The idea is simple enough: build a print-on-demand book printing press that will fit into a space the size of a mall photo booth and give it a net connection. You now have a device that can print a paperback book for anyone who needs something to read.
These devices would be put in places like train and bus stations and airports where folks find they have time on their hands and might want to buy a book with which to occupy themselves. Since they are (relatively) small, these book-o-mats, as I call them, could even be put in places like hospitals.
The design of one of these devices is quite simple in principal, while Xerox and other companies will happily sell you the technology to print a paperback book whenever you wish. There would still need to be a certain amount of engineering required, of course. A customer is likely to get impatient if a book needs more than five to ten minutes to print, so a book-o-mat would need multiple print engines running in parallel.
There will also have to be a status display and a monitor so that a customer could browse through the list of available books and authors. It should also be designed so that the next customer can be checking for which books they want while the previous is waiting for his purchase to finish printing.
The book-o-mat would have a large hard drive for storing .pdf or similar files, but it would mainly be a cache to hold the more popular books. When a book not on the drive is requested, it would be fetched over the net connection. The net connection would also allow the printer to report its status to a central location, either on demand or at regular intervals.
This would ensure that maintenance folks would know when the printer was low on paper or ink, and when it had a paper jam or similar malfunction that required immediate service. The devices would be ruggedized to maximize the mean time between these maintenance visits.
The company that owns and operates the book-o-mats would be a normal publishing house, and sign publishing contracts with the various authors who's books they wish to sell. They would have certain advantages that set them apart though.
To begin with, they would not have a concept of an 'out-of-print' book. Once they had the typesetting file for a given volume, they could print it whenever there was a request. They would also be able to go back to the old system of A-List, B-List and C-List authors, as there would be very little expense in hosting books for authors who are not yet a big name (but might yet become one).
Although print-on-demand systems cost more to produce a single book than do traditional methods, that book does not have any associated shipping costs, and there is no worry about unwanted books being returned. As a consequence this publisher would have a higher profit margin and could afford to give substantially more than the 8% royalty that is a typical author's payment today. 20%, 30% or even as much as 50% starts to become possible as the actual costs of producing a book drop to negligible levels.
These higher returns would allow the publisher to out-bid others for printing rights, and would go a long way to allay the natural fears of authors about what negative impact this new technology might have on their profession. Once the technology was perfected, one might even start to see these machines cropping up in book stores.