Lecture 3: Before the Startup

What this lecture covers

Paul Graham steps back from the tactics of earlier lectures to address a more basic question: what mindset do you need before you even start a startup. Framed as advice he would give his own kids, the lecture lists counterintuitive truths that trip up first-time founders, ending with concrete advice on what to actually do in college if you eventually want to start a company.

After watching, you should understand why startups reward trusting your instincts about people while distrusting your instincts about the business itself, why manipulating a system that worked in school and big companies stops working with startups, and why Graham argues against starting a startup while still a student.

Key ideas

• Startup instincts mislead you: like a beginner skier who leans back instinctively and loses control, first-time founders' natural impulses on strategy are usually wrong, which is why YC exists to warn against them.
• Trust your instincts about people: unlike startup mechanics, your lifelong experience judging people is reliable; don't override a bad gut feeling about someone because they seem impressive.
• Startup expertise isn't what you need: success comes from understanding your users deeply, not from mastering the mechanics of fundraising or "playing house" with the outward forms of a startup.
• Gaming the system stops working: tricks that succeed in school or at a big company fail with users, who only care whether the product does what they want.
• Startups are all-consuming: founding a successful company takes over your life for years, a trade-off Graham compares to having children.
• Don't start a startup in college: you cannot be both a full-time student and a founder, and your early twenties offer a kind of serendipity and freedom you lose once a startup succeeds.
• You can't predict who will succeed: confidence or nervousness before starting has little correlation with how a founder actually performs.
• Good startup ideas come from unconscious side projects: rather than deliberately brainstorming, learn about things that matter, work on problems that interest you, with people you like and respect.

Walkthrough

Startups are counterintuitive (0:01)

Graham opens with the skiing analogy: just as a beginner's instinct to lean back sends them out of control, founders' natural instincts about startups often lead them astray. He describes YC's role as repeatedly telling founders things they ignore until they learn the hard way, since advice only has value when it contradicts your intuition.

Trust your instincts about people (4:05)

The one place founders should trust their gut, Graham argues, is in judging people. Engineers often dismiss a nagging bad feeling about someone because they assume unease is just how business works, but Graham urges founders to work only with people they genuinely like and respect, since impressive-seeming people can turn out badly once interests diverge.

Expertise in your users, not in startups (5:05)

Graham's second counterintuitive point: succeeding requires expertise in your own users, not in the mechanics of starting a company. He warns against "playing house," where founders imitate the outward forms of a startup, raising money, renting an office, hiring friends, while neglecting the one essential task of making something people want.

Gaming the system stops working (11:13)

Graham's third point is that the trick-seeking behavior rewarded by school and large companies fails with startups, since there is no boss to fool, only users who judge purely by whether the product works. He warns that founders can sometimes fool investors for a round or two, but doing so only wastes their own time, since a startup with no real usage is still doomed.

Startups take over your life (14:18)

The fourth point: a successful startup consumes years, even decades, of a founder's life, a commitment Graham compares to having children. He uses Larry Page as an example of a founder whose success brought enormous but largely invisible personal cost, and argues that this all-consuming nature is a real opportunity cost worth weighing seriously.

Why not to start a startup in college (16:19)

Graham argues universities can teach you about startups the way a linguistics class teaches you about language in general, but what you actually need, expertise in your own users, can only be learned by doing, and doing it takes over your entire life. Since you cannot be a full-time student and a full-time founder simultaneously, he advises directly against starting a startup in college, valuing the serendipity and freedom of the early twenties that a successful founder permanently loses.

You can't predict who succeeds (21:28)

Drawing on years of evaluating YC applicants, Graham says confidence or nervousness before starting a company has little bearing on how someone actually performs, since the tests founders face bear no resemblance to the tests they aced earlier in life. If you're only unsure whether you can do it, the only way to find out is to try, just not necessarily right away.

Finding startup ideas without trying to (24:31)

Graham's final counterintuitive point: deliberately brainstorming startup ideas tends to produce plausible-sounding but bad ones. Instead, the best ideas emerge unconsciously from side projects, as happened with Yahoo, Google, Facebook, and Apple. His prescription: learn a lot about things that matter, work on problems that interest you, with people you like and respect, positioning yourself at the leading edge of some technology so that ideas others would miss look obvious to you.

Audience questions (31:36)

In the Q&A, Graham argues business school offers little value for aspiring founders since it trains management for large companies, a problem startups rarely have early on; he advises studying design or simply starting instead. He also addresses fundraising challenges for female founders (a strong growth graph overcomes bias faster than anything else), high valuations versus a genuine bubble, and why he'd study physics if starting college today, purely out of curiosity rather than career calculation.

Before you watch

• Watching Lectures 1 and 2 first helps, since Graham's talk assumes familiarity with the idea-product-team-execution framework Altman introduced.
• Graham references his own essay on how to get startup ideas; reading it is optional but adds detail beyond what this lecture covers.

Check your understanding

1. Why does Graham say founders should trust their instincts about people but not their instincts about startup strategy?
2. What does "playing house" mean in this lecture, and why does Graham consider it dangerous for young founders?
3. Why does Graham argue against starting a startup while still in college?
4. According to Graham, why do deliberate brainstorming sessions tend to produce weak startup ideas, and what does he recommend instead?