“Everything is a Recommendation”: Netflix and the Construction of Blackness Through Algorithms

• In Event: Technology and/in Racial Formations

Abstract

How do algorithms construct our racial identities, and should recommendation systems include race as a variable in order to market to users? In the case of Netflix, spokespersons insist that they do not collect data on users’ race, gender, or ethnicity, making it “impossible” to personalize an individual’s Netflix experience based on these identity markers. Yet in October of 2018, a number of Black Netflix users took to Twitter to air their grievances about images in movie thumbnails featuring black actors and actresses with minor roles, even when the movie itself was a majority white cast. Emergent scholarships within algorithm studies speaks to how user behavior can train the algorithm to produce differences by race; scholars such as Safiya Noble and Latanya Sweeney insist that user behavior can also train an algorithmic system to produce problematic outputs in the context of machine learning. Using the Netflix debacle as a case study, I will further this conversation by exploring how race is still a variable and significant factor in the company’s recommendation system algorithms. Thus, my paper seeks to interrogate the intersections of race, algorithms, and culture, specifically asking questions about how recommendation systems account for (or claim to not account for) race as a variable. By utilizing Erving Goffman’s understanding of “the presentation of self” which focuses on how individuals use different signifiers as a way of constructing a social self, my analysis will show how Netflix transforms “black users” into “black subjects” in order to profit off of the engagement of black consumers.

Author

• Daniel Meyerend, University of Michigan