Nonideal theory and compliance—A clarification

Introduction

What is “nonideal theory,” and when do we have reason to engage in this kind of theorizing? Recently, political theorists have called attention to the conceptual dichotomy of ideal and nonideal theory. So far, however, most contributions to this discussion have focused on the uses and characteristics of ideal theory, either contesting or defending the adequacy of this approach (Farrelly, 2007; Lawford-Smith, 2010; Mills, 2005; Stemplowska, 2008; Valentini, 2009). Much less theoretical effort has been devoted to exploring the concept of nonideal theory. As a consequence, “nonideal theory” is often used in an imprecise way, referring to anything from pure descriptive research to fact-sensitive normative approaches and comparative normative accounts (Galston, 2010; Geuss, 2008; Hamlin and Stemplowska, 2012; Sen, 2006). ¹ This paper aims to clarify the concept of nonideal theory by examining its relation to compliance. Studying this relation not only provides important insights into the various uses of nonideal theory but also suggests a way in which the theoretical gap between “actor-oriented” and “situation-based” conceptions of justice can be bridged.

My argument proceeds as follows. In the first section, I present my understanding of nonideal theory as consisting of two subparts, complementary and substitutive nonideal theory. Section II examines how this twofold conceptualization relates to different kinds of noncompliance. The upshot is that complementary nonideal theory responds to deliberate and avoidable noncompliance, whereas substitutive nonideal theory responds to nondeliberate and unavoidable noncompliance with ideal principles of justice. Section III presents and addresses the possible objection that nondeliberate and unavoidable noncompliance contains no injustice, and therefore is part of ideal, rather than nonideal, theory. Thereafter, I proceed to cases where noncompliance arises due to citizens’ lack of motivation. Sections IV and V explore situations where substitutive nonideal theory provides the only relevant response to this particular kind of noncompliance. Section VI summarizes the paper’s main conclusions.

I. Two subparts of nonideal theory

The terms ideal and nonideal theory were coined by John Rawls in his influential book A Theory of Justice (1971). While ideal theory, he explains, assumes strict compliance and aims at providing principles characteristic of a well-ordered society under favorable circumstances, nonideal theory asks which principles to adopt under less happy conditions (Rawls, 2005: 245). Rawls further distinguishes between two subparts of nonideal theory. For my argument to unfold as clearly as possible, I will here present them in reversed order.

The first subpart of nonideal theory produces principles for dealing with injustice. This comprises “the theory of punishment, the doctrine of just war, and the justification of the various ways of opposing unjust regimes, ranging from civil disobedience and militant resistance to revolution and rebellion.” Further, it deals with compensatory justice, and with ranking different institutional injustices (Rawls, 2005: 8, 245). The first kind of nonideal theory, that is, tackles issues that do not arise in ideal theory as a result of the assumption of full compliance. The prevalence of partial or full noncompliance with the principles of justice gives rise to at least two distinct problems. First, we may ask how our own duties are affected when others fail to comply with theirs (do we, for instance, have a duty to “pick up the slack?”). Second, we may wonder how to deal with those who do not comply. In both cases, nonideal theory in this sense is a complement to ideal theory, which tells us what to do when faced with states and individuals that fail to comply with the rules and regulations derived from the principles of justice governing a well-ordered society. I will call this complementary nonideal theory.

The second subpart of nonideal theory fulfills a different task: It produces principles for governing adjustments to natural limitations and historical contingencies, which, temporarily and under certain conditions, replace or modify Rawls’ two principles of justice. For example, as Rawls explains, it may be justified to adopt an unequal distribution of liberties, given that such measures enable transition of society from less favorable circumstances to a state where equal liberties can be enjoyed (Rawls, 2005: 245–249). In this case, nonideal theory does not concern special issues that do not arise in ideal theory. Rather, it concerns precisely the same issue as ideal theory, namely which principles should guide just institutions. The difference between the two is that while ideal theory acts on the premise of favorable circumstances, nonideal theory takes into consideration prevailing obstacles to achieving perfect justice. Another way of putting this would be to say that while ideal theory assumes only “hard,” or minimal, feasibility constraints, nonideal theory incorporates also “soft,” or less than minimal, feasibility constraints, characteristic of a particular historical situation (Horton, 2010: 436, Simmons, 2010: 6; Gilabert and Lawford-Smith, 2012). It is in this sense, and no other, that nonideal theory offers “guidance here and now” in a way which ideal theory does not (given that circumstances here and now are less than favorable) (Freeden, 2012: 8; Jubb, 2012: 233). ²

