Collective Intelligence

Movements like Top Coder and open source lead us to believe that the intelligence of a group is greater than the sum of its individuals.

I tend to believe this is true, but the problem is that it’s difficult to see the scenarios for which this is not true.

How about when a team comes to the wrong conclusion because the one person who was right was out-voted?
Or the many disciplines that require focus and discipline over a long period of time? Those kind of projects can be more likely to succeed with fewer people.