As machine learning informs increasingly consequential decisions, different metrics have been proposed for measuring algorithmic bias or unfairness. Two popular “fairness measures” are calibration and equality of false positive rate. Each measure seems intuitively important, but notably, it is usually impossible to satisfy both measures. For this reason, a large literature in machine learning speaks of a “fairness tradeoff” between these two measures. This framing assumes that both measures are, in fact, capturing something important. To date, philosophers have seldom examined (...) this crucial assumption, and examined to what extent each measure actually tracks a normatively important property. This makes this inevitable statistical conflict – between calibration and false positive rate equality – an important topic for ethics. In this paper, I give an ethical framework for thinking about these measures and argue that, contrary to initial appearances, false positive rate equality is in fact morally irrelevant and does not measure fairness. (shrink)

John Broome argues that when all claims cannot be perfectly fairly satisfied in outcome, the contribution to fairness from entering claims into a lottery, and so providing them some surrogate satisfaction, ought to be weighed against, and can outweigh, what fairness can be achieved directly in outcome. I argue that this is a mistake. Instead, I suggest that any contribution to fairness from entering claims into a lottery is lexically posterior to fairness in outcome.

John Taurek famously argued that, in ‘conflict cases’, where we are confronted with a smaller and a larger group of individuals, and can choose which group to save from harm, we should toss a coin, rather than saving the larger group. This is primarily because coin-tossing is fairer: it ensures that each individual, regardless of the group to which he or she belongs, has an equal chance of being saved. This article provides a new response to Taurek’s argument. It proposes (...) that there are two possible types of unfairness that have to be avoided in conflict cases, as far as possible: ‘selection unfairness’, which is the unfairness of not giving individuals an equal chance of being saved; and ‘outcome unfairness’, which is the unfairness of not actually saving them, when others are saved. Since saving the greater number generates less outcome unfa-irness than coin-tossing, it is argued that, in many conflict cases, fairness demands that we save the greater number. (shrink)

John Broome argues that fairness requires that claims are satisfied in proportion to their strength. Broome holds that, when distributing indivisible goods, fairness requires the use of weighted lotteries as a surrogate to satisfy proportionally each candidate's claims. In this article, we present two arguments against Broome's account of fairness. First, we argue that it is almost impossible to calculate the weights of the lotteries in accordance with the requirements of fairness. Second, we argue that Broome rules out those methods (...) whose use might provide some resolution to this problem. From these arguments, we conclude that, contra Broome, fairness does not require the proportional satisfaction of claims. (shrink)

According to the nonworseness claim, it cannot be morallyworseto exploit someone than not to interact with them at all when the interaction 1) is mutually beneficial, 2) is voluntary, and 3) has no negative effects on third parties. My aim in this article is to defend the moral significance of exploitation from this challenge. To that end, I develop a novel account of why sweatshop owners have a moral obligation to pay sweatshop workers a nonexploitative wage despite the fact that (...) their relationship is entirely optional. More precisely, I defend two main claims. First, I show that sweatshop owners are morally obligated to pay sweatshop workers a nonexploitative wage even though they have a right not to hire them and even though that will require them to pay sweatshop workers a wage that is higher than the one they voluntarily accepted. Second, I explain why this obligation on the part of sweatshop owners is not defeated by the fact that other individuals not party to the transaction would benefit even more than sweatshop workers from receiving this additional level of pay. (shrink)

Waiting time is widely used in health and social policy to make resource allocation decisions, yet no general account of the moral significance of waiting time exists. We provide such an account. We argue that waiting time is not intrinsically morally significant, and that the first person in a queue for a resource does not ipso facto have a right to receive that resource first. However, waiting time can and sometimes should play a role in justifying allocation decisions. First, there (...) is a duty of fairness prohibiting line-cutting where a sufficiently just queue exists. Second, waiting time has several morally attractive features that can justify its incorporation into allocation schemes. Where candidates are in relevantly similar circumstances, allocating by waiting time is relatively efficient, maximizes distribution equality relative to other Pareto efficient distributions, and treats candidates fairly. The claim that allocation using waiting time is fair is controversial. Some have claimed that formal lotteries are a fairer way to select among equal beneficiaries. We argue that lotteries are no fairer than allocation based on waiting time when it is equiprobable how a prospective queue will be ordered. In practice, lotteries share many of the disadvantages of queues; which is fairer will depend on contingent features of the allocation scenario. The upshot is that first-come, first-served is in fact a just way to allocate resources in many of the cases where it seems pre-theoretically compelling, and waiting time has unique normative properties which frequently justify its incorporation into resource allocation schemes. (shrink)

