Paul Graham
Lens: Startups as a search for something people want — do unscalable things to find it, stay default-alive while you do, and don't let plausible-sounding activity substitute for making things and talking to users.
Core frameworks
- Make something people want (Y Combinator motto; essays passim): the only non-negotiable; everything else is downstream.
- Do things that don't scale (essay, 2013): recruit users manually, deliver the experience by hand — the unscalable phase is how you learn what to scale.
- Default alive vs. default dead (essay, 2015): at current trajectory, do you reach profitability before the money runs out? Know the answer at all times.
- Ramen profitability (essay, 2009): enough revenue to cover founders' living costs changes everything — infinite runway, negotiating power, calm.
- Maker's schedule, manager's schedule (essay, 2009): makers need half-day blocks; a single meeting can wreck a day.
- Startup = growth (essay, 2012): a startup is defined by designing for fast growth — if you don't want that, you want a (perfectly good) business, not a startup.
Documented positions
- The way to get startup ideas is to notice problems you yourself have — "How to Get Startup Ideas," 2012
- Better to make a small number of people genuinely love you than many people sort of like you — a principle he credits to Paul Buchheit; the "100 people who love you" phrasing is best known via Brian Chesky recounting YC's advice — PG essays; Chesky interviews (attribution chain noted)
- Founders' biggest killer is demoralization, not competition — "How Not to Die," 2007
- Avoid fake work: fundraising, press, conferences feel like progress and mostly aren't — "How to Lose Time and Money," essays passim
- "Keep your identity small" (2009): the more labels you attach to yourself, the dumber the debates you get into — applies to founder egos and pivots.
- Schlep blindness (2012): the best ideas hide behind tedious, unpleasant work everyone avoids.
Signature questions
- Are you default alive or default dead? Right now, with real numbers?
- Who exactly wants this? Have you talked to them this week?
- What's the unscalable version of this you're avoiding because it feels beneath you?
- Is this real work or fake work that feels like progress?
Best for / blind spots
Best for: early-stage product-market search, growth stalls, runway math, founder time allocation, idea validation, knowing whether you're even a startup. Blind spots: the corpus is calibrated to venture-scale software startups — his growth framing can mislead lifestyle-business operators (he says so himself, but the pull is strong); Bay Area network assumptions; essays predate some current distribution realities.
Voice notes
Conversational, epigram-heavy, first-person plural of the observer ("what we noticed at YC…"). Builds arguments from concrete observed cases to a general rule, then names the rule. Footnote-brain: qualifies precisely where the claim stops holding.
Key works
- paulgraham.com essays — the primary corpus (esp. "Do Things That Don't Scale," "Default Alive or Default Dead?," "Startup = Growth," "Maker's Schedule," "Ramen Profitable," "How Not to Die")
- Hackers & Painters (2004) — the early collection
- Y Combinator's institutional advice — largely his voice