Information cascades in an artificial cultural market
https://www.princeton.edu/~mjs3/salganik_dodds_watts06_full.pdf
Computational sociologists Matthew Salganik and Duncan Watts investigated the paradox of cultural good success: that hit songs, books and movies are much more successful than average yet experts often fail to predict which ones will succeed. This suggested that the success of the hit products may not depend on the innate quality of the product but initial advantage and information cascades. The researchers created an artificial “music market” with 48 new songs created by C list music bands. They divided the experimenters into two worlds, one in which people will see the ratings of songs by other people, and one in which people cannot see the ratings of songs by other people. The first world is then further divided into 8 different worlds.
The results show that songs which are consistently rated highly (in the second world) can perform anywhere between average to great success when people get to see the ratings by others (first world). The researchers concluded that success in cultural market was only partly determined by quality, and sometimes almost completely depends on the initial ratings and whether an information cascade is created or not.
The music lab experiment in my opinion is one of the most ground breaking experimental studies in computational sociology. It demonstrated the full extend/power of information cascades. Homework 6 distinguished between information based reasons and direct benefit reasons as the primary two reasons why people would imitate other’s behavior. In this experiment we can see that people almost completely imitate others for information based reasons alone. Furthermore, it does demonstrate that information cascades, once formed, can theoretically go on forever. This is proven in theory in the blue ball red ball example in the book but very hard for people to conceptualize in the real world