How Resilient is the War Contract?

In War By Agreement, Yitzhak Benbaji and Daniel Statman argue that the morality of war can be governed by a freely accepted agreement over the principles that apply to it. This war contract supersedes the application of the principles of everyday morality to war, thus defying ‘revisionist’ approaches to war, and it upholds a recognizable version of traditional just war theory. This article argues for three claims. First, the contractarian apparatus Benbaji and Statman deploy is actually inconsistent with the deep reasoning they advance on its behalf, since, unsatisfactorily, the contract is supposed to retain normative force even when it is breached by aggressors. Second, the underlying character of their theory makes it something closer to a consequentialist account. Third, this new understanding of their account renders it less distinct from certain articulations of revisionism than they think.

Yitzhak Benbaji's and Daniel Statman's War By Agreement represents probably the most detailed and meticulous defence we have yet of traditional theorizing about war in the contemporary analytical literature. 1 I tend to think, along with Benbaji and Statman, that traditional just war theory has more flexibility and nuance than most revisionists about war suppose it to have. On this broad issue, then, I stand on Benbaji's and Statman's side of the debate. But I have my doubts about the detailed execution of their account, and about the large role that contractarianism in particular plays in it. This article will attempt to bring these doubts to the surface, and to further indicate in the light of this discussion, though as an independent development of its main thrust, that the basis of division between revisionist and traditional just war theory is less sharply drawn than many have suspected.
The article unfolds as follows. In Section I, I provide some background and describe the main commitments of the Benbaji-Statman account. Section II advances my main objection to their account. In its bluntest form -refinements will be added later onthe charge I press is that the most significant normative work assigned to the contract is reserved for circumstances in which that contract has already been breached. But if the contract has already been breached, then it appears that morally suitable responses to that breach will have to be governed by non-contractual considerations. Having been breached, it is too late for the contract to play a morally active role in directing the contracting parties' subsequent actions. The contract has now in effect dissolved, and appropriate moral remedies must be sought elsewhere. I deepen this line of argument in Sections III and IV, suggesting on the basis of the available evidence that Benbaji's and Statman's actual justifications for following the war convention have a non-contractarian character: in fact, as we shall see, they appeal mainly to consequentialist considerations. In the final section, Section V, I discuss the relationship between the reconstituted quasi-consequentialist Benbaji-Statman account and the concessions to the traditional legal symmetries of combatants that have been offered by revisionist just war theorists such as Jeff McMahan.

I. THE BASIC PICTURE
In War By Agreement, Benbaji and Statman assume a richly structured pre-contractual morality along roughly Lockean-Kantian lines. 2 This picture of morality is not notably different from that which we find in revisionist literature: ordinary deontology-inflected commonsense morality. In the absence of the contract, in everyday contexts, it is these principles of morality that would prevail. Now this moral system can be filled out in different fine-grained ways, depending on how various hard cases and pressure points are dealt with, but the point to emphasize immediately is that Benbaji and Statman share the broad starting points of the revisionist theorists they oppose. Divergence between the two camps soon emerges, however.
Revisionists think that the morality of war is continuous with the moral principles that govern ordinary interpersonal morality. This commitment confirms most revisionists as reductionists: there are no sui generis standards for the moral appraisal of war. Revisionists thus affirm two theses, which Benbaji and Statman refer to as Individualism and Continuity, respectively. They define them verbatim as follows: On their preferred account, by contrast, Benbaji and Statman argue that the permissions and prohibitions of ordinary morality are superseded by the terms of an implicit agreement governing the conduct of war. Thus, the morality of war is not simply a copy of pre-contractual moral principles, applied to the particular circumstances of war. In the Benbaji-Statman contractarian account, the various permissions, prohibitions, and requirements are settled by the terms of the war contract. Different actors are assigned to different roles, and these roles then fix the 'oughts', permissions, and expectations that apply to these role-holders.
Benbaji's and Statman's adherence to a more general claim enjoys priority over their specific views about the war contract: Social Distribution: Under specified conditions, social rules partly determine the distribution of moral rights and duties. 5 Social Distribution gives Benbaji and Statman the impetus to question the force of Individualism and Continuity. The 'specified conditions' alluded to by Social Distribution are those in which conformity to these social rules will produce a higher level of compliance with the aims of pre-contractual morality than a state of affairs in which agents act from an unmediated allegiance to pre-contractual principles.
