II. Utilitarianism

I take utilitarianism to be the classic aggregationist theory. Admittedly, the word ‘the’ would be more appropriate if it were clear just what utilitarianism claims (that is, if there were a single theory accepted by all utilitarians). Alas, it is not so. Jeremy Bentham, who is reasonably regarded as the godfather of utilitarianism, takes pleasure and pain to be the fundamental utilities, and on his view an aggregate of net pleasure is the quantity to be maximized, regardless of how it is composed. But G. E. Moore held that goodness is an unanalyzable quality and that we are to maximize the aggregate of it—though how one does this in the face of what he called the “Principle of Organic Unities” is a nifty question. That principle states that the value of a whole is not necessarily any regular function of the values of its parts. But how could an aggregate be no such function of its parts?Footnote ² In any case, Moore's intent was clearly that something not equivalent to pleasure is what we are really trying to maximize in aggregate, however difficult it may be to do the aggregating in question. And, of course, in more recent times, moral philosophers have cast the discussion in terms of the satisfaction of preferences whatever they may be. But since preferences are intrinsically measured on ordinal rather than cardinal scales, we again have severe problems conceptualizing the aggregation in question. That is one (but not the only one) of the problems with utilitarianism, and more generally with the idea of aggregation for normative purposes.

For my present purposes, it will be crucial to distinguish two ways in which a Principle of Utility might be held. One way would make that principle an intuitively basic one. That is to say, there would be no further reason why we all ought to maximize utility, however that's done—we “just ought.” But a different way would be to hold that this principle is the best way to realize the function of morality, whatever that function is held to be. The latter thesis would be that even though the Principle of Utility is not, a priori, the fundamental principle of morals, embracing this principle is still the best way, or perhaps the only realistic way, of acting morally. (In current parlance, the Principle of Utility would be taken to be a means to an end, rather than having “deontic” status as itself specifying the ultimate end to be pursued.)

Why is this distinction crucial? It is crucial because on the first way of understanding the Principle of Utility, there is in a sense no arguing with the principle—either you see it or you don't. However, that very fact is the problem, as I shall explain in Sections III and IV below. My discussion in those sections will indicate, in a general way, the function of morality, on what I will argue to be the right understanding of that notion. Now, the question is whether utilitarianism or a principle like the Principle of Utility succeeds in performing this function, and I will argue that it does not. However, I will also argue that (a) there is a range of situations in which utilitarian-like reasoning is indeed what we want, and (b) it is especially those situations that provide motivation for embracing utilitarianism as a moral theory.

What will emerge from these discussions is the need to see that morality has a function essentially different from what seems to be assumed by utilitarians, and that embracing utilitarianism is a serious mistake when we get beyond the (fairly considerable) set of cases in which talk of aggregating utilities is not out of place. That there are many cases in which we should not be aggregating is a necessary condition for the rational rejection of a utilitarian moral theory. In order to provide cases of that kind, we need to be able to make sufficient sense of aggregation so that we can see why we should not be aggregating in such cases. And that is not easy, as I will also argue. (I think some people believe it is easy,Footnote ³ and others, though they agree it's not easy, appear to think that we can just barge ahead and reject aggregation anyway.)

III. Intuitive Aggregationism

I will use the expression “intuitive aggregationism” to designate the first of the conceptual attitudes distinguished above, according to which aggregate utility really is the ultimate end of morality. The thesis is that we ought to seek the general happiness, or some other aggregate outcome, not because to do so is conducive to something else that is really prior, but rather for its own sake. We can distinguish, at least rhetorically, three variants of this thesis.

1. (i) It might be said that this aggregate, or the maximization of it, just is “the good” or some such thing.

2. (ii) It might be said that there is simply a direct obligation to maximize aggregate utility.

3. (iii) Finally, it might be said, as Henry Sidgwick argued,Footnote ⁴ that the formal features of morality are such that, when put together, they imply this result.

Each of these variants deserves a brief comment.

1. (i) The problem with asserting that something or other is “the” good is that some to whom this is proclaimed won't see it that way. And certainly, many will not see maximizing any specified aggregate that way. They may assert that some small subgroup of humanity is the true repository of goodness; or that it resides exclusively in themselves; or something else. If we are to persuade these people that they are in error, we need an argument. But the problem with assertions of fundamental goodness is that they resist argument, by their very fundamentality. Fundamental goodness is, after all, supposed to be at the end of the line; nothing further can be said on its behalf. This, it seems to me, is a fatal defect. It might not be so if the result were indeed obvious to all—but the trouble is, it's not.

2. (ii) This is a “deontic” version of variant (i). The idea here is that there is a fundamental duty to maximize utility. And this variant faces the same problem. Why, we ask, are we supposed to devote our efforts to the general good, however constituted? “Well, you just are!” is not a satisfactory answer. Ethical theory is littered with the theoretical corpses of proponents of such views, and the world, alas, is littered with the real corpses of those who have fought for or against some of the many prophets and dogmatists who have pressed, with the sword, for such views.

3. (iii) This leaves us with those who argue for an aggregative principle along Sidgwickian lines. Sidgwick appeals to an abstract principle of justice: “It cannot be right for A to treat B in a manner in which it would be wrong for B to treat A, merely on the ground that they are two different individuals, and without there being any difference between the natures or circumstances of the two which can be stated as a reasonable ground for difference of treatment.” And he couples this with an axiom of prudence: “impartial concern for all parts of our conscious life.”Footnote ⁵ (The idea is that a unit of utility in the future, say, should have as much weight as the same unit in the present. Whether we are also to be impartial as between the pleasures of pushpin and those of poetry, as urged by Bentham, is less clear.) Putting these two supposedly self-evident principles together, Sidgwick concludes “that each one is morally bound to regard the good of any other individual as much as his own, except in so far as he judges it to be less, when impartially viewed, or less certainly knowable or attainable by him.”Footnote ⁶

What should we think of Sidgwick's axioms, and this conclusion? We may certainly say that the good of others is “less certainly knowable or attainable” than that of ourselves, at any rate almost always or at least typically. But that is not what's wrong. What's wrong is that we certainly are “bound” to no such thing as Sidgwick claims we are: we do not and need not have an impartial concern for the good of all, and it is unreasonable for anyone to expect this of us. To be sure, following the leads of any number of moralists down through the ages, we may well agree that we should be concerned about others: namely, we should refrain from imposing significant and uncompensated damages on them. But that is hardly the same as having an “impartial concern for their good.” The concern I have for certain things that I think to be good, and am willing to devote a great deal of time and effort to, is quite unmatched by concern for the similarly enthusiastic pursuit of various goods by just about anyone else. I don't usually share others' enthusiasms, and when I don't, I certainly don't think that I am “morally bound” to do so.

