Silence is Golden: Team Problem Solving and Communication Costs

Numerous experimental studies have found that teams outperform individuals on intellective problems by a large degree. This result is intuitively appealing, as teams can benefit from sharing insights. However, much of this literature considers a task that involves a single clear insight. We instead analyze experiments comparing the performance of individuals and teams of four subjects at solving a series of challenging logic puzzles, where the solution involves a series of incremental steps. Contrary to the existing literature, team performance is statistically indistinguishable from that of individuals when there is no cost to sending a message. Frictions resulting from both congestion and incorrect suggestions substantially derail the solution process for teams. If we impose a very small cost of communication on teams, the performance of teams improves. Underlying these results is a robust negative relationship between frequency of communication and team performance. Using a straightforward prediction exercise, we show that performance would have improved dramatically if the best member of each team had been asked to solve the puzzles alone rather than forming a team.