MORAL DEMANDS IN NONIDEAL
Liam B. Murphy
OXFORD UNIVERSITY PRESS 2000
OXFORD UNIVERSITY PRESS
Oxford New York Athens Auckland Bangkok Bogotá Buenos Aires Calcutta Cape Town Chennai Dar es Salaam Delhi Florence Hong Kong Istanbul Karachi Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Paris Sāo Paulo Singapore Taipei Tokyo Toronto Warsaw and associated companies in Berlin Ibadan
Copyright ? 2000 by Liam B. Murphy
Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016
Oxford is a registered trademark of Oxford University Press.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of Oxford University Press.
Library of Congress Cataloging-in-Publication Data Murphy, Liam B., 1960-Moral demands in nonideal theory / Liam B. Murphy. p. cm. -- (Oxford ethics series) Includes bibliographical references and index. ISBN 0-19-507976-0 (alk. paper) 1. Benevolence. 2. Social ethics. 3. Utilitarianism. I. Title. II. Series. BJ1474 M87 2000 170'.42 -- dc21 99-029459 9 8 7 6 5 4 3 2 1
Printed in the United States of America on acid-free paper
My first thanks go to Barry Taylor for his extraordinary teaching at the University of Melbourne, the inspiring example of his philosophical mind, and his judicious but kind encouragement.
I first wrote about the issues of this book in my doctoral dissertation at Columbia University; scarcely a line of the dissertation remains, but my understanding of these issues has been fundamentally shaped by the suggestions and criticisms of my advisers, Charles Larmore and
Thomas Pogge. I have also been greatly influenced by conversations with Robert Myers that began while I was finishing my dissertation and, happily, have continued since. In writing the book I have also been helped by the comments of many other people, including members of audiences at various colloquia and workshops, I would like to thank in particular Richard Arneson, Ruth Chang, Frances Kamm, Muhammad Ali Khalidi, Lewis Kornhauser, Jeff McMahan, Tim Mulgan, Joseph Raz, Amélie Rorty, Carlos Rosenkrantz, Lawrence Sager, Samuel Scheffler, Gisela Striker, William Talbott, and especially Ronald Dworkin and Thomas Nagel.
Special thanks are due to Sibylle Fischer for her comments on many drafts and for almost ten years of discussions of the themes of this book; I have not managed to convince her of the wisdom of my ways, but the attempt has made the book much better.
The readers for Oxford University Press were Shelly Kagan and Larry Temkin. Their extensive and extremely insightful comments on both details and fundamental issues helped me to improve every chapter in important ways. I am immensely grateful to them both. My greatest debt is to Derek Parfit. From his early enthusiastic encouragement, without which I probably would not have written this book, to his more recent voluminous comments on my manuscript, which cor-
rected several important errors and led to improvements on almost every page, his generosity to me has been literally astonishing.
I am grateful to Columbia University's Society of Fellows in the Humanities and the Mellon Foundation for a Teaching Fellowship, to Harvard University's Society of Fellows for a Junior Fellowship (and to Diana Morse for making life at the Society of Fellows so easy and so happy), to the Filomen d'Agostino and Max E. Greenberg Research Fund of New York University School of Law for financial support, and to my colleagues at NYU for their kind encouragement. Some passages from my article "The Demands of Beneficence", Philosophy
& Public Affairs 22 ( 1993): 267-92 (? 1993 Princeton University Press), appear in different places in this book; I thank Princeton University Press for permission to reprint this material. Finally, I would like to thank Elizabeth Newman for allowing me to reproduce her photograph of the corner of Barkly and Canning Streets, Carlton, Melbourne, circa 1979. New York City L. B. M. September 1999
3 1. Introduction
9 2. Over-demandingness, Alienation, and Confinement
34 3. Doubts about Over-demandingness
63 4. Moderate Beneficence?