The same can be said for a famous argument put forward by R. M. Hare. Hare considers a case in which someone else is suffering—for example, because his neck has been broken in an automobile crash. Hare's question is whether and why I should be motivated to sympathize with the poor man's suffering to the point of being ready to endanger myself seriously by attempting to rescue him. Toward this end, Hare seeks to get utilitarian blood out of analytical turnips by insisting on a distinction between these two propositions:

1. (1) I now prefer with strength S that if I were in that situation, x should happen rather than not.

2. (2) If I were in that situation, I would prefer with strength S that x should happen rather than not.

He asserts that much confusion has stemmed from failing to appreciate this distinction—that is, the distinction between what I do in fact now prefer about some possible situation, and what I would prefer if I were in that situation. He then says, “What I am claiming is not that these propositions are identical, but that I cannot know that (2), and what that would be like, without (1) being true, and that this is a conceptual truth, in the sense of ‘know’ that moral thinking demands….”Footnote ⁷ After he has said this, he invokes his familiar analytical principle that “[m]oral judgments are … universalizable in only one sense, namely that they entail identical judgments about all cases identical in their universal properties.”Footnote ⁸ And he supposes that once we have this in place, there is no option but to (in effect) embrace the Principle of Utility. The preference that I have for myself is, as it were, transmogrified by the universality of moral concepts, so that I have to accept that something should be done to me if I will that it should be done to someone else. “We have established,” Hare writes, “that, if I have full knowledge of the other person's preferences, I shall myself have acquired preferences equal to his regarding what should be done to me were I in his situation; …”Footnote ⁹ And what I want for myself, of course, is that all of my preferences be fulfilled—the very thing that universality compels me to want with regard to the other person. And the other person, of course, is every other person, and so we are off and running.

Or so Hare thinks. But I believe he is wrong. For his supposed move from (1) to (2) is actually fallacious. The problem with “I now prefer with strength S that if I were in that situation, x should happen rather than not” is that being “in that situation,” as Hare says, is being the other person. But since the probability that I am that other person is zero, nothing whatsoever follows from the fact that if I somehow were the other person, I would want this or that. I am not he; so I don't want it; and that is the end of that, if this is intended as an appeal to practical reason.Footnote ¹⁰

For another interesting and much more recent example, consider Elijah Millgram's analysis of John Stuart Mill's famous (or infamous) argument for utilitarianism in chapter four of Utilitarianism. Footnote ¹¹ Millgram proposes that the proof of the apparently dubious thesis that happiness is a “good to the aggregate of all persons,” as Mill's phrase has it, should be understood as leaning heavily on Mill's idea that the appropriate test of the “quality” of pleasure is the majority opinion of what Millgram calls the “experientially privileged who provide the test of desirability.” Thus, Millgram writes:

On this view, Millgram suggests:

As for this reconstruction of Mill's proof, which Millgram claims to be “deductively valid as well,” I should point out that on no interesting reconstruction is this deductively valid, unless what is meant is the following syllogism:

1. (1) The appropriate proof of the desirability of x, to anyone, is the preferences of the experienced.

2. (2) The experienced desire the general happiness.

3. (3) Thus, the general happiness is good, to anyone.

Since both (1) and (2) are, to put it mildly, not obvious, the deductive validity of this argument is not of much interest. Nonetheless, claim (2) is an interesting one. The trouble is, though, that the “experienced” (who in this case, where we are judging arguments, should surely be the logically savvy) do not think that the general happiness has anything like the status that the Principle of Utility wants to credit it with. Those experiencers and judgers, I surmise, will note that far from the general happiness being the super-supreme good desired by all, it is thought to be either a muddle that cannot really be the “object” of any sane person's desire, or an object whose desirability to the agent doing the judging is decidedly dubious.

It is hardly surprising that arguments of this type cannot work. If we are to persuade people to go along with some program, then we need to somehow hook it onto their interests as they are—not as they might be if some highfalutin condition is imposed. We need, then, to look for principles that appeal to people as they are, and the appeal has to be rational, in the general sense that our premises are true or at least plausible, and can, without additional question-begging normative premises, be accepted by those to whom they are addressed.

IV. The Separateness of Persons

The fundamental stumbling block for genuinely aggregative theories is the fact that practical deliberators and choosers are individuals, each different, and each concerned with a range of goals, or values, or utilities that is unique to that person. Thus, for example, John Rawls criticized utilitarianism for insufficiently taking account of this familiar fact (of which we are all intimately aware), namely, that persons are distinct from each other:

Rawls adds:

Some critics have claimed that Rawls himself takes insufficient notice of the separateness of persons. A notable example is David Gauthier, who argues that “[t]he idea of agreement is idle in Rawls's theory because the individuals who are to agree on principles of justice are placed behind a veil of ignorance so thick that the differences among them are obliterated….”Footnote ¹⁵ Impartiality requires acceptability in every position rather than acceptability in ignorance of one's position. The former does and the latter does not “take seriously the distinction between persons”—to turn against Rawls the criticism that he levels at the utilitarians.Footnote ¹⁶

Gauthier surely seems to have a point. The question to be pursued here is: Just what is this “fact” of our individuality and separateness, and how does it affect moral theory? That I am myself and not, for example, Peter Smith over in the next town, or Mikhail Baryshnikov or Catherine Deneuve, or for that matter our cat Widgit, seems terribly obvious. If any moral theory were genuinely to contravene such facts, surely that would be game over for the theory. You don't, one would hope, fly in the face of anything quite that obvious. And yet some theorists are said to have done this. A question that surely should occur to us is, how could they have managed it?

There are, actually, reasons that account for the possible relevance of the charge that some theories, such as Rawls's, ignore or do not do justice to the separateness of persons. I aim to spell these reasons out. I start by addressing a crucially relevant distinction: the distinction between the “fact” of separateness as it affects what we may call the metaethics of a moral theory, and the fact as it affects (or is claimed to affect) the moral substance of a theory—what we might call the theory's “output.” The fact of the separateness of persons is absolutely crucial to the first of these; but, as we will see, it is a matter for discussion how it bears on the second (which is no doubt the more important, from most people's point of view). Individuals who fully recognize the fact of separateness seem somehow to manage, for example, to profess adherence to communism or other kinds of collectivism, and to egalitarianism, and of course to utilitarianism. This is interesting and significant. Yet it should be impossible, on some understandings of how the separateness of persons affects moral theory. Those understandings must therefore be mistaken. What, then, is the truth of the matter?

