The current version of The New Yorker has a wonderful article by Jonah Lehrer called "Groupthink" (you can see the abstract here). It does a great job of showing how creativity is a social process, cites wonderful research by Brian Uzzi showing that when people have experience working together in the past they produce more successful Broadway musicals (up to a point, too many old friends is as bad as too few), and offers research showing that groups where members engage in constructive conflict are more creative -- all themes I have talked about at various times on this blog.
I do however have a major quibble. At one point, Lehrer states flatly that brainstorming doesn't work. He later quotes creativity researcher Keith Sawyer as saying that people are more efficient at generating ideas when they work alone than in groups, something that is well-established. But that is not the same as saying there is conclusive evidence they don't work.
I once devoted way too much time to the question of whether this research shows that brainstorming is useless. In the name of full-disclosure, please note I am a Fellow at IDEO and also a co-founder of the Stanford d.school, which both use brainstorming a lot. But I am not at all a religious zealot about the method. I see it as just one sometimes useful method, and I have often said that the d.school in particular should spend less time teaching brainstorming and more time teaching people how to fight. (And if you want evidence that the d.school believes in more than just brainstorming, look at their Bootleg.)
But please consider several facts about the brainstorming literature, at least as it stood about 7 or 8 years ago when I last reviewed it carefully and which is consistent with a more recent paper from The Academy of Management Review (Here is the abstract, which is quite short):
1. Nearly all brainstorming research is done with people who have no training or experience in doing or leading brainstorming. In fact, there is at least two studies showing that, when facilitated properly, the so called productivity loss disappears. Check this 1996 study and this 2001 study. To me, these two studies alone call into question the approach taken in most brainstorming studies, which don't use facilitation. In other words, the conculsion that brainstorming doesn't work is based largely on studies that use unsupervised brainstorming virgins.
2. As Keith Sawyer's comment implies, nearly all this research looks at only one measure of effectiveness, how quickly people can produce ideas. Because people in groups have to take time to listen to each other, it slows the idea generation process. Most brainstorming studies compare the speed at which people generate ideas such as "what can you do with a brick" when sitting alone and talking into microphone versus doing so in face-to-face groups. In fact, if creativity is about both talking and listening, if you look at the data from these same studies, I once figured out that people are exposed to substantially more ideas per unit of time when you compare group to solo brainstorming -- and I would argue that talking and listening are both key elements of the social process underlying creativity.
3.A key part of face-to-face brainstorming is building on and combining the ideas of others. This comparison is impossible in most brainstorming studies because an individual working alone is not exposed to the ideas of others.
Indeed, one of the very first posts I did on this blog in 2006 dug on this issue. As I wrote then, "To put it another way, if these were studies of sexual performance, it would be like drawing inferences about what happens with experienced couples on the basis of research done only with virgins during the first time they had sex." I also wrote about brainstorming here in BusinessWeek and they started with this setup.
The upshot of my research and my reading of brainstorming experiments is that, if you are just looking at the speed at which an individual can spew out ideas, individual brainstorming is likely superior. But if you look at the range of positive effects has at a place like IDEO -- spreading ideas around the company, teaching newcomers and reminding veterans of solutions and technologies and who knows what, providing variety and intrinsically satisfying breaks for designers working on other projects, creating what I called a functional status contests where designers compete politely to show off their creativity (a key job skill), and impressing clients, brainstorming may have numerous other positive benefits in real organizations where creative work is done -- none of which have not been examined in those simple experiments. If so, those findings about pure efficiency may well be beside the point when it comes to evaluating brainstorming in organizations that use it routinely.
In short, I believe that Lehrer's statement that brainstorming "doesn't work" is too sweeping because it has not been studied adequately in real organizations or with people who have real brainstorming skills. Again, I would describe this as a quibble; the article in The New Yorker is otherwise excellent.
P.S. for the true nerds, here is the 1996 academic article on brainstorming that Andy Hargadon and I wrote: