Finding just the right mix of old friends and new colleagues may be more important than finding the right idea for a new company.


Robert Scoble/Flickr

The standard operating procedure for most start-up incubators is something like this: Bright entrepreneurs submit their million-dollar ideas; a select few get some money, mentors, and time for developing these ideas; at the end, they take the ideas out to the real world in search of capital.

Y Combinator, perhaps the top incubator in the world, is now planning to dispense with one of the core ingredients in that formula: the idea. It will now accept applications from entrepreneurs with no particular idea at all. The Y Combinator team explained their reasoning in an announcement yesterday:

Why are we doing this? Partly because we realized we already were. A lot of the startups we accept change their ideas completely, and some of those do really well. Reddit was originally going to be a way to order food on your cellphone. (This is a viable idea now, but it wasn't before smartphones.) Scribd was originally going to be a ridesharing service.

The other reason we're doing it is that our experience suggests that smart people who think they can't come up with a good startup idea are generally mistaken. Almost every smart person has a good idea in them. A good startup idea is simply a significant, fixable unmet need, and most smart people are at least unconsciously aware of several of those. They just don't know it. And we now have lots of practice helping founders see the startup ideas they already have.

The important thing for success isn't the quality of an idea at the outset but its development. Start-up incubators are premised on this principle, and Y Combinator's decision takes that premise to its logical end.

Research into creativity and how the human brain processes ideas supports this decision. In a recent New Yorker article, Jonah Lehrer argues that brainstorming as we have traditionally imagined it -- an uncritical free-for-all -- is not all it is cracked up to be. But creativity really gets going in groups in which criticism is encouraged, or, as Lehrer writes "dissent stimulates new ideas because it encourages us to engage more fully with the work of others and to reassess our viewpoints."

Even more critically for Y Combinator, Lehrer tells of one study that found that having the right combination of strangers and people who know each other is crucial. The research, by sociologist Brian Uzzi, looked at casting for Broadway productions trying to find the formula for a hit show. He created a metric he called "Q." A high Q score indicated a team that had all worked together before; low Q meant the team was strangers. Lehrer writes:

According to the data, the relationships among collaborators emerged as a reliable predictor of Broadway success. When the Q was low--less than 1.7 on Uzzi's five-point scale--the musicals were likely to fail. Because the artists didn't know one another, they struggled to work together and exchange ideas. "This wasn't so surprising," Uzzi says. "It takes time to develop a successful collaboration." But, when the Q was too high (above 3.2), the work also suffered. The artists all thought in similar ways, which crushed innovation...

The best Broadway shows were produced by networks with an intermediate level of social intimacy. The ideal level of Q--which Uzzi and his colleague Jarrett Spiro called the "bliss point"--emerged as being between 2.4 and 2.6. A show produced by a team whose Q was within this range was three times more likely to be a commercial success than a musical produced by a team with a score below 1.4 or above 3.2. It was also three times more likely to be lauded by the critics. "The best Broadway teams, by far, were those with a mix of relationships," Uzzi says. "These teams had some old friends, but they also had newbies. This mixture meant that the artists could interact efficiently--they had a familiar structure to fall back on--but they also managed to incorporate some new ideas. They were comfortable with each other, but they weren't too comfortable."

This effect is particularly relevant for Y Combinator's experiment in idea-less idea-makers. In their call for these applications, they ask for teams of people, not individuals. But of course, during the Y Combinator process, the team will work with all manner of new faces. The hope, even if not stated explicitly, is that Y Combinator can create an environment that approximates the big-money Q score.

The mechanism underlying the power of the Q score and critical group dynamics might be one and the same: the human mind. A theory advanced last year on why humans reason at all suggests that the purpose is primarily social -- and that as a result the best human thinking happens not when left to our own devices, but when constructing an argument through engagement with other humans. Humans make great ideas by spotting flaws in one other's arguments and refining their ideas to correct for the faults. Finding the right people for that process is more important than beginning it with the right idea -- the ideas, after all, will improve much faster than the humans.

We want to hear what you think about this article. Submit a letter to the editor or write to