In the Communications of the ACM, David Lindley “defined an NP problem as one for which no polynomial-time solution is known”. Scott Aaronson, please call your office (two minutes into the podcast).