Lotteries make science fairer

ABSTRACT

I argue that peer review systems for assigning scientific funding make things less fair than do lotteries that assign funding to research proposals at least partly on the basis of chance. Even though lotteries assign funding randomly, while peer review does so on the basis of morally weighty considerations, lotteries are no more or less fair than peer review as a funding disbursement mechanism. Despite this parity, lotteries make the broader institution of science fairer, as compared to peer review. They do so because they serve to spread the benefits of being a scientist more evenly throughout the scientific community.

KEYWORDS:

• Science funding
• fairness
• public policy
• justice

Introduction

The position defended in this work is that peer review systems for assigning scientific funding make things less fair than do lotteries that assign funding to research proposals at least partly on the basis of chance. ¹ (Strictly speaking, I argue that lotteries make science funding better along a single dimension of moral evaluation that I label fairness, though I identify it below in other terms.) Perhaps surprisingly, the position I defend is not that lotteries are a fairer procedure than peer review. Instead, I argue that lotteries make the broader institution of science fairer. The institution of science allocates special benefits to scientists; when research funding is allocated by lottery such differential benefits become less concentrated, making the institution of science as a whole less unfair (or fairer).

 In arguing for my position, I make a host of assumptions that I do little to defend; I lay them out now. Foremost among the assumptions is that policy decisions are a function of multiple different moral values; in particular, they cannot be decided on the basis of (what I am calling) fairness considerations alone, or utility considerations alone, though both of these sorts of considerations (and others) matter to many policy determinations. ² This means that a less fair alternative may be, on balance, the best one; nothing said here should be read as arguing that we must not, in some cases, dispense funding by peer review. I mean to contribute to the complex question of how research funding should be dispensed by specifying how the value of (what I am calling) fairness contributes to funding policy determinations.

I assume that a peer review system for funding allocation differentially (though not solely) dispenses funds to researchers who have so far conducted successful research. Willis and McNamee characterize sociology as a stratified discipline exhibiting accumulated advantage over time (Willis and McNamee 1990, 373). Graves and colleagues propose a quota limiting the number proposals per applicant to address concentration of research funds allocated by peer review (Graves, Barnett, and Clarke 2011). The relationship between past success and future funding may hold for a number of different reasons. It may hold because the ability to submit successful research proposals also causes successful research. It may hold because so far successful scientists are especially likely to have made contributions to science that their peers regard as orthodox or mainstream, while referees exhibit a preference for, or even bias toward, what they perceive as less risky, orthodox, mainstream science that continues established work (Resch, Ernst, and Garrow 2000; Greenberg 1999; Stehbens 1999; Campanario 1998a, 1998b; Mayo et al. 2006; Fang and Casadevall 2016; Luukkonen 2012; Heackett 1990). Success may cause fame which may in turn cause referees to submit favourable evaluations of famous researchers’ proposals. Even when referees are officially blind to proposers’ identities, they may be able to determine who likely submitted a proposal, especially for the case of proposals from researchers with distinct specializations whose work is widely known (Fisher, Friedman, and Strauss 1994). One way or another, the argument below requires as a premise that peer review causes a concentration of research funding among scientists with a history of success, whether it does so by design or otherwise.

I assume that all differences in social standing could be eliminated by cultural change. I do not assume that differential social standing lacks other causes, even other causes sufficiently ‘powerful' such that they could equally eliminate differences in social standing. This will really only be important to the case of science and scientists. Our society is structured by norms such that science is pursued and those who pursue it are treated as high status individuals who deserve, or at least merit, substantial compensation for their work. By changing or eliminating these norms, we could end the practice of science. Whether we should maintain the practice of science, then, is open to question and justification (we should).

I will also assume that scientists receive substantial support from public sources. Governments, through funding agencies, fund science directly, but public promotion of science stretches beyond grant assignments to include the funding of educational institutions that teach science and the promotion of science as public good. In a complex fashion, we are structuring a broad set of social norms, in addition to dispensing funds from the public purse, to sustain the practice of science, and in doing so we confer substantial benefits upon scientists in the form of remunerative employment and prestige.

