Constructing global justice: a critique

Abstract

This essay criticizes a prominent strand of theorizing about global justice, Rawlsian global constructivism. It argues that the constructivist method employed by cosmopolitan and social liberal theorists cannot grapple with the complexities of interdependence, deep pluralism, and socio-cultural diversity that arise in the global context. These flaws impugn the persuasiveness and plausibility of the substantive conclusions reached by Rawlsian global constructivists and highlight serious epistemological problems in their approach. This critique also sheds light on some broader problems with ideal theory in the global context, showing how it leads to distortions in our thinking about justice and again raising doubts about the epistemological and normative conclusions of global constructivist approaches.

• global justice
• constructivism
• cosmopolitanism
• social liberal theory
• ideal theory
• Rawls

Political constructivism is most familiar as the method of reasoning adopted by John Rawls in his monumental A Theory of Justice. In that work, Rawls introduced the idea of an original position, a hypothetical initial choice situation in which representatives deliberate on and agree to principles of justice to govern the basic structure of their society. Almost immediately theorists inspired by Rawls's arguments adopted this same procedure in thinking about global justice.1 These ‘global Rawlsians’ saw his work as militating in a radical and explicitly redistributive direction globally; for them, political constructivism presented a powerful tool for arguing about the global reach of justice. Many of these thinkers were undoubtedly surprised and disappointed when Rawls himself adopted a different approach, one that applied constructivist reasoning to an agreement among societies or peoples and disavowed extensive global redistribution. The ensuing debate between these ‘cosmopolitan’ and ‘social’ liberals has become a mainstay of international political theory.2

This essay has two aims. The first is to show that global applications of Rawls's constructivist procedure are deeply flawed methodologically and epistemologically. Surprisingly, given the prominence of this debate and the substantive controversies it generates, methodological and epistemological questions have gone largely unaddressed. Cosmopolitan and social liberals, though differing significantly on the correct way to model the global initial choice situation(s), accept that the initial agreement method can address global justice without any particular difficulty. (I shall refer to both cosmopolitan and social liberal arguments as variants of ‘Rawlsian global constructivism’.3) Indeed, the application of constructivist reasoning to problems of global justice is rarely considered at all by its practitioners. While some critics have pointed out flaws in or limits to global constructivist reasoning,4 there has so far been no systematic attention to methodological and epistemological issues surrounding Rawlsian global constructivism.

This essay demonstrates that the structural assumptions on which Rawlsian constructivist reasoning depends in the domestic context cannot simply be transferred to the global domain, as Rawlsian global constructivists assume. The method cannot grapple with the complexities of interdependence, deep pluralism, and socio-cultural diversity that arise in the global context. Specifically, what often appear to be modeling choices actually establish epistemological constraints that determine important normative questions. Without such arbitrary choices, the method becomes indeterminate. These flaws impugn the persuasiveness and plausibility of the substantive conclusions reached by Rawlsian global constructivists, whether cosmopolitans or social liberals. This finding upends several of the most prominent and widely cited conceptions of global justice in today's debates. It also shows, counter-intuitively, that cosmopolitan and social liberal constructivist accounts suffer from the same problems and limitations.

The second aim of this essay is to use the critique of Rawlsian global constructivism to help shed light on some problems of ideal theory. I show that ideal theorizing, as exemplified in cosmopolitan and social liberal constructivism, leads to distortions in our thinking about justice, again raising doubts about the epistemological and normative conclusions of these approaches. This argument thus reinforces and expands on the primary case against Rawlsian global constructivism elucidated in the paper. It also brings much-needed specificity and depth to the discussion of ideal theory, underscoring the need for alternatives.

Before beginning, I want to address two potential objections. The first concerns the charge that this essay flogs a dead horse: Beitz and Pogge have abandoned constructivist arguments, the objection would go, and the most interesting work on global justice today is being done by scholars who adopt other approaches. This objection is unpersuasive. While Beitz and Pogge no longer advance constructivist arguments, they are still widely cited in current scholarship as exemplars of constructivist reasoning and in support of a global difference principle; neither author has explicitly retracted his earlier positions. Further, new work utilizing constructivist reasoning continually appears: Gillian Brock's Global Justice: A Cosmopolitan Account (Oxford 2009) is perhaps the most notable recent example. Moreover, there has been a recent resurgence of interest in defending Rawls's two-stage approach.5 Finally, as noted above, the deficiencies of constructivist reasoning elaborated here are directly relevant to the growing debate on ideal theory.

The second objection concerns my treatment of Rawlsian constructivist reasoning in the domestic context. Rawlsian global constructivists assume that Rawls's method is unproblematic domestically—that the method works more or less the way Rawls says it does. There is a vast literature on every aspect of Rawls's work, including on his methodological choices at the domestic level. Some readers have suggested that I engage with this literature, taking a position on Rawls's domestic procedures. I do not do so for three reasons. First, quite simply, space constraints make even a cursory engagement with this literature impossible. Second, Rawls's domestic method is not directly relevant to my critique of global applications of the Rawlsian approach. I emphasize the distinct and serious problems that arise specifically in the global context: that some of these criticisms might also apply to domestic applications of Rawlsian constructivism does not undermine—it in fact bolsters—my argument. Finally, my position is strengthened rhetorically and argumentatively by showing that Rawlsian global constructivism doesn't work on constructivism's own terms. Even—perhaps especially—if one endorses Rawls's domestic application of the method, one can still accept my critique of its global variants. In bracketing the domestic procedure I do not thereby endorse or defend it; I focus on the global variants without prejudice to domestic questions.

ASSUMPTIONS OF RAWLSIAN CONSTRUCTIVISM

Rawls argues in A Theory of Justice (TJ) that his conception of justice as fairness ‘generalizes and carries to a higher level of abstraction the familiar theory of the social contract’.6 According to Rawls, all contract theories offer an interpretation of an initial choice situation (ICS) and a set of principles that would be agreed to in that situation (TJ 14). The original position, which Rawls describes as the ‘most philosophically favored interpretation of this initial choice situation for the purpose of a theory of justice’, is a ‘purely hypothetical situation characterized so as to lead to a certain conception of justice’ (TJ 11, 16). The principles of justice that would govern the basic structure of society are the object of the agreement reached in the original position (TJ 11). Rawls later clarified that his method could also be understood as an exercise in Kantian constructivism (KC 516). I refer to Rawls's entire method as ‘constructivist’ rather than ‘contractual’ because, unlike most contract arguments, the (actual or hypothetical) agreement of the parties is not a source of binding political obligation for Rawls (TJ 112–114) but rather related to the justification of the principles selected (KC 523, 569).7

In this section, my purpose is to highlight the structural assumptions on which the plausibility and persuasiveness of Rawlsian constructivist reasoning depend in the domestic context. This will necessarily entail some review of familiar aspects of Rawls's Theory, but I shall concentrate on explaining how four key structural assumptions provide constructivist reasoning with the rigor and explanatory power that make its conclusions plausible and persuasive. These assumptions are crucial because, for Rawls, the benefit of constructivist reasoning is that it can transform the selection of principles of justice into a problem of deductive reasoning; the original position admits of a unique solution, ‘a definite answer’ (TJ 16). The ICS and the method of selection within it must, therefore, be carefully specified, because

a problem of rational decision has a definite answer only if we know the beliefs and interests of the parties, their relations with respect to one another, the alternatives between which they are to choose, the procedures whereby they make up their minds, and so on. (TJ 16)

Thus if constructivist reasoning is to explain the choice of principles of justice or provide a justification for their selection, the ICS and procedures must be tightly and carefully defined.

The first key assumption underlying constructivist reasoning is that [1] principles are chosen for a closed system in isolation (TJ 7). Rawls argues that the procedure must treat the membership of society as given (PL 276–277)—or what amounts to the same thing—that individuals must regard their membership in society as fixed. This assumption ensures that parties in the initial situation choose principles for a society in which they have a permanent and abiding interest. The members of this society ‘view their common polity as extending backward and forward in time over generations, and they strive to reproduce themselves, and their cultural and social life in perpetuity’ (KC 536). In addition, there are no significant relations with other societies, and no one enters from without (KC 536).

The ‘closed system’ assumption facilitates a determinate and stable choice by the parties. If they do not regard the polity for which they are choosing as fixed and stable (as a polity extending backward and forward through time) or if they do not regard it as reflecting their (primary) identities, they might choose differently. From the beginning it must be clear that the parties agree both about the society for which they are choosing principles and that they see themselves as having a meaningful and fateful stake in it. This assumption should not be confused with Rawls's now (in)famous interpretation of it: that theories of justice must be worked out for national societies or what Rawls later began calling peoples—a point to which I return below.

