> BAYES_50 should always be zeroed out. BAYES_50 means there's a 50% > chance of it being spam or ham.
There's an adage that says 'if something has a 50% chance of going wrong, 9 times out of 10 it will.' That's partly why I adjusted my Bayes scores in such a way that Bayes_00 scores -5.5, Bayes_50 scores -1 and Bayes_99 scores a mere 2.5 Most of the spam I receive gets caught by Bayes_99 (but would get caught by the other rules in any case), but I used to get a lot of false positives scoring marginally above threshold that had Bayes rated at 50%. And as I prefer users to get a little spam rather than miss an important email I figured that if Bayes isn't sure I'll err on the side of caution. As ever YMMV. Ralph Pickering Crown Business Communications Ltd