The allocation of vaccines and therapeutics for Covid‐19 obviously raises ethical questions, and physicians and ethicists have begun to address them. Writers have identified various criteria that should guide allocation decisions, but the criteria often conflict and need to be balanced against one another. This article proposes a model for thinking about how different considerations that are relevant to the distribution of vaccines and scarce treatments for Covid‐19 could be integrated into an allocation procedure. The model employs the construct of (...) a weighted lottery, which is a construct that has been employed in other contexts that involve the distribution of scarce resources. The article highlights the advantages of applying a weighted lottery to the Covid‐19 context and offers an illustration for how it might work in practice. The primary aim of the article is to articulate the structural features of a weighted lottery for this context and to bring out its advantages over other methods for allocating Covid‐19 medications. (shrink)

Luck egalitarians claim that disadvantage is worse when it emerges from an unchosen risk than when it emerges from a chosen risk. I argue that disadvantage is also worse when it emerges from an unchosen risk that the disadvantaged agent would have declined to take, had he or she been able to do so, than when it emerges from an unchosen risk that the disadvantaged agent would not have declined to take. Such a view is significant because it allows both (...) luck egalitarians and prioritarians to respond to Voorhoeve and Fleurbaey's charge that they fail to accommodate intuitions about the moral relevance of interpersonal boundaries – the so-called separateness of persons objection. I argue that the view is plausible independently of its ability to answer the separateness of persons objection, and is a natural extension of the luck egalitarian concern with the impact of unchosen circumstance. (shrink)

The main body of this paper assesses a leading recent theory of fairness, a theory put forward by John Broome. I discuss Broome's theory partly because of its prominence and partly because I think it points us in the right direction, even if it takes some missteps. In the course of discussing Broome's theory, I aim to cast light on the relation of fairness to consistency, equality, impartiality, desert, rights, and agreements. Indeed, before I start assessing Broome's theory, I discuss (...) two very popular conceptions of fairness that contrast with his. One of these very popular conceptions identifies fairness with the equal and impartial application of rules. The other identifies fairness with all-things-considered moral rightness. (shrink)

With the increasing use of algorithms in high-stakes areas such as criminal justice and health has come a significant concern about the fairness of prediction-based decision procedures. In this article I argue that a prominent class of mathematically incompatible performance parity criteria can all be understood as applications of John Broome’s account of fairness as the proportional satisfaction of claims. On this interpretation these criteria do not disagree on what it means for an algorithm to be _fair_. Rather they express (...) different understandings of what grounds a claim to a good being allocated by an algorithmic decision procedure. I then argue that an important implication of the Broomean interpretation is that it strengthens the case for outcome-based criteria. Finally, I consider how a version of the levelling-down objection to performance parity criteria arises within the Broomean account. (shrink)

What does it mean for algorithmic classifications to be fair to different socially salient groups? According to classification parity criteria, what is required is equality across groups with respect to some performance measure such as error rates. Critics of classification parity object that classification parity entails that achieving fairness may require us to choose an algorithm that makes no group better off and some groups worse off than an alternative. In this article, I interpret the problem of algorithmic fairness as (...) a case concerning the ethics of the distribution of algorithmic classifications across groups (as opposed to, e.g., the fairness of data collection). I begin with a short introduction of algorithmic fairness as a problem discussed in machine learning. I then show how the criticism raised against classification parity is a form of leveling down objection, and I interpret the egalitarianism of classification parity as deontic egalitarianism. I then discuss a challenge to this interpretation and suggest a revision. Finally, I examine how my interpretation provides proponents of classification parity with a response to the leveling down criticism and how it relates to a recent suggestion to evaluate fairness for automated decision-making systems based on risk and welfare considerations from behind a veil of ignorance. (shrink)

While some suggestions have been made as to how the duty to undertake humanitarian intervention should be assigned to specific states, the question of how to assign the duty to carry the economic a...

