It seems to me that it would be useful to have a single repository of false negatives (i.e., stuff that slipped past SA) with some sort of automated process to crunch the messages to produce fodder for rules updates.
This would be most useful for body tests, since people would be using all sorts of methods of re-sending the messages, including manual cut-and-paste and other methods which do not preserve the original intact header. One meta-rule is that we might not want to consider messages that scored above a certain threshold, since most peoples' SAs would have caught them anyway. (There's no point in looking at a message that scored 5.5 which someone sent it because it got through their SA that is set to trigger at 6.0, for example.) One interesting number to know would be what percentage of SA users use the default score (5) as the spam threshold, and how many have raised or lowered it. I think this would help keep the spam-phrases, porn_NN, etc., rules fresh. I'm not volunteering (yet) but I have some ideas about how this might work. -- Michael C. Berch [EMAIL PROTECTED] _______________________________________________________________ Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [EMAIL PROTECTED] _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
