Overwhelmed, Alone, Uncool: Visions of the Listener in Algorithmic Recommendation

Who are recommender systems for? Algorithmic recommendation is a ubiquitous feature of the contemporary internet, filtering social media posts, personalizing the delivery of news, and suggesting music, therapists, restaurants, and countless other choices to users. Recently, these systems have become objects of popular critique: activists and researchers argue that they are designed to addict users and intensify biases, pursuing “engagement” for online platforms at the expense of the broader social world. This talk explores how the people who make recommender systems think about their work, as they weave together vernacular social theorizing and technical design. Drawing on ethnographic fieldwork with makers of music recommendation in the US, I describe how visions of the listener—and the concomitant role of recommendation—have shifted over their relatively short history. At their origins in the mid-1990s, the new “collaborative filters” were envisioned as aids to eager but overwhelmed enthusiasts; a decade later, recommenders were reimagined as tools for tempting otherwise indifferent people into becoming users. Today, recommender systems participate in a distributed ecology of data collection and user experience, shaping the online environments in which people encounter and experience cultural materials. As algorithms become matters of public concern, understanding how the people who build and maintain them think is crucial to reckoning with their power.