There seems to be at least two reasons why unfavorable circumstances call for nonideal, rather than ideal, theorizing. The first reason is provided by the “general theory of the second best,” which states that in cases where we cannot realize all aspects of a complex ideal (like justice), we may well prefer less to more of one of its components (Goodin, 1995: 53–54; Lipsey and Lancaster, 1956). Say that our ideal breakfast is pancakes with maple syrup. If we cannot have the pancakes, we will probably not want the maple syrup either (Estlund, 2011: 216). This shows that what is best under favorable conditions (where all aspects of our ideal is realizable) may not be a good guide to what is best under unfavorable ones (where they are not) (Mason, 2004: 267). ³ The second reason is that in order to embark on a transitory path that will eventually lead to the realization of our ideals, we may have to opt for a less ideal outcome here and now. If what hinders us from establishing just institutions are the unfavorable circumstances we find ourselves in, we should do what we can in order to overcome these circumstances, and this may conflict with the attempt to maximize justice here and now (even when understood in the “nonideal” sense invoked by the general theory of the second best) (Gilabert, 2012a: 48). ⁴ Nonideal theory, in this understanding, does not tell us how to respond to deviations from the rules and regulations issued by just institutions. Instead, it addresses situations where the attempt to establish fully just institutions (since bound to fail) would actually prove counterproductive for promoting justice, now or in the future (Simmons, 2010: 12–13). I will call this subcategory substitutive nonideal theory.

II. Nonideal theory and compliance

Now, the twofold definition of nonideal theory outlined previously gives rise to some troubling observations. First, Rawls sometimes speaks of nonideal theory as synonymous with partial compliance theory, a description that seems to fit complementary, but not substitutive, nonideal theory (Simmons, 2010: 13). Yet, in Rawls’ account, both are part of nonideal theory. How are we to understand this apparent contradiction? As we have seen, substitutive nonideal theory responds to unfavorable circumstances arising from natural limitations or historical contingencies. Stemplowska and Swift term these circumstances “socioeconomic,” a reading that seems to find support in Rawls’ example of a society that is temporarily unable to grant full and equal liberties to all its members (Stemplowska and Swift, 2012: 375). At such point in time, Rawls argues, the society in question could opt for restricted or differential liberties, given that this policy is a necessary step in order to eventually achieve full compliance with the principles of justice. In this understanding, substitutive nonideal theory is a transitional theory, allowing us to improve our background conditions in a way that will maximize future justice. ⁵

This understanding fits well with the claim that substitutive nonideal theory responds to unfortunate circumstances due to historical contingencies. However, the socioeconomic interpretation cannot account for natural limitations of the kind arising from what Rawls calls “accidents of life,” including disabilities and childhood. In these cases, the task of substitutive nonideal theory is not to provide a theory of transition, but to justify differential rights for children and the severely mentally disabled (Simmons, 2010: 15). It thus seems as if substitutive nonideal theory also consists of two, rather distinct, parts. Why, then, does Rawls place them in one single category? Which reasons could we have to bundle together (temporary) socioeconomic constraints and (permanent) accidents of life? In the absence of good answers to this question, we may reject Rawls’ categorization. I do not want to dismiss this view altogether. At the end of the day, it might provide the most consistent conceptualization of nonideal theory. Nevertheless, I will in this paper defend an alternative reading, originally proposed by Simmons (2010), which remains faithful to Rawls’ original idea. According to this reading, substitutive nonideal theory responds to the question of how we are to deal with individuals and societies that fail to comply with the demands of ideal justice, through no fault of their own.