The second assumption underpinning constructivist reasoning is that there is wide consensus on [2a] what restrictions it is reasonable to place on the choice of principles of justice and on [2b] the morally irrelevant contingencies that should be excluded from consideration in deciding them. With respect to [2a], there must be “a broad measure of agreement that principles of justice should be chosen under certain conditions. To justify a particular description of the initial situation one shows that it incorporates these commonly shared presumptions. One argues from widely accepted but weak premises to more specific conclusions.” “The idea,” Rawls explains, “is simply to make vivid to ourselves the restrictions that it seems reasonable to impose on arguments for principles of justice and therefore on these principles themselves” (TJ 16). “The crucial thing,” Rawls writes, “is not to use principles that are contested” (TJ 512). Limiting the structuring assumptions to those that model “conditions that are widely thought reasonable to impose on the choice of principles yet which, at the same time, lead to a conception that characterizes our considered judgments” assures that “the question of justification is settled” (TJ 105). Morally irrelevant contingencies [2b] are addressed structurally through the veil of ignorance, ensuring that no one is advantaged or disadvantaged in the choice of principles by factors widely agreed to have no legitimate bearing on that choice. Again, this requirement should be kept separate from Rawls's substantive, Kantian interpretation of it, which models the original position for a society of free and equal democratic citizens (KC 519ff.).

The third assumption concerns the parties themselves, who are modeled as [3a] rational, [3b] motivated, and [3c] mutually disinterested.8 Rationality is simply the assumption that people can choose instrumentally (they can reason through hypothetical imperatives). Motivation enables the parties to choose among the various options available to them by providing a metric for weighing alternatives and reaching rational preferences about them; Rawls introduced primary goods to provide motivation for the parties (JF 87–88; TJ 54–55, 78–81, 123ff.). The point is that there must be some plausible assumption about what motivates the parties, why they prefer some principles to others. Mutual disinterest means that the parties do not take an interest in one another's interests (TJ 12) as well as that ‘they are not willing to have their interests sacrificed to the others’ (TJ 112).9 Mutual disinterestedness assures that persons engaged in conflicts over moral and religious issues see themselves and others as defending their basic rights and liberties, which secure their legitimate and fundamental interests (JF 85); it facilitates consensus when individuals subscribe to conflicting comprehensive doctrines. In short, it models liberal toleration.10

A fourth assumption informing the selection of principles in the original position is the assumption of strict compliance. As Rawls describes it, ‘one of the stipulations of the original position … [is that] the principles of justice are chosen on the supposition that they will be generally complied with’ (TJ 215). Strict compliance identifies constructivism as part of ideal theory, that type of moral theory that seeks to describe a perfectly just society (TJ 8). It also helps to make the choice of principles determinate: if the parties cannot assume strict compliance, their choice would vary with the nature and extent of non-compliance and their knowledge of it. A corollary of this assumption [4*] is that ‘reasonably favorable conditions’ obtain in society (TJ 216; JF 13). Reasonably favorable conditions refer to the social circumstances that permit the effective establishment and the exercise of basic political liberties such as ‘a society's culture, its traditions and acquired skills in running institutions, and its level of economic advance (which need not be especially high)’ (PL 297; cf. LP 106). Below this threshold, reasoning about justice changes significantly due to the extreme burdens facing society.

My point in identifying these assumptions is to clarify their role in making constructivist conclusions plausible and persuasive. Rawls is quite clear that ‘we shall want to say that certain principles of justice are justified because they would be agreed to’ in a purely hypothetical ICS (TJ 19, emphasis added). We can do with confidence only if ‘the conditions embodied in the description of this situation are ones that we do in fact accept’ (TJ 19, 514).11 Thus the plausibility and persuasiveness of the principles chosen depends crucially on the correct modeling of the ICS and on strict adherence to the assumptions that narrow the choice problem faced there to one of deductive rigor. For example, if there is no consensus on what is placed ‘behind the veil’, the principles chosen will not be acceptable—or even recognizable—to many members of society. Table 1 summarizes the role of the structural assumptions in constructivist reasoning.

Table 1. Assumptions underlying constructivist arguments

Assumption	Structural role
[1] Closed system	Determinacy/stability/identity
[2] Wide consensus on:
[2a] Reasonable restrictions	Models consensus
[2b] Morally irrelevant contingencies	Models consensus
[3] Parties modeled as:
[3a] Rational	Determinacy/rationality
[3b] Motivated	Determinacy/rationality
[3c] Mutually disinterested	Models toleration/pluralism
[4] Strict compliance	Determinacy
[4*] Reasonably favorable conditions	Circumstances of justice

‘RAWLSIAN’ GLOBAL CONSTRUCTIVISM: GUILT BY ASSOCIATION?

Rawls's early cosmopolitan critics challenged his substantive interpretation of the closed system assumption—a closed society in isolation—on normative and empirical grounds. Many saw themselves as more ‘Rawlsian’ than Rawls in utilizing a single, global ICS. Few, however, questioned whether Rawlsian constructivist reasoning could be usefully applied to problems of global justice.12 This omission creates a potential problem for my argument: faulting cosmopolitan constructivists for violating methodological assumptions that few of them explicitly adopt might seem like a charge of guilt by association.

Cosmopolitan constructivists uniformly take Rawls's argument as their point of departure. Amdur asserted that once we relax or reject [1], both Rawls's difference principle and the application of non-ideal theory apply globally. Beitz found no reason to think that principles other than those Rawls articulated for the domestic context would be chosen globally. He later stated that in fully open and interdependent societies Rawls's two principles apply ‘a fortieri’. Scanlon similarly saw no reason why the difference principle would not hold across the global economic system. Pogge proposed a ‘single, global original position’ with nationality behind the veil that ‘leaves intact Rawls's whole argument for the two principles’. In response to Rawls's first attempt to elaborate his views on international justice, Pogge did state that ‘methodologically … it is too early to tell how his idea of the original position … should best be adapted to the complexities of our interdependent world’. By this he meant whether a cosmopolitan or a social liberal approach was warranted; there was no question whether contractualism could aid our reasoning about global justice. Similarly, Moellendorf, in rejecting Rawls's two-stage procedure, states that this rejection ‘does not impugn the political constructivist project either in general or with regard to this particular subject matter. An alternative construction procedure is viable’—namely, the one-stage global original position of persons that Rawls rejects. It seems clear enough, then, that cosmopolitan constructivists—including more recent critics like Kuper, Tan, and Brock—widely and uncritically incorporate Rawls's methodological apparatus in applying initial agreement arguments globally. They criticize Rawls's substantive arguments but never question whether constructivist reasoning is useful and appropriate globally.13

A critic might still object that failure to explicitly reject these assumptions is not equivalent with endorsing them or that the term ‘uncritical’ is unfair. As I have shown, the plausibility and persuasiveness of conclusions yielded by constructivist arguments depend on the soundness of the construction. Given that these cosmopolitan constructivists take Rawls's procedure as their starting point and adopt his method of reasoning, it seems reasonable to infer that they endorse it. Given also that none offers an alternative specification of the procedure or even mentions issues of method and epistemology, the charge of ‘uncritical’ adoption also seems warranted: whether it applies in the sense of ‘without criticism’ or ‘without reflection’ makes no difference for my argument.

COSMOPOLITAN CONSTRUCTIVISM

The question before us is whether the structural assumptions underlying the plausibility and persuasiveness of contractual reasoning are valid globally. In this section I consider this matter in connection with cosmopolitan constructivist arguments, taking up each of the assumptions serially.14 To preview my findings: the assumptions do not hold.

Consider first the requirement that a theory of justice refer only to a closed system in isolation [1]. Rawls's cosmopolitan critics unanimously reject his substantive views on this point. Barry immediately attacked them, calling Rawls's position ‘arbitrary’ and declaring the two-stage contractual process for reasoning about international justice indefensible—judgments in which Amdur concurred.15 Richards found that ‘factual relations among nation-states in the contemporary world’ justified a global application of the original position, a view endorsed by Beitz, Pogge, and others.16 There is thus a consensus among cosmopolitans that Rawls's interpretation of [1] is both morally and factually unsustainable; his recent critics regard this conclusion as almost self-evident.17

One way to resolve this problem would be to treat the entire planet as constituting a single closed system in isolation (no relations with extra-terrestrials, etc.).18 The obvious problem with this approach is that it hardly corresponds to any natural or widely shared conception of identity or belonging. The world's people do not by and large understand themselves as part of a single community (except perhaps in a thin rhetorical sense). There is certainly no global community possessing a common social and cultural life that extends through time and across generations. It is not clear, in other words, that a global society is one in which people can have the kind of meaningful stake that Rawls thinks is necessary to inform the choice of principles of justice—that people regard or could regard the planet as the kind of community for which justice is appropriate.19 That is not to ignore the impact of the global order on people's lives, but rather to note that they do not necessarily see that impact, or their own involvement in the global order, as deep enough to warrant imposing a single conception of justice on the planet.

