Algorithmic Pluralism: A Structural Approach Towards Equal Opportunity
share
While the idea of equal opportunity enjoys a broad consensus, many disagree about what it means for opportunities to be equal. The algorithmic fairness community often relies on formal approaches to quantitatively determine if opportunities are allocated equally. A more structural approach put forth by Joseph Fishkin focuses on the wider network of decisions that determine which opportunities are allocated to whom. In this so-called opportunity structure, decision points represent bottlenecks that are often chained together so that the output of one decision is an input to the next. By evaluating the severity and legitimacy of these bottlenecks, Fishkin offers a qualitative framework to assess whether equal opportunity is infringed upon in a structural way. We adopt this structural viewpoint and use it to reframe many interdisciplinary discussions about equal opportunity in systems of algorithmic decision-making. Drawing on examples from education, healthcare, and criminal justice, we recommend prioritizing regulatory and design-based interventions that alleviate severe bottlenecks in order to help expand access to opportunities in a pluralistic way.
READ FULL TEXT