1. This overview of SHA-1 was only marginally better than reading the spec for it, but the author did warn how technical it was. I was about to ask why AND was the minimum function and OR was the maximum, but I just got it.
2. The author said that a hash function is good only if a change of one bit in the input stream changes many bits in the hash. This of course is a necessary property, but not sufficient for a good hash. Some others that I can think of are: hashes should be uniformly distributed over the inputs, and this goes for each input length, so that for length K and hash length N, there should be generally 2^(K-N) messages of length K that yield the same hash. Can we say anything about the information content (high entropy) of a message and what its hash is? Or maybe that it is generally indistinguishable from the hash of a message with low entropy? That would actually be a desirable property, right?
No comments:
Post a Comment