Astray

16 February, 2026

When you start studying startups, you end up at Y Combinator. I'm not sure how it happens exactly. You start reading about successful companies, you notice the same three letters appearing next to half of them, and within a few clicks you're on ycombinator.com trying to figure out what this thing actually is.

Then, when you start studying YC, you end up at Paul Graham. He's the cofounder, but more importantly, he's the one who wrote everything down. And he wrote a lot. Something like 188 essays over almost three decades. So you start reading.

And the more of his essays you read, the more you notice a phrase that keeps coming back. Lead you astray.

He uses it in Before the Startup, a talk he gave at Stanford in 2014. The line is: “Startups are so weird, that if you follow your instincts they will lead you astray.” His analogy is skiing. When you start skiing and you want to slow down, your instinct is to lean back. But if you lean back on skis, you fly down the hill out of control. You have to learn to suppress the instinct. He says startups are the same. There's a list of things you have to remember, because your gut will give you the wrong answer.

Once I noticed that phrase, I started seeing it everywhere in his work. Not always those exact words, but the same idea. Almost every essay he's written is, in some form, a warning about an instinct that will lead founders astray. So I started keeping a list.

Here's what I have so far.

Trying to consciously think of startup ideas. This is the one he repeats most. If you sit down and try to brainstorm a startup idea, what you'll come up with isn't just bad. It's bad and plausible-sounding, which is worse, because you and everyone around you will be fooled by it long enough to waste a year of your life. The YC team has a name for these. They call them sitcom ideas. The kind of startup a TV writer would invent for a character on a show. The fix, apparently, is to stop trying. Work on things you find interesting, and let the ideas come up on their own. Yahoo, Google, Facebook, and Apple all started as side projects. None of them were supposed to be companies.

Avoiding tedious, messy, or hard problems. Graham calls this schlep blindness. Your unconscious mind filters out the most valuable problems before you even consider them, because they involve unpleasant work. Nobody applies to YC saying “I want to fix payments.” They say “I want to build a recipe site.” The payments problem was right there the whole time. They just couldn't see it.

Trying to game the system. This is the one I find most interesting, because it's the instinct that gets rewarded almost everywhere else. School, big companies, applying for jobs, all of it is about figuring out what the system wants and giving it that. But, as Graham says, startups are as impersonal as physics. There's no boss to trick. Only users. And users only care whether your product does what they want. The dangerous part is that gaming does still work on investors, for a round or two, which is exactly long enough to convince you it works in general.

Relying on expertise in startups instead of expertise in users. Graham puts this bluntly: Mark Zuckerberg didn't succeed in Facebook because he was an expert in startups. He succeeded despite being a complete novice at startups. What he had was expertise in his users, because he was one. The instinct to read more startup books before starting a startup is, itself, leading you astray.

Building something a lot of people want a little bit. The instinct is to go big, address a huge market, talk about TAM in your pitch deck. The reality is that you have to choose. Either a lot of people who want it a little, or a few people who want it a lot. Pick the second one. Almost every good startup idea is of that type.

Avoiding things that don't scale. This is maybe the most counterintuitive of all, because scaling is supposedly the whole point of a startup. But the founders of Airbnb went door to door in New York taking professional photos of apartments. Stripe onboarded its first users by literally typing in their credit card details for them. The instinct to automate too early is, again, leading you astray.

The pattern, once you see it, is hard to unsee. Almost every piece of conventional wisdom about how to be successful, the stuff that works in school and in jobs and in life, stops working the moment you start a company. And that, I think, is what Graham has been quietly saying for thirty years.

You're skiing now. Lean forward.