The core argument proffered here is that allocating scientific funding by peer review, insofar as it directs money to already prestigious and successful researchers, compounds the unfairness of the institution of science writ large. Allocating funding by peer review is not unfair in and of itself; it is no more or less fair than allocation by lottery (so I will argue). But how research funding is allocated impacts the fairness of the institution of science more broadly conceived. It is people with a special status, scientists, who are entitled to receive science funding, and this entitlement to a valuable social good amounts to an unfair benefit. The extent of this unfairness is a function of the distribution of research funding among scientists; the more concentrated this benefit, the more unfair the institution. Lotteries, in contrast to peer review, distribute science funding more evenly across scientists and hence diminish this unfairness.

Differentially assigned normative statuses

Normative interrelationships can be understood in terms of normative status attribution (Brandom 1998). The ways it is appropriate or inappropriate to treat people are a function of their normative statuses. Having a right, permission, entitlement or obligation may be understood as having a normative status, something that alters how the status bearer must or may be treated. Many occupations and offices consist of bundles of such entitlements, obligations, and the like. Normative statuses include legal, social, and moral statuses.

Some normative statuses may be attributed to others by default: all people have a number of inalienable human rights (or at least many rights theorists so contend). Other normative statuses are assigned differentially: physician, judge, president, professor, scientist, grant recipient, and a host of other statuses are assigned only to some people, often at least partly on the basis of other normative statuses those individuals are assigned. I will call differentially assigned normative statuses DANS. Normative statuses may be related to one another by means of relationships of entitlement or obligation. An individual who bears a driver’s license, for instance, is entitled to the status ‘motorist,' a status with its own set of further entitlements and obligations set by traffic laws. Normative statuses that are conferred on the basis of still other normative statuses I call non-fundamental differentially assigned normative statuses (DANS).

Consider now fundamental DANS. No status or set of statuses is sufficient to confer upon the assignee entitlement to a fundamental DANS (hence the name). Instead, fundamental DANS are assigned at least partly on the basis of descriptive facts about the assignees, rather than being licensed wholly on the basis of other normative statuses (as in the driver’s license-motorist case). A simple regress argument is sufficient to show that not all DANS may be of the non-fundamental sort. Indeed, all non-fundamental DANS must ultimately be based upon fundamental DANS, perhaps quite indirectly by extended chains or webs of intermediary statuses.

The assignment of fundamental DANS is morally bad, though not necessarily wrong, all things considered. Fundamental DANS assignments are bad because they involve the conferral of benefits and/or detriments upon the assignees to which they are not entitled or that they are not obliged to take on. Nothing about the assignees makes it appropriate for them to receive the burdens or benefits associated with the fundamental DANS, and yet they receive them anyhow. This is simply true by definition. Were it appropriate to confer some beneficial normative status upon an individual, they would have some normative status making the conferral appropriate, thus making the normative status non-fundamental (and mutatis mutandis for detrimental statuses).

In what follows, I call the sort of badness involved in the assignment of fundamental DANS unfairness. This is the label I use for a sort of badness I pick out in other terms, as the sort of badness involved in assigning people benefits to which they are not entitled, and burdens that they are not obliged to bear. By the use of the term fairness, I do not mean to propose an analysis of that notion; I leave it open whether fairness, as it is understood here, is the same thing as fairness as it is understood elsewhere, such as in distributional theories of justice. Perhaps ‘unfairness' is not a perfect label to use for the sort of badness I pick out, but it is as good a label as any alternative.