Benbaji and Statman think that Social Distribution can and should be applied to warfare, and thus that war can be properly governed by a contract, just as long as three conditions are in place. 6 The first condition is Mutual Benefit: the contract or other arrangement for hosting social rules should state rules, whatever they turn out to be, allegiance to which is expected by these parties to lead to fewer violations of pre-contractual rights in real world conditions. Benbaji and Statman understand the international order as a self-help regime, without a central authority. This order is composed of a collection of states that are mostly decent, but also self-interested. They lack extensive ties of sympathy to other states, or a willingness to sacrifice national interests to protect them. The international order, thus construed, constitutes a 'minimally just anarchy', in Hedley Bull's famous phrase. 7 The second condition is Fairness: the rules of a contract, and the results of complying with it, must not exacerbate background injustice or unfairness. Benbaji and Statman concede, of course, that the international order is characterized by deep and persistent inequalities and power imbalances among different states. But they think that it is unreasonable to expect the war contract, in and by itself, to ameliorate these inequalities, as long as it does not worsen them, and can satisfy security interests that are recognizably possessed by both weaker and stronger parties to the contract.
The third condition, which will play a substantial role in what follows, is Actuality, which insists on evidence that these are the practices actually followed. The utility of the war contract depends not only on what would be the case if everyone followed it, which is consistent with patchy compliance and hugely sub-optimal outcomes. The utility of the war contract also depends on actual levels of compliance with it. A satisfactory war contract must be, not just a contract that is capable of being followed, but one that is actually followed, in order for the separate parties to have strong reasons to continue being governed by it: … the role that Actuality plays in our contractarian framework is… principled: within this framework, the fact that mutually beneficial rules receive habitual obedience makes it presumably true that they are freely accepted. By freely accepting fair and mutually beneficial social rules, members of the society in question lose some of their natural rights. In other words, when people 6 War By Agreement, War By Agreement, p. 72. habitually follow these rules, they waive the relevant rights in exchange for expected benefits, under conditions of fairness. 8 Finally, the normative force of the war contract is not explained simply by the prospect of the benefits of conformity to it in circumstances where these three conditions are satisfied. The normative force of the contract is provided, rather, by the fact that the parties freely agree to it. In freely agreeing to it, they waive their precontractual rights. Individual actors may be motivated to accept the contract because of its expected benefits, but the contract has moral force not because of its expected benefits, but in virtue of the fact that it is freely accepted: The status of an arrangement such that all relevant parties … would accept it ex ante is a major reason for regarding the outlook we propose here as contractarian. Yet the ultimate basis for the legitimacy of a social arrangement and for the distribution of rights and duties it entails lies not in its contribution to overall utility (however defined) but in its free acceptance within society. 9 Now for a bit more on the content of the war contract itself. Traditional just war theory consists mainly of two dimensions. 10 The jus ad bellum dimension forbids first use of force against the territorial integrity of any other state, and authorizes robust defensive responses against any such aggression. The jus in bello dimension makes combatants morally equal, so that it is morally permissible for each side to fight against the other side, and releases them from the duty to question their orders as long as those orders do not direct them to commit atrocities against civilians or against opposing combatants who have surrendered. Jus in bello protects civilians by making them morally immune from deliberate aggression, though not immune from side-effect killings that are proportionate and necessary.
As already indicated, the type of war contract envisaged by Benbaji and Statman strongly overlaps with these provisions of traditional just war theory. Benbaji and Statman are under no illusions that these provisions will neatly conform to pre-contractual morality. 8 War By Agreement, p. 47; emphasis added. 9 War By Agreement, p. 44. 10 There is also jus post bellum, which is concerned with the morality of ending wars, though the writings on jus ad bellum and jus in bello in the contemporary analytical literature are more voluminous. I will focus on these dimensions, since the central revisionist challenge addresses the relationship between them.
The chief philosophical service that revisionism has provided, the prima facie force of which Benbaji and Statman acknowledge, 11 is the oddness of supposing that jus in bello can be dealt with independently of jus ad bellum. Imagine, in a two-sided conflict between A and B, that A has passed the jus ad bellum tests, but that B has failed them. How do combatants from both A and B acquire the right to fight permissibly under jus in bello rules? If it was impermissible for B to send its troops to war in the first place, then it must surely follow that B's troops cannot be fighting permissibly now.
