While I agree on your point about having an R&D department, the reality is that it is often difficult for a society publisher or a not-for-profit to build much infrastructure along those lines. Hiring a staff of top-notch programmers is simply not feasible for many scholarly publishers. Many (if not most) publishers outsource a lot of their technological development and services. A vast number are hosted by outside companies like Atypon and Highwire. There’s good and bad in this hosting. The good is that the companies are stable and costs for R&D get amortized across a large number of journals rather than just your small program. The bad is that they’re often compared to oil tankers in that they’re slow moving and hard to turn in a new direction. The economic savings they provide result in a loss of ability to react quickly to new technologies.

I think it’s also important, as you note, to focus on specific opportunities, rather than just jumping on a new technology because it’s cool or popular. For CSH Protocols, we’ve discussed developing iPhone apps, but I’m skeptical. I know I would never use my precious shiny iPhone on the inherently sloppy biology bench, which is where our material is used. No way am I exposing my phone to organic solvents and acids, or using it with fingers covered in bacteria contaminated gloves. New technology is exciting but investing in it must be tempered with a thought toward functionality that people actually want.

But to go to your broader point. This is why publishers need R&D departments. A way of investing in possibilities in a highly risk managed way. R&D can take those web stats, look at the areas of interest and see how to rapidly prototype something to test the response.
Your points about Lock-in etc are all valid (tho’ come on folks, there are plenty of examples of that particular issue in publishing!) So do not invest big sums upfront now. Recognise that instead it’s time to invest in knowledge gathering capacity so that CSH (or whoever) is able to spot the opportunity and act on it in manner based on the evidence they have collected for their particular set of constituents.

The other side of the problem is technological. Are you willing to tie your textbook program to one company’s device? An e-textbook that works on the iPhone won’t work on the Kindle. Neither will work on an Android phone or a Sony e-Reader, or a Palm Pre or a GameBoy or a laptop. Are you willing to go to the expense of creating 5 or 10 or 20 different versions of your e-textbook so you can cover all the potential devices? Even with a standard file format like ePub, you’re not going to have the same functionality on all devices. The iPhone has built in GPS and a compass, but the Kindle does not. How do you compensate for that? Will each student learn a different lesson depending on which device they own?

What if you invest a huge amount of money to make a Palm-specific version of your textbook and they go out of business or pull their device off the market? What happens when Apple updates the OS on their iPhones and some of your product stops working? Are you willing to hire in-house programmers to keep everything up to date and secure from malware attacks?

It’s not a simple matter of people being stuck in old ways of thinking. You’re asking for an enormous and expensive undertaking that would serve what is currently a very small, nascent and rapidly changing market. People are indeed thinking about new ways to create textbooks and new types of media, that’s not the problem, the problem is in the actual implementation.