Quantum Supremacy
« previous post | next post »
For the past couple of months, the phrase "Quantum Supremacy" has been on my to-blog list, based on points and counterpoints like "Google scientists say they’ve achieved ‘quantum supremacy’ breakthrough over classical computers" (WaPo 10/23/2019) and "IBM Says Google’s Quantum Leap Was a Quantum Flop" (Wired 10/21/2019). My interest, at least on the LLOG dimension, was not in the argument about how difficult a particular problem is for classical computers, but rather in the use of the word supremacy.
Now I can take this one off the stack, because a recent SMBC does a better job than I would have:
The aftercomic:
The mouseover title: "I await your demands for caveats, QC community!"
Frédéric Grosshans said,
December 12, 2019 @ 11:29 am
Note that there actually is an active debate about the use of the “supremacy” term is active inside the concerned research community (i.e. mine), as witnessed by this petition and the associated open letter in Nature , or less seriously by this meme shared over twitter.
As a regular reader of Language Log, I wondered if there could be an objective way to quantify the racial overtone (or its abesence), e.g. thtough corpus lingustics. I did a qucick search of “ADJ supremacy” in the COCA, and it turned out “white supremacy” is more than 50% of the instances. But I don’t know if it’s actually representative.
[(myl) No doubt the exact ratio depends on the source as well as the date, but in the past half-century "supremacy" surely has a strong racist association:
(Sorry about exceeding the margins…)
But beyond that, it's surely an unacceptable level of hype to talk about "supremacy" when Quantum Computing is still not yet usable, much less supreme, for any actual applications. We can hope that over time it will become useful, and maybe even dominant (though that seems less clear), but at best the Google researchers showed that their machine could perform a carefully crafted (and useless) calculation faster than a classical computer. ]
Frédéric Grosshans said,
December 12, 2019 @ 1:29 pm
Thanks for the quick and detailed answer.
I could speak hours on the significance of the benchmark, which was not proposed by Google by the way, but I am obviously biased by my field of research and it is not directly relevant to the theme of this blog. In short, I do not find it unacceptable level of hype, as long as it is clear the computation in itself is useless and that it is viewed as a milestone and not an end goal. The main reasons are: 1/ it is not “merely faster than a classical computer”, it is a really small machine (53 quantum bits) beating by far the largest classical computer (with way more than 10¹² bits); 2/ the crafted computation is well understood theoretically, so there is likely no shortcut, making this a physics experiment testing the computing power of nature.
An alternative term I recently learned an I like for the same concept is going beyond “the HPC horizon”.
About the dominance of quantum computing, it seems pretty clear it will only be dominant for some problems (including in very long term), since classical computers will always be much easier to build. Then some candidates for these problems are combinatorial optimization and machine learning, so they could have (in the long term) very broad applications.
J.W. Brewer said,
December 12, 2019 @ 5:54 pm
The google n-gram trendline myl posted is interesting, especially as compared to the simple trendline for "supremacy" which shows a steady secular decline since c. 1880 – "white supremacy" apparently took up post-1920 a larger percentage of a declining absolute rate of instances of use of "supremacy" in any context. I was disappointed to learn that at all times "supremacy" was still so much more common than Suprematism (coined no later than 1915 by K.S. Malevich as Супремати́зм) that in a graphic comparison the "Suprematism" line remains so close to the x-axis as to appear essentially flat, although when viewed freestanding it has a significant upward slope over most of the 20th century.
Gregory Kusnick said,
December 12, 2019 @ 8:52 pm
Diana Ross definitely achieved R&B supremacy.
Off topic, I'd dispute the claim that "classical computers will always be much easier to build." If quantum computing becomes practical, then the engineering problems will be solved, the usual curves of unit cost as a function of volume will apply, and it will (eventually) be as easy to configure a production facility to produce quantum computers as it currently is to produce classical computers.
Trogluddite said,
December 13, 2019 @ 6:15 am
@Frédéric Grosshans said: "…racial overtone (or its absence)"
The connotation of intolerance for out-groups is also invoked by "autistic supremacy". This refers to the small minority of autistic people who assert that (some expressions of) autistic perception and cognition are innately superior to those of "neurotypical"* people ("NTs"). A few even suggest that autism is the "next stage of human evolution"; in all seriousness so far as I can tell.
In the context of the autism support groups where I see the term used, it is not usually meant kindly, and the connotation of bigoted elitism is a deliberate slur. Such people are often seen as intolerant of both non-autistic loved-ones/allies/people and also of maybe the majority of other autistic people; especially those who perceive autism as debilitating, have greater need of support, or desire an inclusive advocacy movement. The supremacists' experience of being autistic may be very different to mine, and human behaviour often grates on my autistic traits, too; but reciprocal "different not inferior" will do for me!
I have no doubt that all of this has influenced my distaste for "quantum supremacy". I also find that novel compounds with "quantum" easily invoke the kind of pseudo-science woo-woo used to promote "new-age" snake-oil, which may encourage reading more into "supremacy" than was intended.
[*: "Neurotypical"/"NT" veers wildly from the clinical to the insulting: "not diagnosed with a pervasive neuro-developmental condition", "certainly not autistic", "assumed not autistic", "unlike [autistic person(s)]", "exhibiting behaviour confusing or hurtful to [autistic person(s)]", or even "implicitly belligerent towards [autistic person(s)]".]
Rose Eneri said,
December 13, 2019 @ 9:58 am
When I read this post about "Quantum Supremacy," I took the term exactly as it was intended by the authors. The phrase "white supremacy" never entered my mind until I read further.
David L said,
December 13, 2019 @ 10:56 am
When I see the word 'supremacy' the first thing that comes to my mind is Jason Bourne.
D.O. said,
December 14, 2019 @ 1:35 am
Correct me if i'm wrong, but 0.8% at a maximum doesn't seem like a strong association…
James Wimberley said,
December 15, 2019 @ 1:42 pm
"Air supremacy" is a standard if rather hubristic term in the USAF. I'm no expert, but understand that "air superiority" is being generally stronger than the opposition, say above Germany in 1944. Supremacy is when the opposition doesn't show up, as in Normandy the same year.
Garrett Wollman said,
December 16, 2019 @ 12:00 am
For those who perhaps have missed the reference in the cartoon's caption: Scott Aaronson is an outspoken theoretical computer scientist with a popular and opinionated blog (called "Shtetl-Optimized"), who studies computational complexity and quantum algorithms. Formerly at MIT, he moved to UT-Austin a few years ago.