In this article I explore two accounts of the Fallacy of Many Questions made famous by the question "Have you stopped beating your wife?" The accounts are from the works of Lennart Aqvist and Noel Belnap, and the two authors differ in their accounts of the fallacy. Then I give my own account based on understanding a facet of erotetic logic, i. e., the logic of questions.

The joy and significance of serendipity in retail: A report from Faire.

Three recent research reports by Topping and Trickey, by Fair and colleagues, and by Gorard, Siddiqui and Huat See have produced data that support the conclusion that a Philosophy for Children program of one-hour-per-week structured discussions has a marked positive impact on students. This article presents data from a follow up study done three years after the completion of the study reported in Fair et al.. The data show that the positive gains in scores on the Cognitive Abilities Test were (...) still present and had not faded after three years. Given the strength of these confirmations of the positive durable impact of the P4C program of structured discussions and given the relatively low cost of implementing the P4C program, it is recommended that it become a standard part of the school curriculum. (shrink)

This article presents a fairness principle for evaluating decision-making based on predictions: a decision rule is unfair when the individuals directly impacted by the decisions who are equal with respect to the features that justify inequalities in outcomes do not have the same statistical prospects of being benefited or harmed by them, irrespective of their socially salient morally arbitrary traits. The principle can be used to evaluate prediction-based decision-making from the point of view of a wide range of antecedently (...) specified substantive views about justice in outcome distributions. (shrink)

Statistical fairness criteria are widely used for diagnosing and ameliorating algorithmic bias. However, these fairness criteria are controversial as their use raises several difficult questions. I argue that the major problems for statistical algorithmic fairness criteria stem from an incorrect understanding of their nature. These criteria are primarily used for two purposes: first, evaluating AI systems for bias, and second constraining machine learning optimization problems in order to ameliorate such bias. The first purpose typically involves treating each (...) criterion as a necessary condition for fairness. The second use involves treating criteria as sufficient conditions for fairness. Since the criteria are used for both roles, some researchers have treated them as both necessary and sufficient conditions, i.e., as definitions of algorithmic fairness. However, serious problems have been raised for the use of these fairness criteria. Under ordinary circumstances, it is impossible to satisfy multiple criteria at the same time. Moreover, there are counterexamples to both the sufficiency and necessity for fairness of each criterion. I argue that we should instead understand fairness criteria as merely providing evidence of fairness. In other words, satisfaction (or violation) of these criteria should be understood as potential evidence of fairness (or bias). Whether a criterion counts as evidence in a particular case will depend on stakeholders' background knowledge and the specific features of the system's task. This evidence account of fairness conditions provides guidance for recognizing both the appropriate uses and the limitations of fairness criteria. (shrink)

In his paper “Fairness, Political Obligation, and the Justificatory Gap” (published in the Journal of Moral Philosophy), Jiafeng Zhu argues that the principle of fair play cannot require submission to the rules of a cooperative scheme, and that when such submission is required, the requirement is grounded in consent. I propose a better argument for the claim that fair play requires submission to the rules than the one Zhu considers. I also argue that Zhu’s attribution of consent to people (...) commonly thought to be bound to follow the rules by a duty of fair play is implausible. (shrink)

John Broome has proposed a theory of fairness according to which fairness requires that agents’ claims to goods be satisfied in proportion to the relative strength of those claims. In the case of competing claims for a single indivisible good, Broome argues that what fairness requires is the use of a weighted lottery as a surrogate to satisfying the competing claims: the relative chance of each claimant's winning the lottery should be set to the relative strength of (...) each claimant's claim. In this journal, James Kirkpatrick and Nick Eastwood have objected that the use of weighted lotteries in the case of indivisible goods is unacceptable. In this article, I explain why Kirkpatrick and Eastwood's objection misses its mark. (shrink)

This research investigates 3- and 5-year-olds' relative fairness in distributing small collections of even or odd numbers of more or less desirable candies, either with an adult experimenter or between two dolls. The authors compare more than 200 children from around the world, growing up in seven highly contrasted cultural and economic contexts, from rich and poor urban areas, to small-scale traditional and rural communities. Across cultures, young children tend to optimize their own gain, not showing many signs of (...) self-sacrifice or generosity. Already by 3 years of age, self-optimizing in distributive justice is based on perspective taking and rudiments of mind reading. By 5 years, overall, children tend to show more fairness in sharing. What varies across cultures is the magnitude of young children's self-interest. More fairness (less self-interest) in distributive justice is evident by children growing up in small-scale urban and traditional societies thought to promote more collective values. (shrink)