Simmons argues that both complementary and substitutive nonideal theory arise as responses to noncompliance with ideal principles of justice, but whereas noncompliance in the first case is avoidable and deliberate, noncompliance in the second case is largely inevitable and nondeliberate (Simmons, 2010: 14). In order to avoid misunderstandings, it is important to note in which sense substitutive nonideal theory responds to noncompliance. Importantly, neither unfortunate socioeconomic circumstances nor accidents of life are here thought of as the result of unjust behavior. To the contrary, Rawls’ argument presupposes that they are unavoidable products of misfortune. Since the Rawlsian conception of justice is subject to feasibility constraints, this also means that they are not unjust in and by themselves (Gheaus, 2013). ⁶ But poverty at a state level, just like accidents of life at an individual level, may give rise to seemingly unjust acts, where states and individuals fail to comply with the demands of justice as dictated by the choosers in the Original Position (in the case of states) or just laws (in the case of individuals). And just as we do not want to fault a state for undertaking the necessary measures for ensuring future justice, even if this involves sidestepping the principles of justice here and now, we avoid blaming children or mentally disabled persons for breaking laws, including just laws. This is the reason why we should evaluate the acts of states and individuals subject to unfortunate circumstances according to nonideal, rather than ideal, principles.

The distinction between state and individual noncompliance, identified by Simmons, is crucial for my argument that accidents of life may be seen as giving rise to substitutive nonideal theory. I assume this to be uncontroversial, not least since the division also occurs in complementary nonideal theory. Adding this distinction to the one between deliberate and nondeliberate noncompliance yields Figure 1 (Simmons, 2010: 17). OPEN IN VIEWER

If the upper row corresponds to what I here term complementary nonideal theory, the lower row displays two different aspects of substitutive nonideal theory. I think the figure provides a plausible understanding of Rawls’ twofold conceptualization of nonideal theory, as well as an explanation of how accidents of life fit into this picture. For this to constitute a convincing alternative to other conceptualizations of nonideal theory, however, I must also respond to those theorists who hold Rawls’ view mistaken. One of them is Robert Jubb, who claims that the circumstances giving rise to what I have called substitutive nonideal theory can actually be addressed within the ideal theoretical framework, and therefore is not part of nonideal theory. The following section addresses his critique.

III. Feasibility and nonideal justice

Jubb argues that since the unfavorable circumstances to which substitutive nonideal theory responds are not due to (deliberate) noncompliance, substitutive nonideal theory contains no injustice, and is therefore in fact not nonideal at all. His own explanation to why this is so gives a clear account of his position and deserves to be quoted at length:

Either there is some act which is unjust, in which case there is an actor not complying with the demands of justice, or all actors are complying with the demands of justice, in which case there surely cannot be any injustice. Our responses to the natural limitations and historical contingencies clearly can be just or unjust, and a theory of justice will have to say something about such responses. However, to complain that constraints on what we can do and achieve entirely beyond our control make it impossible for us to satisfy the demands of justice seems mistaken. Although we might rightly regret such constraints, they are not themselves directly a source of injustice (Jubb, 2012: 231–232).

As it stands, Jubb’s argument depends on two statements to be correct. First, it must be true that situations that are not the result of anybody’s noncompliance with her duties do not contain injustice. Second, it must be true that if a situation does not contain injustice, it is not nonideal. I will examine these statements below.

The first claim concerns what Anca Gheaus calls “the feasibility constraint on justice,” according to which justice cannot demand the infeasible (Gheaus, 2013). This view is tightly related to an actor-oriented approach, where demands of justice correspond to duties that are in turn owed by agents. Here, the notion of “unavoidable noncompliance,” upon which my conceptualization of substitutive nonideal theory rests, appears like something of a contradiction in terms. This is because the range of possible duties is limited by the Kantian formula “ought implies can”. Since ought implies can, only acts which are avoidable can be characterized as unjust. And since only acts, and not states of affairs, can be just or unjust, injustice cannot arise from unfortunate circumstances alone (Anderson, 2010: 5). Against this view, one could hold something like the luck-egalitarian position that states that every disadvantage that is not the result of an agent’s choice is unjust, regardless of how this disadvantage came about. But such a move would distance us from the Rawlsian outlook which this paper attempts to defend. From a Rawlsian point of view, it seems as if Jubb’s statement that substitutive nonideal theory does not contain any injustice is (at least prima facie) correct. As already noted, rather than dealing with the separate issue of responding to injustice, substitutive nonideal theory sets the standards for justice under unfavorable circumstances.

