As I’ve said before, SpamAssassin 2.50 is quite an improvement over the 2.4x versions in terms of successfully catching spam, even without using the Bayesian filter. I finally felt like I had built up enough “ham,” mail I didn’t want to be treated as spam, spam that SpamAssassin caught automatically, and most importantly, spam that slipped past SpamAssassin, to train the Bayesian filters, so I did so yesterday. I have to say that SpamAssassin is working even better than before now that the Bayesian filtering has kicked in. Specifically, I was getting one or two pieces of spam every day from some kind of art dealer, and now the Bayesian filter is insuring that those messages are being routed into my spam folder. This morning, there was absolutely no spam in my inbox, and no false positives in my spam mailbox either. I haven’t had any problems with false positives in a long time, but I was getting some spam that was slipping through, much to my chagrin. The Bayesian filtering does not seem to have induced more false positives, and is reducing the misses, so I’m more pleased than ever. If things continue along these lines, then I will be pretty darn close to eradicating spam from my life.