Considerations of dynamic inconsistency have figured prominently in debates over the rationality of preferences that violate the separability conditions characteristic of expected utility theory. These debates have mostly focused on risk- and ambiguity averse preferences, but analogous considerations apply to preferences for fairness. We revisit these debates in the context of a specific hypothesis regarding the violations of separability by such preferences, namely that they are potentially both explained and rationalised by attitudes to the chances of goods that motivate (...) a preference for equality in the distribution of chances across people and possible events. Our main aim is to argue that, first, when these violations of static separability are motivated by such attitudes to chances, then they need not result in dynamically inconsistent behaviour. Second, and more generally, considerations of dynamic consistency do not, we argue, undermine the rationality of such attitudes towards the distribution of chances, despite the fact that such attitudes give rise to violations of the static separability assumptions of expected utility theory. (shrink)

From the school yard to the workplace, there’s no charge more damning than “you’re being unfair!” Born out of democracy and raised in open markets, fairness has become our de facto modern creed. The very symbol of American ethics—Lady Justice—wears a blindfold as she weighs the law on her impartial scale. In our zealous pursuit of fairness, we have banished our urges to like one person more than another, one thing over another, hiding them away as dirty secrets (...) of our humanity. In Against Fairness, polymath philosopher Stephen T. Asma drags them triumphantly back into the light. Through playful, witty, but always serious arguments and examples, he vindicates our unspoken and undeniable instinct to favor, making the case that we would all be better off if we showed our unfair tendencies a little more kindness—indeed, if we favored favoritism. Conscious of the egalitarian feathers his argument is sure to ruffle, Asma makes his point by synthesizing a startling array of scientific findings, historical philosophies, cultural practices, analytic arguments, and a variety of personal and literary narratives to give a remarkably nuanced and thorough understanding of how fairness and favoritism fit within our moral architecture. Examining everything from the survival-enhancing biochemistry that makes our mothers love us to the motivating properties of our “affective community,” he not only shows how we favor but the reasons we should. Drawing on thinkers from Confucius to Tocqueville to Nietzsche, he reveals how we have confused fairness with more noble traits, like compassion and open-mindedness. He dismantles a number of seemingly egalitarian pursuits, from classwide Valentine’s Day cards to civil rights, to reveal the envy that lies at their hearts, going on to prove that we can still be kind to strangers, have no prejudice, and fight for equal opportunity at the same time we reserve the best of what we can offer for those dearest to us. Fed up with the blue-ribbons-for-all absurdity of "fairness" today, and wary of the psychological paralysis it creates, Asma resets our moral compass with favoritism as its lodestar, providing a strikingly new and remarkably positive way to think through all our actions, big and small. (shrink)

This report by the WHO Consultative Group on Equity and Universal Health Coverage addresses how countries can make fair progress towards the goal of universal coverage. It explains the relevant tradeoffs between different desirable ends and offers guidance on how to make these tradeoffs.

Holm (2022) argues that a class of algorithmic fairness measures, that he refers to as the ‘performance parity criteria’, can be understood as applications of John Broome’s Fairness Principle. We argue that the performance parity criteria cannot be read this way. This is because in the relevant context, the Fairness Principle requires the equalization of actual individuals’ individual-level chances of obtaining some good (such as an accurate prediction from a predictive system), but the performance parity criteria do (...) not guarantee any such thing: the measures merely ensure that certain population-level ratios hold. (shrink)

The debate over whether ‘fair-play’ can serve as a justification for legal punishment has recently resumed with an exchange between Richard Dagger and Antony Duff. According to the fair-play theorist, criminals deserve punishment for breaking the law because in so doing the criminal upsets a fair distribution of benefits and burdens, and punishment rectifies this unfairness. Critics frequently level two charges against this idea. The first is that it often gives the wrong explanation of what makes crime deserving of punishment, (...) since the wrongfulness of murder is not primarily about unfairness. The second is that it implies that all crimes deserve the same degree of punishment, because all crimes create the same degree of unfairness. These objections are viewed as revealing fatal flaws in the theory. Although Dagger attempts to meet these objections by drawing on political theory, Duff responds that this still draws upon the wrong kind of resources for meeting these objections. This paper argues that these two objections rest on a crucial mistake that has been overlooked by both the defenders and critics of fair-play. This mistake results from failing to distinguish between what justifies punishment as a response to crime (which requires a common element to all crime) and what justifies attaching particular penalties to crimes (which requires making distinctions in the severity of crime). The arguments presented will give reasons to consider fair-play as a viable justification for legal punishment. (shrink)

