Think and write clearly: six habits from Paul Graham

Unofficial. Not affiliated with or endorsed by Paul Graham or Y Combinator. Every teaching links to the original essay or talk.

1. 1

   Why is writing a test of your thinking?

   In "Putting Ideas into Words" (February 2022), PG says writing about something, even something you know well, usually shows you did not know it as well as you thought. "Putting ideas into words is a severe test." He adds that half the ideas in a finished essay come while writing it, and that someone who never writes has no fully formed ideas about anything nontrivial.

   In "Writes and Write-Nots" (October 2024) he warns that AI will remove the pressure to write, so many people will stop. Because writing is thinking, that splits people into those who think and those who do not. He quotes Leslie Lamport: "If you’re thinking without writing, you only think you’re thinking."

   • Before your next big decision (a job, a pivot, a hire), write one page on why. Do not let AI draft it.
   • Reread it as a stranger who only has the page, PG’s test. Fix every gap the stranger would see.
   • Notice which ideas appeared only while writing. That is the point.

   Key idea

   If you cannot write it clearly, you have not finished thinking it.

2. 2

   How do you write like you talk?

   In "Write Like You Talk" (October 2015), PG’s test for every sentence is: "Is this the way I’d say this if I were talking to a friend?" If not, write what you would say. He reads every essay aloud before publishing and fixes anything that does not sound like conversation. Do this, he says, and you will be ahead of 95% of writers.

   His older list "Writing, Briefly" (March 2005) adds the rest of the method: "Write a bad version 1 as fast as you can," then rewrite and cut. If you are stuck, tell someone what you mean and write down what you said. In "Good Writing" (May 2025) he explains why this matters for thinking, not only style: "It’s hard to be right without sounding right."

   Before: "Our platform utilises AI to facilitate seamless interview preparation outcomes for candidates." Said to a friend: "It lets you practise interviews with an AI that asks real questions and tells you what you got wrong."

3. 3

   Why should you keep your identity small?

   In "Keep Your Identity Small" (February 2009), PG asks why arguments about religion and politics go nowhere. His answer: once a topic is part of someone’s identity, they cannot think clearly about it. So the best plan is to let as few things into your identity as possible. "The more labels you have for yourself, the dumber they make you."

   For a founder this is practical. If "I am a mobile-first founder" or "I am an AI founder" becomes who you are, you cannot see when users want something else. PG allows a few identities, like "scientist", that commit you to following evidence rather than to any conclusion.

   • Write down five labels you use for yourself (a tech stack, a city, a school, a politics, a type of founder).
   • For each, ask: if the evidence went against this, could I change my mind without feeling attacked?
   • Drop any label that would stop you.
4. 4

   How do you disagree in a way that changes minds?

   In "How to Disagree" (March 2008), PG ranks disagreement from DH0, name-calling, up to DH6, refuting the central point. In between are attacking the person, responding to tone, plain contradiction, counterargument and refutation. "The most convincing form of disagreement is refutation", and the strongest is refuting the main point, not a side detail.

   This is the skill you need for cofounder arguments, investor objections and user complaints. Quote the exact thing you disagree with, then show why it is wrong. PG adds that moving up the hierarchy also makes conversations less mean.

   1. DH0 Name-calling↓→
   2. DH1 Ad hominem↓→
   3. DH2 Tone↓→
   4. DH3 Contradiction↓→
   5. DH4 Counterargument↓→
   6. DH5 Refutation↓→
   7. DH6 Refute the central point

   Key idea

   Aim every disagreement at DH5 or DH6: quote it, then show where it is wrong.

5. 5

   Why do makers need a different schedule?

   In "Maker’s Schedule, Manager’s Schedule" (July 2009), PG says managers work in one-hour slots, while makers (programmers, writers, designers) need half a day or more. One meeting in the middle of an afternoon can break it into pieces too small for hard work. "For someone on the maker’s schedule, having a meeting is like throwing an exception."

   His own fix at YC: he held office hours with founders at the end of his working day, so they never cut into his focused time. As a founder you are both, so split your week on purpose.

   • Pick two or three mornings a week with no meetings at all. Guard them like appointments.
   • Stack all calls in the afternoon, back to back.
   • If you manage makers, ask before booking a meeting in the middle of their block.
6. 6

   What is the top idea in your mind?

   In "The Top Idea in Your Mind" (July 2010), PG says the idea your mind drifts to when it is free, like in the shower, gets the benefit of all that background thinking. Other ideas starve. Fundraising takes over that spot, which is why startups slow down while raising, and so do disputes: Newton lost years arguing over his theory of colour.

   You cannot pick your shower thoughts directly, but you can pick your situations. PG’s advice: "be careful what you let become critical to you." He also tells himself that grudges and injuries do not deserve room in his head.

   • For three days, write down what you think about in the shower or on your commute.
   • If it is an argument, a grudge or money worry instead of your work, change one situation that feeds it.
   • Keep fundraising, side disputes and online arguments short, so they do not take over the top spot.

In his own words

Straight from the source

What Paul Graham or YC actually wrote or said, taken apart, then what to do with it this week.

Paul Graham

Essayist and cofounder of Y Combinator, who has published essays on paulgraham.com since 2001.

Putting Ideas into Words (essay, February 2022)

1 · Understand

Writing about something usually shows you did not understand it as well as you thought. Putting ideas into words changes and completes them.

