Comment: Re:Secret-Key Cryptography would still work okay (Score 1) 165
Secret-key crypto isn't dependent on NPish-hard problems, just on complex messiness, and it'll work fine even if we've got magic quantum computers. We'd have to go relearn all of those annoying Key Distribution System methods that public-key replaced, figure out what if anything to do about signatures, and have to build a whole lot of new business models for dealing with trust, since we'd have to actually trust the people running the KDC, but we'd live.
This is not quite right. Secret key crypto will be fine if quantum computing becomes ubiquitous (or if we find out that P=BQP), but P=NP is a vastly more powerful result, to the extent that it would shatter secret key crypto as well. P=NP means that you can pluck answers to a question out of the aether with no more difficulty than checking if one random input answers the question. So if you know how to calculate "lambda key: ciphertext.decrypt(AES, key).matches(English)", then by P=NP magic you already know the list of all 256-bit AES keys that satisfy that calculation. (Substitute "English" for any human language or binary file format you prefer.) You would still have the problem of sorting through all the candidate plaintexts, but if the ciphertext is longer than 256 bits then the list of candidates will be very short. Only one time pads (key length equals plaintext length) would remain truly safe if P=NP, because an n-bit ciphertext could represent any possible n-bit plaintext, i.e. the P=NP magic doesn't teach you anything you didn't already know.