It's been said elsewhere in this thread that the textbook publishing industry is nothing more than a racket. The publishers routinely release "new" editions to textbook which only differ in having the problem numbers rearranged to attempt to make old editions useless for current classes. The prices in the textbooks are gouged beyond belief. The publisher makes more profit off of a class of students than the instructor does, which makes no sense on margin.
Part of the problem is that students can't vote with their wallets. The people that make the buying decisions are the professors, or worse the departments. Even without the occasional conflict of interests, these agents have no economic incentive to reduce costs for students. Students' only option to "vote with their wallets" is to look for a different academic program or school, which is rather absurd to regard as an "option". It is only out of the good will of the professor that costs for the students can realistically be minimized, but there are no real incentives to support this at most institutions.
Once upon a time, there was value added in having a publisher. But this is the 21st century, and there are ways for textbook authors to publish without imposing onerous costs on students. There are even some good publishers that will provide manuscript services at fairly minimal cost. So as I see it, the big textbook publishers have become nothing more than rent collectors in the style of the RIAA/MPAA, leeching off of the work of others. The good news is that academic culture seems to be changing. The younger generation of teachers is sensitive to these issues, and open source publishing looks poised to take off. I personally would view it as an ethical imperative to publish any textbook of my own under Creative Commons, and I think this attitude is becoming increasingly popular.