I am getting to a stage in my life where I am willing to pay extra for less clutter. I have come to the point where lack of space is more of a constraint than lack of funds, at least as far as books are concerned. I am therefore an enthusiastic user of my electronic reader, my Kindle 1. But at the same time I share a lot of the concern about DRM that other people have. If I buy a copy of a book, I should own that copy: I should be able to lend it out, I should be able to make a print copy for my personal use. Current DRM regimes appear to me an unseemly mixture of panic and greed.
I think part of the issue is that the current eReader situation is somewhat bizarre. The devices are being sold and supported by booksellers rather than by publishers or groups of authors. I think that this is what is leading to the artificiality of the situation. I've been thinking a lot about what I think the reading industry is going to look like in the coming years.
The easiest way for me to think about it is in terms of the question, "Who adds value (that I find valuable)? What value?" There are several kinds of value. First, the text itself: is it good at what it does? For fiction, does it tell a good story? For non-fiction, is it well-written and factual? For fictitious non-fiction, does it ... provide whatever the hell it's supposed to provide? (Don't look at me, I don't read that kind of thing) There's a lot of value too, though, in guiding selection, value that's increasing all the time. Every minute I spend looking through lists of books is a minute I spend not reading. The cost of missing a book I'd enjoy or find useful is difficult to quantify, but I'm reminded of the 20-odd years of my life spent not knowing that Terry Pratchett existed, and that cost can seem pretty high. That selection applies not only to finding books for yourself, but in finding books to give as gifts.
So, who adds that value? Authors, of course, create the text, and frequently are wonderfully consistent: once you find authors you like, coming back to their work is a great way to skip a lot of searching. Editors and agents help weed out sub-par work, but also work with the author to improve the text. They are less consistent than individual authors, as they work with many more texts, but that can be a very good thing (I'll touch on that again later). Printers (and now device manufacturers) add a lot of value, perhaps not to the text itself, but in producing a comfortable, appropriately-durable reading experience. Marketers perform a needed service, usually badly. (Any occupation that shoves Dan Brown's books in my face and practically hides Charlie Stross's books is irredeemably lost) Libraries, book clubs, and book stores perform similar tasks, of helping you find books that you're likely to find useful or enjoy.
So, thinking in those terms, how do I think things are likely to shake out?
I strongly expect something like the following scheme to surface in the next five years, having seen bits and pieces of it floated in various articles. I think that publishers (particularly strong genre publishers like Tor) and author groups like the Authors Guild or SFWA will begin offering subscription services. A subscriber to this service will pay, say, $50 a year for unlimited digital access to an online catalog.  This access would entail unlimited downloads to mobile devices like Kindle, Nook, iPhone, and Microsoft's inevitable me-too eReader (I'm guessing six months until that announcement, BTW), and probably a PC-based reader with search and limited print capability -- you can print to any printer, but it'll be heavily watermarked. The files would probably DRM'd, because let's face it, you're renting, not owning. There will be grumbling, but I don't think much, especially if they let you buy DRM-free copies of books you particularly like. I mean digital by that, though the savvy services will probably work out exclusivity deals for discounts on pre-printed dead tree versions (with all sort of additional enticement, such as exclusive artwork, leather binding, authors' signatures, gold leaf, charity profit-splitting, a sprinkling of cocaine, etc.)
Now, I've been focusing on eReaders above, but that's not the end of the story by any means. Print on Demand services will come into play here in a big way. If the subscription services are smart, they'll embrace it, as will brick-and-mortar stores. Your corner bookstore (or library or coffeehouse) could very well wind up being basically a front end for a couple POD machines and subscriptions to a lot of different services. You walk in, and use either their subscription or yours or print out a copy of a book you'd like to read, paying an appropriate surcharge. I'm guessing that coffeehouses will wind up with a nice big bookshelf full of "donated" used books (read: left behind, not too soaked in coffee) and that libraries will be willing to eat the cost of a book in exchange for keeping it on the shelves when the user is done.
That suggests to me that the regular BaM stores are likely to keep stocks primarily of nicely-bound books (for gifts, particularly beloved books, or to have signed), books that don't do well as POD (coffee table books, books for small children), and very popular books for which people won't want to wait for the machine (and which can be sold more cheaply pre-printed). As today, they'll probably increasingly try to be combination bookstores/something else -- who wants to bet that Barnes and Noble will be thought of primarily as a coffee chain in 20 years?
As to remuneration: The subscription services would likely be ostentatiously fair to authors, since I wouldn't be surprised to see certain books available from multiple groups, and they will live and die by continued customer trust/support, as well as being dependent on links from authorial blogs. I'm guessing that it'll boil down to most publishers taking a certain % off the top, then distributing the yearly earnings proportionately among the authors whose books are downloaded. And while that seems fair, it's likely to be hotly debated. For one thing, it will basically pit authors against each other. Second, it creates a financial incentive to split books up. The readers won't care (unless it causes an outbreak of Dickens Syndrome), but fellow authors will likely get very angry when they see that happening. I'm confident that something reasonably equitable will get worked out, probably something slightly different for each service. (Authors will likely figure out which service gives them the best deal and steer people that way, much as they currently do with Amazon links)
It would be interesting to see what this does to the short fiction market. Right now it seems that novels rule the roost, and for genres like fantasy, serial novels are giants.  I wonder whether, with something like this, short fiction and novellas will become more popular. As John Scalzi recently pointed out, short fiction used to be a much more reliable income stream for authors. I wouldn't be surprised to see that happen again. Nor would I be surprised if manga/comics become more popular when digital readers become appropriate for displaying them.
One more prediction, by the way: as metadata becomes more important to people trying to skim through thousands of works, I think we're going to see the re-rise of the celebrity editor. There are already editors known for being associated with good work (The name "Neilsen Hayden" springs to mind) after all, and any decent matching algorithm will surely take editor into account when matching readers to new books. I would not be at all surprised if the author's agent also becomes a useful piece of metadata. Moreover, I'm betting that if "editor" and "publisher" are searchable pieces of metadata, then "editor != none & publisher != none" will become a very common search criterion.
 This is not new, by the way. O'Reilly already has something very like this, and it works very well. It is, however, prohibitively expensive for many people. And of course professional societies like IEEE have been doing this for journal papers for years, but they skip over the difficulty of compensation by simply not compensating any of their authors -- indeed, we are charged for membership and conference admission.
 My take on it is that this is a consequence of a couple things. First, readers strive for consistency, and finding a prolific author you like is much more valuable than finding one who isn't. This is not only true for readers but for agents and publishers. Second, world-building is a time-consuming task, and setting multiple stories in the same world saves a lot of time. Moreover, once the world is built, it becomes easier to think of stories in that world than in another. In some ways, this results in better stories set in richer worlds. People like Jim Butcher who are very good at planning things out way in advance gain a lot of natural advantages