Woke up this morning with no spam! Unfortunately my KidRobot newsletter, my Craigslist notices, most of the kronolith list traffic, and about half the IMC list traffic had also been flagged as spam. After retraining it on these false positives, I’ve pushed it off this early fitness plateau, back into the troughs to learn new rules. Has any work been done on visualizing Bayesian nets? I’m thinking of graphs we used to use when playing with genetic programming, that showed a topographical view of the fitness scape.