Hostname: page-component-74656f9884-j7cmm Total loading time: 0 Render date: 2026-11-09T23:28:30.437Z Has data issue: false hasContentIssue false
  • English
  • Français

HISTORY AND PATTERN

Published online by Cambridge University Press:  01 January 2005

David Schmidtz
David Schmidtz
Affiliation:
Philosophy, University of Arizona
Rights & Permissions [Opens in a new window]

Abstract

This essay compares Rawls's and Nozick's theories of justice. Nozick thinks patterned principles of justice are false, and offers a historical alternative. Along the way, Nozick accepts Rawls's claim that the natural distribution of talent is morally arbitrary, but denies that there is any short step from this premise to any conclusion that the natural distribution is unjust. Nozick also agrees with Rawls on the core idea of natural rights liberalism: namely, that we are separate persons. However, Rawls and Nozick interpret that idea in different ways-momentously different ways. The tension between their interpretations is among the forces shaping political philosophy to this day.

Information

Type
Research Article
Information
Social Philosophy and Policy , Volume 22 , Issue 1 , January 2005 , pp. 148 - 177
Copyright
© 2005 Social Philosophy and Policy Foundation

I. Introduction

The agenda for current philosophical work on justice was set in the 1970s by John Rawls and Robert Nozick. Nozick said, “Political philosophers now must either work within Rawls's theory or explain why not.”1

Robert Nozick, Anarchy, State, and Utopia (New York: Basic Books, 1974), 183.

There is truth in Nozick's compliment, yet when it came to explaining why not, no one did more than Nozick.

Rawls spent the next three decades responding first to Nozick, then to a barrage of criticism from all directions. In part because of this, no short treatment can capture every nuance of Rawls's evolving theory.2

A recent book on liberalism by Jon Mandle offers an “overview” of Rawls spanning 133 pages, indicating how hard it would be to summarize Rawls in a few paragraphs. Jon Mandle, What's Left of Liberalism? (Lanham, MD: Lexington Books, 2000).

However, Section II of this essay offers a brief overview of Rawls, and the next four sections reflect on several facets of Nozick's response. Section III explains why Nozick thinks patterned principles of justice are false, and what a historical alternative might look like. Section IV concerns Nozick's skepticism about the very idea that justice is essentially a distributive notion. Section V explains the difference between being arbitrary and being unjust. Nozick accepts Rawls's premise that the natural distribution of talent is arbitrary, but denies that there is any short step from this to a conclusion that the natural distribution is unjust. Section VI notes that Nozick also agrees with Rawls on the core idea of natural rights liberalism: namely, that we are separate persons. However, Rawls and Nozick interpret that idea in different ways—momentously different ways. The tension between their interpretations is among the forces shaping political philosophy to this day.

The remaining sections of this essay are more speculative. Section VII explains how Nozick's tale of the “experience machine” can be seen to illustrate a particular way, Rawls's way, of failing to be serious about the separateness of persons. Section VIII revisits the Rawlsian original position, asking how we would design the thought experiment if our objective were to articulate a genuinely procedural conception of justice. I conjecture that it might end up amounting to a prescription for Nozickian utopia.

II. Rawls

A. An alternative to utilitarianism

According to Rawls, we should think of society as a cooperative venture for mutual advantage. Cooperation enables us all to flourish, but we each want a larger share of cooperation's fruits, so cooperation inevitably involves conflict. One way to resolve the conflict is to distribute the fruits so as to maximize overall utility. Yet this proposal fails to acknowledge that people entering into cooperative ventures are separate persons, contributing to those ventures in pursuit of their own legitimate hopes and dreams. Failing to respect their separate projects and separate contributions is unjust.3

How separate are persons, exactly? Does Rawls presume that individuals are self-contained Robinson Crusoes, as he is sometimes accused of doing by communitarians and feminists? See Andrew Jason Cohen, “Communitarianism, ‘Social Constitution’, and Autonomy,” Pacific Philosophical Quarterly 80, no. 2 (1999): 121–35.

It may even be the paradigm of injustice.

Utilitarianism in its standard form treats character and talent as if they matter only insofar as they affect aggregate well-being. Rawls, though, says that, contra utilitarians, when one person's gain comes at another person's expense, we hardly begin to justify trade-offs merely by making sure winners win more than losers lose. To Rawls, justice is less like an outcome of utilitarian calculation and more like an outcome of a bargaining process in which bargainers have a right to walk away from any proposal they find unacceptable. Rational contractors, meeting to negotiate an institutional structure to govern their future interactions, will want a system that promises to be good for all—a system that sacrifices no one for the greater good.

Accordingly, Rawls asks us to imagine contractors meeting to arrange institutions that will distribute fruits of cooperation in a mutually agreeable way. And if we imagine contractors distributing fruits of cooperation, we may as well imagine contractors distributing inequalities. This is quite a shift on Rawls's part. There is no easy way to explain what would make the shift legitimate, but it occurs quietly over a span of fifteen pages in A Theory of Justice (1971). First, Rawls says that “persons are not indifferent as to how the greater benefits produced by their collaboration are distributed, for in order to pursue their ends they each prefer a larger share to a lesser share. A set of principles is required for choosing among the various social arrangements which determine this division of advantages.”4

John Rawls, A Theory of Justice (Cambridge, MA: Belknap Press of Harvard University Press, 1971), 4.

Here, “advantages” refers to benefits of collaboration. Four pages later, Rawls says that a concept of justice applies “whenever there is an allotment of something rationally regarded as advantageous or disadvantageous.”5

Ibid., 8 (emphasis added).

Not everything regarded as advantageous is a product of collaboration, though, so there is a problem in connecting this sense of advantage to the previous one.6

In passing, what is an “allotment”? Does the sun have an allotment of planets, or are we reserving “allotment” to refer to (potentially blameworthy) allocation decisions? On the one hand, if we use the term's broader sense—allotment as simply the fact of how things are distributed—it becomes true that there is an allotment of advantages, just as there is an allotment of planets, but it becomes false that the concept of justice applies whenever there is an allotment. (It does not apply to the allotment of planets.) On the other hand, if we use the term's narrower sense—allotting as deliberately assigning—it becomes at least arguably true that the concept of justice applies, but then it becomes false that the natural lottery of talent is an allotment, and it is this natural lottery that Rawls wants his theory of justice to cover.

Finally, a few pages later, Rawls speaks of circumstances of justice as those where “none are known to be advantaged or disadvantaged by social and natural contingencies.”7

Rawls, A Theory of Justice, 19. In passing, is it enough that none are known to be advantaged or disadvantaged? What if some of us have disadvantages none of us know about? Does not knowing about a disadvantage suffice to make it fair?

So, by the end of this discussion, advantages have become not fruits of cooperation in particular, or even goods more generally. They have become competitive advantages—the kind of thing we can contrast with disadvantages.

This shift puts Rawls in a position to ask what he wants to ask: Why should anyone have a competitive advantage? Suppose, then, that we treat advantages as if they were fruits of cooperation, and suppose also that we treat inequalities as if their existence requires our agreement. These suppositions imply a startling result: that unless we agree to permit inequalities and on how to distribute them, we have a right, as separate persons, to leave the table not with whatever advantages we brought to it, but with equal shares of the value of the group's pooled advantages.

B. Two principles

We now can say what Rawls is arguing for: the two principles that form the core of his theory of justice.

  1. Each person affected by an institution is equally entitled to the most extensive sphere of liberty compatible with a like liberty for all.
  2. An inequality is allowed only if the institution that allows it thereby works to the greatest advantage of the least advantaged.8

    Ibid., 302. The full statement of the principle is: Social and economic inequalities are to be arranged so that they are (a) to the greatest advantage of the least advantaged, and (b) attached to offices and positions open to all under conditions of fair equality of opportunity. Part (a) is the notorious “difference principle.”

There has been relatively little discussion of the first principle. Despite appearances, the principle does not call literally for a “most extensive system of liberty” but for a modest package of constitutional rights to free speech, to a fair trial, and so on.9

One frustrating thing about reading Rawls is that he systematically takes initially bold, clear, inspiring, unambiguous positions, then later reworks and retracts these statements so as to make them consistent with (or in technical terms, in reflective equilibrium with) his second principle.

We may think the principle should be interpreted more expansively, or we may question Rawls's stipulation that the first principle always takes priority over the second. (In a little-noticed passage, Rawls quietly retracts the idea that his first principle has lexical priority: “While it seems clear that, in general, a lexical order cannot be strictly correct, it may be an illuminating approximation under certain special though significant conditions.”)10

Rawls, A Theory of Justice, 45.

However, few reject the first principle as such. The controversy surrounds the second principle, not the first.

Rawls refers to the second principle as the “difference principle.” What does this principle say? The intuitive idea is this. Contractors initially assume they are entitled to an equal share of the pie, but realize they can make the pie bigger by encouraging each other to work harder. They encourage each other by rewarding people who make the pie bigger: offering a larger share of the pie to those who do a larger share of the work. In effect, we allow inequalities when inequalities make us all better off. This leads to a precursor of the difference principle.