Consider next how the indeterminacy introduced in a global ICS undercuts constructivist reasoning by violating the closed system assumption. In a global ICS the parties remain aware that the world comprises numerous and quite varied national societies, each of which constitutes an ongoing scheme of social cooperation with relatively stable and exclusive membership and a conception of justice of its own. The parties are citizens of the world, choosing global principles of justice, and simultaneously citizens of domestic societies (although they don't know which ones), a duality indicated by the ‘veiling’ of knowledge about their particular nationalities.20 This predicament violates the closed system in isolation assumption, the point of which is to simplify the choice problem facing the parties in the ICS. Without this assumption, the parties have to balance or trade off principles at multiple levels. Surely they would realize that global and domestic principles will condition and potentially conflict with one another, but they cannot know which to prioritize or how either choice would constrain or enable justice at the other level. With multiple levels or layers of justice in play, there is no practical or rational way for the parties to determine which set of principles should take priority, what weight to give to the rival schemes of justice in the global initial situation.21

There is no procedural solution to this indeterminacy. Withholding knowledge of domestic social arrangements or their importance from the parties would distort their understanding of the basic facts about the international system and their choices regarding principles. Providing more knowledge about the constituent societies in the global order or the parties’ membership in them would allow partiality to taint the choice of principles, blocking any agreement. So the parties make essentially a blind choice, one whose implications they cannot fully understand or anticipate. The point is not about which choice is correct; the point is rather that it is impossible to model the ICS in a way that is sufficiently determinate but that does not also foreclose the choice or beg the key questions.22

One example of this problem of indeterminacy is the curious result that most cosmopolitan constructivist accounts yield global principles of distributive justice and national or domestic principles of political justice.23 It is hard to imagine that the parties could rationally settle on global redistributive principles and also decide that no global political structures—even to regulate the economy or enforce distributive rules of justice—are required. Deference to existing national political arrangements seems to govern the choice of political principles but not economic ones. How could this be explained? More importantly, how could it be defended against an interpretation that took the opposite view—that democracy and human rights are important enough to warrant global political arrangements, but that economic matters should be regulated domestically? Again, the issue is not about which decision is correct; it is that indeterminacy undermines the plausibility and the persuasiveness of whatever conclusion is reached. My point is not the skeptical one that we cannot work out moral principles for complex systems. It is rather that constructivist reasoning cannot help us to do so because the ambiguities inherent in such a system preclude the determinacy that constructivist reasoning requires.

Similar difficulties arise in connection with the second set of structural assumptions involving the wide consensus on reasonable restrictions on the choice of principles [2a] and on what morally irrelevant contingencies should be covered by the veil [2b]. Take the second assumption first. Many cosmopolitan constructivists argue that nationality should be excluded as morally irrelevant. Bracket for now the indeterminacy objections just discussed; the problem remains that there is nothing remotely approaching consensus on the moral irrelevance of nationality to a conception of global justice—or, to take another salient example, on the irrelevance of substantive religious doctrines. Even in the rarified world of English-speaking political philosophy, one certainly more cosmopolitan in outlook than the world generally, there is deep and vigorous disagreement on this question.24

The intuition of the cosmopolitan theorists is certainly correct: whether or not we veil nationality significantly affects the choice of principles of justice and shapes the impact that the birth lottery will have on people's life chances. The problem is that, procedurally, placing contingencies behind the veil clarifies our moral reasoning when there is wide agreement on their irrelevance; it vivifies for us the implications of our views. This requirement ensures the plausibility of the conception of justice selected in the initial situation: ‘the conditions embedded in the description of this situation are ones that we do in fact accept’ (TJ 19). The conclusions reached through constructivist reasoning are only persuasive if they are based on assumptions that are widely shared. Veiling controversial features of our moral world, like nationality or religion, achieves hypothetical agreement at the cost of plausibility and persuasiveness. This is not a claim about whether or to what extent nationality or religion should figure appropriately into conceptions of justice; it is a claim about how constructivists are supposed to reason about such questions.

Reasonable restrictions on the choice of principles must also reflect a wide underlying consensus [2a]. One such restriction imposed by cosmopolitan constructivists is that principles of justice should be chosen for free and equal persons. This restriction is modeled through the symmetrical positioning of the parties: all must agree on the principles selected (none can dominate the others). This view—that principles of justice establish terms of cooperation for free and equal persons—is not widely held globally, at least not if those principles are given sufficient content to do real work in the ICS. Numerous patriarchal social and religious systems do not regard women as the moral equals of men, or else they hold that freedom and equality mean different things for women and men (which from a cosmopolitan perspective amount to the same thing). But if freedom and equality are left vague enough to command acceptance across such diverse societies, they would lack sufficient content to meaningfully inform the selection of principles of justice. Crucially, not modeling freedom and equality in the global ICS would also undermine the argument's persuasiveness because any alternative design of the initial situation would also be contentious. Liberals and democrats could hardly accept principles of justice that did not include robust notions of freedom and equality or that were chosen in a way that allowed some agents to dominate others—witness the cosmopolitan outcry against Rawls's limited conception of human rights.

I should emphasize that specific agreement on principles of justice is not required; rather, the method assumes general, unarticulated ‘background’ consensus that the model helps to clarify. Critics might argue that lack of consensus notwithstanding, reason or morality dictates what fair restrictions and moral contingencies ought to be. If one begins from an appropriate moral conception of the person, the ‘constructivist objection’ about the lack of consensus collapses.25 This move appeals to something like Rawls's Kantian interpretation. Such a move is problematic for constructivists, however, because it is not clear that the bare notion of autonomy or equal respect suffices for developing a full account of justice. Even if it did, however, the constructivist reasoning would be superfluous to it: all of the work would be done by the Kantian argument.26

Critics might next object that the point of constructivist reasoning is to resolve deep disagreements such as these: Carens argues that the veil would be useless if it only applied where values are already shared.27 This objection misses a crucial point about what the veil and the ICS do. An initial agreement is not a way to reach moral conclusions when our moral premises are uncertain; it is a way of precisely formulating a conception of justice that reflects and clarifies widely shared but often inchoate ideas about justice prevalent in a society (see PL 102–104). Modeling controversial premises into an ICS does not so much resolve disagreement as banish it by fiat.

The determined critic might make a last stand, insisting that there is no more disagreement globally than domestically about the moral ideas that model the ICS.28 After all, niches of racism and sexism exist even in liberal societies. This point is revealing in two ways. First, it mistakenly suggests that constructivism requires actual consensus or agreement of the kind that niches of racism and sexism would disprove. In fact, what constructivism builds on is, again, widely shared but often inchoate and unarticulated ideas about justice present in a society. In the domestic context this is a vague liberal conception of justice. Second, this objection illustrates precisely my point: that in the global context even such a broad, vague consensus on inchoate or unarticulated intuitions about justice is lacking. It won't do to respond that cosmopolitan constructivism is intended only for liberals, to clarify what liberals should think about the global order. Pogge might be correct in arguing that though our conception of justice is parochial we cannot avoid taking positions on questions concerning the global basic structure or global justice more generally.29 But even accepting this ‘inevitability of global justice’ argument, it does not follow that constructivist reasoning, which presumes agreement, however vague, on basic matters of justice offers a reasonable way to address those questions. Besides, if there actually is no more agreement on such a vague conception of justice domestically than globally, this should count against domestic applications of constructivist reasoning, not in favor of global ones.

Next we must consider the assumptions about the parties to the initial agreement. To facilitate a determinate choice of principles, the parties must be conceived as rational [3a], motivated [3b], and mutually disinterested [3c]. The rationality assumption by itself seems unproblematic if we take it as Rawls does to indicate the ability to formulate and pursue a plan of life in a coherent and effective way. A difficulty does arise in connection with the motivation of the parties, however. Rawls supposed that primary goods constitute a thin conception of the good sufficient to motivate choice in the ICS. His critics charged that they favor individualistic plans of life over more communally oriented ones.30

The thrust of this objection applies globally as well. Rawls's particular primary goods appear more relevant for developed societies than poor ones: wide choice among professions, the powers of offices, and even the social bases of self-respect as Rawls understands them presume a level of social development far beyond what many people enjoy. For much of the world's population, public health, food security, rule of law, and responsive government are lacking (a point to which I return below). It seems doubtful that primary goods would be an appropriate motivation for this population in the ICS. The broader problem, however, is not that any specific list of primary goods is problematic but rather that, given our planet's cultural pluralism and the vast gaps in people's living conditions and experiences, it is difficult to imagine a single motivational account that could adequately and plausibly reduce the choice among principles of justice to a determinate, rational one for all parties in a global ICS.

