Ever since ProPublica published their groundbreaking analysis of Northpointe’s Correctional Offender Management Profiling for Alternative Sanctions Core Risk and Needs Assessment software (COMPAS) in 2016, this web-based decision support system (DSS) has spawned a wide range of critiques and charges of racial bias. COMPAS provides a full suite of decision support applications to the US prison-industrial complex, including algorithmically derived recidivism predictions that increasingly guide parole decisions. The larger conversation surrounding COMPAS raises the question of how we analyze powerful, and yet opaque, data assemblages. In this article, I model an allegorical analysis of data assemblages. I argue the skills of literary analysis can help us surmount the “trap” of analyzing black-box algorithms. Guided by Poe’s “The Purloined Letter,” I analyze how COMPAS purloins our mediated ability to give an account of ourselves under the guise of objective, data-driven decision making.
"Purloined Significance: How Recidivism Algorithms Capture, Transform, and Automate our Intersubjective Unconscious as Data."
CLCWeb: Comparative Literature and Culture
This text has been double-blind peer reviewed by 2+1 experts in the field.
The above text, published by Purdue University Press ©Purdue University, has been downloaded 151 times as of 10/27/23.
American Studies Commons, Comparative Literature Commons, Other Arts and Humanities Commons, Other Film and Media Studies Commons, Rhetoric and Composition Commons, Social and Behavioral Sciences Commons