Is the interesting part of P vs NP (since it seems very unlikely that P = NP) the mathematical machinery that would be required to prove that P != NP? Presumably doing so would require a way to determine a lower bound for _all_ possible algorithms that solve a particular problem.