Mailing List Archive: Re: [spa] Re: [spa] Re: [spa] Semi-invisible font missed by SA

On Thu, 19 Feb 2004, Raquel Rice wrote:
> > That would be my guess. We are now fully into the part of the
> > 'game' where the spammers get hold of spamassassin and run their
> > spew through it*before* trying to mail it, so that they can try
> > 'tricks' like these, and keep trying different ones until
> > something works.
> The problem with your theory, is that your bayes hasn't been trained
> the way mine has, nor has mine been trained the way that Matt's has.
> The likelihood of any given spam getting past two of us, let alone
> all three of us, is very slim indeed.

Unless, of course, you are an ISP, with limited resources, who can't run
Bayes Databases per user, and with such a diverse user base that site-wide
Bayes database might not be a good thing..... :-(

Still, my theory is still not that bad, even with a properly trained
Bayes, because the spammer needs to defeat the rule checks anyways....

- C

From: "Charles Gregory" <cgregory@hwcn.org>
| Unless, of course, you are an ISP, with limited resources, who can't run
| Bayes Databases per user, and with such a diverse user base that site-wide
| Bayes database might not be a good thing..... :-(

In our case, we can only use a central bayes because we are
using a database to store individual user prefs, the alternative
is not to become available from what I understand.

Most spam is so similar that individual bayes dbs may not be
that important as long as each user can adjust their threshold
white lists and black lists.

There are quite a few very nice filters available as well, the best
of all are the ones that block style (for the lack of a better term)
over those that block specific words.

Although, I'm still looking for one that gives me the name, address
phone number and possibly a photo of the spammer.

Greg