V. The Separateness of Persons and Metaethics

To begin again, then, just what is our subject? It is moral philosophy, but of course ‘moral’ is not used in an entirely uniform way: it is often, for example, used synonymously with the term ‘ethical’. In part, one must simply decide which of the subjects identified by the terms is the subject one wishes to discuss, and must try to be consistent in one's usage. The subject or field here identified with morals is, I think, by a small margin the one most usually so identified. The “golden rule” and the Ten Commandments (other than those having to do specifically with religious practice) are entries in this field, as are the Principle of Utility and the Categorical Imperative. The subject or field is, in brief, Rules for Everyone, or, more precisely, normative claims addressed to everyone. Here ‘everyone’ can refer either, literally, to everyone, or to everyone in some lesser group, usually a cultural group of the kind that anthropologists study, that is, a group whose members are basically born into it rather than intentionally joining from the outside. For present purposes, though, the issue of cultural relativism is, I think, a side issue, and I shall simply speak indifferently of everyone as the intended subject of basic moral requirements and recommendations.

VI. Morals Versus Ethics

The subject just described is to be distinguished, perhaps sharply, from another kind of theory, which I shall call the “theory of life,” or “how to live,” or perhaps the “meaning of life.” I take Aristotle's Ethics to be largely about this much broader subject. And because it is broader, it includes the narrower subject: among the things an individual should bear in mind in his or her life are whatever distinctively moral requirements he or she may be subject to. Nevertheless, different considerations come into play when we inquire into the matter of appropriate rules for all as compared with relevant rules for you as an individual. The distinction I have in mind here is, I think, nicely captured by Immanuel Kant in his distinction between what he calls “maxims,” which he characterized as “subjective principles of volition,” and what he calls “practical laws” (or “categorical imperatives,” in the plural), which are the rules for everybody that, he believes, are generated by what he famously calls the “supreme categorical imperative”Footnote ¹⁷ when taken together with the relevant facts about people and their situations. Of course, there is also the frequently discussed question of whether ‘rules’ conveys the right idea. Some hold that virtue, for example, is the more fundamental idea.Footnote ¹⁸ In general, though, I think we can bypass that subject here. All we require is that the system of norms that the moral theorist hopes to come up with provides publicly transmissible guidance to people, in the sense of purporting to enable them to know, at least very often, what to do and what to avoid in general interpersonal contexts.

Still, it might be asked, is there a genuine difference between ethics and morals? After all, those who try to tell us how to live, try to tell all of us how to do so, and that is what I have suggested is the essence of the distinction. The point is well taken, and we need to be clearer. We can sharpen the distinction in the following way. Ethics addresses the individual, as it were, by himself. Other people are quite incidental, though of course relations with others will usually be major matters in an ethical theory. But whatever such a theory says, it says it under the aspect of supplying us with a recipe for filling our lives with as much value as possible, letting the chips regarding other people fall where they may. A moral theory, by contrast, takes it that our relations to others are the central issue: How are we to treat each other? What are we to do in the face of the fact that there are other people out there? A theory of the second type envisions a decided limitation on the set of “chips” allowable by the right theory (or theories) of the first type.

Of course, theories can differ greatly in their answers to these questions, and according to one moderately popular theory, the substantive distinction between ethics and morals simply vanishes. Full moral egoism, for instance, says: Do whatever is best for yourself, regardless of the implications for others. Thoroughgoing utilitarianism, by contrast, says that we ought to treat a given increment of happiness for anybody as on an exact par with a like increment for anyone else. However, these are both moral theories, in the sense intended here. As an ethical theory in the foregoing sense, utilitarianism looks wildly out of place: Is the way to my personal happiness really to be found in judging equal amounts of utility to be of equal value for deciding what to do? Actually, it cannot be found in this approach, for there must be some utility generated by something else before anything can be generated by the utilitarian's rule: utility begins, as it were, at home. Egoism, by contrast, seems redundant if taken as an ethical theory: its object is to find the best life for oneself, and for that purpose it looks as though egoism is not one candidate among others in the field, but a name for the very field itself. However, it could also be construed differently: egoism could be taken to recommend to all agents, positively, indifference to all others. But then it would, again, have become a moral theory, not just an ethical one—and would be rejected overwhelmingly by those to whom relations with others are hugely important, as well as by those (most or all of us) to whom it matters what others do to them. On any reckoning, then, one's life would appear to be far worse if one took this recommendation seriously.Footnote ¹⁹

It is of the essence of morality, in the distinctive sense I am concerned with here, that it be interpersonal. It is also intended to be strong, in the sense that, generally speaking, if some option is clearly the morally required one, then even if another option is quite appealing to the individual in question in some other way, he or she is to do what morality requires, not what he or she would like. However, morality need not (and I think rationally cannot) override any and all “inclinations,” as some statements made by Kant (or at least some interpretations of those statements) may suggest. But if a moral theory does not purport to override any inclinations, then it is just not an entry in the field of moral views, whatever else it may be.

Morality is a set of rules for the group; rules of this kind for each person are to be the same as for anyone else, and they are to be available as a basis of appeal in cases of interpersonal conflict. Given this, what are the rules to be? The issue I want to address is how the separateness of persons affects this question. However, the very formulation of the distinction envisioned here, between ethics and morals, virtually dictates that the separateness of persons will affect the rules of morality, since the fact of group life, life lived in relation to others, is the defining background feature of morality. Without interpersonal relations, morality is not a subject, though what I have called ethics would be.

I have not so far mentioned politics. That is because I take it that normative political theory must follow from moral theory, generally speaking. Morality should tell us what we may or may not do if we occupy positions of political power. Nevertheless, the nature of our political arrangements is bound to make a major difference, as I will discuss later on.

VII. Why the Separateness of Persons Matters

How does the separateness of persons affect moral theory in this narrow sense? That is our main question. But we cannot hope to get any sort of handle on this question without some account of the “rigging” of morals. Where do moral directives come from? Who issues them, and why? How do those addressed by moral directives stand in relation to them? Here, in fact, we have answers, ones that are crucial to our inquiry.

For the project of morality in particular, we can see some things with eminent clarity. First, there are no “moral authorities,” no small subset of persons who get to legislate the moral law, or enforce it, or, for that matter, interpret it. Clearly, everyone is in on the moral enterprise in such a way that no one is an “authority.” All sorts of people propose moral ideas, and the fact that those ideas emanate from Smith rather than Jones, or the pope, or the president, is quite irrelevant—though of course some people's views are more plausible than others, better thought through, better informed, more coherent, and so on. What matters, in short, is the general merits or demerits of a moral idea, alongside which who the idea came from is simply beside the point. Morality has no identifiable set of legislators or enforcers; everyone must “legislate,” everyone must judge. Morality is, in that sense, a strictly informal, decentralized set of directives, to be applied by everyone to everyone (themselves and others).

