I suspect that the random number characteristics for good compression would be nearly identical to those of cryptography. After all a compression algorithm is a form of ecryption. Just a thought...