The precursor:

An inequality is allowed only if the institution allowing it thereby works to the advantage of everyone affected, that is, only if everyone receives more than what would have been an equal share in a more equal but less productive scheme.

Rawls went on to say that the precursor “is at best an incomplete principle for ordering distributions.”11

John Rawls, “Distributive Justice,” in Philosophy, Politics, and Society, ed. Peter Laslett and W. G. Runciman (Oxford: Blackwell, 1967), 58–82. Reprinted in John Rawls, Collected Papers, ed. Samuel Freeman (Cambridge, MA: Harvard University Press, 1999), 135. Page citations are to the latter.

It says departures from equal shares must make everyone better off. However, “there are indefinitely many ways in which all may be advantaged when the initial arrangement of equality is taken as a benchmark. How then are we to choose among these possibilities?”12

Rawls, A Theory of Justice, 65.

A complete theory specifies how to divide the gains.

There are two ways of proceeding from here. First and most obviously, we may doubt we need a theory that is “complete” in this sense. All we need is a theory telling us what is unjust, and the precursor is complete enough for that. It rules out sacrificing some people for the general good. Therefore, it rules out arrangements that fail to respect the separateness of persons. Indeed, we may conclude we do not want a “complete” theory. When we ask what we want from a basic structure, we realize we need incompleteness. Rawls sketches four ways of completing a theory and says his difference principle is what self-interested bargainers would choose, but he never discusses the possibility that choosing among the four is not a theory's job, and indeed is not a basic structure's job. Perhaps any of the four, if freely chosen by people to be governed by it, would qualify as just.

A basic structure's job is to get a political community off the ground, enabling voters and legislators to define and refine their community's norms as they go. Communities whose basic structure evolves toward completion in ways that are good for everyone, as broadly specified by the precursor, will be just. The final sentence of Jon Mandle's summary of Rawls says that when “the design of the basic structure is at stake,” citizens rely “on principles that all reasonable people can share.”13

Mandle, What's Left of Liberalism? 151.

Because the precursor is more open-ended than the difference principle, it is closer to being what reasonable people can be presumed to share, thereby getting a political community off the ground but deferring ongoing elaboration to voters and legislators. In any case, getting from the precursor to a full-blown difference principle is not easy. We easily can imagine the precursor being part of an overlapping consensus among real-world reasonable people. We cannot say the same of the difference principle.

Rawls sometimes says that all we do at the level of theory is pick a framework; societies work out details. This is the right thing to say. Basic structure is merely that; most of what makes a society liberal cannot be guaranteed by the basic structure but is instead in the hands of people and communities working out their own destinies within institutional structures.14

See John Tomasi, Liberalism Beyond Justice: Citizens, Society, and the Boundaries of Political Theory (Princeton, NJ: Princeton University Press, 2001), for the most developed version of this idea.

In a thought experiment, I imagine myself picking something as specific as the difference principle, but if I ask what everyone I know could agree on, an honest answer will be more general, specifying that whatever we pick must be in everyone's interest, as much as possible. No matter how badly I personally wanted the principle to say more, it could not say more without being rejected.15

Rawls acknowledges that we do not need a “complete” theory in the international arena. In fact, in that arena, we do not want a complete theory; we want instead to recognize that “peoples” have a right to self-determination. Interestingly, Rawls shows more willingness in the international arena to compromise liberties protected by the first principle than to compromise the economic protections of the difference principle. (The Rawlsian right to life that any minimally decent society has to respect includes a right to economic security.) See John Rawls, The Law of Peoples (Cambridge, MA: Harvard University Press, 1999), 65.

A second way to proceed, Rawls's way, is to assume that we need a complete theory. We then choose a way of completing the precursor. One way to complete the precursor would be to target one social position and maximize the prospects of persons in that position.16

When we are targeting one particular group as the group whose prospects are to be maximized, it will be hard to characterize the process or the result as an implication of any principle of reciprocity. Reciprocity is a two-way street, not aimed at any particular group, not even the group we think ought to be most favored. It would seem reasonable to give up on saying that this theory has anything to do with reciprocity, but Rawls always insisted there was a connection. For example, see John Rawls, Justice as Fairness: A Restatement, ed. Erin Kelly (Cambridge, MA: Harvard University Press, 2001), 123–24.

Which position should be so favored? One option is the position of those who otherwise are least favored—the group to which life has otherwise been least kind. Roughly, we make the smallest share as large as it can be. We thus arrive at the difference principle: an inequality is allowed only if the institution allowing it thereby works to the greatest advantage of the least advantaged.

Who is Rawls talking about when he speaks of the least advantaged? First, Rawls says the “least advantaged” refers not to the least advantaged person but to the least advantaged economic class. Second, the class is defined by wealth and income, not by any other demographic. In Rawls's theory, “least advantaged” refers in practice to typical representatives of the lowest income class, no more, no less.17

Rawls, Justice as Fairness: A Restatement, 59 n. 26.

Their natural and social disadvantages, and their needs, are not unusual but are instead stipulated to be “within the normal range.”18

Rawls, A Theory of Justice, rev. ed. (Cambridge, MA: Belknap Press of Harvard University Press, 1999), 83.

Why does Rawls use a term like “least advantaged” to refer to people who are not literally the least advantaged? Rawls takes a contractarian approach for a reason: justice is supposed to embody an ideal of reciprocity, unlike utilitarianism. “The first problem of justice concerns the relations among those who in the everyday course of things are full and active participants in society.”19

Ibid., 84.

Rawls envisions a bargaining situation where there is no such thing as a class that other classes do not need. He says, “since everyone's well-being depends upon a scheme of cooperation without which no one could have a satisfactory life, the division of advantages should be such as to draw forth the willing cooperation of everyone taking part in it, including those less well situated.”20

Rawls, A Theory of Justice (1971), 15.

The less well situated, then, are contributors: they are least advantaged workers, not least advantaged people. Thus, what Rawls calls least advantaged is the class with the least amount of bargaining strength compatible with still having a lot. The least advantaged, as Rawls defines the class, have a claim to a share of the social product based not on their needs but on the fact that they contribute to it.21

See Cynthia A. Stark, “How to Include the Severely Disabled in a Contractarian Theory of Justice” (unpublished manuscript, 2004).

The difference principle is meant to acknowledge their contributions, not their needs.

C. Strains of commitment

But is the greatest advantage of the least advantaged the only result to look for? Officially, the answer is yes. Taken at face value, the difference principle says that if we can make the least advantaged better off in the amount of one penny, justice requires us to do so even when the cost to the more advantaged is a billion dollars. As Rawls admits, this “seems extraordinary.”22

Rawls, A Theory of Justice (1971), 157.

So, what does Rawls conclude?

He says, “The difference principle is not intended to apply to such abstract possibilities.”23

Ibid.

Obviously, though, level of abstraction is not the issue. And whether a principle applies does not depend on whether Rawls intended it to apply. (Suppose Joe says, “There are no two-digit prime numbers.” Jane, well-trained in the art of the clever counterexample, asks, “What about eleven?” Joe responds, “The principle is not meant to apply to such abstract possibilities.”)

As it happens, Rawls is right about the principle not applying to such possibilities, but for a different reason: namely, the principle applies not to redistributive decisions case by case, but to the choice of society's basic structure.

Rawls could have noted that he did not intend to impose excessive “strains of commitment.”24

Rawls, Justice as Fairness: A Restatement, 104.

Principles must not ask so much as to make it unrealistic to expect people to comply. Rawls mainly has in mind not asking too much of the least advantaged, but as critics have noted, it is at least as urgent that the system not ask too much of people whose productivity sustains the system. Pushing them so hard that they revolt, or emigrate, would be bad for all, especially the least advantaged.

D. Indeterminacy

What Rawls calls the “veil of ignorance” is supposed to deprive bargainers of information that otherwise might lead them to seek an agreement biased in their favor. In particular, bargainers are supposed not to know what position they personally will occupy in the final distribution.25

Do we really need veils of ignorance as devices for controlling bias? I shall suggest an answer to this question shortly below in the text.

Rawls thinks bargaining behind a veil of ignorance is a fair procedure for picking a basic structure; moreover, if the procedure for picking is fair, the structure picked must be fair. Does this follow? No. We design legal procedures to be fair, but fair procedures do not guarantee just results; fair juries sometimes deliver wrong verdicts. Likewise, fair bargaining procedures may lead self-interested bargainers to select just principles, but there is no guarantee.

There are two related problems here. First, there is no obvious reason to expect the bargaining process to be so determinate as to reach the same conclusion every time we run the experiment. Second, even if the experiment did have a single, robustly replicable result, it is hard to see why the result would be guaranteed to track justice.26

There are two ways to view the procedure. Bargainers are either deciding what is right or discovering an independent moral reality. If our bargainers are merely hypothetical, we had better be able to show that there is an independent moral reality we are tracking when we decide to imagine bargainers agreeing to one thing rather than another. Otherwise, all we are doing is putting our own prejudices in the mouths of imaginary friends. If real bargainers actually agree to live by a principle, then that is altogether different. In that case, they actually gave their word to each other, and if their word does not count for something, they will not be able to live good lives together.

