>
> the AWL adjustment is most likely ignored by the bayes stuff...this make it a score of -5.5
>
> --
> Martin Hepworth
So it was tagged SPAM because of the total hits (32), yet learned as HAM due to AWL?
No, you have it almost exactly backwards.
The AWL will never cause bayes to learn or not learn spam. Bayes ignores the AWL completely and absolutely. The AWL does not exist to it.
The message was tagged as spam because the AWL jacked the score way up.
It was learned as HAM because bayes learns based on the score of the message before AWL (and bayes itself) kicks in. That score was -0.6, and was based on the SARE rule alone.
Maybe I need to ask exactly what is AWL? It's "auto", so not something I setup. tnx
It's the automatic white/blacklist score averaging system. It tracks the score history of a sender, and pushes messages towards that sender's past average.
See
http://wiki.apache.org/spamassassin/AutoWhitelist