Re: My spamassassin rules

[David Gibbs <david@xxxxxxxxxxxx>]

> Are these for your personal email? Or for the mailing lists?

These particular rules are for the entire system.

> I'm not familiar with the syntax, but it looks like you reject any
> email with "friend", "luck", "faith", or other words that in certain
> context would be innocuous.

> I see the points assigned to each rule; only 3 assigned to "f--riend"
> or "lu--ck", but "lo--ttery" scores 9 points. Reu--nion.com scores 20! So,
> perhaps rejection only occurs after a point threshold is reached?

What I'm doing is assigning a SCORE to the various words.  The score is cumulative with all the other spamassassin rules.

So if someone sends me a message that has the word 'frie--nd' in the subject, the score is raised by 3.  If the person who sent the message is truly a friend, I've probably received mail from them in the past ... so they would be in my automatic whitelist, which would reduce the score.  Additional rules in spamassassin will also be evaluated and probably reduce the score.

On my system, a message needs a score of 5 or higher to be considered spam ... any message with a score of 10 or higher gets automatically discarded.

> Based on the convoluted headers and content I've seen that gmail
> flags as spam, it seems to me that spammers work extremely hard to
> overcome anything that spamassassin rules can catch.

Actually, true spammers are loosing the battle against spamassassin (at least in my case) in a major way.  I get very little true spam.

The rules I've added recently have more to do with spam that comes from quasi-legitimate vendors that are spamming me.  Those are the last handful of rules that start RECVD_.  I'm marking some ISP's as spam sources.

FWIW: Your message was not actually delivered to me ... as it contained so many of the 'hot' words that the spam score was 12 ... which caused my system to (incorrectly) discard it.  Thanks for pointing out a weakness in my filtering scheme. :)

david