There is an emerging consensus among political philosophers that state legitimacy involves something more than—or perhaps other than—political obligation. Yet the principle of fair play, which many take to be a promising basis for political obligation, has been largely absent from discussions of the revised conception of legitimacy. This paper shows how the principle of fair play can generate legitimate political authority by drawing on a neglected feature of the principle—its stipulation that members of a cooperative scheme must reciprocate specifically (...) by submitting to the scheme's rules. (shrink)

Although the principle of fair subject selection is a widely recognized requirement of ethical clinical research, it often yields conflicting imperatives, thus raising major ethical dilemmas regarding participant selection. In this paper, we diagnose the source of this problem, arguing that the principle of fair subject selection is best understood as a bundle of four distinct sub-principles, each with normative force and each yielding distinct imperatives: (1) fair inclusion; (2) fair burden sharing; (3) fair opportunity; and (4) fair distribution of (...) third-party risks. We first map out these distinct sub-principles, and then identify the ways in which they yield conflicting imperatives for the design of inclusion and exclusion criteria, and the recruitment of participants. We then offer guidance for how decision makers should navigate these conflicting imperatives to ensure that participants are selected fairly. (shrink)

Distribution of responsibility is one of the main focus areas in discussions about climate change ethics. Most of these discussions deal with the distribution of responsibility for climate change mitigation at the international level. The aim of this paper is to investigate if and how these principles can be used to inform the search for a fair distribution of responsibility for climate change adaptation on the local level. We found that the most influential distribution principles on the international level were (...) in turn built on one or more of seven basic principles: (P1) equal shares, (P2) desert, (P3) beneficiary pays, (P4) ability, (P5) self-help, (P6) limited responsibility for the worst off, and (P7) status quo preservation. It was found that all the basic principles, but P1, P3, and P7, are to some extent translatable to local climate adaptation. Two major problems hamper their usefulness on the local level: (1) several categories of agents need to take on responsibility; and (2) emissions do not work as a base for all principles. P4, P5, and P6 are applicable to local adaptation without changes. P4 is of particular importance as it seems to solve the first problem. P2 is applicable only if the second problem is solved, which can be achieved by using risk of harm instead of emissions as the basis for desert. (shrink)

Inspired by recent breakthroughs in predictive modeling, practitioners in both industry and government have turned to machine learning with hopes of operationalizing predictions to drive automated decisions. Unfortunately, many social desiderata concerning consequential decisions, such as justice or fairness, have no natural formulation within a purely predictive framework. In efforts to mitigate these problems, researchers have proposed a variety of metrics for quantifying deviations from various statistical parities that we might expect to observe in a fair world and offered (...) a variety of algorithms in attempts to satisfy subsets of these parities or to trade o the degree to which they are satised against utility. In this paper, we connect this approach to fair machine learning to the literature on ideal and non-ideal methodological approaches in political philosophy. The ideal approach requires positing the principles according to which a just world would operate. In the most straightforward application of ideal theory, one supports a proposed policy by arguing that it closes a discrepancy between the real and the perfectly just world. However, by failing to account for the mechanisms by which our non-ideal world arose, the responsibilities of various decision-makers, and the impacts of proposed policies, naive applications of ideal thinking can lead to misguided interventions. In this paper, we demonstrate a connection between the fair machine learning literature and the ideal approach in political philosophy, and argue that the increasingly apparent shortcomings of proposed fair machine learning algorithms reflect broader troubles faced by the ideal approach. We conclude with a critical discussion of the harms of misguided solutions, a reinterpretation of impossibility results, and directions for future research. (shrink)

Typically, fair machine learning research focuses on a single decision maker and assumes that the underlying population is stationary. However, many of the critical domains motivating this work are characterized by competitive marketplaces with many decision makers. Realistically, we might expect only a subset of them to adopt any non-compulsory fairness-conscious policy, a situation that political philosophers call partial compliance. This possibility raises important questions: how does partial compliance and the consequent strategic behavior of decision subjects affect the allocation (...) outcomes? If k% of employers were to voluntarily adopt a fairness-promoting intervention, should we expect k% progress (in aggregate) towards the benefits of universal adoption, or will the dynamics of partial compliance wash out the hoped-for benefits? How might adopting a global (versus local) perspective impact the conclusions of an auditor? In this paper, we propose a simple model of an employment market, leveraging simulation as a tool to explore the impact of both interaction effects and incentive effects on outcomes and auditing metrics. Our key findings are that at equilibrium: (1) partial compliance by k% of employers can result in far less than proportional (k%) progress towards the full compliance outcomes; (2) the gap is more severe when fair employers match global (vs local) statistics; (3) choices of local vs global statistics can paint dramatically different pictures of the performance vis-a-vis fairness desiderata of compliant versus non-compliant employers; (4) partial compliance based on local parity measures can induce extreme segregation. Finally, we discuss implications for auditors and insights concerning the design of regulatory frameworks. (shrink)

