I'm just giving in and embracing that we're going to talk about quantum today.

You may recall last month when Project Eleven announced that they were awarding their 1BTC Q-Day prize to someone who had successfully cracked a 15 bit elliptic curve key (#1478012).

I have heard a welter of different opinions about how meaningful this is, including one from Scott Aaronson which was essentially: not meaningful at all.

But then today I came across this very nice post-mortem of the prize:

Current quantum computers experience on the order of one error per thousand gates, but cryptographically relevant instances of Shor’s algorithm require billions of gates. The only known way to cross this chasm is quantum error correction. There are promising quantum error correction experiments being done, but ultimately quantum error correction is still a work in progress. Participants in the competition would inevitably end up using non-error-corrected circuits, which have completely different costs and challenges and scaling properties. In other words, the competition would be measuring something irrelevant.

I hadn't heard it put so neatly before.

it’s too easy for Shor’s algorithm to solve small problems by accident...Basically, for small problems, Shor’s algorithm succeeds regardless of how well your quantum computer works. The computer working well only matters for big problems. This makes judging a Shor’s-algorithm-applied-to-small-problems competition extremely difficult.

And almost immediately after the prize announcement, another researcher claimed that the winning submission hadn't done anything better than a random number generator (#1478156).

It's not that the submission didn't create a quantum circuit, but rather that it is possible to get the correct answer even if the quantum circuit isn't working.

the fact that the circuit construction looks correct speaks to the insidiousness of the falling-with-style issue. You make a correct circuit, you get the expected result, you celebrate… but you got the right answer for the wrong reason. This is a fear that every competent experimentalist knows in their bones. It’s why they don’t just check that something works when it should work, they check that it breaks when it should break. Failing to do that is arguably the most common failure of reasoning in humans, so if you’re running a competition where this is a known possibility then YOU SHOULD BE CHECKING FOR IT.

All in all, this write-up on the Q Day Prize was really nice and helped me understand some of what is at issue here better than most things I've read. Also it was pretty short. Take a look if you want to get a better grasp of why there is so much dispute over quantum progress.