This brings us to the second of Jubb’s claims. If substitutive nonideal theory contains no injustice, why should we consider it a part of nonideal, rather than ideal, theory? Well, precisely because we want our theories to produce normative standards suitable for evaluating the behavior of agents, and not (only) states of affairs. Despite Rawls’ affirmation of the actor-oriented approach, this ambition risks running into conflict with his theory of justice. This is because, for Rawls, ideal justice has a specific content: It is only achieved when the basic structure of society works in a way which grants all citizens maximal and equal liberties (the freedom principle), and guarantees that inequalities are for the benefit of the worst off (the difference principle). But, as we have seen, the unfortunate circumstances that give rise to substitutive nonideal theory are such that the state cannot act on these principles (at least not without ruining future prospects of fulfilling the demands of justice). Recall Rawls’ example of a society where liberties temporarily must be unequally distributed in order for society to move beyond its current deadlock and allow full compliance with the principles of justice. In this case, we have a situation where the state does what it can to achieve justice, and hence could have no duty to do more. Following the actor-oriented approach advocated by Jubb (and commonly attributed to Rawls), we must conclude that the outcome is just. Nevertheless, Rawls’ principles of justice are not met. In order to make sense of this situation, we would need a way to recognize that the state has fulfilled its duties, and hence has not acted unjustly, while acknowledging that the society in question is still not ideally just. And this, I suggest, is precisely what substitutive nonideal theory offers. ⁷ By affirming that unfortunate circumstances justifies a move from an ideal to a nonideal standard of justice, we may evaluate the state’s action according to the nonideal standard, concluding that, given the circumstances, it has fulfilled its obligations. In the long run, however, we recognize that it cannot settle on this nonideal standard, but has to work continuously toward ideal justice. Including substitutive nonideal theory in nonideal theory thus accounts for the intuition that a society where everyone fulfill their duties must in some sense be a just society, while avoiding the mistake of equating ideal justice with that which is currently feasible.

One could object that the distinction between ideal and nonideal justice proposed previously introduces a double standard of justice that risks confusing, rather than clarifying, the debate. Yet, the alternatives seem at least equally bad. We may of course call all situations where all actors comply with their duties ideally just. But this view fails to explain why we should strive toward overcoming the unfortunate circumstances in which we find ourselves, instead of settling for the present situation (it is, after all, fully just). Or we may, to the contrary, conclude that due to the prevailing circumstances, our society is unjust. But this will distance us from the actor-oriented approach, failing to recognize that the state has actually fulfilled its duties of justice. A third possibility would perhaps be to argue that if circumstances are such that the principles of justice cannot be met, the situation cannot be evaluated according to the standards of justice. Since ought implies can, the argument would go, we cannot speak of an injustice unless (ideal) justice is actually a feasible alternative. But apart from rendering justice a more or less inapplicable concept, this understanding confuses the “fortunate circumstances” necessary for realizing the principles of justice with the circumstances of justice, outside of which the concepts of justice and injustice lack meaning. From Rawls’ writings, it is clear that he regards these to be different concepts. Rather than allowing for justice to be realized, the circumstances of justice are conditions that allow for the question of justice to arise. Those are, briefly stated, (1) that individuals coexist in space and time, (2) that none of them is able to dominate all the others, (3) that resources are moderately scarce, engendering (4) conflicting interests (Rawls, 2005: 126–127). In contrast, the unfavorable circumstances giving rise to substitutive nonideal theory arise within the domain of justice. We find ourselves (and others) in a situation characterized by conflicting interests and (only) moderate scarcity, yet we are unable—for the time being—to create fully just institutions. This is a realistic scenario since institutional change, even if possible, may come only at great costs and demand significant societal and cultural changes, which are only realizable in the long term. Hence, our society is not ideally just, nor could it be, given the prevailing circumstances.

