search

tireddy2 / pqc-for-engineers

Other
13 stars 6 forks source link

Is the "10 second RSA break" claim justified? #52

Closed BenS-3 closed 4 days ago

BenS-3 commented 3 weeks ago

Section 7.2 claims that RSA-2048 can be broken with 4099 logical qubits in 10 seconds. It references an academic paper for the 2n+3 figure, but the 10 seconds figure is taken from a company blog, which itself doesn't cite where it got that figure from (the 2n+3 paper doesn't discuss runtimes AFAICT).

I haven't previously heard claims for figures approaching 10 seconds with such a low number of qubits, and my view is that an RFC should aim to cite a credible source that explains where those figures came from.

Otherwise, I'd be in favour of removing that claim. The referenced Gidney/Ekerå paper is already sufficient for proving RSA-2048 can be broken in a short time frame.

BenS-3 commented 2 weeks ago

For what it's worth, I'm still not a fan of the updated text ("10 seconds" -> "a much shorter time"). There's no a priori reason why it might the runtime should be shorter; the paper aims to minimise the number of qubits, which I could believe might actually increase the runtime.

(I lack the knowledge to infer anything from the paper, so I'm happy to be corrected.)

tireddy2 commented 4 days ago

Thanks, fixed text in the latest version of the draft.