If we were to run the experiment with real people, we would have some basis for holding people to whatever they actually agree to—after all, they agreed—but in the real world, different groups would choose different principles.

Less obviously, a similar indeterminacy of result afflicts hypothetical procedures too. In particular, consider the veil of ignorance. If bargainers do not know what position they will occupy, that makes the original position fair, but (as Rawls was aware) what suffices to make the situation fair is not enough to ensure that bargainers will choose the difference principle. Hence, we make more assumptions: bargainers not only do not know what position they occupy; they also do not know what skills they possess, or what skills are prized in their society. They do not know what they personally believe about morality and justice, and so have no basis for choice other than their calculation of what is in their interest. They do not know what probabilities to attach to prospects of being at the bottom, so they have no basis for discounting improbable risks.

Rawls says that if bargainers had to complete the precursor, the difference principle is what they would choose. Why? In Justice as Fairness: A Restatement (2001), Rawls refers to bargainers being “directed” by a maximin rule (according to which one would seek to make the worst possible outcome as good as possible). But why? If bargainers are not pathologically risk-averse, why do they choose as if they were? Why are they “directed” by maximin? Rawls's answer: They are directed by maximin because they do not care what they gain above the minimum, so long as they know that (1) the minimum that would be provided under a maximin rule would be “completely satisfactory,” and (2) the minimums of societies under alternatives to maximin are significantly below that level and may be altogether intolerable.27

Rawls, Justice as Fairness: A Restatement, 98. I argued in “Rationality within Reason,” Journal of Philosophy 89, no. 9 (1992): 445–66, that satisficing choice is genuinely a form of rational choice. In my Rational Choice and Moral Agency (Princeton, NJ: Princeton University Press, 1995), I argued that this opens up possibilities for reconciling elements of rationality with elements of morality. However, I never envisioned defending a suboptimal political regime on the grounds that I would prefer it to intolerable alternatives.

Bargainers also know they are choosing principles for a closed society. The only way to enter is to be born into it. The only way to exit is to die. The more assumptions we add, though, the worse the strains of commitment when we lift the veil and rejoin this world. Can a thought experiment give us reason, here and now, to use the difference principle to assess basic structures? For example, if society would have to be closed for the difference principle to be a rational choice, and if in fact we live in an open society, subject to “brain drains,” then where does that leave the difference principle? Rawls says bargainers “should not reason from false premises. The veil of ignorance does not violate this idea, since an absence of information is not misinformation.”28

Rawls, A Theory of Justice (1971), 153.

There are two points here. First, the veil not only deprives people of knowledge they would have in the real world; it endows them with knowledge—for example, of conditions (1) and (2)—that no real bargainer has. Second, some of that endowed information—for example, that society is closed—is indeed misinformation.

E. “Least advantaged” is not a rigid category

Still, there is much to like about Rawls's theory. One attraction (and one reason to see the difference principle as less tilted than it may appear) is that the position of “least advantaged” is fluid. If Jack originally is least advantaged, then a system that works well on his behalf eventually lifts Jack into a better position than Jane, in which case, the system turns its attention to doing what it can for Jane, until she is sufficiently well off to make it someone else's turn, and so on.

Is this fluidity real? Yes, or at least it would be real in a society satisfying Rawls's difference principle. When Karl Marx was writing in the mid-nineteenth century, Europe was divided into somewhat rigid social classes, defined by birth. (The proper way to greet someone from a higher class was to bow, not shake hands.) When Rawls began writing in the mid-twentieth century, academics still spoke as if things had not changed. (Marxism was prominent in intellectual circles. Incredible though it may seem today, the Soviet Union was held up as a model of justice and efficiency.)

Thus, if we lived in a rigid society where sons of manual workers were fated to be manual workers (and daughters were fated to be wives of manual workers), then what would work to their greatest advantage would be to set a minimum wage as high as it could be without pricing employers out of business. Now suppose that a worker has an alternative: a fluid society where manual work pays less than it might, but where his sons and daughters can go to college and be upwardly mobile. Does he stay or does he go?

People born wealthy tend to prefer the more rigid world with high minimum wages, a world where manual workers are comfortable, secure, and stay in their place. Manual workers often prefer something more fluid, for themselves and for their children. What if people born wealthy want something for poor people that differs from what poor people want for themselves? Should what poor people want out of life affect our thinking about what is to their advantage?

Many people say they find it surprising and ironic that Rawls was a child of the comfortable middle class, whereas Nozick grew up poor. Upon reflection, though, it is not so surprising. If you are poor, the rich are like sports heroes—fantasy figures admired from a distance. It is people born to middle-class comfort who are close enough to the rich to resent the latter's additional advantages, and far enough from the poor to have a Dickens-like image of what being poor is like. Nozick often is accused of defending the rich, but in his own mind what Nozick was defending was the legitimacy of a poor person's dream of a better life.

In any case, an interesting feature of the less rigid world is that higher income classes will consist substantially of people who once were (or whose parents were) manual workers themselves. Higher-income classes, accordingly, will consist substantially of people who are better off because they grew up in a world where classes were fluid, and thus the chance to move up was real. They will be better off precisely because they grew up in a world where people born poor—people like they once were—had a chance to move up.

In a vertically mobile society, there will be a big difference between unskilled workers who have what it takes to move up and unskilled workers who, for whatever reason, do not. But this is a big difference only in vertically mobile societies. Behind the veil of ignorance, we have not yet chosen to create that kind of society, so these subgroups have similar prospects until we decide otherwise. Behind the veil, we are deciding whether to foster a society where talented young people are not held back by accidents of gender or social class.

This fluidity was not what Rawls had in mind. Rawls was not envisioning a world where working-class Jack can acquire skills that will make him wealthy later on. But if that world is best for Jack, then Rawls is right: what works for Jack works for every class, because in that world higher-income classes will contain substantial numbers of people who started (or whose parents started) as Jack did and made the most of the opportunity. If income mobility is to the benefit of younger people who earn less, then it was to the benefit of older people who once earned minimum wages themselves, then moved up.

In the end, there also is something to say on behalf of Rawls's insistence that we talk about classes, not individuals. There is a cliché: a rising tide lifts all boats. This cliché, as Rawls could see, is not literally true, but good societies make it roughly true. Realistically, the tide will never benefit literally every person, but good institutions can and in fact do make it true that the rising tide lifts all income classes. Even the least advantaged class will share in the tide of health benefits (life expectancy, safe water, immunizations, etc.) and wealth benefits (electric power, shoes, etc.) created by cooperation. If all goes well—that is, if Rawls's difference principle (or at least his precursor) is satisfied—then whole classes will not be left behind by the tide that people create when they contribute their talents to cooperative ventures in a free society.

Before we examine Robert Nozick's response to Rawls, we should note that Rawls's closest followers disagree over how Rawls is best interpreted and defended. There are hundreds of theories about how to pull together Rawls's arguments and about which to discard so that the remainder can be presented as an internally consistent whole. In later years, Rawls came to view his work not as a proof that his two principles are true, but instead as a way of articulating beliefs he considered implicit in contemporary Western democracies.29

This view begins to emerge most clearly in John Rawls, “Justice as Fairness: Political not Metaphysical,” Philosophy and Public Affairs 14, no. 3 (1985): 223–51, reprinted in Rawls, Collected Papers, 388–414.

Many of Rawls's followers were distressed by this seemingly colossal retreat. However, Rawls's later interpretation of his objective was exactly right. Rawls gave his readers a vision. It was a vision with grandeur, although it could not withstand scrutiny as the kind of philosophical deduction that some people wanted it to be. Rawls was claiming that, despite differences between their comprehensive moral views, there is an overlapping consensus implicit in how people live together in Western democracies. Rawls sought to identify the elements of that consensus and to explain, at an abstract level, why we believe in them.

Accordingly, our task is not to dwell on the argument's dubious details, but to reflect upon the degree to which we share this grand vision: (a) basic liberties come first, and (under normal circumstances) must not be sacrificed for anything; (b) we evaluate a society by looking at whether it is good for all and in particular by looking at the quality of life attainable by its nonprivileged members; and finally (c) we believe in (a) and (b) in part because they are fair, which in part means they are what we think we would choose if we were choosing impartially.

III. The Problem with Patterns

Nozick advanced the contemporary discussion with his fruitful distinction between historical and patterned principles of justice. The distinction is simple on the surface, but by the time we reach the end of Nozick's discussion, the two categories have become at least three, perhaps four, not as easily separated as we first thought. Some of Nozick's statements are hard to interpret, but the following definitions are roughly what Nozick intended.30

I thank Mark LeBar and John Simmons for encouraging me to be clearer about these distinctions, and I especially thank Richard Arneson for a proposal about how best to draw the distinctions. I follow Arneson's proposal to a significant degree, but not enough to make him accountable for the result.

