FRL asks "How long before publishers stop publishing in paper?" A topic of intense interest among librarians certainly, but phrased this way, the question is so broad as to be meaningless.
As I (and many other writers on diffusion of innovation) have pointed out previously, new technologies rarely replace older, robust technologies. They create new possibilities and shift the older technologies into different niches. Print on paper is a superb technology for many purposes, and publishers will continue to use it for centuries.
The question is only useful when the scope is narrowed. In my professional world, biomedical research and education, there are now electronic versions of just about everything that is published serially in print. The majority of the monographic literature is still print only, but I would expect it to catch up to the serial literature within, say, three to five years. But before publishers can abandon the print versions completely, they need to deal with two primary constraints -- reader acceptance, and global network infrastructure.
Several years ago, reader surveys (and again, I'm just speaking about my world of biomedical research and education) showed that most faculty still preferred using print. By now, that is completely reversed -- the community has become very comfortable with the electronic versions of publications and most people now prefer them. In my library, we quit getting the print versions of our Elsevier journals last January and have not received a single complaint or even comment.
Global networking infrastructure will take longer. While the advances that have been made in the third world in the last few years are truly astonishing, internet access is still far from ubiquitous or reliable across most of the planet. Until that substantially changes, publishers will continue to use paper to reach those audiences.
In general trade publishing (i.e., the books that are piling up around FRL -- and me too!), print on paper is going to be around for a long time, simply because books as objects are such marvelous things. When we think of them merely as containers for text or images, we are neglecting to account for the tremendous esthetic pleasure that we get from a well-designed, well-made book. If we didn't respond as we do to those qualities, publishers wouldn't spend so much time and energy coming up with all of the multitude of variations in book design that grace our shelves.
Cheap paperbacks are likely to disappear (at least in industialized nations) within the next five years or s0 -- electronic versions will move into that niche. But the well-made book will continue to evolve and will be with us for a very long time. And thank goodness for that, despite the ever-escalating danger of being crushed under a collapsing pile of them!