In order to see how fundamental DANS assignments are unfair, consider some examples. In many Western polities, jury duty is assigned as a fundamental DANS. Individuals are picked out by a random process to serve as jury candidates for a particular trial and are required, if ultimately selected, to serve as jurors. This is unfair to jurors. They may legitimately complain that their duties as jurors were foisted on them arbitrarily, that they were selected for no good reason and did nothing to earn or deserve their status as such. Indeed, this is just what makes ‘juror' a fundamental DANS. Instead of their being obliged to serve by their bearing some other normative status, jurors are required to serve for arbitrary reasons, that is, by dint of a descriptive fact about them (their selection in a lottery).

The above should not be read as a criticism of the institution of trial by peers. It would only count as such if it were never appropriate to assign individuals fundamental DANS. But though we must acknowledge fundamental DANS assignments are unfair, we need not regard them as impermissible, provided that the unfairness they involve is remediated by compensatory gains of other sorts. In the case of trials by peers, random jury selection helps to ensure that trials are conducted in a just manner. Trials of this sort are a very valuable social institution and so we stomach the unfairness of forcing random people to serve on juries in order to sustain a critical institution for a just society.

‘Conscript’ is another fundamental DANS much like ‘juror.’ Conscription is unfair to the draftees who are randomly sent off to war and did nothing to deserve their fate, but we stomach this unfairness for the sake of another valuable social institution, a powerful enough military to achieve victory in war. At the very least, we can say this: if conscription is a legitimate social practice, then it is only so for the reasons just given, though we might reject it as a social practice on the grounds that it violates other constraints on state power.

Not all arbitrarily assigned fundamental DANS are assigned randomly, on the basis of a lottery. Assignments of fundamental DANS may be systematic, insofar as they may hinge upon particular descriptive facts about individuals other than their holding of a lot, and yet still be assigned on the basis of morally weightless criteria. Schmidtz (2012) considers the practice of original appropriation and defends it in the same manner as jury duty and conscription were defended just now. In that case, original appropriator is assigned systematically, on a first-to-arrive basis, but Schmidtz acknowledges that being first to arrive is not a morally weighty reason why someone should acquire property. Nonetheless, Schmidtz argues that the institution of original appropriation can be justified because it converts what is typically a negative-sum game into a positive-sum game. The value of the commons is typically destroyed by overuse: no one has a right to exclude others, and everyone has an incentive to grab the most for themselves while foisting the better part of the costs of her usage onto others. Appropriation gives owners a right to exclude others and thus an incentive to preserve and invest in the land. Ownership thereby generates benefits for owners and others through land development and trade. While Schmidtz concedes that the practice of original appropriation may be unfair insofar as individuals cannot supply morally weighty reasons why they, rather than others, should get ownership rights over land previously held by no one, original appropriation is justified as part of a larger institutional package that provides sufficient gains to compensate for this unfairness. Like ‘original appropriator,’ ‘citizen’ is another fundamental DANS assigned on the basis of an arbitrary feature an individual, the place of their birth (see Risse (2004, 93) for a defense of immigration constraints much like Schmidtz’s defense of original appropriation).

Descriptive facts used to assign fundamental DANS may also be morally weighty, in contrast to the weightless criteria of first-to-arrive and random selection discussed already. The state assigns driver’s licenses partly on the basis of age, and age would not seem to be a morally weightless criterion for assigning driver’s licenses. ³ What makes age a morally weighty criterion for driver’s licensing is that would-be drivers’ ages are symptomatic of their reliability and responsibility.

The use of morally weighty descriptive facts to assign fundamental DANS does not, however, make the statuses so assigned less unfair. This is likely at least somewhat counter-intuitive, but the point can be demonstrated easily enough. Consider driver’s licensing again. Suppose, in a doubtless fantastic alternative world, that neither youth nor old age were to bear any relationship to driver competence. Were this so, it would immediately strike us as unfair, or at least wrong in some sense, for the state to require elderly individuals to retake driver’s license testing every year. That constraint is no less unfair in the real world, it’s just that we have morally weighty reasons, ones having nothing to do with fairness, that are sufficient to justify age-based licensing constraints. Such constraints pay off in the real world; the gains from putting them in place overbalance the unfairness involved, in much the same manner as the unfair treatment of jurors, conscripts and original appropriators is overbalanced by gains of other kinds. The moral weight of the criteria used to assign some fundamental DANS may even conceal the unfairness of the assignments they mandate by overbalancing it. But the fact remains that the morally weighty descriptive criteria for the assignment of a normative status remain descriptive criteria: by definition, the individual assigned the status is not entitled to it, or obliged to take it on. It is not fairer that the elderly must retake driver’s license exams on a yearly basis; it’s just safer.

