My ISP recently changed the configuration of its mail servers so that it's no longer possible to forward them to Gmail to take advantage of its huge SPAM database and have the e-mails be sent back to my ISP.
Instead, all e-mails that are deemed SPAM are sent to each domain's spam@ mailbox, and the ISP expects all users to sort through SPAM using some desktop solution (I'm using POPFilter). Since SPAM represents over 90% of the e-mails I get, it's very time-consuming to go through the SPAM folder just to recover the very occasionnal legit e-mail :-/
For those of you managing e-mails for ISP's or big organizations, what are the current techniques available to filter SPAM on the servers so that users spend as little time as possible sorting e-mail?
Off the top of my head:
outgoing mail: Forbid outgoing connections with destination port TCP25 other than the ISP's mail servers, to reduce the impact of viruses that take advantage of compromised computers to send SPAM silently -> all outgoing SMTP connections should go through the ISP's mail servers
incoming mail: When the exact same e-mail is sent to thousands of recipients, mv /dev/null so that the user only has to examin the trickle of SPAM that went through the net
Are there other techniques that have proved to be effective at reducing the amount of SPAM users end up seeing on their desktop?
Thank you.
There are hundreds of techniques for stopping SPAM. A Bayesian filter is one, but there is also DKIM (which is something your ISP should implement), SPF records, Vipul's Razor databases, DCC databases, etc.
Spamassassin's home page discusses some of these techniques if you are interested, but this would be longest post in history to cover all of the ways to stop SPAM. It's truely an exhaustive subject, for obvious reasons.