Benbaji and Statman grasp the force of this challenge, but they still think that, given the real world conditions that are likely to obtain, and in the interests of minimizing expected casualties and rights violations, the war contract offers a more reliable strategy than warfare regulated by the unmediated principles of everyday morality. The rules governing ad bellum apply to office-holders that do not overlap with the office-holders relevant to the rules governing in bello. Politicians and state officials are governed by ad bellum rules, but these rules do not apply to the combatants further down the chain of command, who are governed only by in bello rules. 12 Much of the interest of the Benbaji-Statman account will consist in seeing whether it can satisfy these revisionist scruples. These are the issues, understandably, that revisionist writers are likely to focus on. My point of critical intervention will be different. I want to question the role of the theoretical apparatus Benbaji and Statman deploy in arguing for their version of traditional just war theory. I turn to that task next.

II. CONTRACTUAL FORCE AND CONTRACTUAL VIOLATION
My leading critical claim is that the normative force of the war contract seems at odds with the fact that Benbaji and Statman present it as a contract. The contractarian apparatus is not in satisfactory alignment with what is supposed to be reason-giving about the contract. By the time the war contract comes into its own, it has already been breached, and so has dissolved. Whatever the standards 11 They readily admit that 'the revisionist criticism of traditional just war theory has made the older [traditional] view seem ungrounded and, in a sense, naïve' (War By Agreement, p. ix). 12 Benbaji and Statman emphasize that the moral equality of combatants -the Moral Equality principle, as they refer to it -ensues from the independence of jus ad bellum from jus in bello (War By Agreement, p. 13). that ought to govern morally permissible responses to the breach of this contract turn out to be, we should not expect them to be located in the contract itself.
To see this, we can ask what the war contract does. The contract actually plays two main roles. It is intended, first, to deter parties from violating it; and second, it is intended to provide for robust (but restricted) responses if the contract is violated. I take these points in turn.
The war contract, under the ad bellum dimension, forbids any state's first use of force against the territorial integrity of another state. If this provision is breached, then the contract provides for permissible violent responses from other states. The prospect of such a response serves a deterrence function: states are less likely to engage in military aggression if they can predict that their aggression will be met by violent resistance. But in order to make good on this deterrence function, a contract must also provide for robust defensive responses if deterrence fails.
These defensive responses are actually triggered, however, only if and when at least one side has already broken the contract by violating the ad bellum contract. Suppose that the contract is between states A, B, and C, and that these are the only military powers. The contract provides for violent defensive responses from any of these states should one of them be attacked. But who is going to attack any of these states? Only one (or more) of A, B, or C. A can be attacked by B or C, B can be attacked by A or C, and C can be attacked by A and B. No other attacks are possible, barring combined forms of attack. If the contract prescribes that each of these states ought to refrain from attack unless attacked in the first place, then whoever attacks first, thus supposedly activating the in bello provisions of the contract, has already breached the contract under the ad bellum dimension. If B unjustly attacks A, for example, then B is already in breach of the contract. If B is already in breach of the contract, then it is difficult to see why B's combatants should be covered by the contract under the in bello dimension. The contract encompasses both ad bellum rules and in bello rules, and so if it is violated under the ad bellum dimension, there is no obvious contractually secured progress into the in bello dimension.
Benbaji and Statman may wish to quarrel with this line of thought. After all, since it is possible to distinguish between the war contract under its jus ad bellum aspect and under its jus in bello aspect, it may seem coherent to insist that the in bello contract endures even though the ad bellum contract has been violated. But this 'bifurcation strategy', as I will refer to it, strikes me as unsatisfactory. There are a couple of relevant points.
First, it is difficult to make sense of the relationship between the ad bellum contract (now dissolved) and the in bello contract under the bifurcation strategy. If the in bello contract still endures, then it applies to B's combatants as well as A's combatants. Yet B has already violated the ad bellum contract. How can violation of the ad bellum contract, from which it must follow that B's military aggression is morally impermissible, be rewarded by the in bello contract's assignment of permissibility to B's combatants? The bifurcation strategy sharply exposes Benbaji's and Statman's contractarianism to the very same charge of incoherence that revisionism poses to traditional just war theory. Meanwhile, the permissibility of A's defensive response to B's aggression is over-determined. While the in bello contract does affirm the permissibility of A's military response, A could just as easily claim ordinary pre-contractual grounds for this permissible response to B's aggression.
Second, even if the reasons Benbaji and Statman adduce for operating within the limits of the in bello rules still have application after the ad bellum violation, it need not follow that they enjoy any specifically contractarian provenance. I will have much more to say about these matters in Section III.