Among the most popular strategies for justifying political obligations are those that appeal to the principle of fairness. These theories face the challenge, canonically articulated by Robert Nozick, of explaining how it is that persons are obligated to schemes when they receive goods that they do not ask for but cannot reject. John Simmons offers one defense of the principle of fairness, arguing that people could be bound by obligations of fairness if they voluntarily accept goods produced (...) by a cooperative scheme. Simmons, however, thinks that such a theory will do little work in justifying political obligations, since virtually no one voluntarily accepts state goods. This paper attempts to advance just such a theory by arguing that states are in fact genuine cooperative schemes and that Simmons is overly pessimistic in his appraisal of whether the majority of citizens accept the goods provided by their states. (shrink)

Algorithms can now identify patterns and correlations in the (big) datasets, and predict outcomes based on those identified patterns and correlations with the use of machine learning techniques and big data, decisions can then be made by algorithms themselves in accordance with the predicted outcomes. Yet, algorithms can inherit questionable values from the datasets and acquire biases in the course of (machine) learning, and automated algorithmic decision-making makes it more difficult for people to see algorithms as biased. While researchers have (...) taken the problem of algorithmic bias seriously, but the current discussion on algorithmic fairness tends to conceptualize ‘fairness’ in algorithmic fairness primarily as a technical issue and attempts to implement pre-existing ideas of ‘fairness’ into algorithms. In this paper, I show that such a view of algorithmic fairness as technical issue is unsatisfactory for the type of problem algorithmic fairness presents. Since decisions on fairness measure and the related techniques for algorithms essentially involve choices between competing values, ‘fairness’ in algorithmic fairness should be conceptualized first and foremost as a political issue, and it should be (re)solved by democratic communication. The aim of this paper, therefore, is to explicitly reconceptualize algorithmic fairness as a political question and suggest the current discussion of algorithmic fairness can be strengthened by adopting the accountability for reasonableness framework. (shrink)

Rawlsian justice as fairness is neither fundamentally luck egalitarian nor relational egalitarian. Rather, the most fundamental idea is that of society as a fair system of cooperation. Collective pensions provide a case study which illustrates the fruitfulness of conceiving justice in these latter terms. Those who have recently reached the age of majority do not now know how long they will live in retirement or how well any investments they try to save up for their retirement would fare. From (...) the perspective of the beginning of their working lives, it is therefore rational for each to enter into an agreement with others, who also do not yet know their fates, that, if one turns out to be among the unfortunate whose private pension pots would not have yielded enough for one’s retirement, one will receive much more in retirement, whereas those whose pension pots would have overflowed their retirements will receive somewhat less. These terms are to each person’s expected advantage, which is made possible by a fair sharing of the fruits of social cooperation which arise through the efficiencies reaped by the pooling of the risk of outliving what one could save for one’s retirement on one’s own. It is rational for each to agree to share one another’s fates by pooling risks across both space and time, on fair terms of social cooperation for mutual advantage. Even when collective pensions arise from, and are proportional to, a baseline of unequal income, they can be defended on grounds of reciprocity involving regard for one another as equals. (shrink)

Universal health coverage (UHC) is at the center of current efforts to strengthen health systems and improve the level and distribution of health and health services. This document is the final report of the WHO Consultative Group on Equity and Universal Health Coverage. The report addresses the key issues of fairness and equity that arise on the path to UHC. As such, the report is relevant for every actor that affects that path and governments in particular, as they are (...) in charge of overseeing and guiding the progress toward UHC. (shrink)

To address the rising concern that algorithmic decision-making may reinforce discriminatory biases, researchers have proposed many notions of fairness and corresponding mathematical formalizations. Each of these notions is often presented as a one-size-fits-all, absolute condition; however, in reality, the practical and ethical trade-offs are unavoidable and more complex. We introduce a new approach that considers fairness—not as a binary, absolute mathematical condition—but rather, as a relational notion in comparison to alternative decisionmaking processes. Using US mortgage lending as an (...) example use case, we discuss the ethical foundations of each definition of fairness and demonstrate that our proposed methodology more closely captures the ethical trade-offs of the decision-maker, as well as forcing a more explicit representation of which values and objectives are prioritised. (shrink)

This article discusses how the concept of a fair finite lottery can best be extended to denumerably infinite lotteries. Techniques and ideas from non-standard analysis are brought to bear on the problem.