I have considered three ways in which descriptive facts may be used to assign fundamental DANS and these exhaust the possibilities. We might use random descriptive facts to assign DANS (as in ‘juror’ and ‘conscript’); we might use non-random, but morally weightless, descriptive facts to assign DANS (as in ‘original appropriator’); or we might use morally weighty descriptive facts (as in driver’s licensing). In each case, we found the statuses to be assigned unfairly, though of course not necessarily illegitimately.

It might be thought that granting that it is always unfair to assign fundamental DANS means granting that all the normative statuses they serve to justify are equally unfair, which would threaten the legitimacy of the entirety of our cultural framework. But the unfairness of fundamental DANS does not bleed or leach into the rest of the normative framework they support. The reason the unfairness does not bleed into the rest of the system is that the fundamental DANS are perfectly legitimate normative statuses and, for that reason, are available as justifiers for other normative statuses. The unfairness involved in their assignment has been overbalanced, paid off, and negated. (Otherwise, the DANS really should be unavailable as justifiers, but only because they should not be assigned in the first place.) Whether or not some fundamental DANS may serve to justify other fundamental DANS is a function of the would-be justifying statuses’ legitimacy, not simply their unfairness. Fundamental DANS it is appropriate to assign are legitimate and for this reason may justify the assignment of other DANS.

I want to turn the discussion briefly back to non-fundamental DANS. Non-fundamental DANS consist in entitlements and obligations that, however indirectly, trace their roots to fundamental DANS. Non-fundamental DANS contribute to assessments of the extent to which fundamental DANS are unfair. Fundamental DANS may confer entitlements to benefits, or obligations to assume burdens, by serving as sufficient conditions (or contributors thereto) for non-fundamental normative statuses that leave the assignee better or worse off. The value of these benefits and detriments, in the form of the entitlements and obligations undertaken as a result of a fundamental DANS assignment, impact just how unfair fundamental DANS assignments are.

Consider again jury duty. Jurors are paid for their time; an entitlement for jurors that serves to reduce the unfairness of jury duty. But jurors must also bear the burden of deciding criminal cases, a time-consuming obligation that amplifies the unfairness of jury duty. The extent to which it is unfair to impose upon jurors is determined by the schedule of benefits and losses they accrue, and these benefits and losses take the form of normative statuses to which they are entitled or obligated to take on. Jury duty is undoubtedly a net negative, a deleterious fundamental DANS that must produce sufficiently great benefits to others so as to compensate for the losses of those forced to serve. ‘Original appropriators’ also get entitlements, to exclude others from the land they have appropriated. In contrast to the status juror, ‘original appropriator’ is doubtless a beneficial fundamental DANS precisely because of how it affords the bearer property rights, a valuable non-fundamental DANS. In the property case, too, the special benefit that original appropriators receive in the form of property rights must be paid off by broader social gains. Whether individuals assigned fundamental DANS are made unfairly better or worse off, the extent to which they gain or suffer determines how much unfairness must be paid off by gains made by others. Non-fundamental DANS licensed by fundamental ones determine the extent of the unfairness of fundamental DANS, which in turn contributes to the magnitude of the quantity to be paid off.

