I'm sure I'm just ignorant on this topic, but I am curious. Why is the
UNDISC_RECIPS rule not scored? There is this comment in 20_head_tests.cf:
# NOTE: this is what 100% valid undisclosed-recipients mails look like.
# If this gets a high score, that's a bug!
What does this mean? I think undisclosed-recipients lists are pretty spammy,
actually. There must be something I don't know.
I've been getting the "All the medications..." addressed to
undisclosed-recipients recently. I've adopted antidrug.cf and RCVD_IN_MANY
and trained bayes on them, but some still are getting though (why isn't bayes
learning them faster?) So I thought I'd do something like:
score UNDISC_RECIPS 0.1
meta L_SSSS UNDISC_RECIPS && LOCAL_DRUGS_MALEDYSFUNCTION
score L_SSSS 2.0
Is there something obviously wrong with this?
