Are lotteries fair and if so, when should they be used?

I INTRODUCTION

This paper deals with the issue of allocation processes, especially when there is not enough good for all, generally speaking. Some remits, societies are responsible for are to select conscripts for military service, to decide whom to give access to medical treatment, like transplanting organs, or how to allocate study places. When there are sufficient resources (goods) to satisfy all potential claimants, the implementation of some allocating processes might be quite simple and fair, but what does distributive justice require when resources are scarce? When should we discriminate between people, i.e. decide on basis of special considerations, and when is it better to choose randomly? Philosophers have provided different theories of distributive justice and, in particular, argue in distinct and partly conflicting manners about the use of lotteries. This paper will focus on the following key question, which is not only relevant for individuals and everyday-life decisions, but also in very broader terms for political, societal or economic questions concerning distributive fairness, with probably global impacts:

Are lotteries fair and if so, when should they be used?

In section II, I will clarify what lotteries are and what it means to implement lotteries in allocation processes. Subsequently, I will first present and critically analyze the position that lotteries are sometimes required by fairness, suggested by John Broome (section III). In contrast to that, I will describe and reflect the view that lotteries are never required by fairness, proposed by Tim Henning (section IV). Section V consists of a brief conclusion to Broome and Henning, a final answer to the key question and some practical implications of the results.

II WHAT ARE LOTTERIES?

A lottery is a process, which can generate several possible outcomes, while nobody involved knows who will win and who will lose. Some examples are rolling a dice, tossing a coin or drawing balls from an urn. Applied to allocating processes, this means that the allocator and the claimants do not know in advance who will receive the good and who will not. Thus, a lottery ensures that nobody can be preferred on grounds of special considerations, under the veil of ignorance. Depending on the purpose, the allocator first decides which probabilities to assign to the outcomes before starting to select. (Stone 2007: 279 f.)

III LOTTERIES ARE REQUIRED BY FAIRNESS: JOHN BROOME

John Broome provides a justification for selecting people randomly (which in this context is tantamount to using lotteries) in allocative decision-making and suggests a quite intuitive theory of fairness. Therefore, Broome shows that lotteries are not only a cheap and effective procedure, but additionally and more importantly, that they are fair, even if special considerations about people, as race, gender, social worth, age, level of happiness etc., are not exactly balanced. If a good (or a bad) is indivisible, or cannot be distributed equally, and if all possible candidates have (roughly) equal fairness claims, in some occasions a decision-maker is justified to give everyone an equal chance of receiving the good by using a lottery. (Broome 1984: 40 ff.)

According to Broome, a fairness claim gives the candidates in question some claim to a good and so authorizes them to be treated fairly in relation to others with equal fairness claims, which does not simultaneously imply a right to the good (Broome 1984: 43). Because of Broome's missing definition of "claims", their source and implications, it is not exactly clear what he actually means by "some claim to the good", without implying a right to it, which might be problematic for the proper understanding of fairness claims. As some of the candidates will not receive the good, their claims cannot be satisfied, wherefore it might be more appropriate to define fairness claims as giving all candidates a claim to be treated fairly and to be as strongly considered as all the others, and not a claim to the good itself. Unfortunately, Broome is not entirely clear about that, which will turn out to be the basis for Henning's misunderstanding, analyzed later in this paper.

Fairness, as a relational matter, basically requires to treat all parties concerned equally. That would mean either to divide the good equally, or to give it to nobody. But especially when a good is indivisible, such an equal treatment is not possible. By giving the good to some, and not to others, some unfairness is inevitable. Supposing that a few candidates are in need of an antidote against a poisoning, but the antidote suffices only for one of them, it would not help anybody to divide it equally, so that all would die, only because fairness requires equal treatment. Therefore, the best we can do is to treat all candidates equally in some different way, by at least giving them all equal chances at receiving the good through a random selection process (lottery). Broome's principle to treat (roughly) equal claims equally by using a lottery thus brings at least some fairness into an unavoidably unfair situation and allows a partial satisfaction of claims. (Broome 1 984: 45 f.)

In contrast to duties of fairness, Broome calls all other considerations "the general good" (GG). Such considerations are all "features" which can be differentiated and compared between people. According to utilitarianism, fairness is always determined by (GG), i.e. who promotes it best, receives it; and claims and (GG) are linked. But sometimes favoring one over another may override the fairness claims of all other candidates, which implies that duties of fairness arising from fairness claims and all other considerations can compete. Thus, they should, in contrast to the utilitarian view, be regarded independently of each other in allocation processes. For instance, fairness claims might be determined by superior principles, such as the equal right to life, which has a lot more weight than details about individuals, like their age or skin color. Hence, if fairness is important enough to outweigh the damage done to (GG) by not giving it directly to the person who will promote it best, and if all claims are (roughly) equal, a lottery can increase fairness to some extent without damaging too much (GG) and so reduce the conflict between both. (Broome 1984: 43-48) Surely, in many cases, as for instance selecting someone for a job, fairness is indeed determined by (GG) and there are legitimate reasons to directly favor one over another, based on who will promote the good best. Hence, Broome emphasizes that random selection is not applicable to all allocation processes; only when it is a fair thing to do and when all fairness claims are (roughly) equal, which is not very often the case. In short, the decision whether to choose based on principles of fairness or other considerations depends on their relative importance in particular cases. (Broome 1984: 46-49)

As Broome already mentions himself, his argument could be objected by rejecting his assumption that fairness claims may override all other considerations in some occasions (Broome 1984: 54). One might think that fairness, as treating everyone equally, always requires to give everyone a fair and equal quantity of considerations and choose based on the result who may promote (GG) best. For some people, this kind of decision-making process might seem at least as satisfactory as giving everyone an equal chance of receiving the good, particularly because one could doubt that having equal chances is a valuable improvement or satisfaction. It might even seem unfair for some people to ignore special considerations and only decide by lot. Suppose, a doctor has to make a choice whether to prevent a child from death, or to save an old man who will die soon anyway. By separating fairness from (GG), i.e. basing the decision on the higher principle of an equal right to life, a lottery enables the old man to have an equal chance for surviving as the child, which may seem odd to many people.

