
Sociology professor Duncan Watts at Columbia has done research on consumer preferences that demonstrate this snowball effect. In one study, he asked thousands of people to download songs. When they could see how many times a song had been downloaded by other participants, the popular songs became much more popular relative to the duds. And each time he ran the experiment, different songs came out on top. There were some general trends--good songs were never complete losers and bad songs never won big--but there was a significant element of chaos. So in a parallel universe, might Romney and Edwards still be in the race?