Mutual-disinterestedness poses still a greater problem. Many religious fundamentalists of various stripes do take an increasingly keen interest in the moral lives of others as the rise of Christian, Jewish, Muslim, and Hindu fundamentalisms attests. Recall that mutual disinterestedness makes a solution to the problem of rational choice posed in the initial situation possible by allowing adherents of differing comprehensive doctrines to agree on a conception of justice that is, in Rawls's famous phrase, ‘political not metaphysical’. It models liberal toleration in the initial situation, but, again, as the rise of fundamentalist political ideologies makes strikingly clear, liberal toleration is hardly a global norm. Critics might again point out that such problems equally affect domestic societies. Again, I agree; in fact, the problem seems to have worsened considerably since Rawls first introduced his theory, thanks largely to globalization and attendant declines in economic security, surges in migration, and so on. Rawls's move in PL to address deep pluralism through the idea of overlapping consensus acknowledges these changes, but it is not entirely persuasive—mainly because he simply rules out the hard cases with his ‘reasonableness’ requirement. Since mutual-disinterestedness reflects a liberal norm of reasonableness, it is tempting to think that it is not strictly necessary for constructivist reasoning. The problem is that without it, overlapping consensus among rival comprehensive doctrines—which is essential to the stability of a conception of justice as well as to its plausibility and persuasiveness—is unobtainable. Insofar as mutual disinterestedness poses a problem for constructivism domestically as well as globally, it is a plague on both their houses.

Finally, let us consider the related assumptions of strict compliance [4] and reasonably favorable conditions [4*]. The strict compliance assumption lends determinacy to constructivist reasoning by assuring the parties that whatever principles they select will, on the whole, be respected within society. This assumption is clearly an idealization, as Rawls acknowledges. Yet given the degree of chaos caused by hunger, poverty, and militarized conflicts over territory, religion, and resources (among others), it seems reasonable to question whether this idealization is so extreme as to render the procedure itself problematic. Among the relevant social facts about our world today is that across large parts of it compliance with principles of any kind is virtually non-existent. Many persons routinely suffer injustices; for these people it is not reasonable to expect strict compliance, yet without that assumption a rational and determinate choice of principles of justice becomes impossible. In such conditions an assumption of strict compliance would have a distorting effect on the choice of principles of justice. Ideal theory treats non-compliance as a problem of deviation from principles of justice. When deviation is the norm, the parties cannot reckon the likely effects of various conceptions of justice and thus cannot make a rational choice among them.

The related assumption that reasonably favorable conditions obtain in society [4*] seems equally dubious, as noted above. Recall that reasonably favorable conditions include all those cultural, traditional, and economic factors that bear on the effective establishment and exercise of basic liberties. Such conditions obtain very unevenly across the planet, so that while certain principles of justice might be effective in some contexts, they might prove ineffective or even pernicious in others.31 Still, the problem is not simply that different principles would be appropriate in different conditions; it is that massive poverty, endemic violence, stinging hunger, entrenched patriarchy, religious zealotry, and rampant disease might well be the kinds of circumstance that make justice unobtainable. One might object that the relevant question is whether, on balance, conditions on the planet are favorable enough that justice could obtain. The difficulty with this position, however, is that it treats national or domestic basic structures and social schemes as relatively unimportant factors in determining people's life chances.32 Basing a choice of principles of justice on such a claim seems irrational on its face.33

In sum, there is a sense in which cosmopolitan constructivist arguments are inevitably question-begging. In the absence of wide agreement about where to begin, constructivist reasoning relies on controversial assumptions where by design it requires unobjectionable ones. The very structure of the initial situation becomes contentious, instead of merely being ‘laid out’. The principles of justice yielded by constructivist arguments ‘are not in fact adequately defended by appeal to hypothetical agreement: the correct characterization of the initial choice situation presupposes a substantive view about justice’.34 But in presupposing a view of justice, cosmopolitans eliminate the need for constructivist procedures; they become superfluous.

Readers might doubt whether some of the foregoing criticisms persuade. Recall, however, that all of the structural assumptions contribute to the plausibility and persuasiveness of constructivist reasoning about justice. Thus if any of them does not hold, the constructivist approach is in trouble and the more of them that do not hold, the deeper the trouble gets. Once again, my point is not the skeptical one that effective reasoning about global justice is an impossible or quixotic pursuit, nor is it that cosmopolitan accounts are substantively wrong. It is rather that the constructivist method does not and cannot deliver the sort of plausible and persuasive conclusions about justice that it claims to provide.

RAWLS'S SOCIAL LIBERAL (TWO-STAGE) APPROACH

In light of these conclusions, Rawls's much-maligned two-stage approach to reasoning about global justice might appear more attractive than his critics have allowed. While several theorists have defended versions of a Rawlsian two-stage procedure, it makes most sense to consider Rawls's own social liberal account as an alternative to cosmopolitan models. Rawls was in fact conscious of several important differences between domestic and global initial choice situations and his model appears to be more modest epistemologically than its cosmopolitan rivals, being limited to liberal (and later also ‘decent’) peoples and to matters of foreign policy (rather than questions of justice more broadly; LP 9–10).35 Moreover, contrary to what many of his critics have intimated, Rawls has not abandoned his concern with individuals as the fundamental units of moral concern; rather, like Kant, he has concluded that the international order contributes most to ensuring justice for individuals when it secures the conditions for domestic justice (LP 10).36

Rawls is explicit in envisaging the second original position in relation to the first in his two-stage model. This second ‘running’ of the original position models what citizens of liberal democratic societies would regard as fair conditions under which to specify a Law of Peoples (LP 32). Like the first original position, the second constructs the representatives of peoples as free and equal and as reasonably and fairly situated; as rational; and, as deliberating about the correct subject—here, the content of a Law of Peoples governing the ‘basic structure of the relations among peoples’. Rawls thus assumes equality among peoples (not individuals), an assumption he thinks is well-established in international practice. Parties deliberate using the right reasons as modeled by an appropriate veil of ignorance, hiding from them the size of their territory, its population and relative strength, the extent of its natural resources, level of economic development, and so forth. Parties do know that reasonably favorable conditions permit constitutional democracy at home (LP 32).

Rawls highlights three key disanalogies between the first and second choice situations: in the second, the representatives of peoples have no comprehensive doctrines, since liberal democratic peoples, qua peoples, have none. There is no need for primary goods because the fundamental interests of the parties are given by their domestic conceptions of justice. Finally, in the second running the parties choose among different formulations or interpretations of eight principles of the Law of Peoples (LP 40). One important similarity between the first and second original positions, stressed by Rawls and his defenders, is that Rawls draws only on the shared global political culture in structuring his arguments.37 This restriction dictates Rawls's turn away from cosmopolitanism because ‘there is simply no global public political culture emphasizing that citizens of different countries ought to relate fairly to one another as free and equal in a single scheme of social cooperation—much less that the distribution of global resources and wealth should not be based on factors “arbitrary from a moral point of view”’.38 Wenar maintains that any shared global political culture remains limited, precluding the more ambitious and more cosmopolitan Law of Peoples that many critics would like to have seen. But, he insists, this epistemological modesty allows Rawls to construct a more legitimate global original position.39

Despite these apparent advantages, I shall argue that Rawls's two-stage global application of the initial agreement method nonetheless violates many of that method's structural assumptions. That the pioneer of this method runs into such difficulties is a damning indictment that considerably strengthens the methodological case against global constructivism.

Like his cosmopolitan critics, Rawls relaxes [1] to get leverage on questions of international justice. It appears that representatives of liberal peoples form the fixed membership comprised by his second original position, but this is deceiving. Rawls actually has a three-stage procedure; he runs the second original position twice, first with representatives of liberal democratic peoples only, then a second time (third overall) with representatives of decent peoples included. This last running, Rawls asserts, merely affirms the same principles endorsed in the second. This final iteration, however, is not as benign as Rawls suggests. To see this consider the question of human rights: representatives of liberal peoples in the first global choice situation select principles of fundamental human rights. Surprisingly, they do not opt for ‘liberal’ rights—such as equality, democratic rights, and others (LP 65, 79). This, Rawls says, is appropriate as only ‘urgent rights’ would be chosen, since their violation would be condemned by liberal and decent peoples alike (LP 79).

Freeman defends Rawls's consistency on this point, arguing that liberal people have no interest in enforcing human rights around the world; they are concerned only with securing the conditions of justice for themselves. As Freeman admits, however, this assumption is arbitrary, a function of the way Rawls has defined the second original position.40 Rawls (and Freeman) argue that liberal toleration and notions of equality among peoples require that liberal peoples not inflict their notions about human rights onto others.

