On Wed, 25 Sep 1996 13:46 EDT, E. Allen Smith wrote:

...

Supposing, too, that you know these weaknesses, would using separate algorithms for different portions of the number work well?

That would seem to be a possibility; admittedly, the local copy of AC has been checked out for the past year, so I haven't been able to take a look at it. I'd tend to think that if you _know_ the flaw, one could come up with a better way to deal with it than the generalized method I discussed. For instance, if the MSB of bytes coming out of a scribble window is too low in entropy, only use it XORed or whatever with a more-random bit.

That's what I was thinking. Or possibly even replace the MSB with the MSB from another algorithm with other characteristics. Genetic programing would also be interesting here if you could write a program that would mix-n-match algorithms/seeds and gradually choose one that produces very high entropy! # Chris Adams <adamsc@io-online.com> | http://www.io-online.com/adamsc/adamsc.htp # <cadams@acucobol.com> | send mail with subject "send PGPKEY"