IV. The lack of motivation as a trigger of nonideal theory

So far, I have assumed that the two situations giving rise to substitutive nonideal theory—states subject to historical contingency and individuals subject to natural limitations—are altogether separate matters. In their respective domains, both kinds of circumstances produce nondeliberate and unavoidable noncompliance with the demands of justice, which in turn requires a response in the shape of a nonideal principle of justice. It is now time to complicate this picture. Since states are made up of individuals, acting collectively, we can imagine a case where individual citizens’ noncompliance restricts the state’s opportunities to act justly. What matters here seems to be the extent of individuals’ noncompliance, or foreseen noncompliance, with just rules. A single individual’s noncompliance clearly constitutes a case of complementary nonideal theory, to which the state responds with punishment. But massive noncompliance may not be possible to remedy in this way, and foreseeing that a vast majority of the state’s citizens will break a proposed rule (even if just) may constitute a good reason for the state to postpone its legislation, first taking measures to change public opinion. We may consider this a special case of “unfortunate circumstances,” which takes the shape of motivational deficiencies strong enough to undermine the state’s every (short-term) attempt to put in place just institutions. Such deficiencies may for instance arise due to weak political or cultural traditions or sharp divisions between classes or ethnic groups. Since the state is not capable of creating just institutions, and since (again) ought implies can, it seems as if we, when faced with this situation, would need to evaluate the state’s action according to nonideal, rather than ideal, standards.

Now, the suggestion that lacking motivation may be considered an unfortunate circumstance may seem objectionable, at the very least. Would this not be to let the citizens off the moral hook far too easily, letting their motivational deficit lead us to adopt a less strict standard of justice when evaluating their state? If so, it seems as if we are on very shaky normative grounds indeed. Surely, the mere fact that people do not comply with principles of justice is not a reason for us to abandon those principles (Estlund, 2011: 209; Lawford-Smith, 2013: 256). Nonetheless, as Andrew Mason argues, at some level of analysis we must take into account (i.e. ascribe normative value to) the dangers of trying to instantiate an ideal when it is not motivationally possible (Mason, 2004: 266). ⁸ Though ideal justice could only be acquired by implementing and complying with just institutions, justice could not demand that we implement such institutions if the foreseen noncompliance would actually take us further away from this ideal (Estlund, 2011: 216). In particular, a political theory that aims at prescribing (and not merely evaluating) political action cannot simply ignore the fact of lacking motivation among citizens (Gilabert, 2011; Swift, 2008: 373–374).

Is treating the lack of motivation as an unfortunate circumstance not at odds with the idea that substitutive nonideal theory deals with noncompliance that is nondeliberate and unavoidable? Where noncompliance is caused by lacking motivation, it seems clearly deliberate and, at least in the normatively relevant sense, avoidable for the individual actors concerned. ⁹ But when we think about the acts of a state, noncompliance caused by lacking motivation may sometimes be both nondeliberate and unavoidable. The reason is that feasibility is relative to actors, and the state is an actor only in a metaphorical sense. Thus, when we speak of the state as an agent we actually have in mind some particular actor or set of actors within the state, and depending on who this is, the span of feasible actions varies.

To exemplify: Assume that we want to evaluate the actions of the ruling elite in a state where any attempt to create just institutions would elicit massive noncompliance by the citizens, leading to political chaos. If the elite faced with this situation were to opt for less radical reforms with better prospects of success, this is hardly something we could hold against them from the point of view of justice. This is so because the citizens’ noncompliance restricts the set of feasible actions available to the ruling elite. Since just institutions are no longer part of the feasible set, it can no longer be a duty of the elite to bring them about. By holding constant the noncompliance of citizens, we treat it as a backdrop—an unfortunate circumstance—against which the elite performs. And while recognizing that the elite fulfills its duties, we may still fault the citizens for their unwillingness to comply with (more) ideally just rules and regulations. Though it is admittedly problematic to separate the actions of a state’s governing elite from that of the population as a whole, this operation will sometimes be necessary in order to pursue crucial tasks of political theory, such as assessing a people’s right to rebel against the ruling elite.

