Re: [tlug] giving up on email

[Matt Gushee <mgushee@example.com>]

On Fri, Apr 02, 2004 at 01:32:30PM +0900, Stephen J. Turnbull wrote:
> 
> I would guess that at XEmacs spamassassin gets about 70-80% of the
> spam (with essentially no false positives), with the trend being
> definitely downward (it was getting 90% in October).  I think the
> spamassassin people have gotten tired of trying to keep up with the
> spammers, because they don't seem to be releasing anywhere near as
> often as the spammers come up with ways to avoid spamassassin.  With
> the addition of nearly 1000 lines of procmail filter, the XEmacs lists
> are currently catching about 99.8% of the spam, with total false
> positives running about 0.5% of the genuine posts.
> 
> I think at this point we need to go to a more AI-like approach, even

What would that mean? Isn't Bayesian filtering somewhat AI-like to begin
with?

At any rate, until someone implements that hypothetical better approach,
it might be worth trying SpamBayes (http://spambayes.sourceforge.net/).
Another list I'm on switched to it from SpamAssassin about a month ago,
and so far it has been very close to 100% effective. According to the
SpamBayes site, they achieve these results through a combination of
improved algorithms and rigorous testing.

-- 
Matt Gushee                 When a nation follows the Way,
Englewood, Colorado, USA    Horses bear manure through
mgushee@example.com           its fields;
http://www.havenrock.com/   When a nation ignores the Way,
                            Horses bear soldiers through
                                its streets.
                                
                            --Lao Tzu (Peter Merel, trans.)