Second, morality proceeds in recognition of one significant aspect of the separateness of persons, namely, that people are in fact qualitatively quite different, in various ways. This is not the same thing as sheer separateness of persons as such, and that is important. “Sheer” separateness of persons consists in the fact that all persons are individual entities, no matter how similar they may happen to be to all or any other such entities. Even if ten thousand people are of like mind, it is still the case that what is in the mind of A is not, as such, in the mind of B. Even if A and B think alike, there is still one thought in A's mind and another in B's, and though the two are very similar qua types of thought, nonetheless they are two rather than one. Person A might die or change his mind and person B continue to think just as before; metaphysics does not prevent this. Let us call this “particularity” to distinguish it from difference. The particularity of persons hugely underdetermines personal characteristics. People range from “as like as two peas in a pod” to “they come from different planets”—and yet, somehow, morals are addressed to all. The fact that people are different in all sorts of “ways” matters; for despite all the differences, the idea is that certain rules nevertheless apply to them all. Which rules, and why, and how, are matters that the moral theorist must struggle with. A moral theory which relied on the claim that all people are in fact alike—when they are actually, or at least apparently, quite different in the respects in question—would be almost certain to go wrong. If that theory claimed that all people must do x, then its foundation for that view would have to be rooted in something about people, despite their differences, that gives them all good reason to act (or at least to appraise conduct) together in that particular way.

Third, how do those addressed by moral directives stand in relation to morality? The answer virtually follows from the previous ones. We “expect” moral requirements to be lived up to. If someone fails to do what we suppose to be his duty, then either he must persuade us that our judgment is in error, or there must be a change in his ways (perhaps he must make amends). But in order for him to have this attitude, he must be reachable via his own reason, that is, his own faculty of thinking and deciding about actions. People can consider moral claims addressed to them, and either accept or reject them. If they purport to accept them and yet act in ways that are contrary to what they have allegedly accepted, we may doubt that the claims have in fact been accepted. Hypocrisy, a lack of fit between our claims about what we ought do and our actions, is not tolerable. But it is also not very surprising; on many occasions, it will be easier to dissemble than to do what we ought. The dissembler, though, knows that he is dissembling, and will be discomfited if caught in the act. In any case, if the moral principle he claims to accept is a reasonable one, he will appreciate that he has nothing to say in his own defense. The rules of morals are intended to be the basis of any appeals we may have in the court of general reason—“public” reason, as it is now fashionably known.

The consequence of the foregoing discussion is that a responsible moral theory needs to find a set of principles that everyone subject to them can “live with,” in the sense of finding reason to go along with them—at a minimum, reason to support these principles in the sense of voicing agreement with them and expressing approval and disapproval as appropriate. There has to be something in it for the individual who joins up, as it were, with the moral team. Otherwise, morality would lose its interest, and its grip (or whatever grip it has).

VIII. Agents and Patients

One major point that this puts in perspective is that morality cannot be viewed essentially from the point of view of recipients as such, that is, of persons considered only as affected by the actions of moral agents. Those agents must instead be shown that the recipient point of view deserves to be taken into account. Those who receive the benefits of moral action cannot be nothing but recipients, else we lose the allegiance to principle of those called upon to supply these benefits. This is an error of many moral theories today, and of utilitarianism in particular: curiously, these theories seem, in a sense, to leave the agent out of the picture. Or rather, they seem to present assorted moral considerations as if their quality as authoritative moral directives for agents had nothing to do with the fact that the agents are agents. The considerations are presented as compelling independently of what an agent is interested in, wants to do, or regards as important qua agent.

But this is still misleading as it stands. Of course, morality is always addressed to moral agents, for they, by definition, are the only beings capable of understanding and responding to (or ignoring) moral claims, criticisms, requirements, and assessments. The moralities of which I and others complain (those that emphasize only the receptive as opposed to the active, deliberative, decision-making, responsive, and creative aspect of humans) in effect think that a moral agent can be commanded by the information that somebody somewhere would benefit from a given action, quite independently of whose benefit it is.

This is curious because agents are precisely what create the need, the occasion, for morality, and they are what make it possible. It is we doers and deliberators who are the constituency of morality. A purely passive recipient of this or that action has nothing to say because, insofar as it is such, it cannot say anything and could not do anything about it even if it could say something. Addressing moral rules to such entities makes no sense: we do not reason with vegetables or microbes or even lions (nor with “human vegetables” or infants in arms). Agents might decide to take patients into account in one way or another, for one reason or another (e.g., infants are typically loved by their parents). But to justify requiring agents to do so, some argument will be needed. We cannot just assume from the start, without further explanation, that patients must be treated in some particular way. As Gauthier observes: “There is no way of being acted on which is as such either rational or irrational; there is no rational patient corresponding to the rational agent.”Footnote ²⁰

The fact that we are distinct persons, then, makes the following difference. We each have a (normally very ill-defined) set of interests, desires, and values, which is not only numerically distinct but qualitatively quite different from that of anyone else; we also have a varied assortment of skills, talents, and capabilities, again unique to each of us. Yet despite all this divergence, morality is a set of uniform rules—because that is its point. Morality is to be our guide in going forth into the social wilderness, enabling us to deal with all and sundry despite having never met them. Without such a guide, we would have great difficulty getting through life at all, and certainly any life that is restricted to interaction with a tiny handful of others and no encounters with the wider social world must be extremely impoverished. We need to carry with us a fairly simple set of rules that will equip us to deal usefully with these others. If each person, in addition to all the unique materials mentioned above, is also completely unique in his or her rules of operation, then each person's behavior is utterly unpredictable, and hence a complete mystery to every other person. Interaction under these circumstances would be impossible; with less utterly diverse personal sets of rules, it would become merely chaotic. The point of morality is to impose an order, a uniformity, on the community, as Aquinas puts it,Footnote ²¹ thus enabling us to know, more or less, what to expect, and therefore what to do. But it is a uniformity that is, in the end, imposed by each individual on himself or herself.