Current time-slice principles assess a distribution at a given moment. We look at an array of outcomes. It does not matter to whom those outcomes attach. For example, on an egalitarian time-slice principle, if the outcomes are unequal, that is all we need to know in order to know we have injustice. We do not need to know who got which outcome, or how they got it. History does not matter at all.

End-state principles say something similar, but without stipulating that the outcomes are time slices. So, for example, an egalitarian end-state principle could say that we look at lifetime income; if lifetime incomes are unequal, that is all we need to know. The difference between time-slice and end-state principles is this: Suppose the Smiths and the Joneses have the same jobs at the same factory, but the Joneses are three years older, started working three years earlier, and continually get pay raises in virtue of their seniority that the Smiths will not get for another three years. There is no time when wages are equal, yet lifetime income evens out in the end. We have injustice by an egalitarian time-slice principle, but an egalitarian end-state principle looks beyond the time slice to conclude that the kind of equality required by justice will be achieved in the end.

Patterned principles include both of the above as subsets or examples, but within the broader class are patterns that are neither time-slice nor end-state. They are patterned in virtue of attaching importance to how outcomes track traits of individuals. “Equal pay for equal work” is an example of an egalitarian principle that is patterned but neither end-state nor time-slice; it prescribes how outcomes should track labor inputs, but does not prescribe that outcomes be equal.

Historical principles say that what matters is the process by which outcomes arise. Historical principles are complicated because, notwithstanding Nozick's intended contrast, many patterned principles have a historical element, and vice versa. “Equal pay for equal work” is both patterned and historical; that is, it prescribes outcomes tracking a pattern of what people have done.

Nozick classifies Rawls's difference principle as patterned but not historical. (It prescribes a distribution while putting no weight on who produced the goods being distributed.) By contrast, what Nozick calls “entitlement theory” is historical but not patterned. Nozick says an entitlement theory's principles fall into three categories: (1) Principles of initial acquisition explain how a person or group legitimately could acquire something that had no previous owner. Previously unclaimed land is one example, as are inventions and other intellectual property. (2) Principles of transfer explain how ownership legitimately is transferred from one person (or group) to another. Finally, (3) principles of rectification specify what to do about instances of acquisition or transfer that were not legitimate.31

Nozick, Anarchy, State, and Utopia, 150–53, 157–58, 173, and 208.

Nozick's preferred theory is a version of entitlement theory. Nozick says a distribution is just if it arises by just steps (the paradigm of which is voluntary exchange) from a just initial position.32

Ibid., 150–53, 157–58, and 262–65.

I am not sure Nozick should have said that. It sets the bar high: What can a historical theory say about a world where few titles have an unblemished history? Or perhaps that is simply how it is, and whether we like it or not, there is no way to get from here to a situation where distributions are just. Nevertheless, Nozick may have been mistaken to raise the topic of justice in distribution, since his theory seems to want to go in a different direction. That is, his theory seems to be about justice as how we treat each other, rather than justice as cleansing the world's distributions of original sin. In other words, the substantive core of Nozick's theory is not as previously stated. Nozick's is a theory of just transfer, not just distribution. His real claim is not that a distribution is just if it arises by just steps from a just initial position, but instead that a transfer from one person to another is genuinely just if genuinely voluntary. Nozick's theory ultimately is not so simple, but this is its essence.

Voluntary transfer cannot cleanse a tainted title of original sin, but any injustice in the result will have been preexisting, not created by the transfer. We are fated to live in a world of background injustice, all of us descended from both victims and victimizers, so it is a virtue of Nozick's theory that it does not pretend we might achieve perfect justice if only we could “even the score.”33

Richard Epstein, in as excellent a short discussion of Nozick as one will find, makes a related point: “Any system of property looks backward to determine the ‘chain of title’ that gives rise to present holdings. But this is not because of any fetish with the past but chiefly from the profound sense that stability in transactions is necessary for sensible forward-looking planning.” See Epstein, Skepticism and Freedom: A Modern Case for Classical Liberalism (Chicago, IL: University of Chicago Press, 2003), 130. As Epstein would agree, though, dwelling too much on the past would be as problematic as ignoring the past, and for the same reason: it would reduce stability in transactions. A routine title search when selling a house is one thing; going back as many thousands of years as the land has been held is another.

Still, Nozick thinks, it remains possible for moral agents, living ordinary lives, to abide by his principle of just transfer and, to that extent, to have clean hands. Presumably, Nozick would have agreed that there is no future in evening the score.34

Nozick endorses the undoing of wrongful transfers, but the point of undoing a wrongful transfer is simply that: to undo a wrongful transfer, not to make current holdings match a favored pattern. Nozick has no solution (perhaps there is no solution) to the problem of how to rectify injustice committed not by people now living but by their ancestors. There are places where people have been evening the score for centuries, and it will not stop until people learn to forget the past. For a state-of-the-art argument that successful rectification is about victims and victimizers (or their descendants) getting together to repair damaged relationships and set the stage for a peaceful future, see Linda Radzik, “Making Amends” (American Philosophical Quarterly, forthcoming). I discuss Radzik and add my own thoughts on rectification issues in David Schmidtz, Elements of Justice (New York: Cambridge University Press, 2005).

As an exemplar of the kind of society that would develop in accordance with his own brand of entitlement theory, Nozick offers the ideal of a civil libertarian, free-market society governed by a minimal state (roughly, a government that restricts itself to defending the country's borders and keeping the peace within its borders). In such a society, to the extent that people trade by consent and on mutually agreeable terms, there will be “strands” of patterns; people amass resources in proportion to their ability to offer goods at prices that make people around them better off. Employees tend to be promoted when their talents and efforts merit promotion, and so on. However, although society will be meritocratic to that extent, the pattern will be only one among many. There will be inheritance, gift-giving, and philanthropy too—all conferring goods on recipients who may have done nothing to deserve such gifts.35

Nozick, Anarchy, State, and Utopia, 158.

Is that a problem? Not to Nozick. Nozick joins Rawls in denying that merit is a principle to which distributions (and transfers) must answer. The question, Nozick says, is whether people deal with each other in a peaceful, consensual way.

The problem Nozick sees with patterned principles of justice such as Rawls's difference principle is that, in Nozick's words, liberty upsets patterns. “[N]o end-state principle or distributional patterned principle of justice can be continuously realized without continuous interference with people's lives.”36

Ibid., 163.

Notice that Nozick neither argues nor presumes that people can do whatever they want with their property. Nozick's point is, if there is anything people can do—even if the only thing they are free to do is give twenty-five cents to a basketball player—then that tiny liberty will, over time, disturb the pattern.37

I borrow the point from Edward Feser, On Nozick (Toronto: Wadsworth, 2004), 71.

It is a mistake, Nozick concludes, to think end-state principles give people what entitlement principles do, only better distributed. Entitlement principles recognize rights to choose that end-state principles cannot recognize. None of the resources governed by time-slice principles would ever simply be at a person's (or, for that matter, even a whole nation's) disposal.38

Nozick, Anarchy, State, and Utopia, 167. Rawls's reply: “The objection that the difference principle enjoins continuous corrections of particular distributions and capricious interference with private transactions is based on a misunderstanding.” On the next page, Rawls clarifies: “[E]ven if everyone acts fairly as defined by the rules that it is both reasonable and practicable to impose on individuals, the upshot of many separate transactions will eventually undermine background justice. This is obvious once we view society, as we must, as involving cooperation over generations. Thus, even in a well-ordered society, adjustments in the basic structure are always necessary.” See John Rawls, Political Liberalism (New York: Columbia University Press, 1993), 283–84. Rawls's clarification makes it hard to see why Rawls thinks Nozick misunderstood him. (I thank Tom G. Palmer for this point.) In any case, this is a major challenge in constructing a well-ordered constitutional democracy: how to avoid “necessary adjustments” that tell citizens that ownership of their income is a political football, and they are, to that extent, governed by men, not law.

Although Nozick is right in seeing a huge problem with time-slice principles, not all patterned principles are prescriptions for time slices or even for end states. There are passages where Nozick seems to assume that by arguing against end-state or time-slice principles, he is clinching the case against patterned principles more generally. Not so. Not all patterns are the same, and not all require major interference. If we focus on time slices, we are focusing on isolated moments. We are taking such moments too seriously, when what matters is not the pattern of what people hold at a moment but the pattern of how people treat each other over time. Although even tiny liberties must upset the pattern of a static moment, there is no reason why liberty must upset an ongoing pattern of fair treatment.

A moral principle forbidding racial discrimination, for example, prescribes no particular end-state. Such a principle is what Nozick calls weakly patterned; it is sensitive to history as well as to pattern, and prescribes an ideal of how people should be treated without prescribing an end-state distribution.39

Nozick, Anarchy, State, and Utopia, 164.

