An investigation of the centrality of race, gender, class, and sexuality to Big Data and network analytics that links machine learning tools to discrimination.
In Discriminating Data, Wendy Hui Kyong Chun argues that race, gender, class and sexuality are central to Big Data and network analytics. Chun connects the machine learning algorithms that underlie recommender systems and social media platforms to methods developed for twentieth-century eugenics. Most networks presume homophily-that similarity breeds connection-and so divide users into angry clusters. Chun argues that key concepts of machine learning and data structures, including homophily as well as correlation, authenticity, and recognition, foster acrimony and segregation through their default assumptions and conditions.
Chun, who has a background in systems design engineering as well as media studies and cultural theory, excavates the discriminatory foundations of machine learning. She explains that although machine learning algorithms may not officially include race as a category, they embed whiteness as a default. Trained on selected and often discriminatory or dirty data, these algorithms are only verified if they mirror this data. Facial recognition technology, for example, relies on the faces of Hollywood celebrities and university undergraduates. Predictive policing technology collects data to produce "heatmaps" of crime by zip code, categorizing all residents of certain neighborhoods as potential criminals. How can we release ourselves from the vice-like grip of discriminatory data? Chun calls for alternative algorithms, defaults, and interdisciplinary coalitions in order to desegregate networks and foster a more democratic Big Data.
Wendy Hui Kyong Chun is Simon Fraser University's Canada 150 Research Chair in New Media and Professor of Communication and Director of the SFU Digital Democracies Institute. She is the author of Control and Freedom, Programmed Visions, and Updating to Remain the Same, all published by the MIT Press.
