Why Startup Advice Feels Wrong: Paul Graham on Instincts

TL;DR

Startups are counterintuitive, so following your instincts usually leads you astray, the same way leaning back on skis sends you down the hill out of control. What you actually need is not expertise in startups but expertise in your own users. Mark Zuckerberg succeeded despite being a complete noob at startups because he understood his users very well.

Transcript

That short. Like, long introductions are no good. Sam knows. Alright, ready, everybody? I'm not gonna ask if the mic is working like in every talk so far. I'll just assume it's working. >> No! >> No, fuck! >> All right, well make it work somebody. >> It works, it works ... Read More

Key Insights

• Startups are counterintuitive in the way skiing is counterintuitive: your first impulse when you want to slow down on skis is to lean back, which sends you down the hill out of control, so part of learning is suppressing that instinct.
• The single most useful thing to remember is that startups are so weird that following your instincts will lead you astray. Even that one fact makes you pause before making a mistake, which is often enough.
• Founders persistently ignore advice because counterintuitive ideas contradict their intuitions and therefore seem wrong. Y Combinator partners joked that their function was to tell founders things they would ignore, batch after batch.
• Intuitions about people are the exception and can be trusted. A big founder mistake is not trusting those intuitions enough, meeting someone impressive but troubling, ignoring the misgivings, then saying later that they knew something was wrong.
• Engineers often assume business is supposed to be slightly distasteful, so they excuse smart but distasteful people as normal for business. Pick collaborators the way you would pick friends, people you genuinely like, respect, and have known long enough.
• What you need to succeed is not expertise in startups but expertise in your own users. This makes a startup class unlike a French class, where doing the work teaches you the thing you actually need.
• Playing house is the Y Combinator name for going through the motions: a plausible sounding idea, funding at a nice valuation, an office in Soma, friends hired, and neglect of the one essential thing, making something people want.
• Growth hacks should be mentally translated to bullshit. The way to make a startup grow is to make something users really love and then tell them about it, and the way to convince investors is to be growing fast and say so.

Questions & Answers

Q: Why does startup advice often feel wrong when you first hear it?

Because startups are counterintuitive, and counterintuitive ideas by definition contradict your intuitions, so they seem wrong and your first impulse is to ignore them. Paul Graham says that when he was running Y Combinator, the partners joked that their function was to tell founders things they would ignore. Batch after batch, partners warned founders about mistakes they were about to make, founders ignored them, and came back a year later saying they wished they had listened. By then the problematic investor is on the cap table and there is nothing they can do.

Q: How is starting a startup like learning to ski?

When you first try skiing and want to slow down, your instinct is to lean back, just like in everything else, but leaning back on skis sends you flying down the hill out of control. Part of learning to ski is suppressing that impulse until you build new habits, and in the beginning you hold a list of reminders in your head: alternate feet, make s-turns, do not drag the inside foot. Startups are as unnatural as skiing, and there is a similar list of counterintuitive things to remember so your existing instincts do not lead you astray.

Q: What is the one thing to remember if you remember nothing else about startups?

That startups are so weird that if you follow your instincts they will lead you astray. Graham puts this first on his list deliberately. His reasoning is that even if you retain nothing else from the talk, holding that single fact in mind means that when you are about to make a mistake, you may at least pause before making it. The value is not that the rule tells you the right answer, but that it interrupts the automatic confidence that produces the wrong one.

Q: When should you trust your instincts as a founder?

About people. Graham says your life so far has not been much like starting a startup, but all the interactions you have had with people are just like the interactions you will have with people in the business world, so that intuition transfers. In fact one of the big mistakes founders make is not trusting their intuitions about people enough. They meet someone impressive but feel misgivings, ignore them, and later, when things blow up, say they knew something was wrong about that person but ignored it because he seemed so impressive.

Q: How should you choose the people you work with in a startup?

Pick people the way you would pick people if you were picking friends. Graham calls this one of the rare cases where it works to be self indulgent: work with people you genuinely like and respect and that you have known long enough to be sure. The length of acquaintance matters because a lot of people are really good at seeming likeable for a while. You do not find out otherwise until your interests are opposed, and by then the relationship is already in place.

Q: Why do engineers misjudge distasteful people in business?

There is a specific sub-case that hits people from an engineering background. You think business is supposed to be a slightly distasteful thing, so when you meet people who seem smart but somehow distasteful, you conclude that this must be normal for business and override your misgivings. Graham says it is not normal. The correct response is to apply the same standard you would apply to friends rather than granting business a special exemption for unpleasantness.

Q: What do you actually need to know to succeed in a startup?

Not expertise in startups, but expertise in your own users. Graham notes this makes a startup class different from most classes you take: in a French class, if you do the work, you come out speaking French, close to how a French person sounds. A startup class can teach you about startups, but that is not what you need to know. Mark Zuckerberg succeeded despite being a complete noob at startups, to the point that Facebook was first incorporated as a Florida LLC, because he understood his users very well.

Q: What is playing house and why is it dangerous?

Playing house is the name Y Combinator made up for going through the motions of starting a startup. Founders come up with a plausible sounding idea, raise funding at a nice valuation, rent a nice office in Soma, hire a bunch of their friends, and then gradually realize how completely wrecked they are. While imitating all the outward forms, they neglect the one thing that is actually essential, which is to make something people want. Graham says young founders default to this because gaming artificial measures is what school trained them to do.

Summary & Key Takeaways

• Paul Graham frames the talk as the advice he would give his own kids about startups if they were in college. The core theme is that startups are counterintuitive, an area where you cannot trust your intuitions all the time, and the talk is presented as the beginning of a list of counterintuitive things to remember.

• The skiing analogy anchors the argument: new skiers instinctively lean back to slow down and end up flying downhill out of control, so learning means suppressing instinct and holding a list of rules in mind. Y Combinator functions as business ski instructors, warning founders about mistakes they then ignore and regret a year later.

• People are the exception to distrusting instinct. Founders should work with people they genuinely like and respect and have known long enough to be sure, because many people are good at seeming likeable for a while, and you only see the truth once interests are opposed.

• The second point is that success requires expertise in your own users, not expertise in startups. Zuckerberg incorporated Facebook as a Florida LLC and probably does not know the mechanics of raising an Angel round, yet succeeded because he understood his users very well.

• Young founders default to playing house because school trained them to game artificial measures, the way Graham studied for exams by predicting the questions rather than mastering the material. They ask how do I, and partners answer with just: make something users really love, then tell them about it.