The arbitrariness of the decision salvages Rawls's consistency, but in doing so it damages the method. First, it demonstrates that the most controversial questions are ones to which constructivism must assume arbitrary answers. Second, it highlights the problem of indeterminacy in connection with [1]. The first running of the global original position includes only liberal democratic peoples; there is no reason that anything other than a robust conception of liberal-democratic human rights should be chosen in this setting. Indeed, liberal people would want to create supranational institutions to ‘lock-in’ democracy and human rights, as insurance against political regression at home.41 The only explanation for the choice of the much more circumscribed set of human rights is that the parties are conscious of the requirement that decent peoples be able to accept those same principles in the upcoming iteration of the ICS. Without ‘looking ahead’ to the third running of the original position, the parties would have no conceivable motivation for choosing the truncated notion of rights Rawls prefers. Indeed, Rawls's view that decent societies should be regarded as members in good standing is doubly arbitrary: once he has decided to treat them as members in good standing, why are they barred from participation in the second original position?42

Regarding wide consensus [2], both Wenar and Freeman praise Rawls for limiting his assumptions to those widely shared in the global political culture. Still, at least two difficulties stand out. First, the fundamental concept driving Rawls's account—peoples—is neither familiar nor accepted in global political culture or, for that matter, in academe. Many states and many individuals would have strong reasons to object to this notion. The point is not merely terminological; as Buchanan persuasively argues, the notion of ‘peoples’ actually distorts our understanding of politics, rendering problems like secession or intra-state conflict invisible from the perspective of justice.43 Rawls is certainly correct in asserting that the equality of states is well-established in international law—though in practice this equality is not limited to well-ordered states. But his assertion that ‘in the absence of a world-state, there must be boundaries of some kind’ (LP 39) is not sufficient to justify deference to something like our present Westphalian order, since it does not follow that those boundaries must conform to ‘peoples’, to existing states, or even to a primarily territorial system of government (a point that applies to cosmopolitan accounts as well). Indeed, people might well opt for a looser system of functional governance that cuts across existing national borders.44

Rawls also goes far beyond existing consensus in assuming that the only appropriate subject of justice among such peoples is the content of their foreign policy. Indeed, there is a roiling debate about how, and how extensively and inevitably, the global basic structure affects the life chances of most of the world's people. It might well be that liberal peoples would prefer to limit the discussion in this way, but the salient question is precisely whether this narrow perspective is morally defensible. Many poor countries would prefer to see much greater attention to distributive justice globally. Freeman's defense of Rawls's account once again highlights its question-begging nature: Rawls precludes the parties in the global original positions(s) from deliberating about redistribution because his constructivist method assumes that distributive questions have been resolved domestically.45 This is certainly a consistent position, but it can hardly dispose the question whether there should be redistribution linked with growing global interdependence.46 Rawls's unwillingness to consider the question is strange, given that in a discussion of free trade he presumes fair background conditions whose role is ‘analogous to that of the basic structure in domestic society’ (LP 42 n.52). He also acknowledges the need to rectify the ‘unjustified distributive effects’ flowing from social cooperation in the global economy (LP 43). In light of these contradictory viewpoints, it is hard to know what might count as ‘unjustified’ on Rawls's view, which provides no criteria for making such assessments or for determining in what those fair background conditions might consist.

Rawls appears to be solid on [3]: the parties are rational, they are motivated by their own domestic conceptions of justice (no primary goods are required), and they are mutually disinterested (sovereignty, non-interference). However, as the foregoing discussion illustrates, they are so selfishly motivated and so mutually disinterested as to be unconcerned with whether liberal and democratic rights are realized globally and with the distributive implications of the global order. Again, my criticism here is not that Rawls has failed to reach cosmopolitan conclusions: as Freeman argues, Rawls's rejection of cosmopolitanism reflects his commitment to (his version of) the constructivist methodology, which assumes that justice is primarily a domestic matter. My complaint is that in modeling the parties consistently with this assumption Rawls begs the crucial questions at stake in this debate. Global justice is a new and controversial subject; Rawls, in affirming our rather conservative views on it, risks elevating the rather self-serving anti-distributive prejudices of rich liberal-democratic states to the status of moral principles.

With regard to [4], Rawls deserves credit for recognizing that reasonably favorable conditions do not obtain globally; he appropriately (in a methodological sense) addresses the problems of outlaw states and burdened societies as part of non-ideal theory (LP 89–120). Yet even among liberal and decent societies, it seems dubious to assume compliance when international politics has long been characterized by what Krasner neatly calls ‘organized hypocrisy’.47 The norms of global political culture on which Rawls draws have been routinely flaunted for centuries. In light of this, strict compliance can hardly be a reasonable assumption for the parties in the global original position. This is not merely a technical point. Consider again the question of human rights: if the parties have knowledge of the basic facts of history, they must know that international law and norms have historically been impotent. They should, at the very least, create enforcement mechanisms to ‘lock-in’ these essential protections and guarantee international security.48 Rawls might reply that since political matters are settled domestically, global political arrangements of this kind are unnecessary and in fact precluded. But parties with even a rudimentary grasp of history and politics will be aware of facts such as the shameful and systematic record of disregard for international law; of the history of colonialism and its social, economic, and political legacies; and of rich countries’ failure to meet their pitiful self-imposed target of .7% of gross domestic product (GDP) devoted to foreign aid. In this context, Rawls's proposals to address relations among peoples through international law, a truncated set of human rights, and a vague and sharply limited duty of assistance—‘regulated’ without institutions of any kind—would surely appear risible to the parties, as they have to many of his critics.

In sum, Rawls's two-stage (social liberal) global constructivism encounters many of the same problems as does its cosmopolitan cousin. Beginning with peoples and restricting the procedure to the terms of just foreign policy among liberal peoples is no less presumptuous than beginning with a global contract among individuals.

THE LIMITS OF RAWLSIAN GLOBAL CONSTRUCTIVISM

The analysis of Rawlsian global constructivism in its one-stage (cosmopolitan) and two-stage (social liberal) variants has shown that the method appears ill-equipped to grapple with the complexities of interdependence—with the overlapping multiplicity of institutions, boundaries, and identities that characterize our world. The various arguments advanced by cosmopolitans and by social liberals alike do not meet the high standards of rigor on which constructivist conclusions depend. The key point is that in the absence of background consensus, constructivism necessarily begs key questions: the initial agreement method only gives us persuasive answers when, as shown above, begging them seems ‘natural and plausible’ in light of existing and widespread agreement. Our sense of what global justice might look like is too embryonic for constructivist reasoning to be properly applied in clarifying principles of global justice. (Whether it is natural and plausible to beg these questions domestically is a question about which I've raised serious doubts and that requires further inquiry.)

I suggested at the outset that constructivist reasoning tends to obscure the central points at issue in the debate on global justice. We can now see that it does so by transforming these contentious issues into modeling choices. Constructivism requires that the representatives in the ICS choose the principles of justice; but in practice, the theorist's preferences determine who the representatives are and what choices they will make. Whether the theorist makes the parties represent peoples or individuals determines the nature and scope of the principles of justice they will choose. Representatives of states or peoples cannot choose cosmopolitan principles and individuals cannot choose social liberal principles, because the theorist's choice about how to model the ICS already determines which principles will be chosen.

Many constructivists seem to join Brock in thinking that ‘the thought experiment identifies certain situations as morally desirable because it can be expected that delegates would choose them, when suitably situated’.49 But the parties can only choose what the theorist permits them to choose; their hypothetical ‘choice’ cannot convey legitimacy or even confidence in their choices unless we are certain that the choices have been modeled in an appropriate and uncontroversial way. The problem is, in the absence of consensus, a choice that is supposed to clarify and vivify broadly shared but inchoate beliefs instead simply reflects the predilections of the modeler. The primary objection to Rawlsian global constructivism, to reiterate, is that the method hides controversial determinations on the key substantive points at issue by presenting them as if they were uncontroversial modeling choices. Besides, it is surely a mistake to think that principles of justice have normative force because they are chosen in an ICS.50 Their selection helps to explain and justify them to others who recognize the ICS as an acceptable characterization of their conception of justice, but proves nothing about their truth and validity—as Rawls always understood.

