Hi, On Thu, 09 Jan 2003 11:56:29 +0000 [EMAIL PROTECTED] (Justin Mason) wrote:
> Given that, the aim is to _maximise_ effective classification with > _minimum_ wrong classifications. Razor "gets humans in the loop", which > (theoretically) helps on the difficult spam where automatic classification > is hard. Not only that, the aim is also to drive down the cost of spam detection and rejection. Razor is part of a well-balanced spam defense. If it flags 50% or more of incoming spam without taking too much time or resources, it's a success when used in concert with other tools. I use: - IP-based filtering (firewall rules, DNSBL checks in the mailserver and in SpamAssassin) - Pattern-matching (header & body checks in SpamAssassin; procmail rules) - Distributed message checksums (Razor & DCC) - Statistical comparison to known good/bad email (bogofilter, and eventually SA 2.50) Assuming you can tune each method to reject 50% of incoming spam with no FPs, you've eliminated 93.75% of inbound spam. And that's just a naive analysis. The reason for using a variety of techniques is that each one will pick up messages missed by a previously-applied method. For example, any means of obfuscating content to evade body checks creates a distinctive signature that can be easily detected with a statistical analyzer. The defense-in-depth approach has proven very effective; the trick now is to order the tests so high-accuracy, low-cost tests are made first. As long as the cost of sending spam is lower than the cost of rejecting it, it will continue to be a major problem (assuming rational spammers, etc., etc.) It seems doubtful that the return on spamming will drop to the point that it's not economical to do so, so maybe the best we can hope for is to drop the cost of filtering and rejecting spam. -- Bob ------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com _______________________________________________ Razor-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/razor-users
