We map the recently proposed notions of algorithmic fairness to economic models of Equality of opportunity (EOP)---an extensively studied ideal of fairness in political philosophy. We formally show that through our conceptual mapping, many existing definition of algorithmic fairness, such as predictive value parity and equality of odds, can be interpreted as special cases of EOP. In this respect, our work serves as a unifying moral framework for understanding existing notions of algorithmic fairness. Most importantly, this framework allows us to (...) explicitly spell out the moral assumptions underlying each notion of fairness, and interpret recent fairness impossibility results in a new light. Last but not least and inspired by luck egalitarian models of EOP, we propose a new family of measures for algorithmic fairness. We illustrate our proposal empirically and show that employing a measure of algorithmic (un)fairness when its underlying moral assumptions are not satisfied, can have devastating consequences for the disadvantaged group's welfare. (shrink)

Often machine learning programs inherit social patterns reflected in their training data without any directed effort by programmers to include such biases. Computer scientists call this algorithmic bias. This paper explores the relationship between machine bias and human cognitive bias. In it, I argue similarities between algorithmic and cognitive biases indicate a disconcerting sense in which sources of bias emerge out of seemingly innocuous patterns of information processing. The emergent nature of this bias obscures the existence of the bias itself, (...) making it difficult to identify, mitigate, or evaluate using standard resources in epistemology and ethics. I demonstrate these points in the case of mitigation techniques by presenting what I call ‘the Proxy Problem’. One reason biases resist revision is that they rely on proxy attributes, seemingly innocuous attributes that correlate with socially-sensitive attributes, serving as proxies for the socially-sensitive attributes themselves. I argue that in both human and algorithmic domains, this problem presents a common dilemma for mitigation: attempts to discourage reliance on proxy attributes risk a tradeoff with judgement accuracy. This problem, I contend, admits of no purely algorithmic solution. (shrink)

In this stimulating work of political philosophy, acclaimed philosopher G. A. Cohen sets out to rescue the egalitarian thesis that in a society in which distributive justice prevails, peopleâes material prospects are roughly equal. Arguing against the Rawlsian version of a just society, Cohen demonstrates that distributive justice does not tolerate deep inequality. In the course of providing a deep and sophisticated critique of Rawlsâes theory of justice, Cohen demonstrates that questions of distributive justice arise not only for the state (...) but also for people in their daily lives. The right rules for the macro scale of public institutions and policies also apply, with suitable adjustments, to the micro level of individual decision-making. Cohen also charges Rawlsâes constructivism with systematically conflating the concept of justice with other concepts. Within the Rawlsian architectonic, justice is not distinguished either from other values or from optimal rules of social regulation. The elimination of those conflations brings justice closer to equality. (shrink)

[ Susan Hurley] I argue that the aim to neutralize the influence of luck on distribution cannot provide a basis for egalitarianism: it can neither specify nor justify an egalitarian distribution. Luck and responsibility can play a role in determining what justice requires to be redistributed, but from this we cannot derive how to distribute: we cannot derive a pattern of distribution from the 'currency' of distributive justice. I argue that the contrary view faces a dilemma, according to whether it (...) understands luck in interpersonal or counterfactual terms. /// [Richard J. Arneson] Does it make sense to hold that, if it is bad that some people are worse off than others, it is worse if those who are worse off come to be so through sheer bad luck that it is beyond their power to control? In her contribution to this symposium, Susan Hurley cautions against a closely related fallacy: from the fact that people have come to an unequal condition through unchosen bad luck, it does not follow that, if we aim to undo the influence of unchosen luck, we ought to institute equality of condition. Forswearing the fallacy that Hurley analyses is compatible with answering the question affirmatively, and more generally with holding that principles of distributive justice should be sensitive to the distinction between chosen and unchosen bad luck. This essay explores how this might be done. (shrink)

In this paper I present and defend a highly demanding principle of justice in education that has not been seriously discussed thus far. According to the suggested approach, “all the way equality”, justice in education requires nothing short of equal educational outcome between all individual students. This means not merely between equally able children, or between children from different groups and classes, but rather between all children, regardless of social background, race, sex and ability. This approach may seem implausible at (...) first, due to the far-reaching implications it entails, primarily its requirement to deny better-off children their advantage for the sake of equality. However the paper argues that all-the-way-equality, in fact, does a better job realizing the goals of justice in education than alternative conceptions of justice. It is further argued that at least some of the principle’s most radical consequences, those that make it seem counterintuitive, can be mitigated by balancing all-the-way-equality with competing interests. (shrink)

The paper examines the view that individuals have a claim to the jobs for which they are the best qualified. It seeks to show this view to be groundless, and to offer, instead, a luck egalitarian account of justice in hiring. That account consists of three components: monism, non-meritocracy, and non-discrimination. To demonstrate the coherence of this view, two particular internal conflicts are addressed. First, luck egalitarian monism (the view that jobs are not special) may end up violating the non-discrimination (...) requirement. Second, non-discrimination, it is often suggested, cannot be defined without reference to qualifications, thus violating the non-meritocracy requirement. The paper seeks to address these, as well as other, potential objections, and show that whereas meritocratic accounts are without basis, luck egalitarianism provides a coherent and attractive account of justice in hiring. (shrink)

Does it make sense to hold that, if it is bad that some people are worse off than others, it is worse if those who are worse off come to be so through sheer bad luck that it is beyond their power to control? In her contribution to this symposium, Susan Hurley cautions against a closely related fallacy: from the fact that people have come to an unequal condition through unchosen bad luck, it does not follow that, if we aim (...) to undo the influence of unchosen luck, we ought to institute equality of condition. Forswearing the fallacy that Hurley analyses is compatible with answering the question affirmatively, and more generally with holding that principles of distributive justice should be sensitive to the distinction between chosen and unchosen bad luck. This essay explores how this might be done. (shrink)

Health, Luck, and Justice is the first attempt to systematically apply luck egalitarianism to the just distribution of health and health care.

Does the unequal participation of non-traditional students in higher education indicate social injustice, even if it can be traced back to individuals' choices? Drawing on luck egalitarian approaches,this article suggests that an answer to this question must take into account the effects of unequal brute luck on educational choices.I use a framework based on expected utility theory to analyse qualitative studies on educational choice.This reveals a variety of mechanisms through which differences in background conditions make non-traditional students less likely to (...) apply to university and/or particular institutions; the unequal participation of nontraditional students in higher education remains a problem of social justice. (shrink)