Why Did the Number of E-books on Amazon Triple?

Stephen Dubner reports Joel Waldfogel's finding that the number of e-books on Amazon tripled between 2022 and late 2025, and that AI accounts for essentially the whole increase.

More than half the books released in 2025 used AI, and almost 100 percent of the increase came from authors who arrived after ChatGPT existed. Waldfogel calls the pattern the random long tail. The long tail was access to books that already exist. The random long tail is what happens when making a book gets cheaper and a mass of new books appears. Most of that new work is bad, and some of it is good, including work that would not have existed otherwise. He points to music as the earlier case: between 2000 and 2008, the amount of new music tripled as digitization cut the cost of creating it.

In the book study, quality means whether people want to buy or use a title. The measures are how often a book is rated, its sales rank, and its average star rating. On those measures the AI books are worse on average: lower stars, fewer ratings, and worse sales ranks. They are not all terrible.