It affects the pattern (as would a purely historical principle) without prescribing a pattern (or more precisely, without prescribing an end-state). And if a principle forbidding racial discrimination works its way into a society via cultural evolution rather than legal intervention, it need not involve any interference at all.

If we create a society where Martin Luther King's dream comes true, and his children are judged not by the color of their skin but by the content of their character, then what we achieve is a fluid, evolving pattern that tracks one dimension (merit) rather than another (skin color). In the process, society comes to require less intervention than the forcibly segregated society from which it evolved. Thus, although Nozick sometimes speaks as if his critique applies to all patterns, we should take seriously his concession that “weak” patterns are compatible with liberty. They may even promote liberty, depending on how they are introduced and maintained. So, the problem is not with patterned principles in general but more particularly with end-state and especially time-slice principles.

A weakness in Nozick's critique of Rawls, then, is this: Nozick is right that time-slice principles license immense, constant, ludicrous interference with everyday interaction, but is Rawls defending such a view? In his first published article, Rawls said, “[W]e cannot determine the justness of a situation by examining it at a single moment.”40

John Rawls, “Outline of a Decision Procedure for Ethics,” in Rawls, Collected Papers, 14. The article was originally published in 1951.

Hence, we can doubt that Rawls ever entertained a time-slice conception. Rawls later said, “It is a mistake to focus attention on the varying relative positions of individuals and to require that every change, considered as a single transaction viewed in isolation, be in itself just. It is the arrangement of the basic structure which is to be judged, and judged from a general point of view.”41

Rawls, A Theory of Justice (1971), 87–88.

Thus, to Rawls, a basic structure's job is not to make every transaction work to the working class's advantage, let alone to that of each member of the class. Instead, it is the trend of society as a whole over time that is supposed to benefit the working class. To be sure, Rawls was a kind of egalitarian, but he was not a time-slice or even an end-state egalitarian. The pattern Rawls meant to weave into the fabric of society's basic structure was a pattern of equal status, applying not so much to a distribution as to an ongoing relationship.42

I thank Alyssa Bernstein for her help with this point.

The strength of Nozick's critique lies in how it drew attention to a prohibitive cost of any theory that is time-slice. Recent egalitarian work is an evolving response to Nozick's Wilt Chamberlain example and to the enormity of the flaw in time-slice principles that the example revealed.43

In this example, a million basketball fans choose to pay twenty-five cents each directly to Wilt Chamberlain in exchange for seeing him play—thus upsetting a previously existing pattern of distribution.

Today's egalitarians are realizing that any equality worthy of aspiration will not be a static property of a distribution at a particular moment. It will instead concern how people are treated: how they are rewarded for their contributions and enabled over time to make contributions worth rewarding.

IV. The Very Idea of Distributive Justice

Nozick thought that a bias against respecting the separateness of persons lurks in the very idea of distributive justice. The idea leads people to think of goods as having been distributed by some mechanism for which we are responsible. Generally, Nozick believes, there is no such mechanism and no such responsibility. “There is no more a distributing or distribution of shares than there is a distributing of mates in a society in which persons choose whom they shall marry.”44

Nozick, Anarchy, State, and Utopia, 150.

No one arranges a distribution of mates. Therefore, barring further argument, no one has a duty—indeed, no one has any right—to make sure the distribution of mates is fair.

Suppose for argument's sake that the fair way to redistribute whatever you have a right to redistribute is in accordance with the difference principle. We still face a prior question: What gives you any right to redistribute my income, or anything else? I may agree with Rawls that I do not deserve X, but so what? How does that give you a right to redistribute?45

As Steven R. Smith (approvingly) summarizes, “individuals are seen by Rawls as not deserving their initial endowment packages. Consequently, talents are, to some degree, socially owned.” Consequently? How would we get from the premise that talents are undeserved to the conclusion that talents are community property? Also, which community would we be talking about? Why not the United Nations, or the Church of Scientology? See Steven R. Smith, “The Social Construction of Talent,” Journal of Political Philosophy 9, no. 1 (2001): 31.

Does it beg the question to refer to X (my mate, say, or a wallet I found in a parking lot) as mine? Suppose we say yes, for argument's sake. Still, how do we get from there to a conclusion that you have a right to redistribute X?

Showing that I don't own my mate, or a wallet I found in a parking lot, is not enough; you have to show that you do. Barring more argument, you have no right to redistribute in accordance with the difference principle even if the difference principle is the true principle of justice. To establish that you have a right to distribute, what we need to show is not that what you want to do with the goods is fair. Rather, we need to show that the goods you want to distribute are yours to distribute.

Intuitively, we go some distance toward establishing that a government has a right to rearrange the natural distribution (or what people have done with it) if we can establish that the natural distribution is unjust. But is it? Here is what Rawls says: “The natural distribution is neither just nor unjust; nor is it unjust that persons are born into society at some particular position. These are simply natural facts. What is just and unjust is the way that institutions deal with these facts.”46

Rawls, A Theory of Justice (1971), 102.

What does Rawls have in mind? How should an institution deal with fact X? When should an institution deal with X? What if X is not otherwise unjust? If a distribution is not unjust, why would it need correcting? What exactly would count as correcting a distribution that is not unjust? (At the heart of a theory meant to show respect for the separateness of persons, why would justice mainly concern how institutions deal with facts about aggregate distribution—of talent or anything else?)

Earlier in the same passage, Rawls says, “[T]he difference principle gives some weight to considerations singled out by the principle of redress. This is the principle that undeserved inequalities call for redress.”47

Ibid., 100.

But if Rawls is right that the natural distribution is not unjust, unequal though it may be, and undeserved though it may be, then it does not call for redress either.

And why does Rawls say “undeserved”? Why not simply say that inequalities per se call for redress? Or if the notion of desert is doing work here that Rawls elsewhere seems committed to denying that the concept of desert can do, then why don't undeserved equalities likewise call for redress?

V. The Natural Lottery: Arbitrary but Not Unjust

Rawls says, “We are led to the difference principle if we wish to arrange the basic social structure so that no one gains (or loses) from his luck in the natural lottery of talent and ability, or from his initial place in society.”48

Rawls, “Distributive Justice,” 140. Strictly speaking, this consideration leads only to a precursor form of the difference principle, which specifies that arrangements ought to be mutually advantageous.

Why would anyone wish that? Rawls offers the following reason: “It may be expedient but it is not just that some should have less in order that others may prosper.”49

Rawls, A Theory of Justice (1971), 15 (emphasis added).

Here are three responses.

First, Nozick agrees that it is unjust if some have less so that others may prosper. Where Rawls and Nozick disagree is over the question, “less than what?” Rawls rejects the idea of some having less than an equal share (not of the actual social product, but of what the social product would have been in a regime of strict equality) so that others may prosper. Nozick rejects the idea of some being deprived of their earnings so that others may prosper. Can we know which baseline is more respectful of separate persons? Can we even know which is more generous? How? (To judge from history, the equal shares baseline in a regime of strict equality would never amount to much.)

Second, in the real world, no human being fixes the natural lottery, assigning hardworking characters to some and lazy characters to others. Is there a difference between a lottery Jane wins by luck of the draw and a lottery rigged to make sure Jane wins? Is there a difference between Jack turning out to be less skilled than Jane and a situation where Jack deliberately is held back so as to make sure Jane will be more skilled than Jack? Rawls says, “Once we decide to look for a conception of justice that nullifies the accidents of natural endowment and the contingencies of social circumstance … we are led to these principles. They express the result of leaving aside those aspects of the social world that seem arbitrary.”50

Ibid. (emphasis added).

But “arbitrary” has two meanings. Natural distributions can be arbitrary, meaning random. Or human choices can be arbitrary, meaning capricious.51

Cara Nine argues that arbitrariness is an essentially theory-laden concept, so that X is arbitrary only if X is wrong by the lights of a given moral theory. In that case, if we want to argue for a theory, and want to start with premises about what is arbitrary, the argument is likely to be question-begging. See Nine, “Moral Arbitrariness” (Ph.D. dissertation, University of Arizona, 2005).

In one case, no choice is made. In the second, an unjustified choice is made, unjustified in virtue of being literally unprincipled.52

When we call a choice arbitrary, we are implying not only that it is unjustified, and not only that it is wrong, but also that it exhibits a certain arrogance: there is nothing a person could say, or would even be inclined to say, to justify his or her choice. A person might say, “I can do whatever I want.”

The two are not equivalent. Fair lotteries are lotteries where winners are chosen at random. A rigged lottery is unfair because it fails to be arbitrary in the benign sense. It is by failing to be arbitrary in the benign sense that it counts as arbitrary in the bad sense. What of the natural lottery, then? The natural lottery is arbitrary in the benign sense, but how does that connect to being unfair in the way that capricious choice is unfair?

Rawls says, “Intuitively, the most obvious injustice of the system of natural liberty is that it permits distributive shares to be improperly influenced by these factors so arbitrary from a moral point of view.”53

Rawls, A Theory of Justice (1971), 72.