74 5. Responsibility in Nonideal Theory: The Compliance Condition
102 6. The Distribution of the Effects of Compliance
117 7. The Collective Principle of Beneficence
I. The Puzzle of Beneficence
In some areas of moral and political thought, something like a consensus exists. No one denies that it is wrong to kill a person for the sake of stealing his car. Other areas of normative thought, by contrast, are characterized by great disagreement and uncertainty. The issue of *required beneficence is a salient example. Most well-off people feel that they should "do
something" to alleviate severe suffering, such as that caused by famine. But there is certainly no agreement about the extent of this obligation to promote the well-being of total strangers. Furthermore, and more important, most individuals have no settled view of their own about the question.
Of course, disagreement and individual uncertainty are found in other areas of normative thought as well. The issue of beneficence, however, is distinctive in a further respect, marking a striking contrast with issues such as abortion or euthanasia. For on the question of the shape of required beneficence, people not only tend to be uncertain but also appear to be quite 1content about this condition of uncertainty.
How can this be? If I can greatly reduce people's suffering, and I believe that it can sometimes be wrong for me not to greatly reduce people's suffering, how is it that I can remain contentedly unsure, in any
____________________ *Beneficence is distinct from benevolence. Th e latter is about being well-disposed toward
others, caring about their well-being; the former is about actually promoting others' well-
being. Different accounts of beneficence employ different accounts of wellbeing (see chapter 2, section I) and differ further over whether "others" includes nonhuman animals (I
do not discuss this issue). Strictly, "beneficence" is about doing good, which may involve more than promoting well-being; however, the narrower sense of the term, which I will use throughout, seems closer to ordinary usage.
given case, whether failing to help is or is not wrong? As moral questions go, this one would seem to be very important, for the decision I arrive at is of immense significance for the people who are suffering. Since we typically care rather a lot about not doing even trivially wrong things of other kinds, such as stealing an apple from a market stand, or missing appointments without apology, our complacency about the question of required beneficence needs some explanation.
One possible explanation turns on empirical beliefs. Sidgwick observed that there seemed to 2be no "consensus as to what each man owes to his fellow-men, as such." He also claimed
that "each person is for the most part, from limitation either of power or knowledge, not in a 3position to do much good to more than a very small number of persons." Given that very
badly off strangers lived in Sidgwick's vicinity, and that organized efforts to assist the poor were not unknown in Victorian England, this claim seems wrong even for Sidgwick's time. A hundred years later, when efficient international humanitarian aid organizations and public and private agencies within our own countries can very effectively transform the surplus money of the well-off into greatly improved prospects for the very badly off, the claim is *fairly clearly wrong. But Sidgwick's view about the severe practical limits to any one
person's ability to benefit other people is still widely held, and no doubt at least partly explains the prevailing complacency about beneficence.
However, a much more significant part of the explanation seems to be normative. It is widely believed that when it comes to the well-being of strangers, the question is not so much one of what people should do directly, but rather one of which institutional regime they should support. One way of developing this idea leads to Rawls's division of normative theory into 4individual morality for people and political justice for institutions. I myself reject this
division, but even if we were to grant it, the underlying problem of the extent of people's 5responsibility for the wellbeing of mere strangers would not go away. As I explain in section
III, whether our responsibility for the well-being of others is mediated by institutional structures or not, we still have to know what the shape and extent of that responsibility is. In any case, the thought that beneficence is "for the government" suggests not just an institutional focus; more important in the current context, it also suggests that we tend to view beneficence in terms of collective responsibility. In this respect beneficence is different from other moral obligations, such as the obligation not to kill someone for monetary gain. This further distinctive feature of beneficence may do a lot to explain most people's complacency about it. For though each person is not sure what her responsibility is, what does seem evident enough is that
____________________ *See further chapter 2, sectio n I.
most other people are falling short of whatever their responsibility may be. The practically
relevant question of the shape of required beneficence is thus a question of "nonideal theory" -- a question, that is, of what a given person is required to do in circumstances where at least 6some others are not doing what they are required to do. This question of nonideal theory is
especially puzzling in the context of collective responsibility, and a person may justify her neglect of it by noting the absence of any sign that others are prepared to do their part, whatever it may be.