That the others are others in all this is crucially important. For what it means, above all, is that my set of interests, values, and desires (or to use the modern term, utilities) is wholly separate from theirs. That I want something has no logical connection with whether any of the others wants it. Very likely we are all (or at least many of us) built similarly enough so that we have some common wants. But this observation, in any sense in which it is true, can be extremely misleading. Each other person, like myself, needs food, protection from the elements, frequent rest, and so on. But what he needs is rest for his body, not mine, food in his stomach, not mine, protection of his body from the onslaughts of nature, not mine. And it is my set of utilities, not his, that guides my action. Indeed, to suggest that people's actions “should” be guided by something else, something other than their own sets of interests, is on its face simply unintelligible. By definition, someone else's interests, simply as such, cannot have any effect whatever on my actions: only if they somehow impact on my interests, or my emotions, can they have an effect. And that is not a crossing of the border between us, but rather a transaction of some kind: a receiving by me of a signal from them, which I process in my own way and feed into my decision-making database. And this leads me to act in response to one or another of my own concerns, which now incorporate this new information.

The fundamental significance of the separateness of persons, then, is that it restricts the field of moral theories that can have any chance of being significant to those which can somehow appeal to all persons who are expected to respond to the directives that morality proposes to underwrite.

Having said this, we have to take account of the fact that people have subscribed to irrational or bizarre moralities of all sorts. We can hardly say that all moralities have been rational. The project here is to set forth a view that is, by contrast, rational—a view that has the support of reason. Most moralities, of course, are social: their adherents reinforce each other's behavior. And this makes it arguably rational for people to subscribe to them as well—this being, perhaps, the easiest way to get along in society. Doing what others do, and what they will punish us for not doing, is standard social behavior and can hardly be dismissed as irrational.

Still, the moral codes to which people subscribe may nevertheless be rational or irrational. All moralities that we know of condemn murder, at least at the local, intratribal level, but many fail to condemn it if practiced against the tribe over the hill. All moralities have assorted restrictions on sexual behavior, but the specific restrictions vary a great deal, and it is arguable that some restrictions don't make much sense, while others do. And so on. Thus, our project may admittedly be somewhat narrow: we seek a rational code of behavior, despite the evident lack of rationality in many moral codes that have seen more or less widespread practice. Is this possible? Does it make sense? The assumption is that it does.

This assumption becomes pretty robust for the special case in which groups or individuals with mutually inconsistent moralities nevertheless come into contact with each other. What then? We may now do well to try to reason things out, and find ways in which the existing mores, which bring us into conflict, can be modified to bring about the unity required for mutually productive relations. And if we suppose, as is plausible, that any group anywhere might well come into contact with any other group, given the current era's technological virtuosity, then it also becomes plausible to think that a general moral construction aiming to find the potential common ground for all would be a good idea. Philosophers today do, after all, write as if a universal morality were the subject being explored, by and large—even if some philosophers come to the unhelpful conclusion that the only thing to do is go with the local rules whatever they are. It is in this spirit that I pursue the matter. And it is especially here that the separateness of persons looms very large.

We hear of two supposed types or classes of moral theories, dubbed “consequentialist” and “deontological.” Of these, the first is generally taken to include consequences befalling assorted people in general, while the second purportedly tries to command us independently of our “inclinations” (in Kant's terms), or, more generally, to command us to do things under various descriptions (conspicuously not including the description that you, the moral agent being addressed, care about what is to be done). Whether or not an action would live up to this latter description, we are to suppose, is a matter of indifference to some people. But the separateness of persons, prima facie, reduces both consequentialist and deontological appeals, as they stand, to nonsense. Both sorts of appeals imply what may not be, and prima facie is not, true—namely, that individuals have some sort of interest in acting along the lines indicated. If consequentialist or deontological theories are to have any claim on our attention, this deficiency must be made up. It almost never is; rather, it is only presumed to have been made up. But the need for an account of how people can be assumed to have such an interest looms large if we want to have coherent theories.

The desperate moral theorist may try to maintain that somewhere in the depths of the human soul is a yearning, an impetus, a magnetism, attracting us despite the foregoing considerations to undertake acts of a certain description or acts with certain consequences. But ascriptions of such a yearning cannot be simply arbitrary, as they must inevitably appear to the clear-eyed. A better story must be told. And that further story will inevitably be one that connects your utilities with mine, in that yours lead you to do things which in various ways affect me, with the result that I had better be aware of this fact and consider what to do about it in response.

Thus, I am told not to go around killing people, and that indeed is a plausible rule. But why? Not because it happens that I don't like killing people, or because it doesn't happen to serve my interests to kill people—even though in my case, and no doubt that of most readers, that is perfectly true. Nonetheless, apparently quite a few people differ from me in that regard, and what are we to say to those people?

Nor is the rule against murder plausible because I have a moral sense that murder is a bad idea. For even if I do have this sense, it remains (a) that others may not have it, and (b) that moral senses, as I have noted, can be way off base, and we would want to be sure this one wasn't. We would need, then, a decent account of why it's a good idea to equip people with such a thing as a moral sense that killing people is wrong. Such an account cannot consist in simply reiterating that we have such a sense.

But there is a plausible way to establish a rule against killing people. The foundation of this rule is that I, and all of us, care a good deal about not being killed ourselves; we have enough sense to see that others are quite capable of visiting just that effect on us, and thus we have the basis for a general proscription of this sort of activity among all and sundry. Despite our separateness, indeed because of it, the plausibility of this rule as a socially supported proscription emerges clearly.

Thus far, I have brought out why we should accept that the separateness of persons is crucial in the metaethics of morality. Morality is essentially interpersonal: it calls upon all of us to do this or that. The persons called upon to follow moral prescriptions, however, are separate: one person's beliefs and values are both distinct and invariably different from another's. That is a basic fact about morality, a fact without which there would be no such thing as morals. And that fact restricts, drastically, the sort of arguments that can be effective in this area.

IX. Normative Morality and the Separateness of Persons

What about the substance of morals? Are there some moral prescriptions which not only recognize but are based on some kind of appraisal of or attitude toward the separateness of persons as such? This is a considerably more complicated matter. What we have seen so far is that, owing to the fact that others are different, we cannot simply assume that something we value is something others will value too. And we have noted that people rationally act on their values—not, as such, on the values of other people. If I am to act on “your” values, I have to make your values somehow mine, in the sense that I, upon consideration, adopt values that you have; or I have to value what you do; or I have to value the fact of your having values, or those values in particular, and acting on them. This last seems to be what the theory of utilitarianism says: I am to attach to your values the same weight that I would attach to my own, no matter what your values are. In typical cases, of course, this will be very difficult to do. Indeed, that is an understatement. Doing such a thing would strike most agents as positively crazy. And when it does strike us that way, the question why we should do it comes to the fore. To answer this question, someone will have to explain to me why I should integrate your values into my own.