However, when “arbitrary” means random, as it does here, there is nothing obvious about any connection between being arbitrary and being improper. Capricious choice wears impropriety on its sleeve; the natural lottery does not. Had Jack's mother assigned Jane all the talent, deliberately leaving Jack with none, we might at very least wonder why. In fact, though, Jack's mother did not assign him less talent. It just happened. It was chance, not caprice.

A Rawlsian might respond that everything is deliberately assigned. When we deliberately fail to correct a natural distribution, we deliberately fail to correct what is otherwise unjust; when we deliberately fail to correct what is otherwise unjust, we deliberately commit injustice. Before any such argument can get started, though, we need to show that what we are declining to correct is otherwise unjust. The difference principle may correct a natural distribution, but Rawls understands that before this can seem like justice, we need to show that the distribution being corrected is otherwise unjust. Is it? Rawl himself (as discussed in the previous section) says no.

The third response is this: If we did assign Jack less talent, the reason would not be “so others may prosper.” Making Jack untalented would not help others. If we sought to assign Jack a talent level that would make people around Jack better off, we would assign Jack more talent, not less. What would give us reason to compensate Jack would be to make Jack a talented provider of high-quality services.54

In any system, the main way in which the talented share their talent is precisely by going as far as their talent can take them. What people pay in taxes normally will be trivial by comparison. To take an obvious case, how much did Thomas Edison pay in taxes? The answer is that we do not need to know, because we know that no matter how many millions Edison might have paid, the good done by his taxes was nothing compared to the good done by his inventions.

Ironically, if we did assign Jack his place in a distribution of talent “so that others may prosper,” it is Rawls, not Nozick, who would hesitate to compensate Jack. The worry: we would be compensating Jack not for disadvantages but for services rendered. Rawls allows compensation for services rendered as a way of manipulating incentives on behalf of the least advantaged, but the only compensation Rawls treats as required by justice per se is compensation for disadvantages.

One way to rationalize the idea that Jane's being a more talented worker entitles Jack to compensation is to say that if Jane is more talented, Jane captures more of the pie; therefore, Jack and others necessarily get less. However, it is Rawls's point, after all, that the pie's size is a variable. Society can, if all goes well, be a mutually advantageous venture. Almost everyone can have a better life than they could have had on their own, and the reason is simple: Talented bakers don't capture pie. They make it.55

Even the most self-reliant bakers, of course, cooperate.

Thus, it is an anomaly when Rawls says, “We are led to the difference principle if we wish to arrange the basic structure so that no one gains (or loses) from his luck in the natural lottery….” Gaining is good, not bad, so why prevent it? There is a problem with gaining only if people gain at someone else's expense, and in that case the problem is still with the losing, not the gaining.

No one reasonably thinks we live in a zero-sum world.56

This is not to deny that the assumption is at some level intuitively compelling, or that it tacitly informs a lot of political theorizing.

When a baby is born with a damaged spinal cord, it is not so that healthy babies prosper. When the next baby is born healthy, needing no special care, this baby's health does not come at the first baby's expense. The natural lottery is not zero-sum. Wanting to make sure that no one gains, thereby treating features that individuate us as persons as things to nullify, is, in Nozick's words, “a risky line to take for a theory that otherwise wishes to buttress the dignity and self-respect of autonomous beings.”57

Nozick, Anarchy, State, and Utopia, 214.

From Rawls's perspective as well as Nozick's, any argument for taxing what Jack pays to hear Jane play piano has to come from a different direction. We can and should give up the idea that there is something wrong, crying out for correction, when Jack and Jane reap mutual benefits from the sharing of her talent. To justify taxing what Jack pays Jane, we will have to argue, more simply and more honestly, that what the government will do with the money (helping the least advantaged, say) is important enough to justify taking the money. In truth, however, when we tax income, we are trying to raise revenue, not correct injustice. Why pretend otherwise?

If we think a government needs to tax what Jane pays Jack for services rendered, as a way of financing programs, then we should honestly say so, and reject the premise that the only way to justify a tax is to prove that it is rectifying an injustice. And once we reject that premise, we no longer need to posit that the natural distribution is unjust (or worse, to posit that while the natural distribution is not unjust, it still needs to be redressed). The distribution of talent per se is a moral nonissue. People often lack the money to solve otherwise solvable problems. That may not be a problem of justice, but it is, in any case, the real issue. The problem is not that Jane and Jack have unpaid debts to society, but that society really needs their money.

Let me put the point in a different way. Even if, as Rawls says, there is no injustice in a natural distribution, there may yet be a problem. Being born with a cleft palate is a problem. The problem is not that a cleft palate is unjust but that it is bad. Its badness gives us some reason to intervene so as to fix the problem, assuming we can fix the problem without violating rights or otherwise doing more harm than good.

Note the nature of the real issue: We are not trying to fix an improper distribution of cleft palates. We are trying to fix cleft palates.

VI. The Separateness of Persons

Nozick sometimes is accused of having no foundation, of merely assuming what he needs to prove. Needless to say, it is hard to develop a solid foundation for a theory of justice. No philosopher is widely regarded as having succeeded in doing so. Still, Nozick could have replied that, rightly or wrongly, he was borrowing Rawls's foundation. (And if any philosopher were regarded as having developed a solid foundation, it would be Rawls.) For the sake of argument, Nozick accepted Rawls's premises about the separateness and inviolability of individual persons. Thus, Nozick says, the “root idea, namely, that there are different individuals with separate lives and so no one may be sacrificed for others, underlies the existence of moral side constraints.”58

Ibid., 33. See also A. R. Lacey, Robert Nozick (Princeton, NJ: Princeton University Press, 2001), 25ff.

If this premise is not a foundation, then Rawls does not have a foundation either. Nozick's departure from Rawls was to treat such premises not as imaginary devices designed, defined, then discarded on the road to Rawls's two principles; instead, Nozick asked, when we pay homage to personal inviolability, what if we meant it? When we say people are entitled to the most extensive sphere of liberty compatible with a like liberty for all and that this principle, the first principle of justice, takes priority over all other considerations, what if we actually meant it?59

For a cogent critique of the retreat from the first principle's original maximalist formulation, see Loren Lomasky, “Libertarianism at Twin Harvard,” in this volume.

Rawls says, “[W]hen society is conceived as a system of cooperation designed to advance the good of its members, it seems quite incredible that some citizens should be expected, on the basis of political principles, to accept lower prospects of life for the sake of others.”60

Rawls, A Theory of Justice (1971), 178.

Rawls is complaining about the principle of utility, but as Thomas Nagel notes, the same complaint applies to the difference principle.61

Thomas Nagel, “Rawls on Justice,” in Reading Rawls, ed. Norman Daniels (Stanford, CA: Stanford University Press, 1989), 13.

Some critics think it is worse than that. At least under the principle of utility, everyone counts as one. Peter Singer says:

Nozick is able to make the telling point that the fundamental flaw Rawls finds in utilitarianism—the failure to rule out “even the tendency to regard men as means to one another's welfare”—can be found in Rawls's own principle. The maximin rule treats the better-off as a means to the welfare of the worst-off. Indeed, one could say (though Nozick does not) that the tendency to treat people as a means to another's ends is greater under the maximin rule than under utilitarianism, since a utilitarian would give equal consideration to everyone's interests, whereas the maximin rule forbids giving any consideration to the interests of the better-off, allotting them goods solely in so far as doing so assists the worst-off.62

Peter Singer, “The Right to Be Rich or Poor,” in Reading Nozick, ed. Jeffrey Paul (Oxford: Blackwell, 1981), 48. Rawls likely would reply that utilitarianism is cavalier in its willingness to benefit some at the expense of others, so long as total utility thereby increases. Rawls's theory requires that transfers be justified to each individual. Singer probably would say that if that means the well-off have a veto, then Rawls's theory has all the problems of Nozick's; if not, the distinction between Rawls's theory and utilitarianism has not yet been made.

Routinely, people sacrifice in the name of their overall best interest. Jane sometimes gives up a night at the movies to study for an exam. According to utilitarianism, it is no better and no worse, other things equal, if Jane sacrifices someone else in the name of her overall best interest.63

For excellent discussions of how the separateness of persons issue relates to issues of personal autonomy and agent-relative values, see chap. 5 of Horacio Spector's Autonomy and Rights: The Moral Foundations of Liberalism (Oxford: Oxford University Press, 1992); Eric Mack, “Moral Individualism: Agent-Relativity and Deontic Restraints,” Social Philosophy and Policy 7, no. 1 (1989): 81–111; and Eric Mack, “Self-Ownership, Marxism, and Egalitarianism,” Politics, Philosophy, and Economics 1 (2002): 75–108.

Matt Zwolinski summarizes: “It is because utilitarianism sees no morally relevant difference between tradeoffs within a life and tradeoffs between lives, that it is said to fail to take seriously the separateness of persons.” Zwolinski finds it curious that Rawls and Nozick could end up in such different places, having seemingly started with the same concern.64

One difference, explored by Matt Zwolinski in his “The Separateness of Persons” (American Philosophical Association, 2003), is that while Nozick thinks of separateness as implying restrictions on how people may be treated, Rawls thinks of separateness as a meta-level desideratum. It does not restrict what we can do to people; instead, it restricts what we can offer as a justification for what we do.

