Days after Y Combinator founding partner Paul Graham said he would step down from his position at the startup incubator, he stepped on stage at the Launch Festival in San Francisco to talk about a list of lessons that have come from the past nine years.

First among those, Graham said, was the fact that he hates doing the large amounts of internal work required to keep the organization going, he told the audience at Launch. The annual event brings out more than 9,000 attendees. 

"It's turned into this giant thing, and I'm no good at running giant things," Graham said, explaining why he is leaving. Y Combinator Partner Sam Altman will assume the role of president of the incubator.  

Here are a few more:

Finding a Matter of Months

Since Graham cofounded Y Combinator in 2005, more than 630 startups, including Dropbox, Airbnb, and Stripe, have run through the program. Graham said that not much has changed since then in terms of the format of the program. Three months is still just the right amount of time for the startups to accomplish what Y Combinator is meant to help them accomplish. Essentially, that is finding their focus.  

"The most important things for startups to do is to focus." Graham said. "Because there's so many things you could be doing. One of them is the most important. You should be doing that. And not any of the others." 

For example, you shouldn't be grabbing coffee with investors just because they asked you to. If you're at the point where you're ready to raise money, then launch a full-fledged effort. Otherwise, get back to work, Graham said.  

"It's like optimizing software. There is something that is currently bottlenecked to making the software faster. You work on that. And that's a lot of what we do at YC,"  Graham said. "Sit down with people and just look at all of things they could be doing and asking them, which one is the most important?" 

It's All About Your First Users

And when you're done figuring out what to focus on and have moved onto the question of who to focus on, remember that your original user base can be as small as a single person.

Take a startup in the current Y Combinator batch. The company is developing a mobile email client, basically personalized for soon-to-be Y Combinator president Sam Altman.

"Their goal is to just make Sam Altman happy. Sam Altman uses his email a lot on the go, and he is sufficiently demanding. If they make him happy, they've sort of used him as like the positive for a mold. They can now makes lots of people happy," Graham explained. He said that the true litmus test for the usefulness of that product is whether or not Altman would be incredibly "bummed" if they stopped making the app.

So if the program's format hasn't changed much, and neither has incubator's main objective, what has changed at Y Combinator?

Graham said there's a difference in the type of people they accept into the program now. In short, being smart just doesn't cut it anymore. 

"We used to have more faith in brains," Graham said, adding that the incubator no longer accepts people whose intelligence is their main asset. "It turns out you can be surprisingly stupid if you're sufficiently determined," he said.