by Gerald Gaus
Many worry that, in a diverse society, any attempt to show that almost all perspectives must endorse a moral rule is doomed to failure: we are almost surely to be left with no socially eligible set.148 One response to this worry is, of course, to reflect again on the basic elements of the Open Society, and the great benefits that they provide almost all perspectives. But two other points should be kept in mind. (i) The polycentric (§IV.2.2) nature of the Open Society does not require that all perspectives must share all features of the moral constitution. While no doubt the most fundamental aspects must be shared by all (say, basic understandings of the contours of jurisdictions), we should not suppose that all rules must hold over the entire range of perspectives. Networks crystallize to solve commonly recognized problems—those seeking solutions to shared problems adopt rules to accomplish their aims. Those who worry that diverse societies have “empty eligible sets” take an overly abstract view of moral life and why we participate in it. (ii) However, it is by no means necessarily unfortunate that many moral rules will fail to be justified—we should not seek a maximally large eligible set or a comprehensive moral constitution. As Hayek stressed, the Open Society’s extension of the range and diversity of individuals to whom its moral rules apply inherently brings about a limitation of the content of those rules.149 If we compare the smothering moralization of the Victorian period with our contemporary moral constitution, we are struck by just how much less social morality regulates. It is a thinner, and a far less stultifying, moral constitution. Recent studies of social norms such as female genital cutting, racial discrimination, condemnation of homosexuality, and public gender inequality have made it manifest how many deeply oppressive social rules characterize various societies. That many perspectives conclude that such rules are worse than no moral rules at all on these matters is a great advance, not a handicap, in the development of the Open Society.
Nevertheless, Rawls and Berlin are certainly correct that, in the end, some perspectives will conclude that even the most fundamental elements of the Open Society are worse than no moral constitution at all. Some perspectives are, in the end, unable to share a framework of moral accountability with diverse others. Even the Open Society must be prepared to normalize to some, hopefully to a very small extent. Such “Excluded Perspectives,” which cannot find sufficient space in the Open Society, will almost surely be those that are committed to the optimizing stance, or some near approximation to it. Faced with different rules to live by, the Excluded Perspectives can live only by those that they think best, and so they cannot endorse the characteristic institutions of the Open Society, which seek to provide as much space for all as is possible. Such perspectives may live along with, but are not part of, the Open Society, treating its rules as at best mere descriptive norms rather than moral injunctions. Here, we still can do deals with such perspectives on a purely instrumental basis focusing on mutual benefit150—insofar as effective deals can be done without a supporting moral framework that undergirds trust among strangers.151 Philosophers are apt to worry whether, in the absence of accountability relations, we can still act to defend ourselves against such strangers in our midst, and coerce them if need be. Nothing I have said here denies that the Open Society may exercise such defense, including through the law.152 There is no reason we should leave ourselves at the mercy of those who refuse to live on terms that others can endorse. Yet such Excluded Perspectives are apt to be a source of social conflict and instability. Alienated from the public social world, Excluded Perspectives take refuge in the social world of their comprehensive conception, while viewing the moral constitution of the Open Society as alien oppression. It is a great loss to all when any perspective is, as it were, normalized out of the social contract, though no doubt such loss can be only minimized, not entirely avoided.
4 IMPERFECT COORDINATION ON THE MORAL CONSTITUTION
4.1 Coordination as Diversity Reducing
Thus far the most that could be claimed is that versions of the rules characteristic of the Open Society (§IV.2) will be in the socially eligible set of a very wide range of perspectives, and no effective basis of social cooperation could achieve a more widespread endorsement. Even granting that, I have said nothing about how any rule, specific enough to be part of a practice of responsibility, might be selected. We can, for instance, think of a range of rights of privacy; different perspectives will rank them differently given their understandings of justice. That a contract yields a socially eligible set with no optimal element does not tell us how to choose from this set, only that we have reason to make a choice (§IV.3.1).
On Peter Vanderschraaf’s insightful analysis, this means that the parties to our “contract” are in the true “circumstances of justice.” On his non-Humean account of the circumstances of justice, they can be modeled as a “conflictual coordination game … with multiple strict equilibrium points such that the parties engaged in the game differ over their most favored equilibria. In a conflictual coordination game, the parties have open to them a variety of ways to coordinate, but their interests conflict over which way to coordinate.”153 In our case, they disagree in their rankings of the rules of justice over some matter in the socially eligible set, but they recognize that a moral constitution obtains only when they successfully coordinate on one of them. Figure 4-5 provides a toy game example of the basic problem of our perspectival contractors in the circumstances of justice.
