Quantum supremacy will just be a random number generator

Now that Google is once again bragging that it is about to unveil a Nobel-Prize-worthy quantum computer, guess what its first application will be? Completely worthless, of course.

Quanta mag reports:

But now that Google’s quantum processor is rumored to be close to reaching this goal, imminent quantum supremacy may turn out to have an important application after all: generating pure randomness.

Randomness is crucial for almost everything ...

Genuine, verifiable randomness ... is extremely hard to come by.

That could change once quantum computers demonstrate their superiority. Those first tasks, initially intended to simply show off the technology’s prowess, could also produce true, certified randomness. “We are really excited about it,” said John Martinis, a physicist at the University of California, Santa Barbara, who heads Google’s quantum computing efforts. “We are hoping that this is the first application of a quantum computer.”

This is just crazy. There are 50 cent chips that generate quantum randomness. Why would anyone use a $100M quantum computer?

Measuring the qubits is akin to reaching blindfolded into the box and randomly sampling one string from the distribution.

How does this get us to random numbers? Crucially, the 50-bit string sampled by the quantum computer will have a lot of entropy, a measure of disorder or unpredictability, and hence randomness. “This might actually be kind of a big deal,” said Scott Aaronson, a computer scientist at the University of Texas, Austin, who came up with the new protocol. “Not because it’s the most important application of quantum computers — I think it’s far from that — rather, because it looks like probably the first application of quantum computers that will be technologically feasible to implement.”

Aaronson’s protocol to generate randomness is fairly straightforward. A classical computer first gathers a few bits of randomness from some trusted source and uses this “seed randomness” to generate the description of a quantum circuit. The random bits determine the types of quantum gates ...

Aaronson, for now, is waiting for Google’s system. “Whether the thing they are going to roll out is going to be actually good enough to achieve quantum supremacy is a big question,” he said.

If it is, then verifiable quantum randomness from a single quantum device is around the corner. “We think it’s useful and a potential market, and that’s something we want to think about offering to people,” Martinis said.

I am beginning to think that these guys are trolling us.

If you want some random numbers, just close your eyes and type some junk into your keyboard. Then get the code for the SHA-1 hash function, and apply it repeatedly. Free implementations have been available for 25 years.

SHA-1 has not been broken, but it has some theoretical weaknesses that might be an issue at some future date. If that bothers you, then you can switch to SHA-512. That will have no problem for the foreseeable future.

With all the hype for quantum computing, they admit that (1) quantum supremacy has not been achieved, and (2) even when it is, its most use application might be random numbers.

Update: Peter Shor weighs in with his views:

Horgan: The National Academy of Sciences reports that “it is still too early to be able to predict the time horizon for a practical quantum computer,” and IEEE Spectrum claims we won’t see “useful” quantum computers “in the foreseeable future.” Are these critiques fair? Will quantum computers ever live up to their hype?

Shor: The NAS report was prepared by a committee of experts who spent a great deal of time thinking about possible different ways of achieving quantum computation, the various roadblocks for these methods, and about the difficulties of making a working quantum computer, and I think they give a fair appraisal of the difficulties of the task. 

The IEEE Spectrum article gives a much more pessimistic assessment. It was written by one physicist who has had a negative view of quantum computers since the very beginning of the field. Briefly, he believes that making quantum computers fault-tolerant is a much more difficult task than it is generally believed to be. (Let me note that it is generally believed to be extremely difficult, but that it is theoretically possible if you have accurate enough quantum gates and a large but not impossible amount of overhead.) I don't believe that his arguments are justified.

Horgan: Scott Aaronson said on this blog that “ideas from quantum information and computation will be helpful and possibly even essential for continued progress on the conceptual puzzles of quantum gravity.” Do you agree?

Shor: I want to partially agree.

He also has other opinions about theoretical physics that I may address separately.