The extent to which fundamental DANS assignments are unfair, and hence the magnitude of the quantity to be paid off, is also a function of the distribution of benefits or losses that accrue to those assigned the fundamental DANS, along with the extent of such benefits and losses. Generally, the more that those assigned a fundamental DANS are left better or worse off, the more unfair the assignment (henceforth I will focus on DANS that leave their bearers better off). ⁴ Moreover, unfairness is amplified when fewer individuals are assigned a beneficial status such that the special treatment is distributed more narrowly in society. Finally, if individuals differentially benefit from a fundamental DANS, the more even the distribution of benefits across assignees, the less unfair the assignment. (That last contention about fairness will play a pivotal role in the argument that follows.) These contentions about fairness should be more or less intuitive; the reader is invited to consider the comparisons presented in the histograms in Figure 1 if she needs convincing. ⁵

Figure 1. Less fair (left) and fairer (right) DANS assignments.

The requirement defended here is that the unfairness introduced by advantageous differentially assigned normative statuses must be overbalanced by gains made by those who are not assigned the advantageous status. The approach is broadly speaking Rawlsian, though the constraint is easier to satisfy and social policies meeting Rawlsian strictures will pass the requirement I defend, too. I do not require, as Rawls does, that everyone profit from the social institution, nor that the worst off profit the most (Rawls 2001, 42–43). Instead, I require only that there be sufficient profit for those who are not differentially assigned fundamental normative statuses within some institution to overbalance the unfairness of the differential treatment of the assignees.

The perspective taken here differs from Rawls’ principally in evaluating whether differentially assigned normative statuses are fair, not whether inequalities are. As a result of his perspective on cultural practices and institutions, Rawls does not differentiate between, on the one hand, those who profit from a social institution by dint of being assigned the normative statuses that bring it into existence or sustain it, and, on the other hand, those who profit from an institution indirectly, without being assigned special statuses within it. To illustrate, take the case of breast cancer research. Medical researchers specializing in breast cancer research profit from this institution by finding remunerative employment yielding high socioeconomic status within it. Such research is also beneficial to others, particularly breast cancer patients, but also their family members and loved ones. Rawls would count all the researchers, patients, and their loved ones among the beneficiaries of the practice, ones made unequal by it, and ask whether a practice instituting such inequalities is legitimate. I ask instead whether assigning differential normative statuses to cancer researchers sufficiently benefits others who are not breast cancer researchers to legitimate the practice. The patients and their families are among those whose benefits count toward the justification of differentially assigning the status breast cancer researcher: the families’ gains help to mitigate the unfairness of the ‘breast cancer researcher’ assignment. The benefits made by the patients and families are not part of the inequalities to be remediated or compensated for by gains made elsewhere; instead they are part of the remediating quantity which legitimates the practice of breast cancer research by overbalancing how it privileges some medical researchers.

Fundamental DANS in science

The choice between lottery and peer review for the assignment of the status grant recipient is a choice between two practices, one of which would supply a morally weightless criterion in the place of morally weighty one in the other. Random selection to determine the recipients of research funding is a morally weightless criterion, while selection on the basis of referees’ assessment of research potential as described in a grant proposal is a weighty one (this is true, anyway, if peer reviewers are more successful than chance at funding valuable research). But because it is a choice between weightless and weighty descriptive criteria, the choice between a lottery and peer review is not a choice between an unfair practice and a fair one; the practices are equally unfair, since both assign a normative status on descriptive grounds, the holding of a lot or referees’ assessments of research potential based on the contents of a proposal. ‘Grant recipient’ is a fundamental DANS either way; in neither case is a scientist entitled to research funding by dint of some normative status (or set of statuses) alone.

