by Gerald Gaus
Diversity, I have been arguing, is the source of improvements in the moral constitution of the Open Society. Diverse perspectives can apply their different understandings of justice to explore new rules. In situations of rule ambiguity, those with a revisionary perspective can take advantage of the space of ambiguity to act on what they see as the best rule that will not run head-on into existing normative and empirical expectations. If other perspectives concur on the advantages of the revisionary rule, the moral constitution can develop in ways that a wide range of perspectives will see as a clear improvement. It is important that this process does not require unanimity; as long as the revised rule is within the socially eligible set of the relevant network, there can be movement throughout the socially eligible set (which, of course, itself can expand) without all concurring with this movement. We can expect constant tension between those perspectives seeking to explore modifications of the moral constitution and those seeking to uphold the current one. These tussles, conflicts, and disagreements about the rules are inherent in the Open Society of diverse moral perspectives, each seeking ways of interacting that move us toward its view of greater justice. So long as this movement is within the socially eligible set of rules, changes do not amount to one perspective imposing its view of justice on the others, which would undermine the relations of accountability on which the moral constitution of the Open Society depends.
It would appear, however, that this constant disagreement about the rules and attempts to change that are resisted show that the Open Society cannot achieve the sort of stable coordination on rules of justice that Rawls sought. Although traditional moral and political philosophy may intelligibly insist that stability is not itself a desideratum of morality or a theory of justice,162 when we take up the search for a moral constitution something like stability is of critical importance.163 Stability is the tendency of a moral constitution to return to a justified equilibrium (in which acting on a moral constitution is the best reply to others doing so) in the face of both internal and external shocks that induce deviation from it.164 If we are seeking to evaluate a moral framework, we need to know whether it provides the basis for a stable justified moral constitution in a narrow or a wide range of conditions; if the former, we may well doubt that the constitution can perform its expected function given the vicissitudes of human life. Now Rawls, and almost all political philosophers who have considered the matter, have supposed that stability is induced by homogeneity and endangered by diversity. A society that shares the same basic outlook on justice, it is thought, can weather storms better than one in which people have diverse perspectives on justice. Rawls’s proposed solution to the problem of stability in A Theory of Justice was to argue for a surprising degree of homogeneity not only concerning justice but in our understanding of the good, which would lead us to remain faithful to justice as fairness, even in the face of injustice by others. His later work, acknowledging greater diversity not only about the good but about justice itself, struggled to show how such a deeply divided society could nevertheless be stably just.165
The idea of a society that is apt to maintain just social relations in the face of endogenous and exogenous disruptions can be understood in two ways: stability and robustness. Let us call “stability” the tendency of a system to return to the same equilibrium, and “robustness” the tendency of a system to maintain an equilibrium (on a justified moral constitution)—a robust system returns to an equilibrium, but not necessarily the same one.166 To better see the contrast compare two societies, A and B. Society A is a highly normalized society. All the admissible perspectives share the same view of justice; not only do they agree on what the best rule of justice is, but they agree in their overall rankings. Thinking the same way, they arrive at the same conclusions. They agree on the best choice; if there is anything else in the eligible set, it is Pareto dominated by the single best element. In such a society there is very little pressure to abandon the optimizing stance (§IV.3.3) as everyone can have their optimal choice. There will be no good reasons for people in A to accept that one should live with less than the best rule. As Baier says, the sole moral rule will be identified with the best rule. And so it may well be that the socially eligible set reduces to a singleton. Everyone accepts, and knows that others accept, one correct view, which many will see as the only correct view (it is in this sense that A is homogeneous). Society B, in contrast, has not achieved moral agreement on one and only one constitution as the best; for all rules in the constitution there is a large eligible set with deep disagreement about the best element. Although for any rule in the constitution there is a working coordination on it as the rule they will live by, normative ambiguity and more explicit movements to move within the socially eligible set are a cause of ongoing efforts by many perspectives to change the moral constitution in what they see as a more just direction.