But that is not a good reason to avoid the question. We need to focus more than we traditionally have on the practically urgent problems of nonideal moral and political theory. II. The Problems of Moral Demands
Whatever the reason for most people's contentment with not knowing what they owe other people as such, it is certainly the case that it is very difficult to come up with plausible answers to the question. Of course everyone accepts that it is morally good to promote the interests of worseoff people; "charity" brings social honor. But our question is what beneficent acts are required of us: When do we act wrongly for not promoting someone else's interests? If we limit ourselves to charity, the answer is "never" -- beneficence is always supererogatory. I take it that it is not especially controversial to reject this view; as I have said, most of us believe that well-off people -- or at least their representative institutions -- do have to do something. A "principle of beneficence," as I understand it, is a moral principle that sets out what beneficent action is required of people. The charity view thus presents no principle of beneficence at all -- indeed, it denies that there is such a thing.
One of my two main aims in this book is to suggest a plausible principle of beneficence. The normative question to which such a principle is an answer is urgent and weighty, but there are no natural candidates. And thus my own suggestion -- the "collective principle of beneficence" introduced later in this section -- is offered in the spirit of exploring neglected possibilities. I believe that the collective principle of beneficence is more plausible than the alternatives I am aware of; but this does not mean that I find it overwhelmingly compelling considered on its own. In all parts of normative theory we must make arguments of relative * This cautious methodology is especially important in a case like plausibility
____________________ *See Rawls, A Theory of Justice, 52: "Objections by way of counterexamples are to be made
with care, since these may tell us only what we know already, namely that our theory is wrong somewhere. The important thing is to find out how often and how
this, where disagreement and uncertainty make it abundantly clear that there is no default view on the subject, no view that stands unless rebutted.
In making the argument of relative plausibility for the collective principle of beneficence, much of my attention is devoted to the best-known principle of beneficence -- that embraced in the utilitarian tradition. The "optimizing principle of beneficence" requires agents always to do the best they can for others. It requires us to keep benefiting others until the point where *further efforts would burden us as much as they would help the others. This principle has the
virtue of simplicity, but the demands it makes strike just about everyone as absurd -- as we 7say, a principle that makes such demands "just couldn't be right." Explaining why the
demands of the optimizing principle of beneficence are absurd, if they are, is the second main aim of this book.
The most obvious explanation is that the demands of the optimizing principle are simply excessive. I discuss this problem of over-demandingness in chapter 2, arguing as well that some other objections that can be raised against the optimizing principle may reduce in the end to the same issue of excessive demands. But having thus argued that the simple problem of over-demandingness is most likely what is really of concern even for those who raise more sophisticated objections, I turn in chapter 3 to cast doubt on that concern. The very idea of excessive moral demands turns out to be unclear. Efforts to introduce clarity, and to rescue the idea from the accusation that it simply states a bias in favor of the moral and political
status quo, in the end serve only to undermine the initial intuitive force of the problem of over-demandingness.
This collapse of the most natural explanation for the apparent absurdity of the demands of the optimizing principle of beneficence lends support to the alternative diagnosis I offer in chapters 5 and 6. I argue that
f ar it is wrong. All theories are presumably mistaken in places. The real question at any given time is which of the views already proposed is the best approximation overall. To ascertain this some grasp of the structure of rival theories is surely necessary." *According to utilitarianism, the optimizing principle of beneficence is the only moral
principle there is. Some writers use the word "consequentialism" for what I call utilitarianism. I understand consequentialism more broadly, as any view that assesses the
rightness of action in terms of the goodness of consequences, where the goodness of consequences is not necessarily solely a function of well-being. (On the terminological
issues discussed here and in the note on p. 3. I follow suggestions made by Derek Parfit.)