It is difficult to see how that can be done with any hope of generality. Among the many differences there are between persons, after all, are differences in this very tendency—the tendency to identify with others. Those who do identify with others, what's more, identify with decidedly differing collections of others. For example, people often identify with family members, with fellow residents in their communities, with the people in their province or their nation or their culture. But if you identify with one such group, you do not identify with the others—if I identify with my family, I do not identify with yours. And for each such group, we will find plenty of people who do not so identify, even with those we may think them likely to identify with. The comeuppance of communitarianism as a general philosophy lies in the fact that there are far, far too many communities to identify with, all of them different and many of them more or less at loggerheads with many of the others.

Morality, then, can only work if it is somehow an “agreement” among all. Removing the quotation marks to get a workable theory is hard work, to be sure. But the gist of it is that morality, as a set of general prescriptions addressed by all and to all, will have to cope with the fact of difference, and there is no other way to do this but to assess how our actions, despite or because of our differences, might affect each other, and to look for some principle that will enable each of us to do well given the presence of the rest. The agreement cannot be predicated on falsehoods. It cannot assume that all people are siblings, or friends, or fellow worshippers of the same gods, and so on—the facts show otherwise. The moral agreement is going to have to address us in full recognition of this diversity.

It has long seemed to me that the basic way to do this is the way taken by Thomas Hobbes: what we are rationally impelled to agree to are “articles of peace”—that is to say, what we can agree about is a general proscription against intervening in the lives of others in such a way as to make those lives worse, in the view of the person intervened upon. This leaves each of us with the maximum scope for entering into good relations with others, relations that tend to make life better for those concerned. The “others” with whom we do enter into such relations are, by necessity, various others, who have in common that they are people with whom the agent in question happens to want to have those relations, and who, in turn, want to relate in those ways to that agent. Society, then, is to be an association of consenting adults, adults who consent to various specific kinds of activities in relation to various particular other people. No matter who we are or how different we are from others, we are all liable to damage of readily recognizable kinds, and such damage invites retaliation—and cycles of damage-imposition can only make things worse for all participants. It is reasonable to agree to refrain from such counterproductive activity, and to leave positive activities to the arrangements of people interacting with voluntarily participating others. This approach to morality has the virtue that it stems from realistic observations about people. It requires no one to amalgamate herself with others; interpersonal comparisons of utilities are not made in the “social contract” theory, but are left to the proclivities of individuals. It is difficult to see how we could do better than that for society at large (meaning not our society but society in general—all of society, everywhere).

Any number of theorists have attempted to reconcile liberty and equality by pointing out that it is possible for people to have egalitarian proclivities. And to be sure, the separateness of persons is logically compatible with everyone's wanting to abolish differences among people. But logical possibility is woefully inadequate as a foundation for general morals. It is equally logically possible that we should all be Seventh Day Adventists or fanatical chess players—or, for that matter, Nazis. We must base our moral theories on truths about people, and not only on truths about what is logically possible. Given what we know about people, imposing equality on them is a nonstarter: there is no reason why people should agree to it, given the way they are, and there is ample reason why they should not. As George Bernard Shaw wrote: “A government that robs Peter to pay Paul can always depend on the support of Paul.”Footnote ²² But for the same reason, it cannot expect much support from Peter.

On the one hand, why philosophers, of all people, should side with Paul is an interesting question. On the other hand, there is ample reason for us to want others to refrain from doing the horrible things to us which we all know they can do, and which we know they might have reasons to want to do. The idea of morality is to implant internal dispositions to refrain from these activities. Insofar as this is successful, it enables us to rely on each other to leave us free to go about our business, and thus enables mutually beneficial relations to take place. Those who want to propose a program calling for more substantial sacrifice of one's own prospects must resort to coercion, since people will not do such things voluntarily. A rational moral theory cannot be forthcoming on that basis. But it is also evident that we often, especially among near neighbors, have much to gain from more substantial levels of cooperation than sheer mutual forebearance. The mutual noninterference agreement, of course, leaves us free to negotiate more substantial levels of cooperation.Footnote ²³ For that matter, we can view some local mores as just such agreements. Some of these mores, no doubt, are imposed, remnants from bygone ages with no contemporary relevance, and so on; but many represent what amount to consensus agreements to treat fellow tribespeople (or fellow citizens, etc.) in ways they appreciate, with reciprocation expected among all.

Contemporary philosophers tend to claim that some individuals in a “state of nature” will insist on a stronger agreement than my proposed mutual forebearance. But they also tend to think that these stronger terms are such that they can be imposed on everyone, which means: imposed on people with or without their agreement. It is a mystery how this is to work. Everyone, including the sick and the desperate, can be made yet worse off by anyone of a mind to do so. Most of us will not have minds to do so, and will willingly assist those in grave distress. But if we are not willing, why should we think that the distressed and their friends should be given free reign to extract what they claim to need from us? Especially, one might add, when it is we who are not in distress, but instead engaged in continual useful efforts that benefit those with whom we interact, who set the technological stage for enabling those in distress to be benefited. Widespread coercion will not lead to cures for cancer.

Note that the program envisioned here does not require us to have any particular attitude toward the separateness of persons as such. John Stuart Mill devoted a chapter in his essay On Liberty to “Individuality, as One of the Bases of Well-Being.” And perhaps it is. But then again, perhaps it isn't. Individuality is a right, of course, but so is the opposite. People are free to be joiners as well as individualists or, for that matter, recluses. The general set of rules for us all must not bias things toward one or another of these types. It can accommodate us all by insisting on general freedom for all. I do not see how it can do so in any other way, for whatever other approach we might want to take is going to impose costs on some to the benefit of others, thus defeating the universal agreement envisioned.

One may note that the historically popular sets of codified moral rules or “commandments” tend to contain a core of just the type we have been discussing: they typically forbid murder and damage to others (or, at least, others in the tribe) and enjoin us to keep our agreements. If they say, “Love ye one another!,” this is rarely understood to call for much more than general respect and helpfulness to neighbors in times of need. Beyond that core, moral codes vary considerably, often as a function of a logically extraneous religious component, and frequently as a function of special environmental factors (e.g., mutual aid at a very high level is reasonably made mandatory in the Arctic and the Sahara). And any additional requirements imposed by such special factors are generally manageable by visitors or outsiders without much strain. But the core represents what varying individual people can see to be essential to successful life among their fellows. Anything substantially more than this will be highly variable among groups: it will be the ethics of tribes rather than of humanity. People differ too much to allow more.