If dangerous climate change is to be avoided, it is vital that carbon sinks such as tropical rainforests are protected. But protecting them has costs. These include opportunity costs: the potential economic benefits which those who currently control rainforests have to give up when they are protected. But who should bear those costs? Should countries which happen to have rainforests within their territories sacrifice their own economic development, because of our broader global interests in protecting key carbon sinks? This essay (...) develops an argument from the “principle of fairness,” which seeks to establish that outsiders should pay states with rainforests so as to share the costs of protection. If they do not, they can be condemned for free-riding on forest states. The argument is, I suggest, compelling and also capable of enjoying support from adherents of a wide variety of positions on global justice. (shrink)

This chapter provides a critical overview and interpretation of fair subject selection in clinical and social scientific research. It first provides an analytical framework for thinking about the problem of fair subject selection. It then argues that fair subject selection is best understood as a set of four subprinciples, each with normative force and each with distinct and often conflicting implications for the selection of participants: fair inclusion, fair burden sharing, fair opportunity, and fair distribution of third-party risks. It then (...) defends an approach to navigating the conflicting imperatives of these subprinciples, one which privileges the need to include epistemically distinct groups in research, before concluding by considering the most pressing research questions regarding fair subject selection. (shrink)

Severe poverty is a major global problem about risk and inequality. What, if any, is the relationship between equality, fairness and responsibility in an unequal world? I argue for four conclusions. The first is the moral urgency of severe poverty. We have too many global neighbours that exist in a state of emergency and whose suffering is intolerable. The second is that severe poverty is a problem concerning global injustice that is relevant, but not restricted, to questions about responsibility. (...) If none were responsible, this does not eliminate all compelling claims to provide assistance. The third is that severe poverty represents an inequality too far; it is a condition of extremity with denial of basic needs. The fourth is that there is a need for an approach that captures all relevant cases – and the capabilities approach and the connection theory of remedial responsibilities are highlighted as having special promise. (shrink)

Garrett Cullity contends that fairness is appropriate impartiality (See Cullity (2004) Chapters 8 and 10 and Cullity (2008)). Cullity deploys his account of fairness as a means of limiting the extreme moral demand to make sacrifices in order to aid others that was posed by Peter Singer in his seminal article ‘Famine, Affluence and Morality’. My paper is founded upon the combination of (1) the observation that the idea that fairness consists in appropriate impartiality is very vague (...) and (2) the fact that psychological studies show the self-serving bias is especially likely to infect one’s judgements when the ideas involved are vague. I argue that Cullity’s solution to extreme moral demandingness is threatened by these findings. I then comment on whether some other theories of fairness are vulnerable to the same objection. (shrink)

Is the pool of fair predictive algorithms fair? It depends, naturally, on both the criteria of fairness and on how we pool. We catalog the relevant facts for some of the most prominent statistical criteria of algorithmic fairness and the dominant approaches to pooling forecasts: linear, geometric, and multiplicative. Only linear pooling, a format at the heart of ensemble methods, preserves any of the central criteria we consider. Drawing on work in the social sciences and social epistemology on (...) the theoretical foundations of the wisdom of crowds, we explain how our observations present an exception to the general trend of finding tradeoffs between the accuracy and fairness of forecasts. (shrink)

This essay explores a conception of responsibility at work in moral and criminal responsibility. Our conception draws on work in the compatibilist tradition that focuses on the choices of agents who are reasons-responsive and work in criminal jurisprudence that understands responsibility in terms of the choices of agents who have capacities for practical reason and whose situation affords them the fair opportunity to avoid wrongdoing. Our conception brings together the dimensions of normative competence and situational control, and we factor normative (...) competence into cognitive and volitional capacities, which we treat as equally important to normative competence and responsibility. Normative competence and situational control can and should be understood as expressing a common concern that blame and punishment presuppose that the agent had a fair opportunity to avoid wrongdoing. This fair opportunity is the umbrella concept in our understanding of responsibility, one that explains it distinctive architecture. (shrink)

The principle of fairness suggests that it is wrong for free riders to enjoy cooperative benefits without also helping to produce them. Considerations of fairness are a familiar part of moral experience, yet there is a great deal of controversy as to the conditions of their application. The primary debate concerns whether cooperative benefits need to be voluntarily accepted. Many argue that acceptance is unnecessary because such theories are too permissive and acceptance appears to be absent in a (...) variety of cases where considerations of fairness nevertheless seem applicable. In this paper, I defend the claim that acceptance is necessary by suggesting that these worries can be disarmed, and that theories that deny the necessity of acceptance face deep challenges in articulating an understanding of cooperative benefits that does not also appeal to a person’s acceptance of them. (shrink)

