The Internet makes possible the instantaneous transmission and retrieval of digital text. It is widely assumed that this capacity will lead to the displacement of printed books by digitized books that are read mainly on computer screens or handheld electronic devices. But it is more likely, I believe, that most digital files of books will be printed and bound on demand at point of sale by machines that can quickly and inexpensively make single copies that are indistinguishable from books made in factories. Once most books have been digitized, anyone with access to the Internet will be able to purchase printed books from a practically limitless digital catalog that includes even those books that, under traditional publishing assumptions, would have been designated "out of print."

Also, the digital publication of a book online involves no physical inventory, thereby eliminating the costs of warehousing, shipping books to wholesalers and to retail stores, displaying physical books in retail stores, and returning unsold books to publishers. This would make digital publishing much less expensive than traditional publishing. Given the economic efficiency and convenience for customers of this new digital model of publishing, it is likely to eventually supplant or at least rival traditional publishing—although it will be some time before a catalog of printable digitized books becomes large enough to justify investment in book printing machines at numerous regional sites.

Moreover, the elimination of whole categories of expense means that under the digital publishing model, authors would be responsible for a greater proportion of the value of the final product and would therefore, according to literary agents, be entitled to a larger share of the proceeds. Currently a large percentage of publishers' revenue is absorbed by the costs of printing, selling, and distributing physical books, costs that are irrelevant to digital publication. Literary agents marketing new manuscripts could thus be expected to demand a significantly bigger slice of revenue for their authors than has been traditional. But large, established publishing houses, which are heavily invested in the infrastructure of traditional publishing, initially will be reluctant to accede. So the opportunity to bid for new manuscripts will go first to upstart digital-publishing firms unfettered by traditional practices or infrastructure. Under this competitive pressure, traditional publishers will have to reduce their redundant functions in order to accommodate higher royalty payments to authors or else they will lose their authors. Such adjustments are typical of the interval between a departing economic model and its successor and may help explain the caution with which today's publishing conglomerates are approaching the digital future.