Let the music be your master: Power laws and music listening habits

Abstract

Music preferences have long been studied owing to their importance in the fields of psychology and sociology. However, previous efforts seldom focused on people’s deliberate choices of music in everyday life. In this study, we aimed to analyze music listening behaviors using personal records of music listening activity. We obtained the history of songs listened to by 50 different users of the online database system Last.fm, spanning on average five years of activity. With the use of this data set, we are able to confirm that the number of songs reproduced per artist follows a truncated power-law distribution. The scaling parameter of the distribution varies considerably among users, providing a metric that characterizes the way in which different people explore music. We propose that this pattern is consistent with a preferential attachment model, according to which the probability of listening to a given artist at a given time is proportional to the frequency to which the artist was listened to in the past. These results provide new insight regarding the way in which individual music preferences are built.