I’ve just update mine and Christine’s sites to WordPress 2.0 – no change to you on this end, but the admin interface is much cleaner and prettier.
The other thing I was thinking, as I whacked out all my spam comments, is a modified Bayesian filter to capture spam. In many of my spam comments, I noticed that they contain either a bunch of random nouns or the same ones repeated over and over. What if a spam filter could look at the relevance of topics to each other and determine if the comment fit or was truly random? Say, read in a large corpus as training text and build a topic map of concepts from that, then evaluate a comment against the topic map to determine the degree of relevance between the nouns in the comment.