The objective of this paper is to develop a novel account of how the duty to undertake humanitarian intervention should be assigned to states. It takes as its point of departure two worries about the best existing answer to this question, namely: that it is insensitive to historical considerations, and that its distribution is unfair. Against this background I propose that the duty to intervene should be assigned to states based on the strength of their claim to reject the burden (...) of intervention. Specifically, I argue that the strength of such claims should be determined by ability to intervene effectively, adjusted for historical responsibility for the rights-violations taking place, and the extent to which their capacity for intervention derives from resources tainted by historical rights-violations. This approach, I will argue, can better accommodate the centrality of protecting the victims while also taking into account both historical and distributive concerns. (shrink)

Which harms and benefits should be viewed as relevant when considering whether to launch cyber-measures? In this article, we consider this question, which matters because it is central to determining whether cyber-measures should be launched. Several just war theorists hold a version of what we call the ‘Restrictive View’, according to which there are restrictions on the sorts of harms and benefits that should be included in proportionality assessments about the justifiability of going to war (whether cyber or kinetic). We (...) discuss two such views – the Just Cause Restrictive View and Rights-based Restrictive View – and find both wanting. By contrast, we defend what we call the ‘Permissive View’. This holds that all potential goods and bads should be included in proportionality decisions about cyber-measures, even those that appear to be trivial, and where the various harms and benefits are given different weights, according to their agent-relative and agent-neutral features. We argue further that accepting the Permissive View has broader implications for the ethical frameworks governing cyberwar, both in terms of whether cyberattack provide just cause for coercive responses, including kinetic warfare and cyber-responses, and whether cyber-measures should be governed by just war theory or a new theory for cyber-operations. (shrink)

Many authors in ethics, economics, and political science endorse the Lottery Requirement, that is, the following thesis: where different parties have equal moral claims to one indivisible good, it is morally obligatory to let a fair lottery decide which party is to receive the good. This article defends skepticism about the Lottery Requirement. It distinguishes three broad strategies of defending such a requirement: the surrogate satisfaction account, the procedural account, and the ideal consent account, and argues that none of these (...) strategies succeed. The article then discusses and discharges some remaining grounds for resistance to these skeptical conclusions, as well as the possibility of defending a weaker version of a normative lottery principle. The conclusion is that we have no reason to believe that where equal claims conflict, we are morally required to hold a lottery, as opposed to simply picking one of the parties on more subjective grounds or out of pure whim. In addition to the practical consequences of this skeptical view, the article sketches some theoretical implications for debates about saving the greater number and about axiomatic utilitarianism. (shrink)

We confront the philosophical literature on fair division problems with axiomatic and game-theoretic work in economics. Firstly, we show that the proportionality method advocated in Curtis is not implied by a general principle of fairness, and that the proportional rule cannot be explicated axiomatically from that very principle. Secondly, we suggest that Broome’s notion of claims is too restrictive and that game-theoretic approaches can rectify this shortcoming. More generally, we argue that axiomatic and game-theoretic work in economics is an indispensable (...) ingredient of any theorizing about fair division problems and allocative justice. (shrink)

Philosophical theories of fairness propose to divide a good that several individuals have a claim to in proportion to the strength of their respective claims. We suggest that currently, these theories face a dilemma when dealing with a good that is indivisible. On the one hand, theories of fairness that use weighted lotteries are either of limited applicability or fall prey to an objection by Brad Hooker. On the other hand, accounts that do without weighted lotteries fall prey to three (...) fairness paradoxes. We demonstrate that division methods from apportionment theory, which has hitherto been ignored by philosophical theories of fairness, can be used to provide fair division for indivisible goods without weighted lotteries and without fairness paradoxes. (shrink)

The numbers problem concerns the question of what is the right thing to do in trade-off cases where one can save different non-overlapping groups of persons, but not everyone. Proponents of mixed solutions argue that both saving the many and holding a lottery to determine whom to save can each be morally right in such cases, depending on the relative sizes of the groups involved. In his book The Dimensions of Consequentialism, Martin Peterson presents an ingenious version of such an (...) approach that avoids a commitment to interpersonal value aggregation, which is highly controversial and rejected by many philosophers for a number of reasons. I criticise Peterson’s proposal by first arguing that it cannot account for the idea that holding a lottery is morally wrong if differences in numbers are very large, and by second pointing out that it relies on implausible assumptions about what is good for an individual. Given the shortcomings of Peterson’s non-aggregationist version of a mixed solution, I next address the issue of how problematic a commitment to interpersonal aggregation really is. To this end, I present an aggregationist version of a mixed solution that is reason-based and bypasses most standard objections to interpersonal value aggregation. I conclude that although there is reason to be optimistic, it remains to be seen whether a mixed solution can be worked out in a fully satisfying way. (shrink)

