Elements of Statistical Learning should be on there.
Added: ... so why not do it yourself, man?. Added 2: Duplicate, please disregard.
There are many sources of free books and even more free papers available online:
http://www.archive.org has both reprints of old out-of-copyright books and a lot of government publications, such as http://www.archive.org/details/nasa_techdocs and http://www.archive.org/details/NISTJournalofResearch .
http://www.sciencebooksonline.info/
http://www.intechopen.com/books
http://www.e-booksdirectory.com/listing.php?category=1 for science books http://www.e-booksdirectory.com/listing.php?category=3 for math
I guess I didn't emphasize this, but the difference between "here's a giant list of books" and Hacker Shelf is that the latter implements a karma-like system.
For whatever that's worth.
This website has everything I've ever needed to teach myself basic web programming:
Yes, this a repost from Hacker News, but I want to point out some books that are of LW-related interest.
The Hacker Shelf is a repository of freely available textbooks. Most of them are about computer programming or the business of computer programming, but there are a few that are perhaps interesting to the LW community. All of these were publicly available beforehand, but I'm linking to the aggregator in hopes that people can think of other freely available textbooks to submit there.
The site is in its beginning explosion phase; in the time it took to write this post, it doubled in size. If previous sites are any indication, it will crest in a month or so. People will probably lose interest after three months, and after a year the site will probably silently close shop.
I really wish I had an older version of this book; the newer one has been marred by a Cambridge UP ad on the upper margin of every page. Publishers ruin everything.
The book covers reasonably concisely the basics of information theory and Bayesian methods, with some game theory and coding theory (in the sense of data compression) thrown in on the side. The style takes after Knuth, but refrains from the latter's more encyclopedic tendencies. It's also the type of book that gives a lot of extra content in the exercises. It unfortunately assumes a decent amount of mathematical knowledge — linear algebra and calculus, but nothing you wouldn't find on the Khan Academy.
Hacker Shelf review, book website.
There's just a lot of stuff in this book, most of it of independent interest. The thread that ties the book together is graph theory, and with it they cover a great deal of game theory, voting theory, and economics. There are lots of graphs and pictures, and the writing style is pretty deliberate and slow-paced. The math is not very intense; all their probability spaces are discrete, so there's no calculus, and only a few touches of linear algebra.
Hacker Shelf review, book website.
This is a more fluffy book about the practice of software engineering. It's rather old, but I'm linking to it anyway because I agree with the author's feeling that the software engineering discipline has more or less misunderstood Christopher Alexander's work on pattern languages. The author tends to ramble on. I think there's some good wisdom about programming practices and organizational management in general that one could abstract away from this book.
Hacker Shelf link, book website (scroll down).
I hesitate to link this because the math level is exceptionally high, perhaps high enough that anyone who can read the book probably knows the better part of its contents already. But game/decision theory is near and dear to LW's heart, so perhaps someone will gather some utility from this book. There's an awful lot going on in it. A brief selection: a section on the relationship between game theory and cryptography, a section on computation in prediction markets, and a section analyzing the incentives of information security.