[darkmighty]

Yea I thought of hashes too. Because there are hashes proven (?) to be secure, it follows that it's impossible to make a universally efficient fuzzer (i.e. one that necessarily spends much less than ~exp(parser size) time).

[ynik]

There are no hashes that are proven to be secure. And we aren't likely to get such a proof any time soon: secure hashes can only exist if P != NP.