Glucagon-like peptide-1 (GLP-1) receptor agonists, such as semaglutide, and dual GLP-1 and glucose-dependent insulinotropic polypeptide (GIP) receptor agonists, such as tirzepatide, have been found to be effective for treating obesity and diabetes, significantly reducing weight and the risk or predicted risk of adverse cardiovascular events. There is a global shortage of these medications that could last several years and raises questions about how limited supplies should be allocated. We propose a fair-allocation framework that enables evaluation of the ethics of current (...) practices and could guide governments, professional societies, and physicians in allocation decisions. (shrink)

Social movements often impose nontrivial costs on others against their wills. Civil disobedience is no exception. How can social movements in general, and civil disobedience in particular, be justifiable despite this apparent wrong-making feature? We examine an intuitively plausible account—it is fair that everyone should bear the burdens of tackling injustice. We extend this fairness-based argument for civil disobedience to defend some acts of uncivil disobedience. Focusing on uncivil environmental activism—such as ecotage (sabotage with the aim of protecting the (...) environment)—we argue that some acts of uncivil disobedience can be morally superior to their civil counterparts, when and because such acts target people who are responsible for environmental threats. Indeed, insofar as some acts of uncivil disobedience can more accurately target responsible people, they can better satisfy the demands of fairness compared to their civil counterparts. In some circumstances, our argument may require activists to engage in uncivil disobedienceeven whencivil disobedience is available. (shrink)

The principle of fair play is widely thought to require simply that costs and benefits be distributed fairly. This gloss on the principle, while not entirely inaccurate, has invited a host of popular objections based on misunderstandings about fair play. Central to many of these objections is a failure to treat the principle of fair play as a transactional principle—one that allocates special obligations and rights among persons as a result of their interactions. I offer an interpretation of the principle (...) of fair play that emphasizes its similarities to another transactional principle: consent. This interpretation reveals that playing fair requires one to reciprocate specifically by following the rules of the cooperative scheme from which one benefits, just as consent requires one to act according to the terms of an agreement. I then draw on the comparison with consent to reply to some popular and persistent objections to the principle. (shrink)

Although existing work draws attention to a range of obstacles in realizing fair AI, the field lacks an account that emphasizes how these worries hang together in a systematic way. Furthermore, a review of the fair AI and philosophical literature demonstrates the unsuitability of ‘treat like cases alike’ and other intuitive notions as conceptions of fairness. That review then generates three desiderata for a replacement conception of fairness valuable to AI research: (1) It must provide a meta-theory for (...) understanding tradeoffs, entailing that it must be flexible enough to capture diverse species of objection to decisions. (2) It must not appeal to an impartial perspective (neutral data, objective data, or final arbiter.) (3) It must foreground the way in which judgments of fairness are sensitive to context, i.e., to historical and institutional states of affairs. We argue that a conception of fairness as appropriate concession in the historical iteration of institutional decisions meets these three desiderata. On the basis of this definition, we organize the insights of commentators into a process-structure map of the ethical territory that we hope will bring clarity to computer scientists and ethicists analyzing Fair AI while clearing some ground for further technical and philosophical work. (shrink)

The “impossibility results” in algorithmic fairness suggest that a predictive model cannot fully meet two common fairness criteria – sufficiency and separation – except under extraordinary circumstances. These findings have sparked a discussion on fairness in algorithms, prompting debates over whether predictive models can avoid unfair discrimination based on protected attributes, such as ethnicity or gender. As shown by Otto Sahlgren, however, the discussion of the impossibility results would gain from importing some of the tools developed in (...) the philosophical literature on feasibility. Utilizing these tools, Sahlgren sketches a cautiously optimistic view of how algorithmic fairness can be made feasible in restricted local decision-making. While we think it is a welcome move to inject the literature on feasibility into the debate on algorithmic fairness, Sahlgren says very little about what are the general gains of bringing in feasibility considerations in theorizing algorithmic fairness. How, more precisely, does it help us make assessments about fairness in algorithmic decision-making? This is what is addressed in this Reply. More specifically, our two-fold argument is that feasibility plays an important but limited role for algorithmic fairness. We end by offering a sketch of a framework, which may be useful for theorizing feasibility in algorithmic fairness. (shrink)

Recently, there has been much discussion of ‘fair machine learning’: fairness in data-driven decision-making systems (which are often, though not always, made with assistance from machine learning systems). Notorious impossibility results show that we cannot have everything we want here. Such problems call for careful thinking about the foundations of fair machine learning. Sune Holm has identified one promising way forward, which involves applying John Broome's theory of fairness to the puzzles of fair machine learning. Unfortunately, his application (...) of Broome's theory appears to be fatally flawed. This article attempts to rescue Holm's central insight – namely, that Broome's theory can be useful to the study of fair machine learning – by giving an alternative application of Broome's theory, which involves thinking about fair machine learning in counterfactual (as opposed to merely statistical) terms. (shrink)