My hypothesis is this: there are two “separateness” issues. (1) Rawls cares about our separateness as consumers and says we cannot make up for one consumer having less by enabling others to have more. More precisely, Rawls does think we can make up for one person having less by enabling others to have more. For example, we can make up for the more advantaged having less (than they otherwise would) by enabling the less advantaged to have more, so long as we do not plunge the more advantaged below a baseline of equal shares.65

Rawls in effect holds that the sole implication of separateness is to require a basic structure to justify to each person the assigning of a consumption bundle to that person. See Zwolinski, “The Separateness of Persons,” for more on the distinction between respecting separateness in the sense that rules out involuntary transfers from one to another and respecting separateness in the sense of involuntary transfers having to be justified to each individual. (But I note here that Rawls and his followers are not saying each individual has to agree that the transfers are justified; it is enough that “reasonable” people agree.)

Only then do we use people as mere means. To deem people separate in any other way is to assign undue weight to factors “so arbitrary from a moral point of view.” (2) Nozick respects our separateness as producers. To Nozick, dismissing what we do as arbitrary is disrespectful. Nozick wants to buttress our dignity as producers of goods we bring to the table, not merely as consumers of goods we take from the table.

One factor separating Nozickians from Rawlsians is that Nozickians tend to see rewards (i.e., products) as created by workers, and thus as presumptively belonging to workers (meaning that workers normally are within their rights to sell their products to willing customers on mutually agreeable terms). Rawlsians tend to see rewards as created by society and as society's responsibility to distribute according to principles of justice. Implicit in the latter view: there is no fundamentally real category of separate producers; society has productive components, and produces or fails to produce according to its success in arranging the reward structures of its components.

Rawls does not say why what separates us as consumers (we have separate stomachs, and so on) is less arbitrary than what separates us as producers. Not only is this a difference between Rawls and Nozick; it divides theorists of justice in general. The question is, should we treat production as external to the topic of justice? Is it true that when we think about justice, we need not and maybe should not think about which arrangements empower human beings to undertake the productive activities on which welfare and progress depend? Robert Paul Wolff says that Nozick's criticism is on target: “The veil of ignorance has the effect of making all considerations on the production side so thoroughly hypothetical, so abstract in the bad sense, that inevitably the difference principle comes to be construed as a pure distribution principle, with the distributable goods and services exogenously given.”66

See Robert Paul Wolff, Understanding Rawls (Princeton, NJ: Princeton University Press, 1977), 201.

The problem of cooperation, for Rawls, is a problem of how to distribute what people contribute, which translates into a question of how to distribute inequalities. The Nozickian worry is: Isn't contributing the essence of cooperating? Can we truly have a problem of cooperation that is not centrally a problem of how to respect what people contribute? What if normal people reject the idea that a basic structure's job is to distribute inequalities, and instead say a basic structure's job is to respect the fact that people are separate as producers, not only as consumers?

For Rawls, we reject utilitarianism and undertake a contractarian thought experiment, partly because the latter does not lump us together as utilitarianism does. Rawls acknowledges and even stresses the separateness of persons in his opening statements, but as the theory takes shape, it increasingly reflects an attitude that separateness is regrettable. Having begun with contractarianism, Rawls's subsequent moves seem aimed at mitigating the separateness with which contractarian thought begins. Nozick, by contrast, not only acknowledges our separateness as producers; he celebrates it. Nozick denies that we are obliged to benefit each other per se. Our obligation is simply to deal with each other only on mutually acceptable terms.

If the goods Rawls wants to distribute had no history, that would be different:

If things fell like manna from heaven, and no one had any special entitlement to any portion of it, and no manna would fall unless all agreed to a particular distribution, and somehow the quantity varied depending on the distribution, then it is plausible to claim that persons … would agree to the difference principle rule of distribution. But is this the appropriate model for thinking about how the things people produce are to be distributed?67

Nozick, Anarchy, State, and Utopia, 198.

This shows that Nozick's complaint is not that Rawls has a bad idea about how to distribute manna so much as that Rawls has a bad idea about what counts as manna. To Nozick, advantages are not what bargainers find on the table; advantages are what bargainers bring to the table. And respecting what people bring to the table is the exact essence of respecting them as separate persons.

Does Rawls treats advantages as manna? That is how Nozick reads him. Many read Rawls the same way. Moreover, there is no consensus that this reading is unkind; many embrace Rawls because they like the idea of treating advantages as manna. Rawls, trying to soften earlier statements, said, “[T]he difference principle represents an agreement to regard the distribution of native endowments as a common asset and to share in the benefits of this distribution whatever it may be. It is not said that this distribution is a common asset….”68

Rawls, Justice as Fairness: A Restatement, 75 (emphasis added).

But if we cannot say the distribution is a common asset, what gives us a right to agree (or worse, treat people as if they had agreed) to regard it as if it were?

Nozick concedes that the difference principle applies to manna from heaven, but adds that what we achieve is not manna. Treating achievement as manna fails to respect the separateness of achievers, and, consequently, fails to elicit their willing participation. Rawls accepts the latter, the incentive issue, as a concern. For Nozick, though, the issue is about respect, not incentives. Respecting people's histories, Nozick believes, is a crucial part of respecting their separateness.

VII. The Original Position as Experience Machine

Nozick's story of the “experience machine” illustrates his concern for the separateness of persons as productive agents capable of making their world better. The experience machine lets us plug our brains into a computer programmed to make us think we are living whatever we take to be the best possible life. The life we think we are living is a computer-induced dream, but we do not know that. Whatever experience would be part of the best possible life for us—anything at all—will in fact be part of our felt experience. Nozick asks, “Would you plug in? What else can matter to us, other than how our lives feel from the inside?”69

Nozick, Anarchy, State, and Utopia, 43 (emphasis in original). See also David Schmidtz, “The Meanings of Life,” in Robert Nozick, ed. David Schmidtz (New York: Cambridge University Press, 2002), 199–216.

Nozick's story is among the most talked-about puzzles in contemporary philosophy. Nozick does not present the story as connecting to his critique of Rawls, and I have no evidence that Nozick himself saw a connection, but there is a connection nonetheless. Nozick's idea is that a person is not merely a location of experience. What makes us agents is what we do, what we strive to do, and what we accomplish. The experience machine robs us of moral agency. What about Rawls's original position? Does it do something similar?

If what makes us agents is not experiencing but striving, there could be a problem. Rawls insists that agency in the sense of striving is morally arbitrary, at least when the topic is what we can claim from the basic structure as a matter of justice. Rawls's thought experiment, the veil of ignorance, assumes that what matters—what we focus on when choosing principles—is the sort of bundle we could be given by an experience machine. Rawls says a person behind the veil “has a conception of the good such that he cares very little, if anything, for what he might gain above the minimum.”70

Rawls, A Theory of Justice (1971), 154.

To critics, this looks far-fetched, but it becomes plausible when we see Rawls as using the phrase “a conception of the good” to mean a conception of the good we can have rather than a conception of the good we can do.

What if our conception of the good has to do with what we achieve in life? Is it true that we don't care what we achieve above the minimum? Is it true that we shouldn't care, or that behind the veil, we wouldn't care? Should we suppose that, behind the veil, Michelangelo and Edison do not care what they achieve above the minimum, or that the rest of us would not care? Even if that were imaginable, what difference would it make once the veil is lifted and people find out who they are? Once the veil is lifted, what is to stop someone like Michelangelo from saying, “I'm sorry. You designed the veil to prevent me from knowing what really matters. It wrongly required me not to think beyond my own advantage in material terms. It required me to deliberate as if my talent were not a sacred calling. Given what I have inside me, I have no right not to care what I achieve above the minimum.”

Since a few of us will in fact turn out to be Edisons and Michelangelos, have we any right not to care what we collectively achieve above the minimum? If some groups of bargainers decide they do care, would they be mistaken? Suppose this leads some groups to favor the position of the most talented? Would they be wrong?71

Nozick asks us, how would Rawlsian bargainers assign grades? Nozick says Rawlsian bargainers cannot even consider letting grades track performance. Bargainers want to make the lowest grade as high as possible. They do not care (or do not know they care) about the difference between giving the best grade to the best student and giving the best grade to the worst.

It would be easy to overstate Nozick's point. In a zero-sum game, as where an average grade is fixed, Nozick would be right, but when we speak of positive-sum games, as in real market societies, everything changes. In a positive-sum game, Rawls can rule out giving the best grade to the worst student on strategic grounds, allowing that rewards can be used to manipulate the talented into performing in a way that enlarges the pie (maximizing the smallest slice). Nozick's complaint is not that such rewards would be too small, but rather that their rationale is brutally disrespectful.