2 · Break it down

• Your first words are usually wrong; you rewrite until they are exactly right.
• Ideas turn out not just imprecise but incomplete. Half the ideas in an essay come while writing it.
• Test a draft by reading it as a neutral stranger who knows only the text.
• Someone who never writes has no fully formed ideas about anything nontrivial.

3 · Use it

Write a one-page memo before every major decision, then reread it as a stranger. Rewrite until the stranger would agree it is correct and complete.

Paul Graham

Cofounder of Y Combinator, who ran YC office hours around his own writing and programming time.

Maker’s Schedule, Manager’s Schedule (essay, July 2009)

1 · Understand

Makers and managers use time differently. A meeting costs a manager one hour, but it can cost a maker half a day of real work.

2 · Break it down

• Managers work in one-hour slots; makers need half a day or more.
• A single meeting can break a block into pieces too small for hard work.
• Makers may not even start ambitious work on a day they know will be broken up.
• PG held founder office hours at the end of his day so they never interrupted his focus.

3 · Use it

Block two or three meeting-free mornings this week and move all calls to late afternoon. Compare what you finish.

6 terms

Words to know

Maker’s schedule

Working in blocks of half a day or more, which deep work like coding or writing needs.

Manager’s schedule

A day cut into one-hour slots, where switching tasks and meetings is normal.

Disagreement hierarchy

PG’s seven levels of disagreement, from name-calling (DH0) to refuting the central point (DH6).

Keep your identity small

Let as few beliefs as possible become part of who you are, so you can still think clearly about them.

Top idea in your mind

Whatever your thoughts drift to when they are free. It gets your best background thinking.

Write like you talk

Writing each sentence the way you would say it to a friend, then reading it aloud to check.

1 video · plays here

Watch

Hand-picked from the people who explain this best. The video loads only when you press play.

Do this week

Put it into practice

Write your idea the way you would say it to a friend

1. 1Pick your startup idea, your job pitch or a decision you are stuck on.
2. 2Write it in five sentences, fast. PG: a bad version 1 as fast as you can.
3. 3Read each sentence aloud. Ask PG’s question: is this how I would say it to a friend? Rewrite any that fail.
4. 4Cut every word that does not change the meaning. Count the words before and after.
5. 5Give it to someone who knows nothing about it. Ask them to explain it back. Every place they get it wrong is a gap in your thinking, not in their reading.

Read PG’s "Write Like You Talk"

3 questions · instant feedback

Check yourself

Pick an answer to see if you are right, and why.

Question 1

Question 1 of 3

An investor says your market is too small. Which reply is DH6 in PG’s hierarchy?

A"You clearly do not understand our space."B"That was a rude way to put it."C"No, it is big."D"Your estimate assumes only metro users. 60% of our sign-ups are from smaller towns, so the real market is several times larger."

Question 2

Question 2 of 3

You are a founder who codes. Investors want meetings at 11 am and 3 pm on the same day. What does "Maker’s Schedule, Manager’s Schedule" suggest?

AAccept both, since meetings are shortBMove both to the same block, ideally the end of the day, to protect a long maker blockCCancel all investor meetings foreverDCode during the meetings

Question 3

Question 3 of 3

Your team has argued for a week about whether you are a "B2B company". Users keep asking for a consumer version. What does PG’s "Keep Your Identity Small" suggest is going on?

AThe label has become part of your identity, which makes it hard to think clearly about the evidenceBYou need a better logoCB2B is always the right choiceDUsers do not know what they want

Lesson 11 of 17

Done? Keep going

Mark this lesson complete to keep your place. Progress is saved on this device, no login.

Mark lesson complete

Up next · Lesson 12 · 12 minDefault alive or default deadYou are default alive if, with expenses flat and revenue growing at its recent rate, you reach profitability before the money runs out. If not, you are default dead and your survival depends on investors. Paul Graham says half the founders he asks don’t know which they are. Work it out now, because the most common way funded startups die is hiring too fast, and by the time you notice, it can be too late to raise.← Previous: How to do great work

People also ask

Questions people ask

What is Paul Graham’s writing advice?+

Write like you talk: write each sentence the way you would say it to a friend, and read the draft aloud. Write a bad first version fast, then rewrite and cut. He also argues writing is how you find your ideas, not just how you share them.

What is the maker’s schedule vs the manager’s schedule?+

From Paul Graham’s 2009 essay: managers work in one-hour slots, while makers like programmers and writers need half-day blocks. A single meeting can ruin a maker’s afternoon, so group meetings together.

What is Paul Graham’s disagreement hierarchy?+

A seven-level ranking from his 2008 essay "How to Disagree": name-calling, ad hominem, responding to tone, contradiction, counterargument, refutation, and refuting the central point. Aim for the top two.

10 sources · checked Nov 2026

Sources and further reading

• Paul Graham: Putting Ideas into Words (February 2022)
• Paul Graham: Writes and Write-Nots (October 2024)
• Paul Graham: Write Like You Talk (October 2015)
• Paul Graham: Writing, Briefly (March 2005)
• Paul Graham: Good Writing (May 2025)
• Paul Graham: Keep Your Identity Small (February 2009)
• Paul Graham: How to Disagree (March 2008)
• Paul Graham: Maker’s Schedule, Manager’s Schedule (July 2009)
• Paul Graham: The Top Idea in Your Mind (July 2010)
• How I Write: The King of Internet Writing, Paul Graham with David Perell (March 2024)

Related on this site

How to do great workNext: Default alive or default deadWhat makes a great founderAll Paul Graham lessonsChief of staff guideFounder’s office roles