RAWLSIAN GLOBAL CONSTRUCTIVISM AND IDEAL THEORY

Rawlsian global constructivism reflects an ideal-theoretical impulse that is pervasive in theorizing about justice, both domestically and globally. This is most evident in the strict compliance assumption, which ideal theorists see as ‘necessary to ensure that we know what a commitment to a given set of values entails’.51 Indeed, ideal theorists agree with Rawls in thinking that an ideal conception of justice is necessary because ‘a deeper understanding [of the pressing and urgent matters we confront in everyday life] can be gained in no other way’ (TJ 8, my emphasis). Swift argues that

we need fundamental, context-independent, normative philosophical claims to guide political action; only by reference to philosophy—abstract, pure, context-free philosophy—can we have an adequate basis for thinking about how to promote justice in our current, radically nonideal [sic] circumstances.52

Even critical commentators like Robeyns profess that ideal theorizing is ‘an essential piece of the work to be done’ in thinking about justice because it is a necessary ‘precursor’ of non-ideal theory.53

This claim about the priority or necessity of ideal theory has recently been subjected to powerful criticisms. As Sen writes,

The question ‘What is a just society?’ is … not a good starting point for a useful theory of justice. To that has to be added the further conclusion that it may not be a plausible end point either. A systematic theory of comparative justice does not need, nor does it necessarily yield, an answer to the question ‘What is a just society?’54

To advance justice or reduce injustice does not require, logically or epistemologically, an ideal conception of justice:

Investigation of different ways of advancing justice in a society (or in the world), or of reducing manifest injustices that may exist, demands comparative judgments about justice, for which the identification of fully just social arrangements is neither necessary nor sufficient.55

In this conclusion I shall focus on one particularly worrisome side-effect of ideal theorizing: its tendency to distort our thinking about problems of justice and injustice. Rawlsian global constructivist reasoning, with its emphasis on modeling an ICS, is particularly prone to such distortions. My point in emphasizing this problem is twofold: first, it deepens and extends my general critique of Rawlsian global constructivism, raising additional epistemological and normative concerns parallel and related to those already presented. Second, this discussion links the specific methodological, epistemological, and normative issues raised here to an emerging debate on ideal theory. I hope by concretizing the critique of ideal theory through a discussion of these issues to demonstrate the stakes and the urgency of developing alternative approaches.

What follows is not a rejection of idealization or abstraction. Theorization relies upon abstraction, the simplification of empirical truths for the purposes of clarification and generalization; normative theory, insofar as it is prescriptive, also entails a kind of idealization.56 The problem is distortion—the conscious or unconscious representation of the world in ways that exclude or marginalize some of its uncomfortable or inconvenient features. Some idealizations and some abstractions distort and it is not always clear which idealizations are distortions.57 Consider Rawls's exclusion of race and gender from consideration in the original position. He argues that ideal theory is concerned with identifying those contingencies that lead to troubling inequalities even in a well-ordered society. In such a society, he maintains, neither race nor gender would lead to inequalities because advantages or disadvantages based on them would be ruled out from the point of view of the representative equal citizen; that is, free and equal people would not endorse racist or patriarchal principles (JF 65–66). Any inequalities or injustices linked to race or gender are addressed in non-ideal theory, where we confront deviations from or exceptions to the ideal.

The difficulty with this line of thought is not substantive; Rawls is surely correct that a liberal conception of justice cannot tolerate racist or patriarchal principles. The problem is rather with how constructivism conceives of and responds to such problems. In excluding explicit consideration of race and gender, ideal theory in effect prescribes that they should not matter in a well-ordered society. Yet McCarthy argues that if a culture is structured by racial and gender inequalities, these will in turn shape public thinking (including in an ICS). Ideal theory will prove unable to access these inequalities critically, however, because the ideals themselves will reflect the prejudices and power relationships of society.58 Oppression can be so embedded in societal norms and understandings that abstraction or idealization simply takes it for granted and renders it invisible.59 This is, I think, what McCarthy has in mind when he writes that ‘ideal theory provides no theoretical mediation between the ideal and the real—or rather, what mediation it does provide is usually only tacit and always drastically restricted’.60 The ideal of a race- and gender-blind society is inadequate for identifying and addressing structural forms of oppression like white supremacy and patriarchy because the ideal can only identify specific examples of racism or patriarchy and can only treat them as exceptional rather than systemic.61

Worse, such distortions might be unavoidable. Mills describes ideal theory as ‘a distortional complex of ideas, values, norms, and beliefs that reflects the non-representative interests and experiences of a small minority’ of people.62 This is not to allege a conspiratorial or even conscious element to the distortion that ideal theory entails; it is simply to recognize that idealization often reflects and protects the power and interests of the privileged, whose experience informs and thus approximates the ideal and who thus benefit most from its maintenance.63 In addition, and partly as a result, idealization can hide structural domination and oppression from view, making it hard to recognize them and hard to gain critical leverage on them, even in non-ideal theory.64 From Mills's perspective, the social ontology of ideal theory disguises structural domination by silencing oppression (e.g. by ignoring how systematic oppression shapes the basic structure of society), by idealizing social institutions (e.g. by ignoring their impact on the poor or treating it as exceptional), and by idealizing the cognitive sphere itself (e.g. by ignoring how deeply race shapes our thinking).65

Such problems plague Rawlsian global constructivism as well. Literally billions of people lead lives so dramatically different from those lived by theorists of justice that there is reason to fear the kind of distortions just described. Many theorists, for example, conceptualize the global economy as an uneven playing field; on such a view, fair distribution of wealth, income, and opportunity become the ideals. Yet many people experience the global economy as a system of domination that comprehensively subjugates them to the interests of rich countries, corrupt governments, and powerful corporations. To treat these experiences merely as deviations from or exceptions to the ideal of a fair game can obscure the underlying deformities that structure the global economy. Certainly in the cases of aid, trade, and structural adjustment policies, the assumption of strict compliance and the treatment of injustice as exceptional rather than systemic misrepresent how power operates to uphold the interests of the powerful.

At the very least, it is hard to see how ideals that do not gibe even remotely with most people's experiences of the world could be helpful in identifying or remediating structural domination and oppression of this kind.66 Onora O'Neill is perhaps the chief critic of such distortions, disparaging the idealization of states as protectors of their citizens’ human rights as a ‘grotesque parody’ of reality that precludes theorizing about how to deal with oppressive regimes and the depiction of negative effects of neo-liberalism as unfortunate byproducts of ineluctable market forces when, in fact, they result predictably from economic institutions designed to secure the privileges of the powerful.67

Rawlsian global constructivism is particularly prone to such distortion because it requires that the theorist model the salient facts about the world in an ICS from which principles of justice are derived. Consider two examples. First, recall the problems we have seen with Rawls's concept of peoples. Buchanan rejects the concept as a distortion that conceives peoples as unified political and cultural entities with common sympathies and a shared conception of justice. He objects that ‘it is hard to see how an ideal theory that ignores intrastate conflict by assuming deep political unity within states can serve as the ultimate benchmark for progress from our nonideal [sic] situation’. The problem, he argues, is not simply that ideal theory says nothing about our actual situation in this case but that it also bars us from envisioning any role for international law in intrastate conflict.68 His worry is in effect that one cannot bring back in, at the non-ideal level, what has been denied or ignored in ideal theory. Valentini concurs, arguing that Rawls's Law of Peoples has to deny certain facts about the world as a condition for its own applicability.69

Second, consider again the global constructivists’ curious emphasis on the difference principle and their relative disregard of political principles.70 Even if we accept the primacy of economic issues—a significant concession, given the interdependence of political and economic freedom—the distributive ideal distorts our understanding of the rampant injustices in the global economy.71 Cosmopolitan constructivist theorists almost uniformly recommend transfers (redistribution) as a remedial strategy to move us toward their egalitarian ideal. This recommendation is consistent with their conceptualization of existing inequalities as exceptions to or deviations from a just, egalitarian distribution. I have no objection to transfers, but they do little to address the underlying structural inequities—power asymmetries—that create and sustain perverse distributions of wealth and income in the first place. Put differently, transfers treat the symptoms rather than the underlying disease—a world economy designed to maintain the power and privilege of the wealthy at the expense of the poor. This pathology is obscured by the ideal of egalitarian shares. Domination would, of course, be rejected in any reasonable ICS; the point is that ruling it out of consideration in ideal theory narrows our responses to it in non-ideal theory. Transfers cannot remedy systemic injustices of the kind reflected in trade rules, aid programs, and structural adjustment programs; addressing these injustices in non-ideal theory means treating them as aberrations, in effect ruling out a more fundamental rethinking of the global economic order.

I do not mean to imply that constructivists or ideal theorists would disagree with my substantive points about injustice here. I do strongly imply, however, that their method makes addressing such injustices difficult, epistemologically inconceivable. As such, the procedure they adopt is normatively suspect, at least from the kind of emancipatory perspective that treats freedom and equality for all as its fundamental commitments.

A vigorous debate about ideal theory is underway, but it remains largely abstract and epistemological; so far, there has been much more anti-ideal theory than non-ideal theory.72 Much more remains to be said about what a viable non-ideal theory might look like. I cannot develop such an alternative here, but I hope at least to have shown the urgency of finding one: constructivist reasoning, and ideal theory more generally, cannot deliver plausible and persuasive accounts of global justice or adequately guide our thinking about how to reduce injustice.

ACKNOWLEDGEMENTS

The author is grateful to Anja Görnitz for several thoughtful discussions of an earlier draft of this paper. He would also like to thank the reviewers and editors at Ethics & Global Politics for their constructive and insightful comments.