Nozick's ideal basic structure would give us a chance to live a life, not just experience it. Real life contains danger and failure and episodes of deprivation that a risk-averse person would avoid, but those things go with really living. Rawls wants us to live a life too, of course, yet when selecting principles, we are supposed not to know or care about the life we so far have lived. Until the veil is lifted, we are to focus solely on how much we get, untainted by morally arbitrary questions about who we are and whether we did anything to deserve that much.72

For an argument that we can deserve X in virtue of what we do after receiving X, see David Schmidtz, “How to Deserve,” Political Theory 30, no. 6 (2002): 774–99. I thank Dan Russell for reminding me of this argument's relevance.

What we thereby set aside as arbitrary is doing—precisely what is missing from the experience machine.

It must seem obvious that nothing is stopping Rawls from making room in his theory for less advantaged people to have opportunities to do things, and deserve things, not just experience them. However, if Rawls were to do that, it would have an awkward consequence. If we say bargainers would choose principles that explicitly express respect for what less advantaged people do, perhaps letting respect for their welfare be merely implicit, as in Nozick, we will be agreeing not merely to respect what less advantaged people do from now on, on the grounds that such respect is in their interest. We will also be choosing to respect what they have been doing all along. And if we are to give them the respect that is in their interest, we will also have to regard ourselves as giving them what they deserve and have deserved all along. Otherwise, our attitude toward them is a paternalistic simulation of respect, not the real thing.

If we acknowledge that people command respect in virtue of what they have been doing all along, we are acknowledging that respect is not a matter of us deciding what they reasonably would accept on the assumption that their history does not matter. Nor is respect compatible with insisting that we care about history and about what people deserve, then proceeding to pick principles of justice as if such things did not matter (or as if they begin to matter only after, and only if, we pick principles that underwrite them). If such things matter, then the moment when such things have to be taken into account is the moment when we are picking principles of justice.

VIII. The Original Position as Utopia

Nozick accuses Rawls of devising a procedure guaranteeing that end-state and not historical principles will be chosen. Far from denying it, Rawls goes further, saying his intent was to guarantee that not only patterned principles in general but the difference principle in particular will be chosen. Rawls repeatedly stresses, “We want to define the original position so that we get the desired solution.”73

Rawls, A Theory of Justice (1971), 141.

The conclusion is not that Rawls's project is illegitimate but, rather, that it has to be understood as Rawls later came to understand it. That is, the project is to articulate what “we” implicitly believe, not to force us to a conclusion we would rather reject. Thus, Rawls's admitting why he designed the original position as he did, disconcerting though it may be, does not prove that his principles are wrong, or even that his argument is bad. Perhaps the original position is, after all, a fair test of competing conceptions, and perhaps Rawls's two principles are uniquely capable of passing that test.

How would we decide whether the original position is a fair test? Rawls says, “The idea of the original position is to set up a fair procedure so that any principles agreed to will be just.”74

Ibid., 136.

Yet, needless to say, our imagining a jury reaching a guilty verdict is no substitute for a fair trial. To set up a fair procedure, we first must set up a procedure. What goes on in our imagination is not a procedure, let alone a fair one.

One reason to see the original position as fair is that it would, if it were real, put bargainers in a position of not being able to bias the negotiation in their own favor. Not knowing what position they will occupy forces real bargainers to negotiate in a more impartial way. But if this feature is what marks the original position as fair—as I believe—then the other features (e.g., the assumption that society is closed) are dispensable. Or at least, the other features are dispensable if the objective is to “set up a fair procedure so that any principles agreed to will be just.” The other features are required only insofar as the objective is instead “to get the desired conclusion.”

A. Suppose we do not know what conclusion we desire

Suppose there is no “desired conclusion” and our only goal is to set up a fair procedure and then let people get whatever conclusion they desire. That is, suppose we theorists could set aside our preconceptions regarding what conclusion is “desired.” If we simply wanted to preserve the impartiality (thus the fairness) of the original position, and had no other agenda, what procedure might we set up?

Here is a suggestion. Suppose we theorists put ourselves behind a veil of ignorance. Imagine us trying to construct a fair bargaining game without knowing anything about our own conception of justice. Suppose we do not know what solution we desire, and thus, unlike Rawls, cannot “define the original position so as to get the desired solution.” What would we do? Would we posit that bargainers do not care who brought what to the table; they care only about what they get, yet care little for what they might get above the minimum?75

As Thomas Nagel worries, “Keeping in mind that the parties in the original position do not know the stage of development of their society, and therefore do not know what minimum will be guaranteed by a maximin strategy, it is difficult to understand how an individual can know that he ‘cares very little, if anything, for what he might gain above the minimum’.” See Nagel, “Rawls on Justice,” 12.

Would we posit grave risks in seeking gains above the minimum? Would we assume severely handicapped people are not represented at the bargaining table? Would we imagine ourselves picking rules for a closed society? Would we assume bargainers start with equal claims to (the distribution of) each other's advantages, conceived as a collective asset?

Presumably, we would do none of the above. We might assume bargainers do not know their position in the distribution, since intuitively that has something to do with impartiality, which intuitively has something to do with fairness. But if we did not know our conception of justice—if we had no idea whether we were egalitarians or elitists, no idea which conception we endorsed—then we would not be designing the situation so as to converge on any particular conception. In that setting, we would expect bargainers to try to choose principles that were good for all. We would have no reason to predict anything more specific than that. We would expect different sets of bargainers to converge on different conclusions.

Moreover, that would not trouble us, insofar as we took seriously the idea of procedural justice: the idea that justice is about following a given procedure rather than reaching a given conclusion. On a procedural conception, we conclude that if, after fair deliberation, people agree to bind themselves to each other in a particular way, then by that very fact they are bound to each other in that way. If other people, after similarly fair deliberation, agree to bind themselves on different terms, then they are bound on those different terms. By the lights of procedural justice, the fact that groups of people are bound in different ways is not a problem.

B. Suppose we do not know whether equal shares is the default

Consider another modification of the Rawlsian experiment: Instead of assuming that bargainers begin from a position of strict equality, assume—more abstractly but less controversially—that bargainers begin from a situation where nothing needs rectifying. No one complains about how anyone acquired current holdings. (Some are satisfied that holdings are sufficiently equal. Some are satisfied that no one is in possession of stolen property. Others realize they have not yet picked principles of justice, and therefore lack standards for judging current holdings. Anyway, no one is complaining.) We assume nothing about how things are distributed in this situation, and thus avoid building our preference for egalitarian (or nonegalitarian, as the case may be) principles into the situation's starting point. Such a procedure would have advantages over Rawls's. To mention only one, by not specifying any particular distribution, we avoid giving that specified distribution a position of unearned privilege in the debate about what the distribution ought to be.

Is this thought experiment relevant to a real world where people do have complaints about existing holdings? Maybe not, and maybe that is a good objection. But in that case, we must conclude not that we should reject this thought experiment but that we should reject all such thought experiments, including Rawls's. All such experiments, including Rawls's, assume that we need not return goods to their rightful owners, and that we can focus on distributing goods as if goods were presenting themselves to us in an unowned state. If that assumption is wrong, then all such thought experiments are wrong.

IX. Conclusion: What We Can Learn

We live in a world where goods have histories, and where people care not only about the goods; they care about the histories too. To respect people—to give them their due—is to respect their history, not merely to assign them goods. Nozick argues that no theory of justice is plausible unless it takes history seriously. A theory takes seriously our separate personhood only if it takes seriously our separate histories. For example, as Rawls would have agreed, sometimes justice is about returning a stolen wallet to the person from whom it was stolen. Why return the wallet to that person? Not to restore a previously fair pattern but to restore the wallet to the person from whom it was stolen. We know justice is about returning the wallet, not distributing it. The wallet's history morally trumps any thoughts about how it might best be distributed.

At the same time, taking our histories seriously—taking them every bit as seriously as Nozick takes them—does not rule out taking certain patterns seriously. The larger project of which this essay is a part explains how a theory may integrate patterned and historical principles together to form a pluralistic theory that makes room for principles of desert, equality, reciprocity, and need.76

Schmidtz, Elements of Justice (see note 34).

Footnotes

For comments, I thank Alyssa Bernstein, Geoffrey Brennan, Jason Brennan, Tom Christiano, Andrew I. Cohen, Andrew Jason Cohen, Tyler Cowen, Teresa Donovan, David Estlund, Jerry Gaus, Allen Habib, Alex Kaufman, Mark LeBar, Loren Lomasky (especially Loren, for insight and inspiration over a period of many years), Cara Nine, Ellen Frankel Paul, Guido Pincione, Thomas Pogge, Dan Russell, Michael Smith, Horacio Spector, and Matt Zwolinski. I thank the Earhart Foundation for financial support in the fall of 2002 and Australian National University's Research School of Social Sciences for its wonderful hospitality during a ten week stay in 2002. The support of the folks at Liberty Fund in Indianapolis during the final stages of this project goes beyond anything I will ever be able properly to thank them for.