The Bestseller Code

Two researchers fed 20,000 novels into a machine-learning algorithm and came back with answers about why some books sell millions of copies and others don't. The findings are genuinely surprising — the themes and narrative rhythms that reliably produce bestsellers turn out to be counterintuitive, and the authors explain the science without losing the reader in jargon. The writing is warm enough that you forget you're reading about computational linguistics. The honest caveat: knowing the formula doesn't help you use it, and writers looking for a practical edge may finish the book enlightened but no closer to a breakthrough. Still, as a meditation on how mass taste actually works, it earns its keep.