Notes

1. Robert Amdur, ‘Rawls’ Theory of Justice: Domestic and International Perspectives’, World Politics 29, no. 3 (1977); Brian Barry, The Liberal Theory of Justice (Oxford: Clarendon Press, 1973); Brian Barry, ‘Humanity and Justice in Global Perspective’, in Ethics, Economics, and the Law: Nomos 24, eds. J. Roland Pennock and John W. Chapman (New York: New York University Press, 1982); Charles R. Beitz, Political Theory and International Relations (Princeton, NJ: Princeton University Press, 1979); Charles R. Beitz, ‘Economic Rights and Distributive Justice in Developing Societies’, World Politics 33, no. 3 (1981); Charles R. Beitz, ‘Cosmopolitan Ideals and National Sentiment’, The Journal of Philosophy 80, no. 10 (1983); Charles R. Beitz, ‘Justice and International Relations’, in International Ethics, ed. Charles R. Beitz et al. (Princeton, NJ: Princeton University Press, 1985); Thomas W. Pogge, Realizing Rawls (Ithaca, NY: Cornell University Press, 1989); and Thomas W. Pogge, ‘Cosmopolitanism and Sovereignty’, Ethics 103, no. 1 (1992).

2. Charles R. Beitz, ‘Social and Cosmopolitan Liberalism’, International Affairs 75, no. 3 (1999).

3. I don't know of any global constructivists working in political theory today who are not Rawlsian global constructivists. Still, this terminology helps to clarify that we are not discussing constructivism as international relations specialists understand it, nor are we concerned with the kind of social constructivism deployed by scholars in sociology and anthropology.

4. Charles Jones, Global Justice: Defending Cosmopolitanism (Oxford: Oxford University Press, 1999): 14n9; Martha C. Nussbaum, ‘Beyond the Social Contract: Capabilities and Global Justice’, Oxford Development Studies 32, no. 1 (2004); and Janna Thompson, ‘Reasoning About Justice in Global Society’, Global Change, Peace & Security 15, no. 5 (2003).

5. E.g. Samuel Freeman, ‘The Law of Peoples, Social Cooperation, Human Rights, and Distributive Justice’, in Justice and Global Politics, eds. Ellen Frankel Paul, Fred D. Miller, Jr., and Jeffrey Paul (Cambridge: Cambridge University Press, 2006); and Leif Wenar, ‘The Legitimacy of Peoples’, in Global Justice and Transnational Politics, eds. Pablo De Greiff and Ciaran Cronin (Cambridge, MA: MIT Press, 2002).

6. John Rawls, A Theory of Justice, rev. ed. (Cambridge, MA: Harvard University Press, 1999): 11; hereafter TJ. I shall use KC for John Rawls, ‘Kantian Constructivism in Moral Theory’, The Journal of Philosophy 77, no. 9 (1980); PL to refer to John Rawls, Political Liberalism (New York: Columbia University Press, 2003); LP for John Rawls, The Law of Peoples; with, ‘The Idea of Public Reason Revisited’ (Cambridge, MA: Harvard University Press, 1999); and JF for John Rawls, Justice as Fairness: A Restatement, ed. Erin Kelly (Cambridge, MA: Harvard University Press, 2001).

7. See ‘Contemporary Approaches to the Social Contract’ in the Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/entries/contractarianism-contemporary/, section 5. This usage also follows the convention of using ‘contractualism’ to describe Scanlonian views; see ‘Contractualism’ in the Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/entries/contractualism/#HowDoeScaConDifOthSocConTheConKanConRawCon.

8. Rawls also assumes that the parties have ‘the two moral powers’: the capacity for a sense of justice (to understand, apply, and act from principles of justice) and the capacity for a conception of the good (to have, revise, and rationally pursue a conception of the good; JF 18ff.). These assumptions, while crucial for a theory of justice, are not unique to constructivist theories.

9. This assumption was famously criticized by Michael J. Sandel, Liberalism and the Limits of Justice, 2nd ed. (Cambridge: Cambridge University Press, 1998), generating a huge controversy; see Allen E. Buchanan, ‘Assessing the Communitarian Critique of Liberalism’, Ethics 99, no. 4 (1989); and Amy Gutmann, ‘Communitarian Critics of Liberalism’, Philosophy and Public Affairs 14, no. 3 (1985).

10. There is also a Kantian interpretation of this assumption; see TJ 223; KC.

11. The next sentence reads: ‘Or if we do not, then perhaps we can be persuaded to do so by philosophical reflection’ (TJ 19). This point underscores that agreement underlies the constructivist method of reasoning, contrary to the claims made by Joseph H. Carens, ‘Aliens and Citizens: The Case for Open Borders’, The Review of Politics 49, no. 2 (1987). I return to this point below.

12. I have written ‘few’, but I know of none who did so.

13. Amdur, ‘Rawls’ Theory’; Beitz, Political Theory and International Relations, 152; Beitz, ‘Justice and International Relations’, 285; T.M. Scanlon, ‘Rawls’ Theory of Justice’, in Reading Rawls: Critical Studies on Rawls’ a Theory of Justice, ed. Norman Daniels (Stanford, CA: Stanford University Press, 1989), 75; Pogge, Realizing Rawls; Thomas W. Pogge, ‘An Egalitarian Law of Peoples’, Philosophy and Public Affairs 23, no. 3 (1994): 197, 247; Darrel Moellendorf, Cosmopolitan Justice (Boulder, CO: Westview Press, 2002), 28; Andrew Kuper, ‘Rawlsian Global Justice: Beyond the Law of Peoples to a Cosmopolitan Law of Persons’, Political Theory 28, no. 5 (2000): 654; Kok-Chor Tan, Justice without Borders: Cosmopolitanism, Nationalism, and Patriotism (Cambridge: Cambridge University Press, 2004), 61; Gillian Brock, Global Justice: A Cosmopolitan Account (Oxford: Oxford University Press, 2009); Wenar, ‘The Legitimacy of Peoples’; and Freeman, ‘The Law of Peoples …’

14. I want to stress that the assumptions sometimes overlap and work together—for instance, to provide determinacy to the conclusions reached through constructivist reasoning (see again Table 1).

15. Barry, The Liberal Theory of Justice, 33, 129; and Amdur, ‘Rawls’ Theory’.

16. David A.J. Richards, ‘International Distributive Justice’, in Ethics, Economics, and the Law: Nomos XXIV, eds. J. Roland Pennock and John W. Chapman (New York: New York University Press, 1982), 282; Beitz, Political Theory and International Relations, 29, 148–49; and Moellendorf, Cosmopolitan Justice. Barry, Beitz, and Pogge all also claimed that even self-sufficient national societies would choose more robust principles of international justice in the second (international) original position than Rawls imagined; see Barry, The Liberal Theory of Justice; Beitz, ‘Justice and International Relations’, 305; and Pogge, Realizing Rawls, 241ff.

17. E.g. Allen Buchanan, ‘Rawls's Law of Peoples: Rules for a Vanished Westphalian World’, Ethics 110, no. 4 (2000); Stéphane Chauvier, ‘Justice and Nakedness’, in Global Justice, ed. Thomas W. Pogge (Oxford: Blackwell Publishers, 2001); Kuper, ‘Rawlsian Global Justice’; Moellendorf, Cosmopolitan Justice; and Tan, Justice without Borders.

18. Pogge, Realizing Rawls, 247.

19. For different, academic versions of this argument see Michael Blake, ‘Distributive Justice, State Coercion, and Autonomy’, Philosophy and Public Affairs 30, no. 3 (2001); and David Miller, National Responsibility and Global Justice (Oxford: Oxford University Press, 2007).

20. E.g. Carens, ‘Aliens and Citizens’; Kuper, ‘Rawlsian Global Justice’; and Pogge, Realizing Rawls.

21. Cf. Kuper, ‘Rawlsian Global Justice’.

22. Consider the apparent analogy with the domestic case, where individuals do not know to what families or associations they belong. The key difference is that principles of justice, which govern the basic structure, do not obtain in families or in associations (though the larger conception of justice might). The problem globally is that two different but overlapping, interdependent basic structures exist; the principles that govern each will necessarily impinge on those governing the other. (I am grateful to a reviewer for suggesting this illustration.)

23. To me the most striking feature of the cosmopolitan constructivist literature—indeed, of the entire literature—on global justice is its almost exclusive focus on redistributive questions, despite the priority Rawls gave to the first principle (though see Luis Cabrera, Political Theory of Global Justice: A Cosmopolitan Case for the World State [New York: Routledge, 2004]; and Véronique Zanetti, ‘Global Justice: Is Interventionism Desirable?’, in Global Justice, ed. Thomas W. Pogge [Oxford: Blackwell Publishers, 2001], who argue that the logic of cosmopolitanism and a global ICS leads inexorably to a global state).

