What do you think of my idea of producing paper books on demand. Let's say you go on Amazon to purchase a book (print edition). Amazon has 1,000 copies available in inventory. Once inventory has run out, when a new sales takes place, the book is printed in real time, automatically, and shipped the next day.
This allows publishers to save costs on unsold inventory, by producing fewer copies than they expect to sell, for a given book. Now of course, if the cost of printing one book on the flight is prohibitive, my idea makes no sense. But probably there's an optimum: maybe print 1,230 books, sell 1,230 from inventory, and another 675 on the flight (or in batches of 10 books). This also means being pretty good at under-forecasting sales by the right amount, possibly by selling a small batch first (testing the market) to decide how many books need to be printed.
My idea of a book produced on demand has another advantage: you can customize the book, add customized ads in the book, maybe even customize the title. Are publishers doing A/B testing to choose the title of a book, to increase sales?
Precision: A few people mentioned that print-on-demand already exists. I know that (see also the note below, and see the above picture - it's a machine that does just that). My question is two-fold:
Note: I recently purchased Switzerland: And the Adjacent Portions of Italy, Savoy, and the Tyrol on Amazon. The book, which was last published by Baedeker in 1923, arrived the next day, brand new, with a publication date of June 2, 2013 (the date when I purchased it). The author was Anonymous, and it had no publisher. The only information available about the publisher was "Made in the USA, San Bernardino, CA". Clearly, this book is printed on demand.