On Monday, August 16, 2004, 1:36:07 PM, Andrew wrote: <snip/>
CA> I rarely get a complaint from my users about this kind of spam; I call it CA> "self-inflicted", where someone signs up for a "joke a day" or "daily CA> horoscope" or "diet advice" but they don't read the fine print. They CA> continue to get signed up for similar nuisances in perpetuity; the CA> unsubscribe is real, but only for the current newsletter. They can not get CA> off the "master list". CA> Usually these mailers are in SBL http://www.spamhaus.org/ and also listed in CA> Sniffer http://www.sortmonster.com/MessageSniffer/ as return code 60 CA> "greylist/experimental". Sorry, I must correct you here. (This is Pete (Madscientist)) I'm guessing you missed our announcement. ;-) We have retired the gray hosting rules at group 60 due to false positive rates that nearly matched accurate spam captures. We have replaced group 60 with the Experimental IP rules that are coded from messages reaching our traps. Accurate rule group assignments are available here: http://www.sortmonster.com/MessageSniffer/Help/ResultCodesHelp.html CA> If you choose to weight those two tests high enough, you'll successfully CA> block them. SBL is much slower than Sniffer at picking up the same kind of CA> mailing lists from new addresses, so once in a while, they will get through. We have significantly improved the accuracy and coverage of Message Sniffer over the past few months. As a result, many tests that might have been weighted more cautiously in the past should probably be revisited. A very good guide to the accuracy of spam tests in general, including all of the Message Sniffer rule groups can be found here: (Thanks to Markus) http://www2.spamchk.com/public.html While group 60 in the past may have been good for capturing the class of spam in this discussion, there is no current rule group that corresponds to the original "Gray Hosting" rules. When a spam case involving Constant Contact, XTM, Akamai et al we are now coding specific rules rather than the generalized rules of the past. We can always code specific white and black rules for registered users upon request. Hope this clears things up a bit. Thanks, _M PS: We have found that the kinds of messages being discussed here tend to be a source of conflict. Frequently by coding a rule for a joke-of-the-day or similar "implied subscription" we find we are inviting false positive reports down the line. Based on this, I recommend applying such rules to the smallest group that is both possible and practical... the balance will always depend on your users and policies. Hope this helps. --- [This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)] --- This E-mail came from the Declude.JunkMail mailing list. To unsubscribe, just send an E-mail to [EMAIL PROTECTED], and type "unsubscribe Declude.JunkMail". The archives can be found at http://www.mail-archive.com.
