Style guide · Essayist compression

How to write like Paul Graham

Paul Graham's essays inform his X posts. The same patterns appear in both: plain-English compression, observation-then-implication structure, founder-as-protagonist framing. Here are the 5 most-imitable patterns and the one trap most imitators fall into.

Public X handle: @paulg · Co-founder of Y Combinator. Public essays and tweets on startups, founders, and clear thinking.

The 5 observable writing patterns

1

Observation, then implication

Opens with a specific observation (something he saw at YC, in a meeting, in his data), follows with the implication. The observation is concrete; the implication is general.

2

Plain-English compression

No business jargon, no startup vocabulary. "Make something people want" beats "achieve product-market fit" because the words are smaller.

3

Founder empathy without coddling

Writes as if to a specific founder he respects. Empathetic but expects the reader to do hard things. "You should" framings without softening hedges.

4

Counterintuitive truths as 1-sentence claims

"The most successful founders are not the ones who think the most about success." Reads as obvious in retrospect, surprising before reading.

5

Essay-as-tweet structure

Many tweets are clipped essays — 280 chars containing setup + observation + implication + payoff. Tight structure inherited from 20 years of essay writing.

Pattern shapes (NOT verbatim quotes)

These are illustrative structural templates derived from public writing patterns. Use them as scaffolds for your own specifics — the structure is universal, the words should be yours.

[Specific observation about founders]. [The general implication].

Why this works

Two-sentence essay. The specific opening earns trust; the general implication earns the share.

If you [common founder mistake], you [predictable bad outcome]. The way out is [specific small action].

Why this works

Conditional + counsel. Names the problem precisely + gives a concrete fix.

The reason [common observation] is that [unexpected underlying mechanism].

Why this works

Causality reveal. Reframes a familiar pattern with a new mechanism. Earns saves because it teaches.

When [situation], most people [common response]. The successful ones [contrarian response].

Why this works

Setup-then-flip. Common response → contrarian response. The structure is the whole post.

Do this

  • +Use small words. "Help" not "facilitate." "Use" not "utilize."
  • +Start with a specific observation, not an abstract claim
  • +End with the implication or counsel, not a CTA
  • +Write as if to one specific founder you know personally
  • +Read your draft aloud — if it sounds like a meeting, cut the meeting words

Avoid this

  • -Trying to sound smart — Graham's whole style is the opposite
  • -Business jargon ("synergize", "leverage", "orchestrate")
  • -Generic startup advice — every claim should come from a specific observation
  • -Long setups before the observation

Common questions

How does Paul Graham's tweet style differ from his essay style?+

It's the same patterns, compressed. His essays have the room to develop an idea over 1,500 words; his tweets compress the same observation→implication arc into 280 characters. If you've internalized the essays, his tweets will read as natural distillations.

What's the one trap imitators fall into?+

Sounding smart. Graham's style is deliberately UN-smart-sounding — plain words, concrete observations, no jargon. Imitators often substitute jargon for compression, which produces the opposite of his voice. If your draft uses "leverage," "optimize," or "synergy," you've fallen into the trap.

Will this style work outside startups?+

Yes — the observation→implication structure transfers to any field. "I noticed [specific thing] in my [field]. That suggests [general truth]." Works in fitness, finance, parenting, art. The startup angle is Graham's domain; the structural pattern is universal.

Can AutoTweet generate Paul Graham-style content?+

Yes — the Educational tone profile produces the closest structural match. Add Graham's essays to your voice profile in /dashboard/settings/voice as reference samples — the AI will learn the plain-English compression habit and the observation-then-implication structure.

Generate tweets in this style

AutoTweet's AI uses the Educational tone profile (closest match) and your voice samples to produce output in this structural style. Add reference posts in Settings → Voice training, then generate.

Cancel anytime