[Paper Review] Escaping the Impossibility of Fairness: From Formal to Substantive Algorithmic Fairness

Ben Green|arXiv (Cornell University)|Jul 9, 2021

Ethics and Social Impacts of AI15 references11 citations

TL;DR

This paper challenges the 'impossibility of fairness' in algorithmic decision-making by critiquing formal fairness definitions that isolate decisions from broader social contexts. It proposes 'substantive algorithmic fairness'—a methodology rooted in legal and philosophical theories of substantive equality—to evaluate algorithms based on real-world justice outcomes, enabling meaningful equity in public policy applications despite mathematical constraints.

ABSTRACT

Efforts to promote equitable public policy with algorithms appear to be fundamentally constrained by the "impossibility of fairness" (an incompatibility between mathematical definitions of fairness). This technical limitation raises a central question about algorithmic fairness: How can computer scientists and policymakers support equitable policy reforms with algorithms? In this article, I argue that promoting justice with algorithms requires reforming the methodology of algorithmic fairness. First, I diagnose the problems of the current methodology for algorithmic fairness, which I call "formal algorithmic fairness." Because formal algorithmic fairness restricts analysis to isolated decision-making procedures, it leads to the impossibility of fairness and to models that exacerbate oppression despite appearing "fair." Second, I draw on theories of substantive equality from law and philosophy to propose an alternative methodology, which I call "substantive algorithmic fairness." Because substantive algorithmic fairness takes a more expansive scope of analysis, it enables an escape from the impossibility of fairness and provides a rigorous guide for alleviating injustice with algorithms. In sum, substantive algorithmic fairness presents a new direction for algorithmic fairness: away from formal mathematical models of "fair" decision-making and toward substantive evaluations of whether and how algorithms can promote justice in practice.

Motivation & Objective

• To diagnose the limitations of current formal algorithmic fairness methodologies that restrict analysis to isolated decision procedures.
• To address the paradox where mathematically 'fair' algorithms still perpetuate systemic oppression.
• To propose a new framework—substantive algorithmic fairness—that evaluates algorithms based on their actual impact on social justice.
• To shift the focus from formal mathematical fairness to practical, context-sensitive evaluations of algorithmic justice in public policy.
• To provide a rigorous, theory-driven alternative that enables algorithms to genuinely support equitable policy reforms.

Proposed method

• Drawing on theories of substantive equality from law and philosophy, the paper redefines fairness beyond mathematical definitions.
• It advocates for a broader analytical scope that includes historical, structural, and contextual factors shaping algorithmic outcomes.
• The methodology evaluates algorithms not by compliance with isolated fairness metrics, but by their role in advancing or undermining social justice.
• It emphasizes the importance of power structures, historical marginalization, and institutional contexts in assessing algorithmic impact.
• The approach integrates critical race theory and critical legal studies to inform the evaluation of algorithmic systems.
• It frames algorithmic fairness as a normative, context-dependent inquiry rather than a purely technical optimization problem.

Experimental results

Research questions

• RQ1 Why do formal fairness definitions fail to prevent systemic oppression in algorithmic decision-making?
• RQ2 How can algorithmic fairness be redefined to move beyond mathematical incompatibilities and actualize justice?
• RQ3 What role do historical and structural power dynamics play in shaping algorithmic outcomes despite compliance with fairness metrics?
• RQ4 In what ways can legal and philosophical theories of substantive equality inform the evaluation of algorithmic systems?
• RQ5 How can algorithms be assessed not just for fairness in isolation, but for their real-world impact on social equity?

Key findings

• Formal algorithmic fairness leads to the 'impossibility of fairness' because it isolates decisions from broader social and historical contexts.
• Mathematical fairness definitions are insufficient to prevent harm when applied without attention to structural inequities.
• Substantive algorithmic fairness enables a more rigorous, context-aware evaluation of algorithms that can identify and mitigate systemic injustices.
• The framework allows for the identification of 'fair' algorithms that still perpetuate oppression when evaluated through formal metrics alone.
• By grounding fairness in substantive equality, the approach provides a practical path for using algorithms to support equitable public policy.
• The paper demonstrates that algorithmic fairness must be redefined as a normative, justice-oriented inquiry rather than a technical optimization task.