Being distinct individuals with distinctive profiles of interests and capabilities has not, we must admit, prevented some from trying to take up communism, others from proclaiming utilitarianism, and yet others from embracing many other outlooks that involve the reduction or steamrolling of our differences. We claim to understand these outlooks somehow, but not in such a way as to offer any real hope of general success. They are all doomed to unreality—laced often enough with intolerance and totalitarianism—by the ultimate fact of our individual distinctness, at least as that distinctness functions in the broadly familiar range of environments, physical and technological, in which we live today.

These points seem not to be well understood by many of today's prominent philosophers in the Anglo-American world. These philosophers, in a variety of ways, sound the trumpets for egalitarianism, but much as I love the sound of the trumpet, I do not regard that as a legitimate use of the instrument. One might have hoped that the voice of reason is what would be heard in the land on these matters—and that voice must be concerned with the truth, hard as it may be. The truth is that people are different. They are not only different particulars, but they are different in an indefinite variety of substantive ways. The problem for egalitarians is to explain why we should accept any program of general equalizing—on the kind of abstract grounds one would hope that social philosophy would appeal to, rather than on the grounds of promoting the self-interest of those who stand to gain by steamrolling differences.

I am inclined to think that a major confusion of redistributionists is an assumption that “distributive justice” is concerned, just like that, with the “distribution” of stuff, that is, of natural resources as such. This mistake is indulged perhaps more eloquently by Henry George than by anyone else, but he has won many adherents in his insistence that the land is not created by people and belongs equally to all, from which he infers that a kind of tax imposed on land as such—the “single tax”—is the naturally just way to fund public undertakings for the good of all.Footnote ²⁴ Now, undertakings for the good of all, if there are any, would indeed be plausibly supported by all. But that support would come from relevant inputs from all, not from inert stuff. What George did not appreciate is that natural resources are, in fact, of no value whatever, and the attempt to divide them up “equally” is based on an illusion. There is, on occasion, not much difficulty in dividing a lump of matter into two or n more-or-less equal parts. But what could ever be the point of doing so? For what we humans do is to use nature, and to use it in innumerable different ways, all of which involve understanding and effort on the part of the users—who are also different in respect of the kind of understanding and effort they are able to put forward, as well as the particular interests that impel them to utilize various resources that they find in their environment.

X. The Shape of Morality

So where does that leave us? The sort of moral principles we should end up with will offer each person the best deal compatible with a like deal for everybody else, and compatible with, as we might put it, our personal operating systems. I have for some time thought that the basic principle to emerge is a general non-harm principle: we are each to refrain from imposing losses on others (more precisely, net losses, that is, losses not compensated, from the point of view of those affected, by some benefit that they accept as sufficient to justify the cost). But as to doing good for others, that is largely up to us. We will do good for those we like enough to do good for, or who are such that there is an indirect payoff to us on account of our good offices; but we cannot be compelled to do more than that, generally.

We should, as I have argued elsewhere,Footnote ²⁵ accept a general (and rather vague) principle of mutual aid. We should be ready to render help of a major and needed kind when we can do so at small cost to ourselves, and where the help is broadly of a kind that we ourselves might well stand in need of some day. To accept such a principle is to hook into a sort of informal social-insurance network. The terms are such that we are nearly all virtually certain to benefit from its general observance. It is hard to do better than that if we seek rationally acceptable principles. But this seems very far from the sort of duties that a fundamentally aggregationist stance would generate.

The general drift of this approach to morality is toward a strongly individual-oriented ethic. Each person can expect to do better from subjecting himself to the requirements of morality—“better” on his own terms rather than on the basis of some theory he does not accept.

The previous comparison to insurance is rather narrowly intended. We buy insurance despite the fact that we will pay more per unit of benefit than in the absence of the scheme. What makes it worth doing? The general answer is provided by the word itself: what we get until we actually need it (if ever we do) is “assurance,” that is, the sense or awareness that if the unhappy eventualities against which we take out insurance should actually materialize, then we will be able to cope well enough. Instead of devastation, there will be (on net) minor nuisance. We could even say “minor loss,” except that we do not really lose. For we are willing to pay something (the specified insurance premium) in order to be in this condition, even as we fervently hope that we will never be in a position to collect.

Altogether, the approach I have described is a program by which everyone's individual agency is protected from depredations, in general. But it is not a program for aggregate maximization.

XI. Back to Aggregation

How, then, can we ever aggregate, consistently with the foregoing arguments? We are who we are; we reason from our own points of view as distinctive, individual persons. Robert Nozick famously says:

Still, on Nozick's view, there is a general prohibition on force and fraud, which amounts to a general right of all to liberty of action in the absence of behavior that violates others' rights. As Nozick himself notes, things like air pollution can generate problems—problems whose solution, though of course addressing particular people, nevertheless derives from considerations of diffuse effects, and the probabilistic reasoning necessary to assess them. Does my right to liberty and bodily integrity imply a right to make my way across a busy street in the legitimate expectation that none of the fast-moving automobiles on it will collide with me, even despite considerable effort on the part of the drivers? It does not, of course. But then, just what does it imply? Certainly it implies that those same drivers are to refrain from veering onto the curb and running me down where I stand waiting for the traffic to clear. But what else? Does it imply that they are not to pollute the air as they drive by, thus giving me diseases? Perhaps. Is this result, and similar ones, reachable via the idea of particular agreements between particular people about particular things?

To see why it is, we need to consider more carefully the concept of harm. There are two matters, especially, to be considered more closely. First, there is the distinction between (1) the intent to harm, as a direct result of one's actions, and (2) the subtler form of harm in which what one does, essentially, is to increase the probability that a certain negative result will occur. The second matter concerns the ancient legal distinction between harm that is intended and harm that is a function of an agent's actions though not even foreseen, let alone intended, by the agent. (Roughly, this concerns the distinction between the criminal law and tort law.)

The first distinction leads us in the direction of aggregation because probability concerns groups. If person A does something that only raises the probability that person B will be harmed in some definite way, how are we to decide whether A is responsible for the harm to B, if it ensues? There is no alternative to pinning on A a charge of “risky” behavior. The riskier the behavior, the worse the charge, but translating this into damages to particular individuals is difficult.

The second distinction makes us observe that when A's actions harm B as a side effect, unintended and unwanted by A, and unknown to him as well, it becomes problematic to say, just like that, that A has “harmed” B. If we do say that he has, it is often because the marginal contribution of his particular act to the losses (say in terms of ill health) experienced by random members of some large group is appreciable. Unintended as these losses are, there seems to be no alternative to invoking something close to strict liability: what the agent did, though no part of his intention and likely unforeseen or even unforeseeable by him, nevertheless does cause some kind of damage, and he is accordingly held responsible. In the future, hopefully, he will know better.