It may well be counterintuitive that peer review is no fairer than a lottery. Economist Adam Jaffe writes that it seems intuitively obvious, to scientists anyhow, that the peer review system is best (quoted in Mervis 2020, 597). But one can accept that peer review is best without accepting that this is so because it is fairer. We can see that it is really not fairness considerations at work behind the intuitiveness; the judgment that peer review is best if we imagine that peer review allocates funding no more efficiently than does a lottery. Such a scenario is not so fanciful: there is evidence that a considerable element of chance affects the peer review process with respect to both scientific publication and funding allocation (Campanario 1998a, 1998b; Rothwell and Martyn 2000; Cole, Cole, and Simon 1981; Mayo et al. 2006; Mervis 2020; Fang and Casadevall 2016; Roumbanis 2017). Imagine a scenario where the role of chance is so large as to be entirely determinative. That is, despite reviewers’ efforts to determine which proposals ought to be funded so as to allocate funding in a manner that maximizes gains in the form of valuable scientific insights, reviewers are no better than chance at determining who should be funded. In such a circumstance, fairness provides no reason to maintain a peer review process over a lottery. But this would not be so were peer review systems genuinely fairer, for fairness would operate as a consideration independent of efficiency and in favour of peer review.

I conclude, then, that lotteries and peer review equally assign the status funding recipient as a fundamental DANS and, hence, introduce unfairness to be paid off. Moreover, neither approach distributes funding more narrowly than does the other, nor does either approach dispense greater benefits. Thus, neither funding mechanism is more unfair than the other. Accordingly, it might be thought that, with respect to fairness considerations at least, the two alternatives were equivalent: both unfairly (though ultimately justly) distribute benefits to grant recipients in the same sort of way (if not to the exact same people). However, this is not the case. The distribution of scientific funding among scientists is relevant to the fairness of the broader institution of science. It might be no fairer or less fair to proposers to assign funding by lottery or by peer review, but the distribution of funding within science matters to the fairness of that broader institution.

Consider first that though ‘grant recipient' is a fundamental normative status, insofar as no set of normative statuses is sufficient to assign it, the normative status of scientist is necessary to do so. It is only scientists who may apply for science funding. The institution of science, just like any other distinct social institution, ultimately depends on the practice of assigning individuals fundamental DANS. Again, a regress argument is sufficient to establish that the various differentially assigned normative statuses involved in the institution of science cannot all be non-fundamental ones. Moreover, science is a distinct institution, not an appendage or fragment of some broader practice such that all the fundamental DANS upon which the institution of science rest lie within the purview of another encompassing institution. It is difficult to say, exactly, what are the fundamental DANS that ground the institution of science; a natural place to look at is the credentialing procedures (tests, examinations, peer review procedures, other evaluations, etc.) that are used to assign individuals fundamental DANS proper to science. In these sorts of practices, we will find how descriptive facts about individuals, measurements of their prowess and performance, are used as bases to assign DANS (perhaps together with other DANS serving as necessary conditions).

For simplicity’s sake alone, I will treat the status ‘scientist’ as a fundamental DANS proper to the institution of science. It is difficult to say what makes an individual a scientist. Perhaps someone becomes a scientist when gaining employment at an academic institution in a science department. Conceived as a definition, the characterization seems overly restrictive; perhaps one might work in an academic context but outside a science department and still do science; moreover, there certainly have been, in the long history of science, gentleman and – women scientists who worked independently of any particular academic institution. Perhaps, instead, one becomes a scientist upon receipt of a PhD in science from some suitably accredited academic institution, allowing that there are scientists who do not practice science. Doctoral degrees are assigned, at least in part, on the basis of descriptive facts about their recipients, their fulfilment of various credentialing requirements, and this descriptive basis is sufficient to make at least ‘PhD recipient’ a fundamental DANS. It doesn’t matter to the following fairness argument that ‘scientist’ is what I pick as the fundamental DANS proper to the institution of science, leaving somewhat indeterminate who exactly counts as such. The argument could be made using any alterative fundamental DANS that is proper to science and necessary for receipt of research funding.

