I just got a spam which scored 13.80 but had a -4.8 point AWL adjustment. I don't understand how it could have been whitelisted because it is obviously porn spam. It even rates a BAYES_70. How is it possible that it was whitelisted?
This is a very common misunderstanding, which results from a general lack of understanding of what the AWL really does.
The AWL is NOT a white or a blacklist. It is a score averaging system. It's sole purpose in life is to try to push the score of an email towards the average for the past emails from that sender.
It is perfectly normal for the AWL to assign negative points to spam, and positive points to nonspam. This is NOT a bug, and is not a problem unless the message gets pushed to the wrong side (ie: a spam fails to get tagged or nonspam gets tagged).
If the AWL never assigned negative points to spam messages it would not work correctly. A negative score does not indicate that the AWL thinks a particular sender is "whitelisted", it merely indicates that this message is "more spammy than average for this sender".
Say for example a spammer sends me a message that scores a +10 and this is the first message he's ever sent so the AWL learns this user as having an average score of 10.
The same spammer then sends me another message which scores +20. The second message is "more spammy" than average, so the AWL "splits the difference" between the average score of +10 and the current score of +20, and subtracts 5 points to make the message score a +15.
Note that there's no problem here, because if the same spammer sent me a message that scored a +4 instead of +20 it would have added 3 points instead, making the email score a total of 7.
The only way the AWL will never assign negative values to spam is if the same sender is always decreasing his spam score, and never sends an email that scores higher than his past emails have.
------------------------------------------------------- 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
