The scale of the catalogues currently provided by music streaming services is vast — as of 2018, Apple Music advertise providing 45m songs. Catalogues of this scale are unnavigable to human users, therefore digital tools, such as recommendation algorithms and individualised playlists, have been developed to ease discovery and navigation. These tools utilise vast amounts of data collected on the millions of users — detailed accounts of listener behaviour and interaction — an attempt to form a deeper understanding of the users, theoretically creating the ability to better provide individual song recommendations.
Even with the best of intentions, there is potential for negative consequences. With increasingly personalised experiences, users may become intellectually unaware of the wider music environment as hearing the unexpected becomes impossible. Serendipity, by definition, being unprogrammable. Furthermore, rather than working solely for the benefit of users, algorithm design is likely heavily influenced by financial and business pressures — services having an acute need to maintain good working relationships with content providers and advertisers.
Although the opaque nature of these digital tools makes satisfactory investigation of the design impossible, this chapter addresses many of the underlying processes behind the recommendations, and explores the consequences of personalised listening.
Full chapter included within Popular Music in the Post-Digital Age, Bloomsbury (2018)