I should point out here that talk of moral "theories" and "principles" in this book is meant to imply neither any particular metaethical commitments, nor any assumption that normative thought can be reduced to something like a decision procedure. As I understand
it, a moral principle is just a representation, at whatever level of specificity is possible, of a particular area of moral thought; and a moral theory is just a representation, at whatever level of specificity is possible, of the whole of moral thought.
the optimizing principle of beneficence imposes its demands unfairly in situations of partial compliance; the true problem with the demands of the optimizing principle thus does indeed arise in the realm of nonideal theory. If I am complying with the optimizing principle but others are not, I not only have to do my own fair share, I have to take on as much of the shares 8of the noncomplying others as is optimal as well. This, I argue, is unfair. I suggest a
"compliance condition" on principles of beneficence: very roughly, the idea is that the demands on a complying person should not exceed what they would be under full compliance with the principle. My diagnosis of the apparent absurdity of the optimizing principle of beneficence is thus that it violates the compliance condition. As I explain in chapter 5, the compliance condition does not apply to socalled deontological obligations, such as the obligation not to kill for monetary gain, but rather only to principles that can be understood in terms of collective responsibility. Thus this diagnosis explains why it is never argued that extremely demanding deontological requirements are for that reason absurd. The collective principle of beneficence is designed to satisfy the compliance condition; roughly, it requires agents to promote the well-being of others up to the level of sacrifice that would be optimal under full compliance. More accurate formulations of the principle are quite complicated, and I devote chapter 7 to sketching how the principle might be employed by agents in practice.
The other main alternative to the optimizing principle of beneficence is found in a group of principles I call "limited principles of beneficence." These principles, all of which hold that there is a limit to the size of the burden people need to sustain for the sake of benefiting others,
are appealing if we accept that the problem of over-demandingness provides the right diagnosis of the apparent absurdity of the optimizing principle of beneficence. But their familiarity makes them worth discussing even if that diagnosis is rejected; I discuss limited principles of beneficence in chapter 4.
III. Politics, Morality, and Nonideal Theory
Though most of my discussion will focus on questions of what people might be required to do to benefit others directly, examples from politics are also invoked at crucial points. Indeed, I believe that almost all of the argument in this book is as relevant to political theory as to moral theory.
As I have indicated, I believe that principles of justice apply not just to institutions but also to personal conduct. I am also inclined to believe, furthermore, that the best theoretical account of political justice is grounded in a principle of weighted beneficence -- one that gives addi-
9tional weight to benefits to worse-off people. In this account of justice, the principle of
weighted beneficence requires us both to design institutions and also to act so as to promote overall weighted well-being. Rawls would reject not just the content of this view -- weighted beneficence -but its very structure. On his view, it is institutions alone that have responsibility for promoting the aims of justice; people's responsibilities in respect of justice are to support 10and promote just institutions. From this perspective, this book's discussion of people's
responsibilities of beneficence could be thought to be relevant only to the subject of individual morality, and indeed the discussion could be thought to be rather unimportant, on the ground that most of the work of making others' lives go better should be done by institutions, the design of which requires distinct normative principles.
But even if we accepted Rawls's dualist view about normative theory, the problems about the demands of beneficence discussed in this book would nevertheless arise in closely analogous form. The point is clearest and strongest if we focus on what Rawls deliberately did not focus on: nonideal theory. In Rawls's view, political theory must first produce an account of just institutions as they would be under conditions of full compliance, and then, having 11established the appropriate ideal, turn to questions of nonideal theory. Rawls himself has
not devoted much attention to nonideal theory. But it is clear that he believes that doing so would involve attention to the issue of the sacrifices that could be demanded of people. In discussing the operation of the "natural duty of justice" in nonideal circumstances, Rawls says that "we are to assist in the establishment of just arrangements when they do not exist, at least 12when this can be done with little cost to ourselves." Obviously enough, there are many
possible views about the shape of the responsibilities of people to bring about just institutions. Any discussion of what Rawlsian justice requires of people in an unjust world will need to cover pretty much the same territory as this book's discussion of the demands of beneficence, 13including, especially, the discussion of the problem of demands under partial compliance.
Since we do live in an unjust world, and since indeed we are doomed always to apply nonideal theory only, this is hardly an insignificant political issue.