To fix ideas, take just one rule in the moral constitution of each society, R, say, a rule that marriage is a long-term monogamous commitment between two adults. Suppose further that the current rule in both A and B about this matter is R, but in society A rule R is the not only the universally agreed-on best rule, but the only rule in the socially eligible set, and all have settled on it; in contradistinction, in B rule R is in the eligible set but some perspectives are actively seeking to replace it with R*, another rule in the socially eligible set. Now consider what occurs if, say, under a wave of religiosity (such as has occurred periodically in America) or immigration, a perspective arises that strongly believes that polygamy is the best form of marriage, and certainly is an acceptable form of marriage. Should this perspective grow, society A’s moral constitution will be under stress; it has been premised on the assumption that all acceptable perspectives must agree on what is the best option. It thus must either simply deny that the new perspective is “reasonable” or begin searching for a new consensus on the best marriage arrangement. Its stability now looks illusory—it was based on an assumption that the constitution is a stable response to a certain array of views, but is unable to adjust to the inclusion of a new one. In this sense it is a liberal, but not an open, society (§IV.1.4). In contrast, society B has a number of marriage rules in the socially eligible set, which the perspectives order differently. It thus witnesses debate and contestation; perhaps our polygamous perspective seeks to exploit rule ambiguity, advocating lifelong polygamous cohabitation with claims to common law rights. Society B thus has the resources to adjust to the new perspective, and give it social space to press its case for change. Society B is robust as it can maintain its justification by either maintaining the R equilibrium or moving to a new, R*, equilibrium. The moral constitution is more fragile in A; it has fewer moral resources to adapt to changing perspectives (given that it was based on normalization, this should not be surprising). A society that has significant moral disagreement within a socially eligible set has greater resources to maintain a basic charter for its social world that all can live with. Such a society will often exhibit a sort of punctuated equilibrium, converging on an equilibrium for a sustained period and then, after disruption, gravitating to a new one, but always within the set of the type of rules all can live with.
The worry arises, however, whether this robustness actually invites instability. Consider representative persons Alf and Betty in, respectively, societies A and B, again both of which have justified constitutions with rule R. Suppose at some point in society A people are overwhelmingly acting according to this R-including constitution; Alf, a member of A, will act on it so long as his commitment to maintain a practice of accountability (and the coordination benefits it helps secure) plus his expectation of being punished outweigh his temptation to defect in order to better pursue his perspective’s ideals. Radically simplifying, then, for Alf to continue acting justly it must be the case that (letting p indicate the relevant probability):
(EQ. 1) p[benefits (practice of responsibility)] ≥ p[benefits (defection)] − p[costs (punishment)]
The probable benefits of continued moral relations based on the R-including constitution only need outweigh the probable benefits of defection discounted by probable punishment. Contrast this to Betty, in society B. She has an additional incentive to defect on the moral constitution—the expected payoff that her defection might drive B to, say, her favored R*-including constitution. So for her to have reason to conform to the current R-including constitution, it must be the case that:
(EQ. 2) p[benefits (practice of responsibility under R-constitution)] ≥ p[benefits (defection)] + p[benefits (practice of responsibility having achieved R*-constitution)] − p[costs (punishment)]
The probable benefits of moving society to what she sees as a better moral constitution can provide her with a moral reason to violate the present moral constitution. And even if probability of [benefits (of getting R*-constitution)] is very low, she would still have incentive to pursue it when, given rule ambiguity, others would accept both R and R* as meeting their normative expectations. (Again, rule ambiguity is highly conducive to rule innovation.) In contrast, when it is clear that only action in conformity with R will meet normative expectations, and so Betty can expect to be held accountable and perhaps punished for R*-based action, her incentive to act on R* will be much less (as eq. 2 indicates). To be sure, if Betty thinks there are enough like-minded others so that they could actually move to the R*-constitution and this action will be important in doing so, then she may still act on R*. Other things equal, Betty in society B will thus have more incentive than a comparable Alf in A to defect on the R-including moral constitution. We face the prospect that the very possibility of change to another eligible constitution, which is required for robustness, will tend to destabilize the current moral constitution, inducing people to defect in order to achieve a moral constitution they consider superior. Here then, is our problem: how do we achieve sufficient stability while also allowing us to exploit the moral resources that promote robustness and moral change?
As equation 2 shows, increasing punishment certainly can induce stability on a specific equilibrium by discouraging those who would seek to move to another moral constitution in the eligible set. Although recent analysis shows that punishment is indispensible in maintaining equilibrium on norms and moral rules (§IV.2.1.2), the problem with punishment is that it can potentially stabilize any equilibrium, in or out of the eligible set.167 And, of course, liberals rightly recoil at the prospect of a social order that can be sustained only by high levels of force.
