As good as it gets
This book makes for terrible machine learning study material -- it's entirely equations, with no intuitive explanations or even real-life examples. It might be a good reference for someone who already has an intuitive understanding of the algorithms.
Unfortunately, I've done some research and haven't found any books that are better overall. Tom Mitchell's book has a far better explanation of traditional machine learning methods, but it doesn't cover SVMs, which should be the meat of any modern machine learning course. I'd tell a friend to buy Mitchell's book instead, then using various papers, the site svms.org, and the tutorial by Burges to learn SVMs.