Figure 4-5. An Impure Coordination game
Unless they coordinate on one member of the socially eligible set, they do not have a mutually acknowledged rule, which provides the basis of a practice of responsibility in which they share the same normative expectations (§IV.2.1.2). Either Rule RA or RB is better than lack of coordination (no shared rule at all, the “z” option), and thus both are Nash equilibriums: if they coordinate on RA or RB neither can achieve a better outcome by unilateral defection, acting on the alternative rule. If Alf acts on RA, Betty’s best response is to also act on RA; if Betty acts on RB, Alf’s best response is also to act on RB. If Alf justifies his action because it is called for by RA it must be the case that not only is RA in the socially eligible set, but it is the rule on which they have coordinated, and so both have empirical and normative expectations that they will act on it. A justified appeal to RA presupposes that it is the rule from the socially eligible set on which they have coordinated (even though Betty believes it is not the best, given her understanding of justice). As we saw earlier, the functional nature of the moral constitution implies an existence requirement (§IV.2.1.3). This points to a fundamental revision in theorizing about public reason and justice, which has not been appreciated (indeed, it has not been glimpsed) by most of those who are content with justifying a “family” of liberal conceptions. The justification of a rule of justice depends not simply on the reasons of the parties for specific rules, but on their coordination on some single eligible element. The justification of an appeal to a rule of justice R as a basis of a practice of moral accountability in a group requires (i) it is in the socially eligible set of that group and (ii) they have been able to coordinate on it.
Note that the very need for the group to coordinate on a rule is inherently diversity reducing: the value of coordination simultaneously decreases diversity within a group (by inducing them to select within the socially eligible set), leading members to converge, while increasing the diversity (of moral constitutions) between groups.154 Different societies are apt to coordinate on different constitutions, even if they share the same socially eligible set. Thus we can expect significant diversity among the moral constitutions of open societies even when each has the same range of perspectives represented.
This choice from the eligible set can be made in a wide variety of ways. Sometimes we can envisage a procedure that all would accept, say, some form of democratic vote. But here too we are likely to find disagreements about the best procedure—one thing democratic theory after Arrow must realize is that there is no flawless way of voting
over three or more options.155 “When sincere and good people differ,” Nozick observed, “we are prone to think they must accept some procedure to decide their differences, some procedure they … agree is reliable and fair. … [But] this disagreement may extend all the way up the ladder of procedures.”156 However, the core lesson from Sen’s work is not that we must justify a meta–choice procedure to determine what, from some metaperspective, is the best in the eligible set, but that rational agents will see the need that a choice is made. In our case, they will see the need to coordinate on some member of the socially eligible set as the rule that they will employ in their moral constitution. Experimental evidence in many-generational-iterated impure coordination games indicates that, indeed, people settle on a tradition of playing one or the other (pure) equilibriums, despite their disagreement on which is the best way to coordinate. In deciding what equilibrium to play, the current generation draws on both the history of play and, more importantly, a socialization process in which the present generation teaches and passes on current conventions to the next generation.157 In another work I have shown how social evolution can secure a similar result.158
Philosophers generally assume that this variety of reasonable ways to practically overcome the indeterminacy of justification is regrettable. Surely, they reason, it would be best if there were some aggregation method that unequivocally and uncontroversially determined what the rational choice to be made from a set of eligible options is. Without such a determinate and uniquely best decision procedure (Condorcet voting, a rational bargain, etc.), even if a society manages to solve the coordination game in figure 4-5 (say by a social process that has produced a constitution with RA), Betty will continue to push for her favored RB. The matter, after all, was not definitely, rationally settled. Thus the never-ending pursuit of the philosopher’s Holy Grail of THE DETERMINATE SOLUTION. I believe this is a fundamental mistake. It is the very absence of an unequivocal and determined best solution to the problem of equilibrium selection that is the engine of moral improvement in the Open Society.
4.2 The Changing Moral Constitution
To see this, consider first a sort of indeterminacy that I have thus far been ignoring. I have been supposing that the relevant perspectives concur on the classifications that constitute the public rules, and thus when they coordinate on Ri they coordinate on what they see as the very same rule (§IV.3.1). To be a bit more formal, we have supposed that they agree on Ri in the following sense: in circumstances C, Ri prohibits (requires) (ϕ-type actions. We also have supposed that they agree in their interpretation of the classification of circumstances (states of affairs) that constitute C, and the classification of actions as ϕ-type. However, given that perspectives see the social world very differently, we might expect that they will disagree in how they understand the relevant circumstances and classifications. Even supposing that they concur on the core or paradigm cases, they might differ on “how to go on” in less obvious cases. To a great extent, the judgments of the large majority are apt to dictate what the proper interpretation is; if we wish to coordinate, then we will typically follow the large majority’s belief on what that involves. Because participants in a practice of responsibility have empirical and normative expectations as to how others should act, when these expectations are not met, they will call each other out. So long as (i) it can be observed what action Alf (occupying perspective ΣA), thinks R calls for, and (ii) the majority concludes that this action clearly disappoints their expectations about R-compliant behavior, they will be able to call Alf out and reestablish coordination. Thus the specification of a rule is by no means a onetime act, but an ongoing process. Again, note how important it is that the rule be part of a practice of moral accountability, rather than simply a descriptive norm (§IV.3.2).