But when thinking further, it is also not that clear that it may be fairer in general to choose based on (GG). If humans were completely responsible for their situation, like their social worth, age, intellect, health etc., then we could reasonably discriminate on grounds of such considerations. But it is not the case that all people have the same initial positions, opportunities, freedom or financial resources to be fully responsible for what they are. To differentiate between considerations humans are helpless against and those considerations over which they have full control is a different philosophical question. But as long as we do not have a complete objective overview of all causes and effects and how they are related to everything else, as a stronger form of the "principle of insufficient reason" to favor one over another (Broome 1984: 51), this task might even be impossible. Nevertheless, if we could perfectly differentiate all these considerations properly, we could clearly evaluate what fairness requires and which of both prevails: fairness or (GG).

Finally, I shall explain Broome's view about the applicability of random selection. For instance, when A is slightly needier of food than B, so their claims are not exactly equal, A should consequently be given slightly more than B, in order to treat them proportionally to their claims. But again, if the food (good) is indivisible, it would be unfair to give it directly to A, as her claim is only slightly stronger than B's. Thus, random selection would still be the fairest thing to do, as it at least gives both an equal chance, which is a lot fairer than giving в no chance at all. This argument increases the applicability of random selection in allocative decision-making (Broome 1984: 48 f.), without defining clear borders at which point it becomes fair to use a lottery and when other considerations outweigh. Again, if there existed the possibility of properly and objectively weighing up all considerations, it might theoretically be possible to define logically the borders of applicability and proportional distributions of claims. That in turn would require more philosophical work on the meaning of claims.

IV LOTTERIES ARE NOT REQUIRED BY FAIRNESS: TIM HENNING

As a counterpart to Broome, Tim Henning analyzes three strategies to defend "the Lottery Requirement" (LR) and rejects all three, concluding that there is no moral requirement for using lotteries in conflict cases. Such a conflict case means that one can either save A or B, but not both, when everything else is equal. Henning further stresses the problem of counting (LR) as a valid moral principle, for that results in tending to sanction people for failing to use such chance procedures. Yet, he stresses that using lotteries is not false. (Henning 2015: 169 f.)

First, Henning rejects the "Surrogate Satisfaction Account", which states that when candidates have equal claims to an indivisible good, a lottery has the unique ability to divide the good in some way, by giving everyone an equal chance of receiving it. Hence, all claims cannot be satisfied, but the candidates are at least provided with a surrogate satisfaction. As already mentioned, one can reject such an argument by assuming that giving someone a chance to a good does not satisfy the claim to the good itself, for it cannot be divided. According to Henning, chances are not valuable in themselves and thus cannot generate surrogate satisfaction. That results in the conclusion that there is no reason to prefer lotteries over all other allocation procedures and lotteries are not required by distributive fairness. (Henning 2015: 171-177) But it seems that Henning misunderstands Broome by claiming that lotteries do not satisfy claims, because this is not what Broome means, although I must admit, that Henning's interpretation is understandable, because Broome is not perfectly clear about his account of claims. Broome takes fairness claims as given and argues that giving everyone equal chances satisfies the fairness claims of the candidates, not the claims to the good themselves. I do not think that Broome assumes that equal chances generate surrogate satisfaction of claims, but rather satisfaction of fairness claims. Nevertheless, the question remains, if chances are valuable or not and here, Broome and Henning have different positions.

The second (LR) which is rejected by Henning is the "Procedural Account", which demands that a fair decision procedure must be impartial, i.e. not choosing for "bad reasons" in conflict cases. Henning argues that a lottery may be sufficient to satisfy the demand for impartiality, but not necessary, for there are other fair procedures, which are not based on bad reasons either. Examples would be using some independent and impartial means, such as picking the one whose t-shirt you like most or who blinks next, without having bad intentions or reasons for discrimination. (Henning 2015: 178-194) I am not entirely sure, if it makes a difference if someone chooses only whites, because she likes white skin color more, and if someone chooses the person who has the brightest shirt, because she just likes bright shirts. In my view, both parameters imply discrimination in some sense, no matter if intended or not. Certainly, the procedure of just picking based on some simple parameter, as Henning suggests, depends on the preferences of the decision-maker and thus does not seem fairto me.

Henning further argues that the (LR) can sometimes even lead to worse decisions, which are based on no reasons, although there might be good reasons which would make a difference (Henning 2015: 184). But this cannot be regarded as an objection to Broome, for he also stresses that the decision-maker should first do her best to weigh up all considerations and only if he found no reason for favoring one over another, a lottery should be used as a second-best solution. It seems as if Henning and Broome have similar ideas about these other considerations, but Henning somehow misses that point. However, Henning concludes that lotteries are not required by procedural fairness (Henning 2015: 195).

Finally, Henning questions the justification of (LR) based on the "Ideal Consent Account", according to which a lottery is able to give sufficient reason to consent to and makes it permissible not to satisfy the claim(s) of the loser(s). For claims cannot all be satisfied in conflict cases, according to contractualists, lotteries can at least change the candidates' claims, so that all get equal chances of receiving the good. Henning's reply is that a consent to a lottery would require the existence of moral reasons for holding a lottery.

[...]