Explanatory Writing

The test of explanatory writing isn’t whether the reader understands what you said. It’s whether they can now think with the idea — apply it, extend it, use it to make decisions.

This is the difference between reading about compound interest and being able to spot it in disguise. Between nodding at “sunk cost fallacy” and catching yourself mid-mistake.

Great explanation is high-fidelity compression.

Compression: you reduce a messy domain into a smaller mental model the reader can hold. The world is too complex to reason about directly.

High-fidelity: the compression still cashes out. The model makes predictions, guides decisions, reveals what to notice. A false simplification is worse than none — it gives confidence without capability.

The reader should feel smarter. They leave equipped with something they can use.

Five components recur across writers who pull this off:

A sharp question. Don’t start with content. Start with the puzzle that forces structure. Feynman didn’t explain quantum electrodynamics. He explained why light behaves in ways that seem impossible.

A governing model. A small set of principles that explains many things. The model is what survives after the reader forgets the details.

Cognitive handholds. Concrete examples, analogies, thought experiments, crisp definitions. These are what the mind grips when abstraction gets slippery. Einstein’s elevators. Dawkins’ selfish genes.

Epistemic posture. How confident you are, what you admit you don’t know. This signals whether the reader should update their beliefs or just file information.

Edit-level craft. Sentence-by-sentence choices that reduce reader load. Short clauses, active voice, zero filler.

Different writers emphasize different components.

Feynman‘s power is mechanism over labels. If an explanation is just a name, it doesn’t count. He shows the moving parts.

Orwell treats clarity as ethics. Sloppy language enables sloppy thinking.

Paul Graham writes to discover, then edits to communicate. Default to spoken-language structure. Revise until it sounds like a person.

Sagan balances wonder with skepticism. Awe without gullibility.

Kahneman teaches by having the reader experience the failure mode. You feel the bias before he names it.

Common failure modes:

Faking understanding. Using technical terms as explanations. “It works because of quantum tunneling” — but what is tunneling?

Compressing too hard. Millennia-hopping that loses nuance. If you compress this hard, be explicit about uncertainty.

Decorative metaphor. Metaphors that illustrate but don’t constrain. A good metaphor is a contract.

No epistemic signal. Presenting contested claims with the same confidence as settled facts.

Wall of text. Making the reader hold too much at once. Never make the reader hold more than one new concept per paragraph.

The meta-skill is knowing which move a given explanation needs.

Sometimes the reader needs a governing model. They have details but no framework. Sometimes they need cognitive handholds. They have the framework but can’t apply it. Sometimes they need epistemic posture. They don’t know how much to trust what they’ve read elsewhere.

Diagnose first.

Go Deeper

Books

The Sense of Style by Steven Pinker — Cognitive science applied to prose.
On Writing Well by William Zinsser — The classic on nonfiction craft.

Essays

“Politics and the English Language” by George Orwell — Clarity as ethics.
“Putting Ideas into Words” by Paul Graham — Writing as thinking.