There are a lot of parallels between diapers and books, and I expect those parallels will continue. When the original (really quite awful) Pampers were created, I don't think anyone was envisioning that section of Costco that had massive units of diapers for sale, much less that Amazon would buy Diapers.com and a lot of people would be kind of impressed by that.
I'm not necessarily interested in trying to predict the book equivalent of Costco diapers much less Diapers.com. I am, however, trying to understand what a dramatic change in the demand for paper books will mean for the physical manufacture of those books. I'm completely convinced that the Big 6 and a wide variety of many (but not all) other publishers are going to transition to electronic books successfully. But I don't think they're going to be running their physical plant unaltered.
So what's going to change?
If this guy is to be believed:
"The Big Six NYC publishers use LSI to keep much of their backlist in print. ... LSI will get your books listed in the Ingram catalog, which allows all those bookstores and libraries to order them. Plus, the major online retailers (Amazon, B&N, Powell’s, etc.) will list your books on their sites and can order directly from LSI. The other major US book wholesaler, Baker & Taylor, will also order books for libraries and bookstores directly from LSI."
This is LSI:
Publishers too small to justify their own printing plant already use this service or contract to an outside printing plant. The size of publisher "too small to justify their own printing plant" will likely scale up in a relatively smooth path, with larger publishers already transitioning backlist titles or using LSI to reduce risk on books likely to have a very short life.
The terms on which physical books are offered for sale in stores reflect a world in which gadgets were made cheaply by production in vast quantities. In that world, it was common for producers to not demand payment from retailers until the retailers had had lots of time in which to sell the widgets, and to accept returns without fees. The manufacturers only had one setting (make a job lot) and they had no other choice in terms of moving the goods. In a world in which physical books are produced in very small quantities (perhaps one), it is reasonable to expect that the terms would change.
In terms of impact on corporate structure and jobs and so forth, the transition from pbooks to ebooks is close to a manufacturing change that, through automation, other technological changes and so forth, results in the loss of factory jobs.
So when bloggers speculate wildly about print-on-demand (POD) producing pbooks in a future world dominated by ebooks, they are Not Wrong -- but they are also Not Right. Visualizing POD as something that Kinkos might do, or a library might operate to serve patrons (possibly making a book that would survive one or a few readings and then be readily compostable -- hey, I'm not making this stuff up. Someone else is.) is a _massive_ misunderstanding of how POD interacts with an ebook dominant future. The POD in question is a printing plant indistinguishable to someone ignorant of printing plants from the printing plants currently run by the Big 6. It just doesn't happen to be owned by one company (because it no longer makes economic sense to vertically integrate this piece) and it uses a very different printing technology (I'm betting digital offset or direct image offset or whatever it's called, but I don't know for sure).
I've pieced this together from a small number of people who weren't attempting to answer the question I had, so I'm sure some crucial part of this is wrong. I'd be perfectly happy to hear more information on the subject.
ETA: This is an area in which books are really _not_ like music. As a manufacturing business, the transition from vinyl to CD was comparatively smooth, and while automation eroded jobs in CD manufacture, DVDs and so forth have lengthened the lifetime of those factories. OTOH, it's _exactly_ like the disk industry in general, which is falling off a cliff as music, movies and other media, software and so forth are all switching over to digital fulfillment.