Some moral principles require agents to do more than their fair share of a common task, if others won’t do their fair share – each agent’s fair share being what they would be required to do if all contributed as they should. This seems to provide a strong basis for objecting to such principles. For it seems unfair to require agents who have already done their fair share to do more, just because other agents won’t do their fair share. The (...) philosopher who has written most about this issue, however, Liam Murphy, argues that it is not unfair to do so, at least in the standard sense of that term. In this paper, I give Murphy’s reasons for saying this, explain why I think he’s wrong, and then say a little about why this issue might be important. (shrink)

In Free Market Fairness, John Tomasi tries to show that ‘thick’ economic liberties, including the right to own productive property, are basic liberties. According to Tomasi, the policy-level consequences of protecting economic liberty as basic are essentially libertarian in character. I argue that if economic liberties are basic, just societies must guarantee their fair value to all citizens. And in order to secure the fair value of economic liberty, states must guarantee that citizens of roughly similar dispositions and talents (...) are roughly equally able to use their economic liberties to develop and pursue a conception of the good. This, I will argue, is a very demanding standard that requires aggressive taxation and redistribution. (shrink)

Data-driven predictive algorithms are widely used to automate and guide high-stake decision making such as bail and parole recommendation, medical resource distribution, and mortgage allocation. Nevertheless, harmful outcomes biased against vulnerable groups have been reported. The growing research field known as 'algorithmic fairness' aims to mitigate these harmful biases. Its primary methodology consists in proposing mathematical metrics to address the social harms resulting from an algorithm's biased outputs. The metrics are typically motivated by -- or substantively rooted in -- (...) ideals of distributive justice, as formulated by political and legal philosophers. The perspectives of feminist political philosophers on social justice, by contrast, have been largely neglected. Some feminist philosophers have criticized the local scope of the paradigm of distributive justice and have proposed corrective amendments to surmount its limitations. The present paper brings some key insights of feminist political philosophy to algorithmic fairness. The paper has three goals. First, I show that algorithmic fairness does not accommodate structural injustices in its current scope. Second, I defend the relevance of structural injustices -- as pioneered in the contemporary philosophical literature by Iris Marion Young -- to algorithmic fairness. Third, I take some steps in developing the paradigm of 'responsible algorithmic fairness' to correct for errors in the current scope and implementation of algorithmic fairness. I close by some reflections of directions for future research. (shrink)

We introduce a fairness criterion that we call Spanning. Spanning i) is implied by Calibration, ii) retains interesting properties of Calibration that some other ways of relaxing that criterion do not, and iii) unlike Calibration and other prominent ways of weakening it, is consistent with Equalized Odds outside of trivial cases.

In this essay, Johannes Giesinger comments on the current philosophical debate on educational justice. He observes that while authors like Elizabeth Anderson and Debra Satz develop a so-called adequacy view of educational justice, Harry Brighouse and Adam Swift defend an egalitarian principle. Giesinger focuses his analysis on the main objection that is formulated, from an egalitarian perspective, against the adequacy view: that it neglects the problem of securing fair opportunities in the competition for social rewards. Giesinger meets this objection by (...) expressing two basic theses: First, he argues that Brighouse and Swift themselves fail to give an adequate account of fair competition; and, second, he shows that the adequacy view provides the theoretical resources to face this problem. (shrink)

One of the main lines of research in algorithmic fairness involves individual fairness (IF) methods. Individual fairness is motivated by an intuitive principle, similar treatment, which requires that similar individuals be treated similarly. IF offers a precise account of this principle using distance metrics to evaluate the similarity of individuals. Proponents of individual fairness have argued that it gives the correct definition of algorithmic fairness, and that it should therefore be preferred to other methods for (...) determining fairness. I argue that individual fairness cannot serve as a definition of fairness. Moreover, IF methods should not be given priority over other fairness methods, nor used in isolation from them. To support these conclusions, I describe four in-principle problems for individual fairness as a definition and as a method for ensuring fairness: (1) counterexamples show that similar treatment (and therefore IF) are insufficient to guarantee fairness; (2) IF methods for learning similarity metrics are at risk of encoding human implicit bias; (3) IF requires prior moral judgments, limiting its usefulness as a guide for fairness and undermining its claim to define fairness; and (4) the incommensurability of relevant moral values makes similarity metrics impossible for many tasks. In light of these limitations, I suggest that individual fairness cannot be a definition of fairness, and instead should be seen as one tool among several for ameliorating algorithmic bias. (shrink)