In the recent literature across philosophy, medicine and public health policy, many influential arguments have been put forward to support the use of randomization procedures to allocate scarce life-saving resources. In this paper, I provide a systematic categorization and a critical evaluation of these arguments. I shall argue that those arguments justify using RAND to allocate SLSR in fewer cases than their proponents maintain and that the relevant decision-makers should typically allocate SLSR directly to the individuals with the strongest claims (...) to these resources rather than use RAND to allocate such resources. (shrink)

The difference between the unity of the individual and the separateness of persons requires that there be a shift in the moral weight that we accord to changes in utility when we move from making intrapersonal tradeoffs to making interpersonal tradeoffs. We examine which forms of egalitarianism can, and which cannot, account for this shift. We argue that a form of egalitarianism which is concerned only with the extent of outcome inequality cannot account for this shift. We also argue that (...) a view which is concerned with both outcome inequality and with the unfairness of inequality in individuals‘ expected utilities can account for this shift. Finally, we limn an alternative view, on.. (shrink)

We ask why pharmacological cognitive enhancement (PCE) is generally deemed morally unacceptable by lay people. Our approach to this question has two core elements. First, we employ an interdisciplinary perspective, using philosophical rationales as base for generating psychological models. Second, by testing these models we investigate how different normative judgments on PCE are related to each other. Based on an analysis of the relevant philosophical literature, we derive two psychological models that can potentially explain the judgment that PCE is unacceptable: (...) the “Unfairness-Undeservingness Model” and the “Hollowness-Undeservingness Model.” The Unfairness-Undeservingness Model holds that people judge PCE to be unacceptable because they take it to produce unfairness and to undermine the degree to which PCE-users deserve reward. The Hollowness-Undeservingness Model assumes that people judge PCE to be unacceptable because they find achievements realized while using PCE hollow and undeserved. We empirically test both models against each other using a regression-based approach. When trying to predict judgments regarding the unacceptability of PCE using judgments regarding unfairness, hollowness, and undeservingness, we found that unfairness judgments were the only significant predictor of the perceived unacceptability of PCE, explaining about 36% of variance. As neither hollowness nor undeservingness had explanatory power above and beyond unfairness, the Unfairness-Undeservingness Model proved superior to the Hollowness-Undeservingness Model. This finding also has implications for the Unfairness-Undeservingness Model itself: either a more parsimonious single-factor “Fairness Model” should replace the Unfairness-Undeservingness-Model or fairness fully mediates the relationship between undeservingness and unacceptability. Both explanations imply that participants deemed PCE unacceptable because they judged it to be unfair. We conclude that concerns about unfairness play a crucial role in the subjective unacceptability of PCE and discuss the implications of our approach for the further investigation of the psychology of PCE. (shrink)

This article investigates the relationship between distributive injustice and political obligation within the confines of the fair-play theory of political obligation. More specifically, it asks ho...

This article investigates the relationship between distributive injustice and political obligation within the confines of the fair-play theory of political obligation. More specifically, it asks how the distribution of benefits and burdens of a cooperative scheme affects people’s fair-play obligations to that scheme. It argues that neither a sufficiency-based nor a proportionality-based approach is capable of answering that question singlehandedly. However, the two approaches can be combined in a plausible way. Noting that some of the duties that go into our (...) fair-play obligations are ‘gradual’ and others are ‘non-gradual’, the article argues that the sufficiency-based approach explains non-gradual duties such as the duty to obey the law, and that the proportionality-based approach explains gradual duties like political participation and civic responsibility. Thus, the article contends that the fair-play theory should be extended beyond the question of legal compliance and be applied to the duties of good citizenship more generally. (shrink)

In this article I criticize the non-consequentialist Weighted Lottery (WL) solution to the choice between saving a smaller or a larger group of people. WL aims to avoid what non-consequentialists see as consequentialism's unfair aggregation by giving equal consideration to each individual's claim to be rescued. In so doing, I argue, WL runs into another common objection to consequentialism: it is excessively demanding. WL links the right action with the outcome of a fairly weighted lottery, which means that an agent (...) can only act rightly if s/he has actually run the lottery. In many actual cases, this involves epistemic demands that can be almost impossible to meet. I argue that plausible moral principles cannot make such extreme epistemic demands. (shrink)