Expected applications include DoS-proof hash tables and random generators.

i think that "DoS-proof" may be kept here since it doesn't mean any guarantees, but establish the expected application - when we need not a just another hash function, but one that can be used to protect from DoS attacks - unlike murmurhash3a, for example

Thanks for pointing to all the misleading spots. I have now completely rewritten each, so we can no longer apply this patch. Hopefully all concerns are now addressed ("DoS-proof" now gets an entire section); if not, please comment in #29.