with the Google experiment, I wanted to point out a certain aspect which Kalai had mentioned in a blog post or so. I had figured out something similar on my own this year (by April/ early May), but it only thereafter that I had found that Kalai had mentioned something essentially similar, *years* ago. However, now that I wanted to refer to it, I cannot locate it. ... It's a different matter that Kalai had *also* put forth some idea (regarding QC) in the same blog post (or a connected post), a point with which I did not agree. ... But then, the first point (the one related to the Google experiment) was, practically speaking, more important between the two. And, I anyway cannot locate either of the two! [If I come to write a paper about it, I will contact Kalai for help in locating it.]
Anyway, just in case you haven't done so already, ear-mark (e.g. bookmark) the following two posts by him:
1. https://gilkalai.wordpress.com/2025/02/17/robert-alicki-michel-dyakonov-leonid-levin-oded-goldreich-and-others-a-summary-of-some-skeptical-views-on-quantum-computing/
2. https://gilkalai.wordpress.com/2025/02/11/roadmap-for-the-debate-about-quantum-computers/
Bye for now.
--Ajit
This comment has been removed by the author.
DeleteWell, I am happy to add that I've managed to locate the post by Kalai (mentioned above). I now found that it was cached at my HDD. It came up when I was looking for something else --- something which was related, but not exactly the same as the Google experiment.
DeleteAnyway, the relevant post is this:
https://gilkalai.wordpress.com/2024/04/02/random-circuit-sampling-fourier-expansion-and-statistics/
The reason why I'd found this blog post interesting was this passage:
``E) One of our findings is that the effect of gate errors towards larger effective readout errors is witnessed in data coming from Google’s simulator (and other simulators) but not in the samples of the Google quantum supremacy experiment.''
And no, the point on which I didn't agree with Kalai didn't come from the same post (mentioned in this reply); it came from the following:
https://gilkalai.wordpress.com/2026/03/06/the-fully-depolarizing-noise-conjecture-for-physical-cat-states-is-twenty-years-old/
--Ajit
[Deleted the previous reply to reply just to correct for typos and to make the wording more clear.]