> I think SpamPal and K9 use the same Bayesian algorithm.
The basic algorithms are the same, but it's the way that they're
implemented that makes the difference. Maybe it was the specific type of
spam that I was getting when I did my trials, but K9 was really fast --
much faster than SpamPal -- at picking up new twists that spammers add to
try and defeat the filters.
I'm also pretty sure that K9 was one of the first to consider things such
as contrast of words/background or font-size in HTML emails so that the
good/bad databases don't get poisoned by microscopic lemon-yellow words on
a white background.
But as you say, it's whichever floats your boat, and we're really only
talking about minor percentage points at the top of the league. They're
both way better than most of the others. :)
> I switched from K9 to SpamPal because SpamPal can do more than Bayesian
> filtering. I have turned off the public blacklists because they were
> giving me too many false positives. I'll have to try the P2P plugin.
> That sounds interesting.
Visit <URL http://www.rhyolite.com/anti-spam/dcc/
> for background info.
Oh, and the P2P filter also catches new viruses as soon they start because
as the virus spreads, lots of people start getting the same email and the
DCC servers notice it and then so do you... <g>