Most reading lists are status reports. "Here is what I read; impressive, eh?"

A useful reading list does something else. It names the few books that changed how you think, says briefly what each changed, and trusts you to find the rest.

I'll take a stab at the latter.

Genghis Khan and the Making of the Modern World rewrote a semester of schooling. Most of what I'd learned about Genghis was that he was a butcher who killed his way to an empire. The book argues he was something stranger — a logistician who happened to be ruthless, whose actual contribution was the spread of literacy, religious tolerance, and post-roads across a continent that had none. The lesson isn't about Genghis. The lesson is that the version of history you've absorbed is almost always the loser's. The winner moves on. He has more important things to do than ensure that his narrative persists in the record books.

Everything Is Predictable is the friendliest available primer on Bayes' theorem, which is the closest thing the world has to a complete theory of how to update your beliefs in the face of new information. I read it the year I was clerking, and I started noticing how much of the law is built around refusing to use it. That is because the legal system is inherently backwards looking — there is little room to revisit your original hypothesis to add an additional constraint when it is binding precedent.

Kingfish, the Richard White biography of Huey P. Long, is the best book on American political ambition I've ever read. Long was a populist demagogue from Louisiana who would have been president if he hadn't been shot in the lobby of the Capitol building he'd built for himself. The lesson is that the people who change a state aren't the ones who play the rules well. They're the ones who write a new set. But that tack isn't perfect, it just might get you killed.