24. E.g. Will Kymlicka, Politics in the Vernacular: Nationalism, Multiculturalism and Citizenship (Oxford: Oxford University Press, 2001); Will Kymlicka, ‘Citizenship in an Era of Globalization: Commentary on Held’, in Democracy's Edges, eds. Ian Shapiro and Casiano Hacker-Cordón (Cambridge: Cambridge University Press, 1999); David Miller, Citizenship and National Identity (Cambridge: Polity Press, 2000); David Miller, ‘The Limits of Cosmopolitan Justice’, in International Society: Diverse Ethical Perspectives, eds. David R. Mapel and Terry Nardin (Princeton, NJ: Princeton University Press, 1998); Charles Taylor, ‘No Community, No Democracy, Part II’, The Responsive Community 14, no. 1 (2003/2004); Charles Taylor, ‘No Community, No Democracy, Part I’, The Responsive Community 13, no. 4 (2003); and Michael Walzer, Spheres of Justice: A Defense of Pluralism and Equality (New York: Basic Books, 1983).

25. Beitz, ‘Cosmopolitan Ideals’, 596; and Carens, ‘Aliens and Citizens’, 256.

26. I am grateful to a previous reader for help on this point. For an interesting but—in my view—unsuccessful attempt to resolve this problem, see John Charvet, ‘International Society from a Contractarian Perspective’, in International Society: Diverse Ethical Perspectives, eds. David R. Mapel and Terry Nardin (Princeton, NJ: Princeton University Press, 1998).

27. Carens, ‘Aliens and Citizens’, 257; and cf. M. Victoria Costa, ‘Human Rights and the Global Original Position Argument in the Law of Peoples’, Journal of Social Philosophy 36, no. 1 (2005). Also see note 10 and the surrounding text.

28. See Beitz, ‘Cosmopolitan Ideals’, 596. I am grateful to a reviewer for suggesting how this example might clarify my argument.

29. Pogge, Realizing Rawls, 267–68; and Thomas W. Pogge, ‘Human Rights and Human Responsibilities’, in Global Justice and Transnational Politics, eds. Pablo De Greiff and Ciaran Cronin (Cambridge, MA: MIT Press, 2002).

30. See Amdur, ‘Rawls’ Theory’, 460; and cf. Nussbaum, ‘Beyond the Social Contract’.

31. Some cosmopolitan constructivists acknowledge that in unfavorable conditions Rawls's two principles can be suspended with the priority of liberty giving way to economic measures needed to ensure a decent minimum for all; see Amdur, ‘Rawls’ Theory’, 444–45; and Beitz, ‘Justice and International Relations’, 298–99. I am doubtful about this proposal on its merits; political freedoms are essential for economic development; see Rhoda E. Howard, ‘The Full-Belly Thesis: Should Economic Rights Take Priority over Civil and Political Rights? Evidence from Sub-Saharan Africa’, Human Rights Quarterly 5, no. 4 (1983); Amartya Sen, ‘Freedoms and Needs: Why Political Rights Are Primary, Even in the Face of Dire Economic Need’, The New Republic 213, no. 4.121 and 4.122 (1994); and Amartya Sen, ‘Human Rights and Economic Achievements’, in The East Asian Challenge for Human Rights, eds. Joanne R. Bauer and Daniel A. Bell (Cambridge: Cambridge University Press, 1999).

32. Cf. Freeman, ‘The Law of Peoples …’, 61.

33. Cf. Mathias Risse, ‘How Does the Global Order Harm the Poor?’, Philosophy and Public Affairs 33, no. 4 (2005).

34. Jones, Global Justice, 14 n.9.

35. Charles R. Beitz, ‘Rawls's Law of Peoples’, Ethics 110, no. 4 (2000), has argued that Rawls's Law of Peoples would, despite these restrictions, be quite radical and progressive given current conditions. Cf. LP 116.

36. Rawls's extension of his Law of Peoples to ‘decent’ peoples (see LP 4–5, 62ff.) seems more problematic in this regard.

37. Cf. Freeman, ‘The Law of Peoples...’, 42–43; Wenar, ‘The Legitimacy of Peoples’, 62.

38. Wenar, ‘The Legitimacy of Peoples’, 62–63.

39. Ibid., 63.

40. Freeman, ‘The Law of Peoples …’, 48–49.

41. Andrew Moravcsik, ‘The Origins of Human Rights Regimes: Democratic Delegation in Postwar Europe’, International Organization 54, no. 2 (2000); and Jamie Mayerfeld, ‘The Mutual Dependence of External and Internal Justice: The Democratic Achievement of the International Criminal Court’, Finnish Yearbook of International Law 123 (2001).

42. For a powerful substantive critique of Rawls's account of human rights and their selection, see Kuper, ‘Rawlsian Global Justice’, 664.

43. See Buchanan, ‘Rawls's Law of Peoples …’ for a thorough critique on this point.

44. Kuper, ‘Rawlsian Global Justice’; Friedrich Kratochwil, ‘Of Systems, Boundaries, and Territoriality: An Inquiry into the Formation of the State System’, World Politics 39, no. 1 (1986); and Pogge, ‘Cosmopolitanism and Sovereignty’.

45. On this assumption see Freeman, ‘The Law of Peoples …’, 53, 58–59.

46. Several of Rawls's critics have argued that even in his two-stage approach, parties would select some distributive principles; see Pogge, ‘An Egalitarian Law of Peoples’; Buchanan, ‘Rawls's Law of Peoples...’; and cf. Leif Wenar, ‘Contractualism and Global Economic Justice’, in Global Justice, ed. Thomas W. Pogge (Oxford: Blackwell Publishers, 2001). This disagreement nicely illustrates the problem of indeterminacy I have been highlighting throughout.

47. Stephen D. Krasner, Sovereignty: Organized Hypocrisy (Princeton, NJ: Princeton University Press, 1999).

48. See note 41 and the surrounding text.

49. Brock, Global Justice, 181. Brock's reliance on constructivism reasoning is frustrating; she is most persuasive when relying on her argument for basic human interests, 58–75. Indeed, given this argument, it is not clear what additional work the constructivist procedure does in her account.

50. Ibid., 304–5.

51. Zofia Stemplowska, ‘What's Ideal About Ideal Theory?’, Social Theory and Practice 34, no. 3 (2008): 334.

52. Adam Swift, ‘The Value of Philosophy in Nonideal Circumstances’, Social Theory and Practice 34, no. 3 (2008): 82, 363.

53. Ingrid Robeyns, ‘Ideal Theory in Theory and Practice’, Social Theory and Practice 34, no. 3 (2008): 352.

54. Amartya Sen, ‘What Do We Want from a Theory of Justice?’, The Journal of Philosophy 103, no. 5 (2006): 226.

55. Ibid., 217. For an attempt to refute Sen's critique see Swift, ‘The Value of Philosophy’, 372ff.

56. Onora O'Neill, ‘Bounded and Cosmopolitan Justice’, Review of International Studies 26 (2000): 67ff.; and Charles W. Mills, ‘ “Ideal Theory” As Ideology’, Hypatia 20, no. 3 (2005): 167.

57. Mills, ‘Ideal Theory’, 168; and Onora O'Neill, Bounds of Justice (Cambridge: Cambridge University Press, 2000): 68. O'Neill distinguishes between abstraction that she finds necessary and useful, and idealization that she describes as false assumptions, 151. I follow Mills in thinking that prescriptive theory necessarily idealizes, so I use the term ‘distortion’ to capture what O'Neill means by idealizations or false assumptions. For a critique of O'Neill's distinction see Lisa H. Schwartzman, ‘Abstraction, Idealization, and Oppression’, Metaphilosophy 37, no. 5 (2006).

58. Thomas McCarthy, ‘Political Philosophy and Racial Injustice: From Normative to Critical Theory’ (Evanson, IL, Northwestern University: 2001): 8–10.

59. Schwartzman, ‘Abstraction’, 578.

60. McCarthy, ‘Political Philosophy and Racial Injustice’, 15.

61. Mills, ‘Ideal Theory’, 173ff.

62. Ibid., 172.

63. McCarthy, ‘Political Philosophy and Racial Injustice’; Mills, ‘Ideal Theory’; and O'Neill, Bounds of Justice, 152.

64. McCarthy, ‘Political Philosophy and Racial Injustice’, 10.

65. Mills, ‘Ideal Theory’, 168–69.

66. Cf. ibid., 172.

67. O'Neill, Bounds of Justice, 165–80.

68. Buchanan, ‘Rawls's Law of Peoples …’, 717–18.

69. Laura Valentini, ‘On the Apparent Paradox of Ideal Theory’, The Journal of Political Philosophy 17, no. 3 (2009): 22. Beitz (‘Rawls's Law of Peoples …’, 680) correctly observes that mis-description is not the issue; in his view, the point of idealization is to establish worthy ideals toward which reformers can aim. He rejects peoples as an undesirable or unworthy ideal. This position, however, fails to distinguish relevantly different kinds of mis-description; namely, abstraction, idealization, and distortion. Peoples is an undesirable concept in part because it distorts reality.

70. See note 22 and the surrounding text.

71. On this interdependence see note 30.

72. I owe this felicitous phrase to Robert Lepenies.