The status scientist, then, is a fundamental DANS that contributes to the entitlement of its bearer to research funding. Because ‘scientist’ is a fundamental DANS, it may only be assigned if sufficient compensatory benefits accrue to non-assignees as a result of the assignment of ‘scientist.’ There can be little doubt that such benefits obtain. Arguably, nearly everyone has benefitted from the practice of science, especially when the concept of science is understood as capaciously as possible so as to include the entire history of the institution. These days, one must consider individuals who die extremely young or very remote parts of the globe in order to find individuals who do not profit from science. Nonetheless, even though the institution of science clearly passes muster as one conferring sufficient benefits on non-scientists, the extent to which the institution is unfair remains a matter of moral concern. Ideally, the unfairness of the institution is minimized even if there are ‘excess' gains on the part of non-scientists such that the institution could be substantially more unfair than it is and still pass muster as providing sufficient compensation to non-scientists.

As noted earlier, the extent to which fundamental DANS are unfair is a function of the distribution of the gains or losses that accrue to those assigned the status. The more equally scientists profit from counting as scientists, the less unfair it is to assign them the status scientist. How science funding is assigned has an influence on exactly this distribution. Assigning science funding by lottery produces a more even distribution of funding than does allocating funds by peer review. This is true, at least, given that so far successful scientists will disproportionately gain funding if money is assigned by peer review, whether or not by design. Such individuals have already profited from their status as scientists in form of increased prestige, remuneration, or both. The mechanism of peer review will allot them still greater gains, diminishing the fairness of the institution of science overall.

It is perhaps worth emphasizing here that even if research funding is allocated by peer review rather than by lottery, there should be little question that the institution of science deserves to persist: we are nowhere near the threshold where compensation for non-scientists becomes too slim to justify the enterprise. Nevertheless, more fairness is better and the determination of whether funding should be allocated by lottery or peer review should be made in a manner that takes fairness into account. To see how this might go, suppose that an all-things-but-fairness-considered judgment concerning the other dimensions of disbursement policy yields a draw between a lottery and peer review for some specific imaginary case. As it turns out in our fictional example, peer review allocates funding more efficiently: it helps to produce more higher quality research in comparison to a lottery. However, peer review is costly in terms of the efforts that scientists must devote to the process, leading to decreased productivity (Goldsworthy 2009). In the end, in our imaginary scenario, the costs and benefits of peer review vis-a-vis a lottery cancel out, fairness aside. In such a circumstance, a lottery is a morally superior disbursement mechanism. A lottery spreads one of the benefits of the status scientist more evenly across scientists than does peer review, which will tend to concentrate grant receipt among so far successful scientists. Assigning the status grant recipient by lot makes the practice of assigning the status scientist fairer, making assigning ‘grant recipient’ by lottery morally preferable when all other morally relevant considerations cancel out.

Conclusion

Above, I drew attention to a morally important theoretical quantity, one I had no better name for than ‘fairness’. I evaluated two scientific practices, lotteries and peer review, in its terms. In conclusion, I would like to (humbly) advocate for the inclusion of this quantity within evaluations of whether scientific innovation is being carried out responsibly. Assessing whether scientific innovation is being carried out responsibly should involve an assessment of the fairness of the institutions that make it possible.

Schomberg (2011) suggests a helpful contrast between a process dimension and a product dimension in evaluations of scientific innovation. The scientific research process has long been assessed in binary terms: scientific research complies with constraints on animal usage in experimentation or it doesn’t, conflicts of interest have been confessed or not, human subjects have been treated with respect or otherwise. Many of von Schomberg’s criteria for the evaluation of whether the innovation process is carried out responsibly have equally taken a binary character. They involve adherence to codes of conduct and standards, the introduction of ethics into the design of technology, the development of models to allocate responsibility, and the proper involvement of the public (Schomberg 2011, 11–14). These are important forms of ethical evaluation. But they have a distinctive feature: they are yes/no, pass/fail: they invite assessments of responsible innovation as acceptable or not (Schomberg 2011, 10). Moreover, they are not the sorts of criteria that involve us in tradeoffs. When von Schomberg advocates for the constructive involvement of societal actors, he does not suggest a criterion for responsible science that he is willing to trade off for gains elsewhere. This attitude is shared. Consider Jean-Pierre Alix’s response to a representative of the French General Confederation of Labour (CGT) union center at the 2008 European Conference in Paris, ‘Science in Society: Dialogues and Scientific Responsibility.’ The union representative noted an imbalance among participants at the conference, with too few representatives from civil society. In response, Alix admirably admits to failing to involve sufficiently many important societal actors in the conference (quoted in Felt 2009, 52); while the failure in no way scuppered the event, Alix makes no suggestion that the fault is overbalanced by positive aspects of the conference.

