Writing & Expression: Masters · Paul GrahamWrite Usefully · The Essay as Thinking · Shorter Is Better · Write Like You Talk

PG doesn't ask "is it well written?" He asks "is it useful?" Useful writing does one thing: it tells the reader something true, important, and that they didn't already know — and says it as unequivocally as possible.

PG breaks "useful" into four parts you can sharpen separately: important, novel, correct, strong. The first two decide whether it's worth reading; the last two decide whether it can be believed. The most counterintuitive is strength — he argues you should make claims as strong as you can without making them false. Most writers instinctively stuff sentences with "perhaps," "to some extent," "I feel," thinking it's safe. It isn't — it just dilutes the information and dodges accountability. Useful writing wants precision, not the false safety of vagueness: push the claim to the edge, but not past the line between true and false.

• ✓ Decision memos, tech-stack choices, postmortem conclusions, public opinions — anywhere the reader expects you to make the call
• ✓ Self-check after writing: which is my strongest claim? Can I make it stronger without lying?
• ✗ Trap: using "maybe / perhaps / to some extent" as a disclaimer, mistaking vagueness for rigor
• ✗ Trap: chasing only "correct" — empty truisms are correct too — while forgetting important and novel

The word essay comes from the French essayer — "to try." PG says an essay isn't copying down a conclusion you already have; it's figuring it out in the act of writing.

This overturns the common belief — "think it through first, then write." PG's experience is the opposite: you think you understand, and the moment you start writing you find the holes. Ideas can coexist vaguely in your head, even contradict each other, without you noticing; but force them into linear sentences and every leap, every unresolved link, gets exposed. So writing is a lie detector for thinking: it rewards not eloquence but honestly thinking something all the way through. For engineers and leaders this is especially valuable — the real payoff of writing a design doc or decision memo is often not the document, but the edge case the writing forced you to discover.

• ✓ Any complex problem you "think you've figured out": architecture decisions, strategy, postmortems — write a draft first to force the holes into the open
• ✓ Treat writing as a thinking tool, not a reporting chore: the first draft is for you to think with, not for others to read
• ✗ Trap: stalling until you've "fully thought it through" — that moment never comes; writing is what brings it
• ✗ Trap: letting AI smooth out your draft — the prose gets smooth, but you skip the most valuable thinking

PG deliberately writes with the plainest words and simplest sentences. Not because his vocabulary is poor, but because the energy readers spend decoding your sentences should have gone to your ideas.

Complex sentences and big words are often taken as signs of intelligence; PG says they're more often a sign that the writer is lazy or trying to bluff. His logic is economic: the reader's attention is a fixed budget, and every obscure word, every winding clause, levies a tax. The more it taxes, the less is left for the idea you actually want to convey. This shares a root with Orwell's "if it's possible to cut a word, cut it," but PG adds a layer of motive — in "Writing, Briefly" he compresses the whole craft into one near-total piece of advice: cut out everything unnecessary. Short isn't the goal; short is the means of returning the reader's cognitive budget to the ideas. But note: simple ≠ simplistic. Complex thoughts can ride on simple sentences — the hard part is whether the writer is willing to digest the complexity for the reader first.

• ✓ All writing whose purpose is to be understood; especially for cross-team, cross-background readers
• ✓ Run a "tax audit" before the final draft: for every big word and long clause, ask whether it's worth the tax
• ✗ Trap: using big words and long sentences to manufacture an air of expertise — most readers feel only the effort, not awe
• ✗ Trap: mistaking "simple" for "simplistic," so you never dare simplify complex ideas — but that's exactly the real skill

PG writes like he's chatting with a smart friend — no posturing, no showing off learning. He has a sharp image for it: informal language is the athletic clothing of ideas.

Formal prose is like a suit — crisp, presentable, and restrictive. PG says that for ideas to run, probe, and pivot, they need athletic wear: spoken language. He offers a wonderfully practical test — after writing a passage, imagine reading it aloud to a friend, and change anything you'd never actually say out loud. This isn't lowering the bar; it's using "would I say this to a friend?" as a sincerity detector: written stiffness is the easiest place to hide vagueness and bluster, because it dresses emptiness in a suit. But keep your sense of proportion — "like talking" means natural, close, unpretentious, not dumping your verbal tics, filler, and rambling onto the page. It's pruned speech: keep the warmth, cut the noise.

• ✓ Essays, newsletters, personal brand, internal comms — anywhere you want to be read as "a person," not "a document"
• ✓ The test: read the passage to a friend; which line could you not say out loud? Change it
• ✗ Trap: taking "like talking" as license to ramble and pile on filler — it's pruned speech, not a transcript
• ✗ Trap: forcing casual tone where formality is due (contracts, public notices, compliance docs), losing all sense of proportion