That's not what I said. Generating a predictable distribution by hashing is old hat.
What's awesome is you can then make inferences about the original data from those hashes -- something that good hash functions are supposed to be resistant to, in isolation.
Comments
That's not what I said. Generating a predictable distribution by hashing is old hat.
What's awesome is you can then make inferences about the original data from those hashes -- something that good hash functions are supposed to be resistant to, in isolation.
It works here because we only care whether two bits of data are equal -- and a hash function had better preserve that relationship!