OVER-DEMANDINGNESS, ALIENATION, AND CONFINEMENT
In the next few chapters my aim is not to solve the familiar problem of over-demandingness but rather to undermine its force. As a first step, I offer in this chapter a detailed interpretation *of the problem. This involves describing two other potential problems concerning the demands of the optimizing principle of beneficence. These are, first, the problem of alienation, which concerns the idea that compliance with the optimizing principle alienates agents from ethically significant aspects of their own lives; and, second, the problem of confinement, where the idea is that the optimizing principle narrows the options available to complying agents in an objectionable way. I argue that both of these problems may be compelling only to the extent that they are aspects of the problem of over-demandingness.
I. The Demands of the Optimizing Principle of Beneficence
The over-demandingness problem has its origin in a traditional objection to utilitarianism. Richard Brandt puts that objection succinctly: "Act utilitarianism makes extreme and oppressive demands on the individual,
____________________ *I see the problem of over -demandingness as arising within morality, rather than as a
problem about the authority of moral reasons. For the latter view, see Susan Wolf, "Moral Saints", Journal of Philosophy 79 ( 1982): 419-39, and "Above and Below the Line of
Duty", Philosophical Topics 14 ( 1986): 131-48; David O. Brink, "Utilitarian Morality and
the Personal Point of View", Journal of Philosophy 83 ( 1986): 417-38. For defense of the
approach I adopt, see Samuel Scheffler, Human Morality ( New York: Oxford University
Press, 1992), esp. chap. 4.
so much so that it can hardly be taken seriously; like the Sermon on the Mount, it is a morality 12only for saints." This objection to utilitarianism has been familiar at least since Sidgwick,
but it has been much revived in recent decades, due in large part to the work of Bernard 3Williams.
Now our interest is not limited to utilitarianism. Utilitarianism presents the optimizing principle of beneficence as the whole of morality; but a pluralist moral conception that includes the optimizing principle of beneficence, such as that of W. D. Ross, is likely also to 4be very demanding. Indeed, extreme demands can also come from parts of morality other than beneficence. This fact will be important in the next chapter, where I note that if extreme demands are objectionable, they presumably are so whatever their provenance. In the meantime, however, I will limit my discussion of the problem of over-demandingness to the context of beneficence. One reason for this is just that the optimizing principle of beneficence gives us the clearest case of an extremely demanding moral principle. More important, this focus allows us to pursue at the same time our project of developing a potentially plausible principle of beneficence.
The remainder of this section describes the demands that would be made by the optimizing principle of beneficence standing alone -- as the whole of morality. But the aim is to provide a rough outline of the demands made by the optimizing principle of beneficence wherever it is found. The reason our outline can be only rough for principles of beneficence that are part of pluralist moral theories is that the principles of a pluralist theory may be ranked in various ways that effectively reduce the demands of some of them. For example, the optimizing principle of beneficence may make lesser demands on a person also subject to various prior special obligations. Furthermore, in a pluralist moral theory the requirements of the optimizing principle of beneficence may overlap with those of various other principles of the theory, and in such a case it would be arbitrary to assign demands to one principle rather than another. These points show that the proper object for investigation is the demands of entire moral theories, not moral principles considered in isolation. However, it would be tedious to discuss in turn the demands of each of the various possible theories that might contain the optimizing principle of beneficence. Moreover, I do not believe that anything in the subsequent discussion is undermined by my simplifying assumption that the impact the optimizing principle of beneficence has on demands is roughly the same whatever the overall 5 moral theory.
What, then, are the demands of the optimizing principle of beneficence? Such a principle requires each person to act such that she will produce as great an expected overall benefit, given what she has reason to believe, as she would acting in any other way available to
*her. So the optimizing principle requires each agent to go on promoting well-being until the point where further efforts would burden the agent as much as they would benefit others. Depending on the circumstances, the loss sustained by the time this point is reached could be very great; moreover, the agent's resulting level of expected well-being could be very low in absolute terms. In certain circumstances, then, the optimizing principle is extremely demanding.
