As a data point: I'm reluctant to pay too much for ebooks, because they have no resale value. There are plenty of print books that I've happily spent $30-100 for, because I know that I can resell them for most of the cost (typically $5 less, at most) if I don't consider them worth keeping. Arguably, this makes their effective cost more like $5-10, give or take - mostly shipping. I'm quite willing to pay more for books that are worthwhile, but it's hard to tell upfront. I really like the MEAP subscription model of, "free trial chapter(s), subscribe and get further chapters as they're written, get the book when it's done". Plus, it works even better for serial fiction.
Also: I have a kindle, and I use the hell out of it (citeseer!); I'm in no way opposed to ebooks. Just, pricing them is tricky. I've bought several. I also worked in a library for several years. Make what of that you will.
I actually consider physical books to have negative resale value. I recently moved to Japan so I wanted to get rid of all my heavy belongings. I listed all my books on an auction site and on bookmooch. All in all I ended up spending hours thinking about those books, mailing them, meeting people who came to pick them up etc. and all for some insignificant sum compared to the time spent. It would have been cheaper to throw them in the trash, but being a physical thing potentially useful for someone, I didn't have the heart to do that.
Who pays (sticker - $5) for a used book? Add the shipping cost and you're more expensive than a new one! Plus, what kind of book do you buy that have resale value? They surely aren't programming/IT books, because those become worthless after 3 years or so.
They don't all become worthless after 3 years or so, though the ones that don't tend to be the ones you aren't going to want to sell.
Some categories of programming books whose utility and/or interest needn't drop much years after they were written: Algorithms books (Knuth, Cormen/Leiserson/Rivest/Stein). Books on fundamentals and high-level principles (SICP, Code Complete, Pragmatic Programmer, Design Patterns). Books whose tool-specific content relates to tools that aren't changing much (Paradigms of AI Programming, Thinking Forth) or at least that have a large enough stable-enough core (Effective C++).
(Note: "hasn't lost much value" isn't quite the same thing as "has a great deal of value". Some of those books are better than others.)
Not necessarily; that says more about Ruby and Javascript than anything. I have plenty of tech books from ten or twenty years ago that are still valuable resources. Actual computer science books have a much longer shelf life than "$Tech_of_the_moment In Action".
Also: I have a kindle, and I use the hell out of it (citeseer!); I'm in no way opposed to ebooks. Just, pricing them is tricky. I've bought several. I also worked in a library for several years. Make what of that you will.