Hacker News Reader

The context Bernstein is talking about is very meaningful here. The mathematics of public-key cryptography provides a rich tapestry for covering up hidden backdoors. ARX ciphers are very simple compared to elliptic curves, and the complexity of round constants isn't really comparable to that of curves.

View on HN

Right, I should've added I agree with GP that ARX ciphers are incredibly simple and the ability to backdoor them would be a very novel (and concerning) discovery.

My point is that "the values come from pi" is not necessarily proof that the constants really are "nothing up my sleeve". Bernstein was discussing this in the context of NIST curves (which could be backdoored), but the same one-in-a-million maths works for any constants (so long as you happen to know a weak-constants vulnerability that isn't known by the public).

Reply on news.ycombinator.com