I'm currently trying out a unix script called spamgard(tm) from
wje@netcom.com. Since we are on the same system, I could grab it as
follows:
/u4/wje/release/sg-0.4.tar.gz <--- generic, plain-wrap
/u4/wje/release/sg-0.4n.tar.gz <--- adapted to Netcom
Sorry, don't know where it is otherwise stored, but assume it is since he
is supporting a non-netcom version (the netcom version is mostly a
different path for perl).
It is quite versatile. You can set up files for all major header lines
that let you accept via "regular expressions" (for non-unix types, you can
put in full e-mail addresses, words, or parts of words). You can also set
up a file that just drops the mail (like anything from Cyberpromotions or
savetrees, for example!)
Anything that doesn't fit is dropped (but logged), a "nice" letter goes
out (you can alter it) that says something to the effect that "we'd love
to hear from you, please put the following password into your subject
line."
He says that for as long as he has been using it nobody (or maybe just
one) person was clueless enough not to figure out what to do. My own
experience (of one week) is that the few people who needed to get through,
figured it out and a LOT of junk has been eliminated.
BTW, when the letter goes out and bounces due to forged returned address,
the script does NOT send out another "nice letter" (thus there is no
endless loop).
The biggest drawback I see is that it vastly increases net traffic. But
so far it seems to be working pretty well.
Just adding the topic of your FAQ will into the accept.subject file will
probably let a lot of real messages through. He also recommends adding
"re:" to the accept.subject file, thus people replying to usenets posts
are also likely to get through since most readers are set up to add that
to the subject line.