According to the principle that ‘ought’ implies ‘can’, it is never the case that you ought to do something you cannot do. While many accept this principle in some form, it also has its share of critics, and thus it seems desirable if an argument can be offered in its support. The aim of this paper is to examine a particular way in which the principle has been defended, namely, by appeal to considerations of fairness. In a nutshell, the (...) idea (due to David Copp) is that moral requirements we cannot comply with would be unfair, and there cannot be unfair moral requirements. I discuss several ways of spelling out the argument, and argue that all are unsatisfactory for a variety of reasons. (shrink)

If deployed, aerosol geoengineering (AG) could involve unfairness to both present and future parties. I discuss three broad risks of unfairness that an AG deployment policy might carry: (1) causing disproportionate harm to those least responsible for climate change, (2) burdening future parties with the costs and risks of AG, and (3) excluding some interested parties from contributing to AG decision-making. Yet despite these risks, it may be too hasty to reject AG deployment as a potential climate change policy. I (...) argue that since it is very unlikely that a completely fair climate change policy will be pursued, we have ethical reason to prefer some “incompletely fair” policy. Given various facts about our world, it might be the case that some AG policy is ethically preferable to many other feasible climate change policies, even if AG carries deeply problematic risks of unfairness. (shrink)

The increasing application of artificial intelligence (AI) to healthcare raises both hope and ethical concerns. Some advanced machine learning methods provide accurate clinical predictions at the expense of a significant lack of explainability. Alex John London has defended that accuracy is a more important value than explainability in AI medicine. In this article, we locate the trade-off between accurate performance and explainable algorithms in the context of distributive justice. We acknowledge that accuracy is cardinal from outcome-oriented justice because it helps (...) to maximize patients’ benefits and optimizes limited resources. However, we claim that the opaqueness of the algorithmic black box and its absence of explainability threatens core commitments of procedural fairness such as accountability, avoidance of bias, and transparency. To illustrate this, we discuss liver transplantation as a case of critical medical resources in which the lack of explainability in AI-based allocation algorithms is procedurally unfair. Finally, we provide a number of ethical recommendations for when considering the use of unexplainable algorithms in the distribution of health-related resources. (shrink)

The move from open outcry to electronic trading added another responsibility to futures exchanges—that of matching orders between buyers and sellers. Matching systems can affect the level and speed of price discovery, the distribution of revenue, as well as the level of price efficiency of a given market. Whether the matching system is procedurally fair is another important consideration. I argue that while FIFO (First In First Out) is a fair procedure in principle and is perceived as the default matching (...) system, it is not a fair procedure in practice. Likewise, while pro rata is a fair procedure in principle, it is not so in practice. Nevertheless, both FIFO and pro rata are relics of an open outcry system. Instead, I propose an alternative approach to matching systems that builds on the strengths of electronic trading—the ability to randomize in real time. I introduce random selection for service (RSS) as a matching system that is procedurally fair both in principle and in practice. (shrink)

Sam Berstler defends a general moral advantage for misleading over lying by arguing that liars, but not misleaders, act unfairly toward the other members of their linguistic community. This article spells out three difficulties for Berstler’s account. First, though Berstler aims to avoid an error theory, it is dubitable that her account fits with intuitions on the matter. Second, there are some lies that do not exhibit the unfairness Berstler identifies. Third, fairness is not the only morally relevant difference (...) between lying and misleading. (shrink)

In a paper recently published in this journal, Navin and Largent argue in favour of a type of policy to regulate non-medical exemptions from childhood vaccination which they call ‘Inconvenience’. This policy makes it burdensome for parents to obtain an exemption to child vaccination, for example, by requiring parents to attend immunization education sessions and to complete an application form to receive a waiver. Navin and Largent argue that this policy is preferable to ‘Eliminationism’, i.e. to policies that do not (...) allow non-medical exemptions, because Inconvenience has been shown to maintain exemption rates low while not harming parents by forcing them to do something that goes against their beliefs. We argue that it is at least doubtful that Inconvenience is ethically preferable to Eliminationism: while the latter disregards the value of liberty, Inconvenience disregards the value of fairness in the distribution of the burdens entailed by the preservation of a public good like herd immunity. We propose a variant of Inconvenience, which we call ‘Contribution’, which we think is preferable to the versions of Inconvenience discussed by Navin and Largent in that it successfully strikes a balance between the values of parents’ liberty, fairness and expected utility. (shrink)