all
Tinkering with a new box for my SA and just noticed that the bayes + net score in 3.02 is a lower than in 2.64 and lower for bayes only.
why?
As best I can tell it's score dilution by the URIBLs from SURBL.
See this post from the archives, which covers a similar note:
http://article.gmane.org/gmane.mail.spam.spamassassin.general/61489
