> the following is from paul graham's "A Plan for Spam" web
site... what paul
> describes is known as Bayesian Classification and is basically a statistical
> analysis of the contents of messages...
> http://www.paulgraham.com/spam.html
> warning! this article (and thus this message) is over 600 lines in length...

Heh, spamassassin does a good bit of this now. I'm running it here
(dimstar.net) and it catches about 98% of the spam sent to my system. 

Russ

--- BBBS/LiI v4.01 Flag-4