In some circumstances C, though, it may not be manifest to the large majority what the relevant rule is, and so Alf may be able to exploit this ambiguity by choosing the rule that allows him to act as he prefers (ranks best), and yet still meet the normative expectations of others. In an important series of experiments, Bicchieri and her co-investigators have found that exploitation of such norm ambiguity is easily induced. In a notable experiment with Alex Chavez, subjects played Ultimatum Games (§IV.2.5). In the Bicchieri-Chavez study 106 college students played Ultimatum Games with limited options and different amounts of information provided to the Responders. In what they called the “full information” condition, Proposers had three choices on how to split $10: (5, 5), (8, 2) (the 8 going to the Proposer), or a flip of a coin between the (5, 5) and (8, 2) options. Responders knew that these were the Proposer’s options, and which option the Proposer chose. Assuming that the Responders think the coin toss is fair and will accept its outcome, the expected payoff of the coin flip for Proposers is .5(5) + .5(8), or 6.5, higher than the expected 5 from the equal-split option (recall that simple [8, 2] offers are very likely to be rejected, leaving the Proposers with nothing). Participants were quizzed on their normative expectations, and it was found that coin toss was widely seen as fair by Responders, and most Proposers correctly believed that Responders thought so. Thus in the full information condition we would expect that Proposers would tend to select “coin,” as it gave them a higher expected payoff than equal splits (it is important that they correctly expected Responders to accept the outcome of the toss). This would be a case of Proposers exploiting rule ambiguity: since both equal splits and equal chances seem like fair ways to divide money (both were thought by Responders to be rules that apply to this circumstance), Proposers could take a more selfish choice (with the expected payoff of 6.5 rather than 5) and yet still meet the normative expectations of Responders. In the “limited” information condition, Responders knew that the three options were available to the Proposers, but they did not know which was actually chosen; all they knew was whether the offer was (5, 5) or (8, 2), but they did not know how the (8, 2) offer came about (whether it was the result of a coin or a selfish offer). In this condition Bicchieri and Chavez expected that more Proposers would select (8, 2). Because the Responders did not know how the outcome was selected, and so it could have been selected by a toss of the coin in which (8, 2) was selected, Responders could not judge whether or not the choice was selfish or in conformity with the equal chances norm. Consequently the Proposers would expect that Responders may well accept (8, 2), and so be tempted to make the low offer. In the “private information” condition, the Responders did not know that coin toss was available to Proposers, and Proposers were aware that Responders thought that direct (5, 5) and (8, 2) splits were the only options. In this final condition Bicchieri and Chavez did not expect Proposers to choose coin; even if it was truly the coin that selected (8, 2), Responders would take this as simply a selfish choice on the part of Proposers, and so be highly apt to reject. The predictions were well borne out. Subjects did seek to exploit norm ambiguity, systematically selecting the action most advantageous to them when doing so satisfied—or at least did not clearly flout—normative expectations.159
When there is normative ambiguity individuals can choose what rules to follow, and so can act on their preferred rule which, in this situation, is competing with another. Bicchieri and her coworkers’ studies were designed to highlight the choice of norms in light of self-interest, but we should not forget that interest often tracks broader normative commitments: those who seek normative change often do so because they do not think their interests are sufficiently catered for. More generally, normative convictions given one’s perspective on justice may also figure into rule choice in ambiguous situations. This is significant. Because it is important to coordinate on a rule from the socially eligible set, and each coordination outcome is a Nash equilibrium, and because, further, the rules of the moral constitution are part of a practice of responsibility and so one is apt to be called out for violation, it would seem that once we coordinate on a rule we are locked into it. Rule ambiguity loosens this. Individuals can explore and promote alternativ
e rules, not simply by flouting the current rule and so risk being held accountable and perhaps punished, but by exploiting the ambiguity of normative situations to act on what one’s perspective deems to be superior rules (in the socially eligible set) while still satisfying the normative expectations of others. Indeed, in one study Bicchieri and Hugo Mercier found that when exploiting normative ambiguity, a person is apt to be concerned with providing a “public justification” that shows that the action is reasonable.160 Norm innovation and improvement are consistent with acting within the scope of a public moral constitution; the very ambiguity that results from the inevitable lack of complete agreement on what the moral constitution requires provides the opportunity for improvement of the moral constitution of the Open Society. As Hayek observed, “it is, in fact, desirable that the rules should be observed only in most instances and that the individual should be able to transgress them when it seems to him worthwhile to incur the odium this will cause. … It is this flexibility of voluntary rules which in the field of morals makes gradual evolution and spontaneous growth possible, which allows further modifications and improvements.”161 In cases of rule ambiguity one may do so without risking the “odium” normally associated with breaking the rules of a practice of accountability.
4.3 How Diversity Maintains the Open Society