building a true RNG

[David Wagner]

Barney Wolff  wrote:
>This leads me to ask what may be a laughably naive question:
>Do we even know that the popular hash functions can actually generate
>all 2^N values of their outputs?

It seems very unlikely that they can generate all 2^N outputs
(under current knowledge).  However, they satisfy the next-best
thing: their output appears to be indistinguishable from uniform to
computationally-bounded observers, hence it's "as good as" if they
could generate all 2^N outputs for most purposes.

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at wasabisystems.com