The importance of this fact for moral theory depends in part on the nature of the relevant circumstances. If extreme demands occurred under only very exotic circumstances, it could plausibly be thought that no pressing normative problem was raised. But of course the circumstances under which the optimizing principle of beneficence imposes extreme demands need be no more exotic than those that currently obtain.
The most obvious aspect of the problem is world poverty. Determining the best way to promote the well-being of the poorest people in the long term is of course an extremely 6difficult and complex matter. Nevertheless, individual agents need not, and, if they accept
the optimizing principle, ought not, wait until the various issues involved are definitively settled. Each moderately well-off person is already in a position to save many thousands of lives by donating money to efficient aid organizations that provide cures for various 7childhood diseases. And there is money that could provide food for people who would
otherwise die of starvation. Beyond saving lives, each person who has resources in excess of what is needed for health, food, shelter, and those extras that might be needed to keep income flowing could best promote overall well-being by giving the surplus to those who are sick, hungry, or homeless. Here it should be clear, moreover, that there is actually no need to bring up world poverty. A restricted optimizing principle, requiring agents to promote the good in
their own country, would also be very demanding in most countries, including some Western industrial countries at the end of the twentieth century.
____________________ *This states the principle's subjective criterion of right action, which, in practice at least, is
the only criterion we need. I follow Derek Parfit, "What We Together Do" (unpublished ms., 1988), and Allan Gibbard, Wise Choices, Apt Feelings ( Cambridge, Mass.: Harvard
University Press, 1990), 42-43. The phrase "expected overall benefit" masks the extremely
difficult issue of whether agents should choose the action that will lead to the greatest increase in the expected total level of well-being, or the greatest increase in the expected
average level of well-being. See, on this question, Parfit, Reasons and Persons, pt. IV.
Note, finally, that though I take principles of beneficence to be about the promotion of expected well-being, for the sake of simplicity I will from now on typically write simply about the promotion of well-being, or benefiting people. On the importance of choosing the
action productive of the optimal increase in overall expected well-being, rather than the
action that has the best chance of bringing about the greatest benefit that any action available to the agent could bring about, see Frank Jackson , "Decision-Theoretic
Consequentialism and the Nearest and Dearest Objection", Ethics 101 ( 1991): 461-82.
The demands of the optimizing principle are not just financial. In addition to the loss of the pleasures both high and low that extra money can bring, there are the losses involved in not being able to live the life of one's choice. For most of us it would be hard to show that there is no alternative to our way of earning a living that would better promote the good, perhaps just because it would bring in more money. But switching to that alternative could bring a great loss in well-being. It might do this because it would make it not feasible to have children, for example, or simply because we would have to give up what we enjoy doing for something we do not. And even if it would be best if I kept my current job or project, I have leisure time that could be spent volunteering in homeless shelters or in political campaigning. Of course I need some time to relax, or I might run out of steam and be able to do less good on the whole, but it is obvious enough that most moderately well-off people could absorb a substantial loss of leisure time without any reduction in their efficiency in promoting the good. I should emphasize that I am not claiming that it would best promote overall well-being if all
of us gave away all our surplus money, changed jobs, gave up on leisure time, and so on; rather, I am claiming that each moderately well-off agent could now best promote overall well-being if she did these things -- the assumption being, of course, that not many others are 8likely to do the same. The question of what the better-off as a group could do to promote the best outcome raises distinct issues about the overall economic effects of widespread changes in the behavior of skilled people and widespread changes in what the rich do with their money. So the question raises all the problems facing development policy generally, and more. Nevertheless, it is safe to assume that the greater the number of people trying to act optimally, the lower the sacrifice required of each person trying to act optimally. That this is so is most clear for the purely financial demands of the optimizing principle, and for the demands on leisure time, but it is also plausible for the demands on careers and projects -- it would very likely not make the outcome best if everyone in the industrialized world gave up their jobs and tried to promote overall well-being in some more direct way. Thus we can assume that with full compliance the demands of the optimizing principle of beneficence on each agent would drop drastically, and they would continue to drop as this high level of beneficence