I also bougth collective intelligence in action and algorithms of the intelligent web. Both are books from manning.com. I love those 2 books more than programming collective intelligence because I find them more complete and as easy to read.
This is a good read for those who just want a broad overview and working code for some interesting machine learning problems. It's not too realistic, but will get the ball rolling. For those who want to delve further, I would suggest reading through the Netflix Prize forums and perhaps even try implementing some of the code thrown around there. You will learn a lot by tackling a problem of that scale.
Would make a neat combination with newly-re-available Wikipedia data dumps: <a href="http://news.ycombinator.com/item?id=1265138" rel="nofollow">http://news.ycombinator.com/item?id=1265138</a>