This post isn't all that spam-related, but I think it's an important topic because it represents a fundamental trend.

I've always said (well, I say it sometimes), that if I wasn't involved in the anti-spam industry, other than the stock trading arena, my next industry of choice would be something to do with Web 2.0.  I came across a good introductory article to Web 2.0 here.   I like the summary of the internet phenomenom, now all it has to do is start making money.

Both ourselves (Exchange Hosted Services) and Hotmail make use of Web 2.0 concepts in our spam filtering, although we piggyback off of Hotmail, they didn't borrow it from us. End-users volunteer to classify mail into spam and not-spam and the filtering algorithm learns from these classifications to classify future messages.  I call it Pseudo-Bayesian for lack of a better term as it based on probability.  Granted, users don't own the content nor do they own the filtering algorithm, but they do own the "material" for generating the decision.

For example, if one person classifies a message as non-spam but 75 others say it's spam, then the algorithm calls it spam.  So, this lets users (a lot of them) own the decision on spam classification.  The weakness, as I've pointed out in the past, is that users are not particularly accurate.  This is fine for an okay spam filter but insufficient for a great one.