How do I actually choose my next non-fiction book?

Most non-fiction reading decisions are made by accident. Someone on a podcast mentions a book; a friend posts a photo of a cover; an algorithm surfaces it as "people who liked X also bought." Three weeks later you've bought the book, and a year later you've finished a third of it and can't remember why you wanted it in the first place.

The result is the pile problem — the one I wrote about in why your reading pile keeps growing. But the pile problem starts upstream, at the purchase decision. If you choose better, you read more of what you buy and you waste less time on books that don't actually deserve it.

Here are five criteria worth applying before the next purchase. None of them are about the book's marketing. All of them are cheap to apply once you have them in hand.

1. Does the book have an argument you can summarize in one sentence?

The single strongest predictor that a non-fiction book is worth reading is that the author can compress its central argument into one sentence. Not "this is a book about happiness" — that's the topic. The argument is the claim: "happiness rises sharply with income up to about \$75K/year, then plateaus, because the goods money buys become substitutes for goods it doesn't."

Books with a tight single-sentence argument: Atomic Habits ("identity precedes behavior; tiny daily votes compound"). Thinking, Fast and Slow ("two systems run your cognition; System 1 is much louder than you think"). Essentialism ("the disciplined pursuit of less, but better, beats the undisciplined pursuit of more").

Books without a clear single-sentence argument tend to be loose collections of related observations. They can still be worth reading — but the per-page value is lower and you'll forget more.

The check: read the book's flap copy or first chapter. Can you state the argument in one sentence without using marketing words? If yes, proceed. If no, look harder before buying.

2. Is the author qualified by experience, not just by writing about the topic?

Non-fiction has a credibility problem that gets worse every year: the proliferation of authors who write authoritatively about topics they've never directly worked on. The book gets researched, synthesized from other sources, and presented as expertise — but the author is essentially a journalist with a strong opinion, not someone who's been in the arena.

This isn't always a deal-breaker. Some of the best non-fiction is written by skilled synthesizers (Malcolm Gladwell, Outliers, is the classic example — he's not a psychologist, but he's a great explainer). The check is whether the synthesizer adds insight beyond the underlying research.

But for any book promising to tell you HOW to do something — invest, lead, negotiate, write, build, parent — strongly prefer authors who have done the thing at scale. Daniel Kahneman did the original research. Charles Duhigg spent five years reporting and interviewing. Chris Voss was the FBI's chief international hostage negotiator. The credibility you get from someone who has DONE the thing is qualitatively different from the credibility of someone who has merely WRITTEN about it.

When you can't tell from the cover, look at the bio. If the bio is mostly "is a frequent contributor to X" and "writes for Y," you're probably reading the journalist-version. Adjust your expectations accordingly.

3. Does it pair with something else you've already read?

Books read in isolation are forgotten. Books read in pairs reinforce each other. The single highest-leverage choice in your reading life is to pick books that share an underlying framework with something you already know.

If you've read Atomic Habits, the next move is The Power of Habit. Clear and Duhigg share the cue-routine-reward framework; reading them together gives you two reps on the same idea, which is the cognitive condition for retention.

If you've read Thinking, Fast and Slow, the next move is something that applies System 1 / System 2 thinking to a domain. The Psychology of Money applies it to financial decisions. Pre-Suasion applies it to persuasion. Each one anchors Kahneman's framework in a specific application — which makes you remember Kahneman better AND makes the new book stick.

This is the entire logic behind the curated reading stacks on this site. The build-better-habits stack connects Duhigg → Clear → Newport → McKeown around the unifying claim "habits compound, and the right habits compound the right things." Reading those four books in that order produces dramatically better retention than reading any one of them alone, because each one reactivates the others.

When you can't pair, at least adjacent. The criterion is: does this book share a framework, a vocabulary, or a thinker-tradition with something you've recently read? If yes, the pairing will compound. If no, you're making a fresh start and will probably forget the new book within six months.

4. Does the recommendation pass the credibility filter?

Where the recommendation comes from matters more than the recommendation itself.

High-credibility sources (worth following): - A specific recommendation from someone who has independently arrived at expertise in the relevant domain - An academic citation in a book you already trust ("Kahneman cites this paper repeatedly...") - A pattern across multiple unrelated recommenders ("Naval, Tyler Cowen, AND Patrick Collison all rate this in their top 10")

Low-credibility sources (worth ignoring): - Generic "best books of the year" lists (curated for breadth, not for you) - Bestseller lists (correlated with marketing budget, not quality) - "What Bill Gates is reading" (publicly performed for brand reasons) - Single podcast mentions where the host hasn't read the book carefully - Influencer reviews on social media (often sponsored, often performative)

The honest test: would I have heard about this book if it weren't this month? If yes, the recommendation has compounding signal. If no, you're being swept by recency bias.

5. The 10-page test (and the chapter summary test)

The cheapest way to filter is to read the first 10 pages. Most non-fiction books reveal in those 10 pages whether the writing is going to be substantive, whether the argument is genuinely there, and whether the author respects your time. Books that pad in the first 10 pages are going to pad in the next 240; books that are precise from page 1 tend to stay precise.

Or, if you don't want to commit to 10 pages of standing in a bookstore: read a chapter summary. A 100-word summary of the book's first chapter tells you whether the book is going where you want to go. If the summary makes you want more, the book is worth your time. If the summary leaves you flat, the book wasn't going to teach you anything you didn't already know.

(This is, of course, exactly what Read Stacks is for. Honest disclosure of self-interest noted. But the principle holds even if you use a competitor's summaries instead.)

The meta-point

The choice problem is solvable. Most non-fiction readers don't have a willpower problem or a memory problem — they have a SELECTION problem. They buy books they were never going to finish, then feel bad about a pile they didn't deserve.

Apply the five criteria for a year and the pile shrinks. You read more of what you buy. You retain more of what you read. You stop chasing the next hot recommendation and start building actual knowledge across related books.

If you want the selection problem solved for you, the curated stacks on this site are the answer: four books in the right order, with editorial framing, designed to compound. Each stack is a single decision instead of four.

Either way: the next book you read should be one you can defend by the five criteria above. If you can't defend it, it's not the next book. Try again.