Machine learning algorithms pervade contemporary society. They are integral to social institutions, inform processes of governance, and animate the mundane technologies of daily life. Consistently, the outcomes of machine learning reflect, reproduce, and amplify structural inequalities. The field of fair machine learning has emerged in response, developing mathematical techniques that increase fairness based on anti-classification, classification parity, and calibration standards. In practice, these computational correctives invariably fall short, operating from an algorithmic idealism that does not, and cannot, address systemic, Intersectional (...) stratifications. Taking present fair machine learning methods as our point of departure, we suggest instead the notion and practice of algorithmic reparation. Rooted in theories of Intersectionality, reparative algorithms name, unmask, and undo allocative and representational harms as they materialize in sociotechnical form. We propose algorithmic reparation as a foundation for building, evaluating, adjusting, and when necessary, omitting and eradicating machine learning systems. (shrink)

Measures of algorithmic bias can be roughly classified into four categories, distinguished by the conditional probabilistic dependencies to which they are sensitive. First, measures of "procedural bias" diagnose bias when the score returned by an algorithm is probabilistically dependent on a sensitive class variable (e.g. race or sex). Second, measures of "outcome bias" capture probabilistic dependence between class variables and the outcome for each subject (e.g. parole granted or loan denied). Third, measures of "behavior-relative error bias" capture probabilistic dependence between (...) class variables and the algorithmic score, conditional on target behaviors (e.g. recidivism or loan default). Fourth, measures of "score-relative error bias" capture probabilistic dependence between class variables and behavior, conditional on score. Several recent discussions have demonstrated a tradeoff between these different measures of algorithmic bias, and at least one recent paper has suggested conditions under which tradeoffs may be minimized. -/- In this paper we use the machinery of causal graphical models to show that, under standard assumptions, the underlying causal relations among variables forces some tradeoffs. We delineate a number of normative considerations that are encoded in different measures of bias, with reference to the philosophical literature on the wrongfulness of disparate treatment and disparate impact. While both kinds of error bias are nominally motivated by concern to avoid disparate impact, we argue that consideration of causal structures shows that these measures are better understood as complicated and unreliable measures of procedural biases (i.e. disparate treatment). Moreover, while procedural bias is indicative of disparate treatment, we show that the measure of procedural bias one ought to adopt is dependent on the account of the wrongfulness of disparate treatment one endorses. Finally, given that neither score-relative nor behavior-relative measures of error bias capture the relevant normative considerations, we suggest that error bias proper is best measured by score-based measures of accuracy, such as the Brier score. (shrink)

Data‐driven algorithms are widely used to make or assist decisions in sensitive domains, including healthcare, social services, education, hiring, and criminal justice. In various cases, such algorithms have preserved or even exacerbated biases against vulnerable communities, sparking a vibrant field of research focused on so‐called algorithmic biases. This research includes work on identification, diagnosis, and response to biases in algorithm‐based decision‐making. This paper aims to facilitate the application of philosophical analysis to these contested issues by providing an overview of three (...) key topics: What is algorithmic bias? Why and how can it occur? What can and should be done about it? Throughout, we highlight connections—both actual and potential—with philosophical ideas and concerns. (shrink)

In controversies about technology and society, there is no idea more pro vocative than the notion that technical things have political qualities. At issue is the claim that the machines, structures, and systems of modern material culture can be accurately judged not only for their contributions of efficiency and pro-ductivity, not merely for their positive and negative environmental side effects, but also for the ways in which they can embody specific forms of power and authority. Since ideas of this kind (...) have a persistent and troubling presence in discussions about the meaning of technology, they deserve explicit attention... (shrink)

This book continues and revises the ideas of justice as fairness that John Rawls presented in _A Theory of Justice_ but changes its philosophical interpretation in a fundamental way. That previous work assumed what Rawls calls a "well-ordered society," one that is stable and relatively homogenous in its basic moral beliefs and in which there is broad agreement about what constitutes the good life. Yet in modern democratic society a plurality of incompatible and irreconcilable doctrines -- religious, philosophical, and moral (...) -- coexist within the framework of democratic institutions. Recognizing this as a permanent condition of democracy, Rawls asks how a stable and just society of free and equal citizens can live in concord when divided by reasonable but incompatible doctrines? This edition includes the essay "The Idea of Public Reason Revisited," which outlines Rawls' plans to revise _Political Liberalism,_ which were cut short by his death. "An extraordinary well-reasoned commentary on _A Theory of Justice_...a decisive turn towards political philosophy." -- _Times Literary Supplement_. (shrink)

This volume draws on a range of approaches in order to explore the problem and determine what ought to be done about allegedly harmful speech.Most liberal societies are deeply committed to a principle of free speech. At the same time, however, there is evidence that some kinds of speech are harmful in ways that are detrimental to important liberal values, such as social equality. Might a genuine commitment to free speech require that we legally permit speech even when it is (...) harmful, and even when doing so is in conflict with our commitment to values like equality? Even if such speech is to be legally permitted, does our commitment to free speech allow us to provide material and institutional support to those who would contest such harmful speech? And finally, and perhaps most importantly, which kinds of speech are harmful in ways that merit response, either in the form of legal regulation or in some other form? This collection explores these and related questions. Drawing on expertise in philosophy, sociology, political science, feminist theory, and legal theory, the contributors to this book investigate these themes and questions. By exploring various categories of speech (including pornography, hate speech, Holocaust denial literature, 'Whites Only' signs), and attending to the precise functioning of speech, the essays contained here shed light on these questions by clarifying the relationship between speech and harm. Understanding how speech functions can help us work out which kinds of speech are harmful, what those harms are, and how the speech in question brings them about. All of these issues are crucially important when it comes to deciding what ought to be done about allegedly harmful speech. (shrink)

Fricker shows that virtue epistemology provides a general epistemological idiom in which these issues can be forcefully discussed.

Argues that the War on Drugs and policies that deny convicted felons equal access to employment, housing, education and public benefits create a permanent under-caste based largely on race. Reprint. 12,500 first printing.