Music streaming platforms are determinant of the listening experience today. Their ability to profile users and to predict behaviours and tastes is key as their business-models are based on the loyalty of users. Drawing on a study of The Echo Nest, a music recommendation engine acquired by Spotify in 2014, which claimed to combine the analysis of the music signal with monitoring of consumer behaviour via the collection of their data for the first time, this essay interrogates automatic taste-profiling as a transformation of the philosophical concept of taste, opening up new perspectives on music and language.

This chapter provides an overview of the role of algorithmic recommendation in contemporary music streaming services, describing how they work, how they relate to other algorithmic applications, and problems that have emerged from their use. Against a dominant discourse that pits algorithms against humans, it argues that contemporary recommender systems are best understood as ‘ensembles’, comprising a variety of algorithmic and human parts working in conjunction with each other. This suggests new directions for research, focusing not on the intrinsic character of human or algorithmic mediation, but rather how this sociotechnical ensemble is composed and conducted. The issues raised by algorithmic music recommendation are not new but variations on past concerns including payola, the treatment of so-called world music and the power of cultural intermediaries.