But if we treat lacking motivation as a circumstance when we evaluate the actions of the elite, it seems as if we should do the same when evaluating individual citizens’ actions. Of course, the mere fact that I am unwilling, or too lazy, to fulfill a duty does not release me from this duty. But even if my own lack of motivation cannot count as a relevant restriction on the actions available to me, the lack of motivation in others may well do so if the obligation we are under requires collective action (Lawford-Smith, 2013: 256). This, as Holly Lawford-Smith explains, is because my duty as a member of a group to do my share of a collective obligation is conditional upon the belief that the other members will do their share (Lawford-Smith, 2012: 457–459).

Assume that I and my fellow citizens together have a duty to implement just institutions, and that I as a consequence have a duty to do my share in implementing them. Let’s further assume that I have good reason to believe that my fellow citizens, due to their lacking motivation, will actually fail to do their share, with the result that just institutions will not be put in place even if I do mine. In this case, my primary obligation will be replaced by a secondary obligation that will probably entail, but may also exceed, trying to change the attitudes of my fellow citizens. Since the move from a primary to a secondary obligation is the result of others’ (expected) behavior, their lack of motivation constitutes an unfortunate circumstance that restricts the set of actions available to me (Lawford-Smith, 2013: 256). But when we turn to the behavior of the others, nothing restricts their set of available options, which is why we should rightfully blame them for their failure to comply. The moment we shift object of evaluation, one actor’s lacking motivation becomes an unfortunate circumstance restricting another actor’s set of feasible options. And while the first actor’s noncompliance is deliberate and avoidable, the second actor’s noncompliance (within the requirements of ideal justice, that is) is nondeliberate and unavoidable. As a consequence, the second actor could not be faulted for her noncompliance. Further, while our recommendation to the first actor will simply be to fulfill her original obligation, the prescriptions we give to the second actor must acknowledge the restrictions that the noncompliance of the first places on the possibilities available to her. This move from primary to secondary obligations constitutes, I believe, a simultaneous shift from ideal to nonideal theory.

It may seem as if the only difference between the situation invoked here and the noncompliance of complementary nonideal theory is that in the former case, noncompliance is collective. But surely the fact that many, as opposed to a few, disobey does not transform a moral failure into an “unfortunate circumstance?” I admit that it seems strange that quantity should make the difference between substitutive and complementary nonideal theory. But even so, there is no way of dealing with an entire population’s disobedience through the institutions of correction and retribution outlined by complementary nonideal theory. The proper response to such a situation would instead consist in finding ways of turning the public opinion in favor of just institutions. If theories of transition are part of substitutive nonideal theory, a government’s response to (foreseen) massive disobedience against just institutions clearly belongs to this category.

But it is one thing to let a ruling elite represent a state, quite another to treat the acts of specific citizens or groups of citizens as representative for the states they live in. Regarding the latter, one may object that as long as the political community taken as a whole is capable of fulfilling its obligations, nothing motivates a shift to a nonideal standard of justice. If this is true, the only way the lack of motivation could be considered an unfortunate circumstance is when the noncompliance it produces is in some sense nondeliberate and unavoidable for all citizens of a state. It is admittedly hard to think of such cases. Still, it is possible. In particular, the members of a society may all recognize their duty to comply with the requirements of ideal justice, given that most others also comply (an assumption which, it is important to recall, is a prerequisite for the duties of ideal justice to arise in the first place). Yet, knowing that they live in a corrupt society, where rule followers are bound to be the “suckers” of the social game, all eventually end up defecting. In this case, citizens find themselves in what is sometimes called a “social trap,” where low trust and expectations of disobedience with rules render reformatory work futile (Gambetta, 2000: 217; Rothstein, 2005: 4). ¹⁰ The following section explores this case in further detail.

