on Wed, May 05, 2004 at 11:47:49AM +0100, Colin Watson (cjwatson@debian.org) wrote:
> On Wed, May 05, 2004 at 01:34:25AM -0700, Karsten M. Self wrote:
> > on Tue, May 04, 2004 at 11:49:20PM -0700, William Ballard
> > (40414.nospam@comcast.net) wrote:
> > > Is it fixable? Do I need to start giving up on Debian being able to
> > > filter the spam?
> >
> > You might start taking note of significant contributors to the list,
> > whitelisting these, and possibly filtering out messages from unknown or
> > spammy sources with no reference to existing posts. Something of a
> > harsh measure, but seems we're headed this way.
>
> That's a *particularly* harsh measure on debian-user, where many of the
> messages starting threads are from "unknown" users looking for help.
It assumes that a new user will prompt _some_ response from somebody,
and that not everyone will be running this filter. Probably safe
assumptions.
Less harsh would be to find some good mutt scoring hooks, though I've
had limited success with these. The inability to score on message
content is really annoying, particularly as it *is* possible to do
index highlighting on same, mooting the mutt doc's justification for not
providing the capability for scoring.
For myself, I've recently whitelisted pretty much all the major
contributors to debian lists, and will be using this to ham-train mail.
I've been running sa-learn on all spam, but am concerned I'm not getting
a good representation of ham in my filters to differentiate.
Peace.
--
Karsten M. Self <kmself@ix.netcom.com> http://kmself.home.netcom.com/
What Part of "Gestalt" don't you understand?
There is a time for everything.
Attachment:
signature.asc
Description: Digital signature