When our actions affect miscellaneous others, aggregation is rational. Why is this true? Not because everyone's good “matters equally,” just like that. Rather, it is true because of the fact that we ourselves are members of many aggregates. John Taurek, in a well-known essay, addresses the matter of trade-off situations, in which “we must choose between bestowing benefits on certain people, or preventing certain harms from befalling them, and bestowing benefits on or preventing harms from befalling certain others.” He then raises the question “whether we should, in such tradeoff situations, consider the relative numbers of people involved as something in itself of significance in determining our course of action”Footnote ²⁷ He argues persuasively that all sorts of personal relations outweigh any such considerations, and appears to hold that the answer to his question is that numbers should not be considered a matter of moral significance at all. Thus, in a case where we must choose between giving all of a lifesaving drug to one person, David, who will not be saved with less, and splitting it among five people, each of whom would be saved (though David would then die), Taurek supposes that many people would think that “the death of five innocent persons is a worse thing, a greater evil, a greater loss, than the death of one innocent person.”Footnote ²⁸ But this, he argues, is not so. Suppose David is a good friend of Taurek's. He may choose to save David's life rather than the others, and many would agree with this choice; but he does not think that the fact that David is his friend makes David's death, absolutely, a “worse thing” than the deaths of the others. He goes on to suggest that “we should either agree that it would be wrong for me to save David in this situation or admit that there are no grounds for a moral requirement on anyone, special obligations apart, to save the five instead of David.”Footnote ²⁹

Taurek appears to share the individualistic point of view pressed earlier in this essay. Does that viewpoint lead to the conclusion just stated? Not quite, I think. Taurek considers many cases which involve what we might call special obligations or special considerations, and, in these cases, it is indeed plausible to think that considerations of sheer numbers are pretty readily outweighed. But should we think that there is literally no moral weight to be attached to numbers? No.

The reason for this is easy to state. We are, as I have noted, members of many aggregates. A policy of maximizing a good for an aggregate will result in better results for you, in the long run at least, if you are simply a random member of that aggregate, and if the acts or policies being considered benefit that aggregate generally, in the probabilistic ways described above. The reason we should consider numbers is that the greater the number in a certain affected group, the greater the likelihood that I, or someone I care about, is a member of such a group who will be affected by the action in question.

Many actions do not have the potential to affect random members of groups in this way, and, in many other cases, as in several that Taurek considers, the other, personal considerations he mentions may and should dominate. But, after all, we very often are in a position to act so as to affect random groups of strangers. We do not need to sacrifice much, or perhaps anything, to affect them more favorably, statistically speaking. But if we sacrifice very little, the relevance of this factor comes into view. Not all of our interactions with others are personal. Indeed, depending on what you count as an interaction, perhaps most of them are not personal. But it is not as though we have nothing to gain or lose on these fronts. If we relate to a whole lot of others, either via an institution or otherwise, we have an interest in doing as well as we reasonably can as a result of those relations. And since we relate to one another as members of aggregates of greater or lesser size, the rule proposed seems an obvious one: maximizing the value or values, whatever it or they may be, that the aggregate in question is concerned with leaves us, as random members of the group, as well off as we can reasonably hope to be from that kind of interaction.

Why not, after all? This seems to be a context in which one has a considerable amount to gain and little to lose. The proposal here, I should note, is that we regard numbers as significant, in the sense that, in the absence of more special considerations, we should prefer the option in which fewer people are worse off, or the one in which more are better off. This is a weak principle, but it does not assign numbers a moral weight of zero, which appears to be what Taurek is proposing. And, of course, it requires judgments about when people are better off. On that point, however, the general premises of morality are such that we should normally use the standards of judgment of the individuals concerned as our basic guide, rather than some other standard. This is an antipaternalistic stance, one that is dictated by individualism.

The proposal here, then, is certainly not utilitarian, except in the very modest sense that it attends to the specifically relevant utilities of those concerned. But it does not attend to them by making them, as they stand, in aggregate the basic criterion of right and wrong. It attends to them, rather, in the spirit of approaching the problem of how any individual should orient himself or herself in relation to the miscellaneous people in his or her environment (which can be taken to include, possibly, even the global environment). The general idea is that we want to be free to run our own lives as we see fit, and, in doing so, we recognize the requirement that we allow others to do the same.

XII. Economic Aggregation?

I turn now briefly to the economic realm. Individualists have a way of supposing that it would be a good idea to maximize the gross domestic product. Is this an aggregationist view, and therefore inconsistent with individualism? I would say it is not, though it is very easy to succumb to the lure of trying to promote GDP by various interventionist initiatives in national economies.

Nonetheless, GDP, properly so-called, is not actually aggregationist in any interesting sense. What we call the “domestic product” is just people doing the best they can, given the constraints imposed by our general rights, and specifically our rights to property. We suppose that most people prefer greater incomes to lesser, though no generalization to that effect can stand unqualified, since plenty of people forgo greater incomes for the sake of other things (familial peace, say, or the fact that they like to live in places where their economic prospects are lower). A greater GDP for a given society results from more people doing better. No individual needs to be interested in promoting GDP as such, and almost none are. And governmental policies aimed at compelling people to attain higher levels on their preference scales are nonsensical: your higher position is defined by the fact that you prefer it, so that compulsion is necessarily counterproductive.

We do not, so far as I know, have a way of incorporating the satisfactions of values that do not automatically translate into monetary ones into a measure of “gross domestic product” that would be relevant here. I have read of a farmer in Africa who resisted the blandishments of a government official who urged him to plant new crops, take out loans, and so on. “I'm happy sitting under my banyan tree,” the farmer said, “thank you very much!” If our measure of GDP were capable of counting this man as maximizing his utility by not bothering to promote his income, that would be, as we might say, very cool! Lacking that, we have to agree that GDP, reckoned in dollars, is a very imperfect measure of the sort of social utility that is relevant for social philosophy. Nevertheless, it is a pretty good measure, and the point is that it does not have the defects I have ascribed to truly aggregative theories.

A policy of general freedom will leave us with the most plausible moral worldview that we can assemble—not one in which each person attempts to promote the sum of human happiness as far as possible, but one in which each is free to do, and generally does, as well as can reasonably be hoped, given the way the rest of us are. Of course, the freedom of each must, necessarily, be limited by considerations of the effects of otherwise free actions on the freedom of the others.

That the results of a general rule requiring us to refrain from imposing known and knowable harms on other individuals will also be for the general good is, I think, very clear. I believe we cannot realistically do better.