Spam, spam, spam, spam.
After a stint last night of getting all my bayes database up to date with all the mail I have, I've been working for the longest while getting almost every mail I have learned, either good or bad. It's fun to see this included:
5.4 BAYES_99 BODY: Bayesian spam probability is 99 to 100%
And it's pretty darned accurate, too. So, now I'm more familiar with how spamassassin deals with bayes stuff.