Hashed hash (and Kent's games)

[Berzerk]

On Sun, 17 Jul 1994 Ben.Goren at asu.edu wrote:
> think, the desire to beat up on careless users. Berzerk suggests a 0.1 S/N
> ratio (and in an earlier note a couple useable algorithms for the multiple
> encryption process); that would not be practical for any decent sized
> database, and I might have 100K or so people to deal with. But I almost

It depends on the size of the noise.  If the noise could be a simple 
4-6char number(compressed name, with pointer to trash adresses or real 
mismatched ones), giving a 16 char hash and the rest of the information was 
much larger, say 100chars, a signal to noise of 1 would only be a 15% ish 
increse in size, and this improves if you have more data.

Berzerk.