Before I get there, I need to mention other fallback possibilities. Perhaps, one might think, C is still regulated by the contract, in the sense that C owes it to A to comply with the terms of the contract. B has not directly aggressed against C, unlike A. But even this seems doubtful. If there is a question mark over whether we can reasonably expect A to accept that it is still governed by its contract with B, what stops C from coming to a similar view? B has violated the terms of a contract that it originally held with A and C, even if there is a narrower sense in which B has aggressed only against A. In any case, we should expect the war contract to retain its full force for each of A, B and C, if the contract does genuinely continue to hold.
The main critical challenge, then, is this: how can a major breach of the contract be rendered consistent with the endurance of the contract? If B has invaded A, then why should A and C be required to respect contractual terms with B, as opposed to falling back on precontractual moral rules of engagement in their subsequent dealings with B?
Now Benbaji and Statman go out of their way to situate their version of the war contract in a world of limited sympathies and moral ambitions. They are not assuming a rosy-coloured picture of the moral or political world. This is a contract meant for our world, warts and all. Yet there is still a problem, because a contract is still a contract. However unenviable the conditions in which the parties to the contract agree to it, the parties to whom it applies must be construed as those who are not in substantial breach of it. We can certainly make the terms of the contract less ambitious to suit the various moral limitations or limited sympathies of the parties governed by it. It would be a serious mistake to expect too much of them, given their character and the situations they are in. But the contract itself should work like any other contract, where major material breaches of it are consequential, and serve to release nonoffending signatories from the contractual obligations these signatories would owe to the other parties in conditions of full compliance.
Perhaps there would be residual pressure to act within the terms of the contract if there was no real moral alternative to the terms specified by the contract. But Benbaji and Statman have already admitted that there is a richly characterized pre-contractual morality that should function as the moral default. When contracts lapse, then the original parties to those contracts should then presumably revert to the detailed principles offered by this pre-contractual morality. To do so, however, will leave us with squarely revisionist resources, and no obvious way of resisting Continuity.

III. DOES THE WAR CONTRACT NEVER REALLY COLLAPSE?
How, in more detail, do Benbaji and Statman characterize the significance of breaches of contract? They tend to depict the contract as being peculiarly resistant, or super-elastic, and so as able to reconstitute itself as soon as it has been breached: … the right granted to states to wage war against violators of their territorial integrity is not a result of the collapse of the war contract -of a withdrawal, so to speak, to the state of nature. Rather, it is a term in the contract itself. The victim of the aggression may address it by weakening the aggressor's army in a way that blocks the aggression, restores any violation of its territorial integrity, and deters the aggressor from further violations of the prohibition in the future. We suggest that the same applies at the in bello level… the war contract never really 'collapses '. 13 This point is repeated elsewhere: The normative implications of breaching the war contract are themselves part of the contract. 14 Breaches of the war contract never push the parties back to the state of nature, under the umbrella, as it were, of pre-contractual morality. Rather, the parties to the contract agree not only on the rules regulating war, but also on the rules regulating the responses to violations of these rules. 15 An insistence on the resilience of the war contract risks being an exercise in table-thumping unless Benbaji and Statman can explain away the conceptual problem uncovered in Section II. Contracts do not standardly endure when there are major breaches of them. To suggest that there are exceptions to this truth that are explicable by the peculiar super-elasticity of a contract does not explain how any contract can acquire such super-elasticity.
Benbaji and Statman do not, in fact, have anything directly to say about how the war contract can acquire such super-elasticity. Their attention and critical energies are focused elsewhere. What primarily drives their argument is the concern that, if the terms of the war contract are suspended, then outcomes are likely to be substantially worse: there will be dangerous tendencies towards escalation, reprisal, and a movement towards total war. We need to dig deeper. As I interpret it, the heart of Benbaji's and Statman's reasoning is implicitly governed by the following principle: Intactness: If there are morally compelling reasons, based on the likelihood of severe future losses and rights violations, for not reverting to pre-contractual morality following a major breach of a contract, but for continuing instead to be governed by the terms laid out in the contract, then the contract can be considered intact.