V. The case of “social traps”

The problem of social traps arises because trust and cooperation tend to be mutually reinforcing social phenomena. While experiences of cooperation yield trust, in most cases some minimal level of trust is a precondition for cooperation to arise in the first place. Hence, where trust is excessively low, this will often hinder precisely the kind of interaction necessary for increasing the level of trust. Distrust, we may say, is “self-fulfilling” (Gambetta, 2000: 234; Platt, 1973: 645). ¹¹ In the political domain, this problem is perhaps even more pertinent. In modern societies, trust is both a prerequisite for, and a result of, successful political action (Gambetta, 2000: 224; Misztal, 1996: 7; Putnam, 1993). Furthermore, trust on a societal level often follows on institutional and socioeconomic development (Misztal, 1996: 199). If the lack of trust in a society constitutes an “unfortunate circumstance” restricting the feasible option set for this society, this may then be a result of other unfortunate circumstances, such as poverty. ¹²

The mutually reinforcing dynamics of distrust and noncooperation provides the link between social traps and the unfortunate circumstances that give rise to substitutive nonideal theory. As we saw previously, we are only under an obligation to do our share of a collective duty if we expect others to do theirs. But when caught in a social trap, we precisely do not expect that the others will cooperate. Our distrust need not be based on the belief that the others are unconditionally uncooperative. Instead, we may fear that they will not trust us to cooperate, and defect as a result of their distrust (Gambetta, 2000: 216). Unlike the cases examined previously, there seems to be no one to blame for the noncooperation resulting from this situation: All would comply, given that the others did, too, but all have good reason to believe that the others will not. This is what renders the social trap an unfortunate circumstance.

How does this relate to my earlier claim that substitutive nonideal theory deals with noncompliance that is nondeliberate and unavoidable? The fact that all citizens would prefer a situation where the primary obligation (to put in place just institutions) was fulfilled shows in which sense the noncompliance resulting from a social trap is nondeliberate. But to which extent is it unavoidable? It would be tempting to suggest that the noncompliance resulting from social traps is unavoidable since we cannot choose to trust other people when in fact we do not. As Bo Rothstein explains:

You cannot simply rationally decide to trust people whom you have mistrusted for a long time, even if you fully understand that the development of mutual trust would be very advantageous. However rational it may seem, it appears impossible to decide to forget a thing like the experience of severe injustice or betrayal of trust. Once a society, an organization, or a group has been caught in a social trap situation, it usually stays there (Rothstein, 2005: 18). ¹³

When caught in a social trap, our individual or collective experiences render us psychologically incapable of trusting others. Worse, the only way of overcoming this obstacle is likely to pass through the kind of collective action that is ruled out by our very lack of trust. Still, this does not release us from our obligations, since even if incapable of trusting others, we are able to act as if we trusted them. Diego Gambetta argues that this is what we should do, since this strategy may be the only way to break the deadlock created by mutual distrust. Especially where stakes are low, he explains, it is rational even for moderately forward-looking egoists to “trust trust and distrust distrust” (Gambetta, 2000: 228, 234).

Does this mean that we have a duty to act as if we trusted one another? First, it is important to note that Gambetta’s argument stipulates a situation where you simply do not know if the other will cooperate or defect. Where chances are 50/50, we have good reasons to act as if we trusted the other (since it is possible, even probable, that the other will copy our behavior). But in the kind of social trap situation I am talking about here, chances are not 50/50. In order for my belief that others will defect to be reasonable, I must have at least some evidence that points in this direction. Precisely at which point my belief becomes reasonable is a matter of discussion, but 50/50 will clearly not do. And where the estimated prospects for cooperation are, say, 30/70, or even 1/99, to “trust trust” no longer seems like a rational strategy. ¹⁴

