The Junk Mail filtering in Thunderbird didn’t seem to be functioning correctly after I did some random poking around so I reset the junk mail training to start over. I had 500+ junk e-mails to start the training off, but apparently that wasn’t near enough. Ugh. Bayesian filters work pretty good once they’ve been trained, but until then getting e-mail is a royal pain in the rear! :)

Web definition for Bayesian from Google:

a variation of the maximum likelihood classifier, based on the Bayes Law of probability. The Bayesian classifier allows the application of a priori weighting factors, representing the probabilities that pixels will be assigned to each class.

Update: Or in other words it compares the e-mail to a list of rules and decides whether or not it’s junk depending on how many rules it triggers. If you tell it something is Junk or Not Junk, it will learn from that.