A moral constitution requires significant stability if it is to perform its coordinating task, helping to settle expectations about future interactions while, at the same time, it should possess sufficient flexibility to be capable of responding to disruptions by switching to a new equilibrium, and encouraging the search for a better constitution. Now we can imagine ideal members of a moral order that have precisely the correct trade-off rate between valuing stability and inducing change, but, of course, we do not know at any particular time what this trade-off rate is. In environments with a low rate of change, stability is generally appropriate; in times of storm and stress, flexibility is apt to be more valuable. And, in any event, we should no more expect homogeneity on this value than on any other. But that is not such a great worry, for we do not really need individuals to agree on the optimal trade-off rate. Recent studies in cultural evolution, the philosophy of science, organizational theory, and democratic theory converge in showing that diverse populations—those that are divided between more reformist/innovative agents and those that incline toward conservative or conformist values or behavior—often arrive at better collective outcomes than those characterized by a single type.168
Consider a society such as BDIV, divided between those who are critical of the existing constitution, searching for ways to improve it, and those who place high value on stability and so are very reluctant to move to a new equilibrium. Contrast this to society BCON, an orderly society whose members all value stability, and society BREF, a society of reformists whose members all place high value on achieving what they see as the best constitution (though, of course, the various reformists will not concur on what that is). There is strong reason to think that under a range of environmental conditions, BDIV will outperform BCON and B REF in the sense of better maintaining a justified moral constitution over a sustained period. As Page demonstrates, there are two lines of analysis that support this: averaging and decreasing returns to type.169 (i) Homogenous BCON populations will perform very well (a) in environments with minimal disruptions and (b) assuming that it has initially achieved a justified equilibrium. BCON will perform badly when (c) there are severe and regular disruptions that render the current equilibrium difficult to maintain and (d) the current equilibrium is not in the socially eligible set. On the other hand, BREF does well under (c) and/or (d), but worse than BCON under conditions (a) and (b). In a range of environments homogenous BCON and BREF groups will experience wide variation in their ability to maintain a justified constitution; in contrast the diverse BDIV is almost certain to have less variation in its performance, and it can be shown that systems such as BDIV, with less variation, generally outperform less diverse systems such as BCON and B REF.170
(ii) A similar result can be shown by appealing to decreasing returns to type.171 Suppose we start out with BCON and replace n conservative members with reformist members, where n constitutes a small proportion of the society. This new group, BCON *, will almost certainly outperform BCON; given that there is still a large proportion of conservative members, the small n of conservatives lost will not much reduce the impact of the prostability perspective, but the small n of reformist citizens will make contributions that otherwise would not exist, alerting the other citizens to new possibilities and problems (for example, that the current equilibrium is flawed in ways not previously appreciated). If we think in terms of one’s marginal value to achieving a long-term justified moral constitution, the new n reformist members have a higher marginal value than the conservative members they replaced.172 Both lines of reasoning support the conclusion that, while a diverse society may depart from its current equilibrium, it possesses underlying features that enhance its ability to maintain itself in the face of external and internal changes.
4.4 The Perspectives of Reform and Order
The moral and political constitutions of the Open Society require both stability to maintain an equilibrium, and dynamic exploration of better rules. Jonathan Haidt, drawing on moral psychology rather than diverse system dynamics, has recently come to much the same conclusion: different types of perspectives focus more heavily on one or the other of these tasks and thus complement each other. Haidt’s research provides support for the idea that those devoted to a reformist ideology (which he associates with liberals) and conservatives do indeed have different basic moral outlooks, and that these different moral outlooks yield different orientations to alterations in the current moral constitution. Haidt’s hypothesis that moral reasoning is grounded in six “foundations” or “dimensions” is given in figure 4-6. Haidt finds that liberal subjects display responses and justifications that more strongly focus on liberty/oppression and care/harm dimensions. We must be careful; this is not to say that they are without “intuitions” based on the other foundations, but that their reactions are more inclined to those two foundations (or dimensions) as are, especially, their justifications. Thus on their view, morality is essentially about treating all as free and equal, avoiding harm, and ensuring that needs are met.173 In contrast, Haidt argues, those associated with conservative political views have a stronger tendency to rely on all foundations, both in their reactions and justifications.174
Figure 4-6. Haidt’s moral foundations. Source: Haidt, The Righteous Mind, pp. 153–54, 181–85.
The range of foundations on which Haidt’s conservatives draw bears out the long-held view that conservatives tend to be guardians of the current moral constitution. Conservative subjects are far more apt to be loyal to
current moral rules and respect their authority even to the point of seeing them as sacred (they are not understood as mere social rules).175 And they are ready to expend resources in policing them, ensuring that others do not cheat. As Haidt concludes, their moral outlook supports the importance of protecting our current “moral capital,” our practices that ensure a cooperative and peaceful social life. “Moral communities are fragile things, hard to build and easy to destroy. … If you don’t value moral capital, then you won’t foster values, norms, practices and identities, and technologies that increase it.”176 He continues:
If you are trying to change an organization or society and you do not consider the effects of your changes on moral capital, you’re asking for trouble. This, I believe, is the fundamental blind spot of the left . … It tends to overreach, change too many things too quickly, and reduce the stock of moral capital inadvertently. Conversely, while conservatives do a better job of preserving moral capital, they often fail to notice certain classes of victims, fail to limit the predation of powerful interests, and fail to see the need to change or update institutions as times change.177
Haidt is led to the nineteenth-century idea of the parties of order and of change: “Here’s the most basic of all ideological questions: Preserve the present order, or change it? At the French Assembly of 1798, the delegates who favored preservation sat on the right side of the chamber, while those who favored change sat on the left. The terms right and left have stood for conservatism and liberalism ever since.”178
Haidt’s work in moral psychology leads him to a conclusion that I have reached by another route: that a moral and political order composed of diverse perspectives performs better—from the moral point of view of the Open Society—than a homogeneous order of either alone. As Mill stressed,