However, rationality is one thing, duties another. What we need to know is whether or not we have a duty to act as if we trusted others. This seems important, because if we have such a duty, it looks as if my earlier claim that social traps trigger substitutive nonideal theory was mistaken. If our duties remain the same, then perhaps the social trap does not constitute an “unfortunate circumstance” after all? Well, that depends on what we mean by saying that our duties remain the same. I will not deny that we have a duty to act as if we trusted others. I hold that it is possible, even probable, that we do. Instead, I want to defend my earlier claim by arguing that if we have such a duty, it is a nonideal duty. Recall that the reason why we are released from the primary obligation to do our share is not that we are incapable of doing that—even if all others defect, we could always still do our share. Rather, the reason is that, whatever we do, we are incapable of fulfilling our collective duty, given (our reasonable belief) that others will defect. Now, what Gambetta suggests is that acting as if we trusted others may in fact impact their behavior, and therefore we have good reasons to do so. Our duty to act as if we trusted others is thus dependent on the possible success of this strategy: If we did not believe that our cooperation might increase the level of trust in society (i.e. if our fellow citizens were unconditionally noncooperative), we would not consider it a duty in the first place. Rather than a duty to do our share, then, it is a duty to increase trust, and only if acting as if we already trusted each other is a means to this end (which is, of course, an empirical question), do we have a duty to do so. What this shows is that the obligation to act as if we trusted others is really a secondary obligation, aimed at bringing about a transition away from the unfortunate circumstances in which we find ourselves. Of course, if all members of society fulfill this secondary obligation, the primary obligation will, as a result, also be fulfilled (in which case the need for increasing trust curiously seems to disappear). But this peculiar outcome is a result of the specific content of the duty to act as if we trusted each other, and does not undermine my claim that it is a secondary duty (though an outcome where all citizens fulfill this secondary obligation may indicate that the original belief that others would defect was perhaps not reasonable after all).

One important question still remains. In which sense are the acts of citizens who are (conditionally) willing to cooperate, acts of noncompliance? If, in the face of others’ (expected) noncompliance, my obligations truly change from one set to another, how can I fail to comply with the first set of obligations that no longer apply to me? (Lawford-Smith, 2012: 463, 466). ¹⁵ This seems confused, but I believe that the confusion stems from the ambivalence within the Rawlsian framework between a situation-based and an action-based account of justice, treated earlier in this paper. Ideally, we have certain obligations and we are able to fulfill them. But if unfortunate circumstances render it impossible for us to fulfill those obligations then other, secondary, obligations will apply. Since ought implies can, fulfilling those secondary obligations is all that justice can demand from us here and now. Yet, as long as the basic structure does not work according to the two principles of justice, our society remains less than ideally just. What this seems to indicate is that our primary obligations linger somewhere in the abstract, waiting for the right circumstances to arise. When I say that the social trap produces nondeliberate and unavoidable noncompliance, it is noncompliance with those primary, or ideal, obligations that I have in mind.

VI. Conclusions

This paper set out to clarify the meaning of the frequently used concept “nonideal theory,” and in particular its relation to different kinds of noncompliance with ideal principles of justice. While acknowledging the problems arising from the Rawlsian division of nonideal theory into two subparts, my paper shows that we might nonetheless have reasons to stick to this twofold definition. I have suggested that we should think of complementary nonideal theory as responding to deliberate and avoidable noncompliance with the principles of justice (in the case of societies) or with the rules and regulation of just institutions (in the case of individuals), whereas we should think of substitutive nonideal theory as dealing with nondeliberate an unavoidable noncompliance due to unfavorable circumstances. The advantage of including substitutive nonideal theory in our account of nonideal theory is that it allows us to preserve the term “ideal justice” for justice under minimal feasibility constraints, while still making sense of the widespread intuition that a society where all agents fulfill their duties of justice (though these are subjected to less than minimal feasibility constraints) must in a (nonideal) sense be just. I have further argued that substitutive nonideal theory may sometimes be triggered by a lack of motivation on the part of citizens. The lack of motivation, I have proposed, calls for nonideal theorizing (1) when our aim is to evaluate the political actions undertaken by specific individuals or groups in a society (in particular, the ruling elite) whose set of feasible options is constrained as a result of the lack of motivation in others, and (2) when a situation of mutually reinforcing distrust and noncooperation among citizens—sometimes called a social trap—constrains the feasible option set of the entire population.