On Sun, 13 Sep 2009 06:56:27 +0100 Clunk Werclick <mailbacku...@googlemail.com> wrote:
> Well crafted accurate rules - that should really catch this very > common type of spam - will always be very much preferable to > something that can be easily broken by feeding it a mail full of junk > words. I don't particularly want to get into an argument about the merits of Bayes, but I can't let that piece of disinformation go uncorrected. Bayes poisoning is 99% myth and 1% misnomer. It does no significant harm to a database except make it larger, and slightly more aggressive, so in no sense is it genuine poisoning. This is pretty obvious if you think about the mathematics. Adding irrelevant text to a spam may make it less likely likely to be caught, but it seems to me that it's mostly the use of 0bfu5cation and multiple sources that gets these through. Spams that use these techniques are typically caught by obfuscation rules, BOTNET, URLBLs, Pyzor, Razor etc.
