Machine learning classifiers are increasingly used to inform, or even make, decisions significantly affecting human lives. Fairness concerns have spawned a number of contributions aimed at both identifying and addressing unfairness in algorithmic decision-making. This paper critically discusses the adoption of group-parity criteria (e.g., demographic parity, equality of opportunity, treatment equality) as fairness standards. To this end, we evaluate the use of machine learning methods relative to different steps of the decision-making process: assigning a predictive score, linking a classification to (...) the score, and adopting decisions based on the classification. Throughout our inquiry we use the COMPAS system, complemented by a radical simplification of it (our SAPMOC I and SAPMOC II models), as our running examples. Through these examples, we show how a system that is equally accurate for different groups may fail to comply with group-parity standards, owing to different base rates in the population. We discuss the general properties of the statistics determining the satisfaction of group-parity criteria and levels of accuracy. Using the distinction between scoring, classifying, and deciding, we argue that equalisation of classifications/decisions between groups can be achieved thorough group-dependent thresholding. We discuss contexts in which this approach may be meaningful and useful in pursuing policy objectives. We claim that the implementation of group-parity standards should be left to competent human decision-makers, under appropriate scrutiny, since it involves discretionary value-based political choices. Accordingly, predictive systems should be designed in such a way that relevant policy goals can be transparently implemented. Our paper presents three main contributions: (1) it addresses a complex predictive system through the lens of simplified toy models; (2) it argues for selective policy interventions on the different steps of automated decision-making; (3) it points to the limited significance of statistical notions of fairness to achieve social goals. (shrink)

Fairness is a central, but under-theorized, notion in moral and political philosophy. This paper makes two contributions. Firstly, it criticizes Broome’s seminal account of fairness in Proc Aristotelian Soc 91:87–101, showing that there are problems with restricting fairness to a matter of relative satisfaction and holding that it does not itself require the satisfaction of the claims in question. Secondly, it considers the justification of lotteries to resolve cases of ties between competing claims, which Broome claims as support for his (...) theory, and contrasts random procedures to contests of skill, which may also be considered lotteries in a broader sense. I offer no alternative account of fairness of my own, but hope to point the way for future research on the nature of fairness. (shrink)

To what extent can we as a community legitimately require individuals to contribute to producing public goods? Most of us think that, at least sometimes, refusing to pay for a public good that you have enjoyed can involve a kind of 'free riding' that makes it wrong. But what is less clear is under exactly which circumstances this is wrong. To work out the answer to that, we need to know why it is wrong. I argue that when free riding (...) is wrong, the reason is that it is unfair. That is not itself a very controversial claim. But spelling out why it is unfair allows us to see just which forms of free riding are wrong. Moreover, it supplies a basis from which some more controversial conclusions can be defended. Even if a public good is one that you have been given without asking for it or seeking it out, it can still be wrong not to be prepared to pay for it. It can be wrong not to be prepared to pay for public goods even when you do not receive them at all. And furthermore, it can be right to force you to do so. (shrink)