[Cryptography] updating a counter

[dj at deadhat.com]

>
> Precisely. If I understand things correctly, there is (or ought to be)
> zero correlation between encrypt(x) and encrypt(x + 1) for a good block
> cipher.
>
> But of course, not being a cryptographer, I'd love to learn otherwise.
>

There is one relation for a block cipher E:
  E(key, x)  != E(key, x+1)

This is not true for a hash. There may be collisions, but they are hard to
find with a secure hash.