spamassassin is running almost fine, but seems to give too much weight to one factor. The factor is ALL_TRUSTED - which means all the MTA's the mail has gone thru are trusted, or conversely are not known spam senders/relayers. ALL_TRUSTED gives a negative score (ie less likely to be spam)
So I am getting messages with BAYES_99, ie where there is a 99% bayesian probability of spam, but not gone thru an untrusted MTA, so it is tipping it back below my limit of 6.0. I have manually set BAYES_99 to give a score of 5,5, but I cannot find what figure the defaults are assigning to ALL_TRUSTED. If anyone understands that ramble and can help, please let me know! -- Nick Rout
