I got Ruby Best Practices and the Pragmatic Bookshelf's 'Programming Ruby 1.9' from the library last summer, and had an incredibly productive time. I would sit in front of my laptop reading RBP, and any time I was confused by a class or anything, flip there in the reference and see what the method did. I could keep my terminal up on screen and not switch into "web browser + search engine" mode at all.

Having a book is like having an additional monitor full of information that is uncompromising on your task. It's also wonderful to read, and easy to pop open even when your laptop isn't out, and you're just curious "I wonder if Array has a method to do ...."

A final note, because of that, having a paper reference (especially an annotated one) lends itself strongly to actually reading large swaths of it, instead of just finding the method you need and putting it away. With it being clunky, you don't want to do a lookup twice, so you open it and keep it open, and often read through a module/class' entire method list. That really improves your knowledge.

So in short, because it encourages good work and work practices.

Obviously, I want one, or I wouldn't have asked!

The back of the book is not just a copy and pasted version of the standard python docs, it is more of an annotated reference. People pay money for this stuff because they either want a different viewpoint or more indepth coverage on certain topics.

I could write a book just answering this question. I find that most online reference sites, while extremely useful, lack a certain style, as well as, you know, in-depth treatment of language concepts.

It is also harder to flip back and forth between 2 or more screens to check related details than it is between multiple pages. Unless you are just checking a fairly specific, narrow detail, books are just more efficient in time and effort.