Beyond case studies: Teaching data science critique and ethics through sociotechnical surveillance studies

Abstract: Ethics have become an urgent concern for data science research, practice, and instruction in the wake of growing critique of algorithms and systems showing that they reinforce structural oppression. There has been increasing desire on the part of data science educators to craft curricula that speak to these critiques, yet much ethics education remains individualized, focused on specific cases, or too abstract and unapplicable. We synthesized some of the most popular critical data science works and designed a data science ethics course that spoke to the social phenomena at the root of critical data studies -- theories of oppression, social systems, power, history, and change -- through analysis of a pressing sociotechnical system: surveillance systems. Through analysis of student reflections and final projects, we determined that at the conclusion of the semester, all students had developed critical analysis skills that allowed them to investigate surveillance systems of their own and identify their benefits, harms, main proponents, those who resist them, and their interplay with social systems, all while considering dimensions of race, class, gender, and more. We argue that this type of instruction -- directly teaching data science ethics alongside social theory -- is a crucial next step for the field.