Excellent Paul Graham Interview
Startups, determination, co-founders, “market” matters most.
F|R: What is the mathematical function from which Y Combinator takes its name, and why did you choose this?
Graham: It’s a function that builds recursive functions without them needing to have names. The Y Combinator is one of those things that seems miraculous when you first encounter it. You wouldn’t necessarily have expected such a thing to be possible. We named the company after it partly because we thought it was such a cool concept, and partly as a secret signal to the kind of people we hoped would apply.
F|R: How quickly can you now tell whether a start-up will make it? And what are the key characteristics that indicate potential success to Y Combinator?
Graham: We can never tell for sure. No investor can. But we are trying hard to get better at predicting.
I think the key quality is determination. The founders who do the best are the type of people who just refuse to fail. Most startups have at least one low point where any reasonable person would give up. That bottleneck is the reason there are so few successful startups. The only people who get through it are the ones who have an unreasonable aversion to failing.
F|R: Several copycat incubators have sprung up since Y Combinator launched in 2005 (TechStars, Y Europe, Seedcamp, BoostPhase). Can your model be replicated?
Graham: There are a few things they haven’t copied correctly, but really it’s not our model that distinguishes us. It’s the people that make the difference—not just us, but the 250 or so founders we’ve now funded. The amount of knowledge accumulated in all these heads is remarkable.
F|R: I read that when you call Y Combinator winners, the founders have only five minutes to accept. (“If people turn us down,” he says, “as far as we’re concerned they’ve failed an IQ test.”) Have startups turned you down? Are there any that have turned Y Combinator down and still gone on to succeed with a liquidity event?
Graham: You’re confusing two separate things. The reason people are supposed to decide quickly whether or not to accept is that they already know everything except the percent we’ll ask for. They’ve already seen the deal terms, and they already know as much as they’re going to know about YC before actually working with us. So they should already know when we call what percentage they’d be ok with. Since all they have to do is subtract one integer from another, five minutes should be enough.
The “IQ test” quote refers not to how fast they have to decide, but the amount of equity we usually ask for. In the median case it’s 6%. If we take 6%, we have to improve a startup’s outcome by 6.4% for them to end up net ahead. That’s a ridiculously low bar. So the IQ test is whether they grasp that.
There was one startup that turned us down because they received an acquisition offer during the weekend when we did interviews. It was a pretty good offer. I’d have taken it in their position, and they did. But other than that I don’t know of anyone who turned us down and went on to succeed. There have only been about three others who turned us down.
F|R: How did you determine the 12-week term of each Y Combinator class? Why is two months too short, or six months too long to “incubate” a startup?
Graham: We discovered it by accident. When we first started YC, we began with a summer program. We were trying to learn how to be investors, so we invited college students to come to Cambridge and start startups instead of getting summer jobs.
Now we’re looking for founders who consider the startup as a real job, not just a summer one. But we kept the 3 month cycle because it is a good length of time to build a version 1. Some startups may not be able to launch in such a short time, but they should all be able to build something impressive.
F|R: Name one thing founders can do to increase their odds of being selected by Y Combinator. Is Andreessen right that “the market” matters more than the idea, the tech, and even the talent?
Graham: Get good cofounders. You can’t change who you are, at least not in a short time. And the idea doesn’t matter to us as much as the people. So the best thing any individual can do is find good people to work with.
I think Marc may be right that market is the biggest determinant in the outcome of successful startups. But that’s not unrelated to the qualities of the founders. Smart people will find big markets.
UPDATE / COMMENT: RE Tom’s comment: I hear you, but when making decisions with incomplete and imperfect information and under time constraints, it’s (usually) not possible to “know” what the best alternative for anything is. For most (many) people, going with YC is probably going to do more good than harm. It’s (certainly) probably not going to be a fatal mistake. [I put my comment in as an update because of a technical difficulty with WordPress.]