Contrast the binary quantities used for the evaluation of the process of science with those von Schomberg suggests for the evaluation of scientific products: high level of protection to the environment and human health, sustainability, and societal desirability (Schomberg 2011, 10). The creation of economic and social value has long been considered the responsibility of science (Owen, Macnaghten, and Stilgoe 2012, 754); what is important here is that such creation is a matter of degree. The quantities used for ethical evaluation of scientific products are continuous; moreover, making tradeoffs among them is all but inevitable. Unfairness is also a continuous quantity available for trade off, but it has to do with the process of science, or at least its institutional structure, rather than its products. More generally, aspects of the innovation process may be more or less good or bad, just like aspects of the products of innovation may be more or less good or bad.

Philosophers seized upon seemingly radical epistemological implications of anthropological/sociological/historical investigations of scientific practice (e.g. Latour 1987). But social scientists who lay bare how normative statuses are conferred in science provide tremendous raw material for the ethical evaluation of scientific research using continuous quantities, such as fairness, that are available for trade-off. A critique of scientific practice that exposes it as a locus of circulation of privilege, power, and prestige need not be aimed at a radical undermining of the institution and yet retain ethical relevance to questions of institutional design. We are used to thinking this way when evaluating non-scientific practices. We are used to the question of whether CEO pay is so great as to be ethically impermissible. To decide whether this is so, we will want to know whether ever-greater CEO pay brings ever-greater benefits in the form of business innovation. I have tried to show that the institutional structure of scientific research involves us in tradeoffs of a similar sort: In cases where certain values, such as efficiency and productivity cancel each other out, fairness might tip the balance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes on contributor

Peter Gildenhuys is a philosopher of science working in the Philosophy Department at Lafayette College in Easton, Pennsylvania USA. His primary research interests are in philosophy of biology, with a focus on evolutionary theory. His scholarship also includes work in bioethics. He has a doctorate from the History and Philosophy of Science Department at the University of Pittsburgh.

Notes

1 Gillies (2014) defends a lottery mechanism for funding allocation while Avin (2015) argues for a mixed approach in which a lottery is combined with a preliminary screening process. See also the discussion in Roumbanis (2019).

2 Elster writes that ‘to be rejected by fortune [is] less dishonourable than to be rejected by the community’ quoted in Roumbanis (2019, 1012); perhaps honour or dignity are moral values relevant to funding allocation, too.

3 Indeed, ‘juror’ and ‘conscript’ are also assigned partly on the basis of the morally weighty criterion of age (thanks to an anonymous referee for pointing this out).

4 It is an interesting question, though moot in the present context, whether fundamental DANS’s fairness is a function of whether their assignment is beneficial or deleterious. It seems at least prima facie plausible that deleterious ones require greater compensation than those which improve the assignees’ welfare to the same degree.

5 Sufficientarians might reject the intuition that the more unevenly distributed a benefit conferred by a fundamental DANS, the more unfair (or at least bad) the assignment of the DANS. Instead, they might contend that, provided those assigned the fundamental DANS all profit to some extent (to be determined by their current welfare and the minimum they’re owed), any ‘excess’ benefits conferred to the assignees may be distributed in any fashion without the excess distribution being unfair. Accordingly, depending upon how the threshold of minimum benefit is specified in the case at hand, sufficientarians might well reject the conclusion defended here that lotteries make science fairer.