Government is key, even among morally perfect people.

There are at least four kinds of reasons that morally perfect people might have disagreements with each other about practical matters.8 The first and most obvious of these is their cognitive limitations. Given human beings' finite, limited, and differing intellectual capacities, and their differing experiences and information, different morally perfect people would have different factual beliefs about the world and the causal processes characterizing it. These factual disagreements would inevitably lead to serious practical disputes, even among those who qualify as angels according to the same moral theory.

This is most evident in the case of utilitarian or other consequentialist moral theories. Where lightness or wrongness is determined in a straightforward way by the value of the consequences of alternative courses of action, disagreement on what those consequences would be leads directly to disagreement about what it is right and wrong to do.

The same holds true of the other types of moral theories, however, since they too allow that facts about actual or hypothetical consequences are relevant to determining right and wrong in many contexts. On Robert Nozick's libertarian theory, for example, whether a present distribution of property is just depends on whether that distribution sufficiently resembles what would have come about if neither force nor fraud had ever been exercised in determining property holdings in the past.9 Even libertarians agreeing exactly about what constitutes force and fraud could hardly be expected to agree on the implications of this counterfactual test of distributive justice, given the enormous cognitive complexity involved in applying the test. A similar point holds with respect to communitarian rights theories, liberal pluralism, and other moral doctrines that attach importance to the notion of moral rights. Whether a right of mine has been violated or infringed will often depend, among other things, on whether any of my interests have been damaged. But this opens up a whole range of practical disputes among perfect rights-respecters, based on their differing factual beliefs. Angel A may disagree with angel B, for example, about whether smoking in public violates others' rights to life and well-being, because they have different beliefs about the health effects of breathing the fumes of other people's cigarettes.

A second source of practical disagreement among the morally perfect is the truth of the philosophical doctrine of moral pluralism or, as it might more accurately be labeled, "incomplete objectivism." This doctrine allows that while there may be some objective moral truths (e.g., that torturing young children to death simply for the fun of it is wrong), there are also substantial issues (e.g., the morality of abortion) about which different people may hold differing views without any of them being guilty of any sort of error. Persons living in the same society holding opposing, but non-erroneous, beliefs on such matters may well need a government to ameliorate the practical conflicts that arise from these opposing beliefs.

Nor are these practical disagreements necessarily linked to commitment to different types of moral theories. Equally irresolvable disputes may arise among angels adhering to the same general moral theory. This can happen in several ways. There may be irresolvable differences about the scope of a moral theory. Thus, one utilitarian angel may disagree with another about the ethics of animal experimentation or factory farming because one seeks to maximize human happiness while the other aims to maximize the happiness of all sentient creatures. Moreover, the key terms of a moral theory (e.g., "utility," "liberty," or "rights") may be open to a variety of non-erroneous interpretations, each having its own policy implications. Thus, for example, pluralist angels may differ among themselves over whether the important right to liberty requires others to actively assist you in your exercise of certain options, or only requires others not to actively interfere with your exercise of those options. Further, even when there is theoretical agreement about the meaning of key terms, the application of theoretical principles to particular situations may involve matters of judgment about which people may differ without any of them being in error. We would not, for instance, expect every perfect libertarian to draw the line between fraud and legitimate salesmanship in precisely the same place. In particular, we would expect some to endorse the principle of "caveat emptor," while others do not.

A third source of practical disagreement among the morally perfect is the structure of certain of their interactions. Given moral pluralism and/or cognitive limitations, even angels believing in the same type of moral theory will often disagree in their value-rankings of outcomes that may be brought about by their joint actions. As a result, they will find themselves facing interaction situations—such as Prisoners' Dilemmas or public-goods problems—in which each person's moral aims will be better achieved if all are mutually coerced by government than if they act independently in the light of their flawless moral beliefs.

Suppose, for example, that you and I are angels believing in moral theories of the same general type, but because of different factual or moral beliefs,10 I believe that act X (which only you can do) is quite morally valuable, while act Y (which only I can do) is slightly morally undesirable. You have opposite beliefs about the moral value of these two acts, but we share the belief that promises to do what is immoral are not binding. We exchange promises: you agree to do X in your faraway location at the same time I do Y. Without enforcement, neither of us will keep the agreement, for neither will pointlessly do what is morally undesirable, and neither can affect the other's act. But if we have a government to enforce compliance (by imposing sanctions that we regard as "morally overriding"), we will both comply and bring about an outcome that is better from each of our points of view: production of the quite valuable at the cost of the slightly undesirable.

Before leaving the topic of interaction structures, it is worth noting briefly that government can also help alleviate interaction problems that are not the result of disagreement. Such problems of coordination involve independent agents combining their actions in ways that will produce mutually valued outcomes. Since such problems do not presuppose divergent evaluations, angels would be subject to them just like the rest of us. Typical examples of coordination problems include what side of the street we should drive on, what items should be used for currency, which of our offices you and I should meet at, and what time it is (or should be called) now. Coordination problems can be quite complex, and they may be difficult to solve efficiently without some enforcement mechanism, especially if they are not problems of pure coordination, but involve some elements of conflict. (Consider Southern California traffic flow as a salient example.)

Motivated belief—the fourth and final cause of angelic disagreement may be viewed as a special case of our first cause: factual disagreement due to cognitive limitations. But it is a sufficiently important, and controversial, source of factual disagreement, to deserve to be separated out for comment. Motivated beliefs are beliefs that are not determined solely by evidence, but are instead influenced by motivational states of the agent.11 This category is intended to include such familiar phenomena as wishful thinking (believing something because one wants it to be true), sour grapes (believing some object lacks desirable features simply because the object is unattainable), and various forms of self-deception (including holding beliefs because they are convenient or psychologically comfortable rather than because they are supported by evidence).

Since motivated beliefs may allow us to do what we want, while believing it morally right (whether it is or not), they are an important potential source of disagreement among the conscientious—including angels, who are, by definition, perfectly conscientious. It might be objected that to hold motivated beliefs always constitutes a moral failing, and hence that angels could never have such beliefs. I doubt that this is so. Whether holding a given motivated belief constitutes a moral failing, a cognitive error, both, or something else, depends on the details of the particular case. Indeed, motivated beliefs may sometimes be the result of cognitive simplifying or motivational processes that are rational at the meta-level. It may, for example, help us get on with life in a difficult world if we often engage in wishful thinking or have excessive confidence in our judgments about facts.12 Otherwise, we might have more serious morale problems, or waste time and energy trying to resolve uncertainties that cannot be resolved.

Without going further into the ethics of belief, we may say this much. If, as I believe to be the case, having motivated beliefs is not always a moral failing, it can constitute a significant source of practical disagreement among the morally perfect. Together with factual disagreements derived from other cognitive limitations, moral pluralism, and the structure of some interactions, it provides us with ample grounds for expecting even a community of angels to suffer from a considerable degree of conflict and discord, in the absence of an authoritative procedure for settling disputes.