Though they do not explicitly espouse Intactness, Benbaji's and Statman's commentary on these matters suggests their commitment to it, or at least a principle in its immediate vicinity. Imagine, as before, that the ad bellum contract has been breached by B, who has unjustly attacked A. Since aggression has already broken out, we are now dealing with the part of the war contract concerned with jus in 13 War By Agreement, pp. 164-165; emphases added. 14 War By Agreement, p. 179. 15 War By Agreement, p. 189.
bello. The revisionist tendency is to deny permissions to the unjust side, and to continue to affirm, under the in bello dimension, the asymmetry of moral status between the just side and the unjust side that was established by the ad bellum evaluation. But Benbaji and Statman contend that the ascription of wrongdoing to B's combatants will only make the outcome worse, especially if the ascription of wrongdoing is then partnered by the further ascription of criminality. In the absence of a symmetrical legal status between A's and B's combatants, the standing conditions for deterrence would be impaired, because states could no longer count on obedient armies. The provocations for intensified conflict would also be fuelled, either because each side would judge itself to be in the right, or because each side would unscrupulously unleash enough propaganda and misinformation to sustain such a conviction among its armies and citizenry: … it is in the interest of individuals who live in a minimally just symmetrical anarchy to be protected by states that control obedient armies. Any asymmetrical restrictions, like a right to fight just wars only, or a right to kill only Unjust Combatants, would undermine the main objective of the contract, which is to enable states to efficiently address ongoing aggression and to deter potential aggressors from unlawful use of force. Asymmetrical rules would compromise the obedience of combatants and thus the ability of states to act in self-defence. 16 In wars governed by… asymmetric rules, each side would regard itself as entitled to retaliate for what it takes to be the enemy violation of the in bello code: each side would take the other as violating the rule that allows killing only Unjust Combatants, say. Such retaliations would aggravate the apparent injustice in the eyes of the other, which would lead to more violence, and so on, in a dangerous spiral. 17 These passages, which are centrally concerned with the Mutual Benefit condition, state the main lines of the Benbaji-Statman argument in compressed form. For my purposes, I need not quarrel with them, because it is the conditional structure of Intactness in which I am interested. Even if these considerations provide us with decisive reasons for treating the combatants on both sides of a conflict as legally and morally symmetrical, does it follow that the contract has endured? My fear is that this inference is a non sequitur.
True, Benbaji and Statman strenuously argue that there are compelling reasons for not falling back on pre-contractual rules of engagement. These reasons for not being guided by the principles of pre-contractual morality demonstrate, for them, that the contract must have endured through breaches of it. The endurance of the reasons for sticking to the terms of the contract must therefore, in turn, explain the contract's super-elasticity. But this line of thought, which plainly reflects their allegiance to a principle like Intactness, is not to be trusted. Even if we accept the force of these reasons after the initial violation of the contract and agree that unmediated guidance by pre-contractual morality would make the outcome morally worse, it still does not follow that the contract has endured. What it shows instead is that original parties to the contract ought to proceed as if the contract still holds, in order to minimize casualties and avoid an escalation to all-out war. The relevant lesson is that the considerations arising from the combination of Mutual Benefit, Fairness, and Actuality can still be reason-giving, even if they can no longer claim a contractarian basis.
I now want to make some further points about the role played by Actuality in particular. Perhaps Actuality is meant to paper over the conceptual problems arising from violation of the contract by encouraging the following thought: the contract endures, despite violations of it, because the rule-governed practice of war endures, and this is just what we would expect if Actuality belongs to the set of conditions that need to be satisfied if the contract is to be upheld. On this view, it will not be possible to draw any sharp distinction between the endurance of the contract and an 'as if' interpretation of the contract. If the parties' actions subsequent to the violation of the contract continue to conform to what the contract apparently tells them to do in these circumstances, then this supports Benbaji's and Statman's contention that this contract is indeed a self-repairing one that can survive its own violation. Now Actuality is already playing an important double role in the Benbaji-Statman system. First, it provides a mechanism demonstrating what acceptance of the contract consists in; and second, it explains the thought that the war contract must actually be followed by most of the parties to it at most times in order to be able to claim any party's ongoing allegiance. The considerations rehearsed above envisage a third role for Actuality: this condition can also help to demonstrate that the contract is a self-repairing or self-renewing one.
I doubt, however, that Actuality can deliver on this promise. We cannot insert into the war contract the stipulation that this is a self-healing contract. Contracts are such that they ought to be followed, and contracts dissolve upon major breaches of them. These are general conceptual truths about contracts, and Benbaji and Statman have not repudiated them. There are two more particular points to make about this issue.
