Duncan Watts, who was highlighted in a Fast Company article I linked to yesterday, also was the brains behind Music Lab, a social experiment that attempted to measure how people reacted to bands based on the merits of the music compared to their perceived popularity among others:
The control group was asked to rate the songs with no input from the other members. Seven other groups were also asked to rate the songs, but were allowed to see how other members of their group voted.
Two findings emerged. First, the variation in popularity in the control group was much smaller than in the other groups. In the groups where they could see their peers’ opinions, people voted for the same acts that other people voted for, suggesting that popularity breeds more popularity, which should be no surprise to anybody who’s been through high school.
The more interesting finding was the complete randomness of the songs that became popular in each group. There was almost no connection between “objective quality” (as measured by the control group) and popularity–the song “Lockdown” was ranked number 26 in terms of quality, but its popularity in the other groups ranged from number 1 to number 40. True, the very bad songs almost never finished near the top, and the best songs seldom finished near the bottom. But overall, finishing in the top five in quality only guaranteed a 50% chance of finishing in the top five in popularity in any given group.
