> I'd say that with all checks (bayes, rbl's, razor, dcc etc) enabled, that > 5.0 is pretty good for "straight laced" email acounts, and 7 is reasonable > for "most people". Any higher than 8 and you start missing a lot of spam. > Given the commentary in this thread and the fact that my bayes seems to be working better now (deleted the .spamassassin directory, did a rebuild, and then fed it 650+ spam messages), I have adjusted the settings for my users as follows for their "sensitivity" options: Severe = 0.5 (unchanged) High = 2.0 (changed from 1.5) Medium = 6.0 (changed from 4.0; this is the new default) Low = 10.0 (changed from 7.0) None = 999999 (unchanged; what's the max allowable?) I'll test with these levels and see how we do with false positives and false negatives. I was motivated to make this change after noticing that with the reloaded bayes, scores were tending to be much higher.
cheers, -ben -- Benjamin Tomhave, Senior Systems Engineer [EMAIL PROTECTED] Sofast Communications www.sofast.net ------------------------------------------------------- This SF.Net email sponsored by: Parasoft Error proof Web apps, automate testing & more. Download & eval WebKing and get a free book. www.parasoft.com/bulletproofapps _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk