Hacker News new | past | comments | ask | show | jobs | submit
I still think P != NP, I would assume 2^{n/log(n)} is an elegant optimal boundary.