First, if one of the functions of Actuality is to show that the contract is self-mending, then Benbaji and Statman face the embarrassment that breaches of the contract actually offer crucial evidence that the contract has been dissolved, for the simple reason that Actuality is no longer satisfied. The original significance of Actuality is that it tells us that the war contract endures when it is actually heeded. Accordingly, if the contract is breached, then we have substantially less reason to think that it can have endured in this particular case. Second, the only reason that the parties conform to what the contract tells them to do after the contract is breached is due to the costs of not doing so. The normatively binding role played by free acceptance of the contract has now lapsed, simply because the contract has been breached. If the parties nonetheless continue to act in ways specified by the contract, their reasons for acting in these ways cannot be because this is what the contract tells them to do, but because they wish to avoid certain losses. The parties act, in other words, as if the contract still endures. This falls short of demonstrating that the contract genuinely endures, or has proven itself to be, unusually, a self-repairing one.

IV. THE 'AS IF' WAR CONTRACT AND CONSEQUENTIALISM
I have suggested that the war contract no longer endures when there is a substantial breach of it. Rather, and at best, the parties ought to act as if the contract still holds. In particular, the parties ought to act as if the contract still holds if the considerations associated with Mutual Benefit still obtain.
The main point here is that, if an unjust war has begun, it is better for it to proceed in an orderly way, as disciplined by symmetrical in bello rules, than to embrace asymmetric moral and legal standards. According to the logic of the revisionist asymmetrical view, unjust combatants act wrongfully, and so there is at least a presumptive case (if not a decisive case, taking all relevant factors into consideration) 18 for labelling them as criminals. This might lead the unjust side to be inattentive to civilian casualties, if everything the unjust side does is deemed wrongful, and to pursue victory at all costs, as victory may offer its only route for escaping criminal prosecution. Moreover, each side will be tempted to claim the moral high ground, thus encouraging the formation of contemptuous, heavily moralized attitudes to both combatants and non-combatants alike on the other side. The likely result is further escalation and reprisals and an increasing erosion of respect for civilian immunity.
I want to comment on two upshots of this adjusted understanding of the Benbaji-Statman account in the final sections of this article. The first of them, addressed in this section, is concerned with the normative basis of their account, as it should now be understood, and in particular the possible connection between this normative basis and consequentialism. The second of them, covered in Section V, is concerned with the relationship between the Benbaji-Statman account and certain articulations of the revisionist project.
As we know, Benbaji and Statman think that the force of their account consists in the parties' free acceptance of a contract. Even if these parties are motivated to accept the war contract because of the prospect of the benefits it offers, or the losses it avoids, it is not the benefit-loss sheet itself that endows the contract with normative authority. Now I have argued that the war contract does not endure when it has been violated. So, if the parties have reasons to act as if the contract still obtains, that can only be due to the contents of the benefit-loss sheet. 19 But it is then tempting to conclude, once the relevant considerations are detached from the normative force of contractarianism, that it must be a species of consequentialism, or at least quasi-consequentialism, which is now taking the normative strain. In particular, the normative basis of the 'as if' contract is plausibly constituted by something like a 'consequentialism of rights', such as the one advanced by Amartya Sen. 20 By acting as if the contract still held, the warring sides will end up violating fewer rights and causing less damage to lives, property, and landscape. 18 See, further, Section V. 19 Subject, as usual, to the further conditions stated by Fairness and Actuality. 20 See Amartya Sen, 'Rights and Agency', Philosophy and Public Affairs 11 (1) (1982): pp. 3-39.
One of the features of such consequentialist rules is that we cannot immediately count on their invariance. If we can refine the rules to achieve better outcomes, then there is a pro tanto reason to do so. There are, of course, limits to how elastic the rules can be, and there may be benefits in making certain rules, including the rules of war, relatively hard-edged or invariant. These are issues much debated among consequentialists. 21 However, consequentialists will not dogmatically insist that the rules, once established, are forever fixed. There must at least be a readiness among consequentialists to adjust and refine existing rules in order to promote better outcomes. If so, there will then be some novel theoretical pressure on Benbaji and Statman to come to terms with the problematic independence of jus ad bellum from jus in bello. Having obedient armies may indeed be a source of benefit, because the existence of obedient armies deters future aggression. But more than one consideration is in play if we are compiling a list of benefits. Having armies that refuse to follow orders to engage in unjust aggression may also be a source of benefit, because combatants' refusal to participate in what they reasonably judge to be unjust wars will ensure that those unjust wars are not actually fought. 22 Benbaji and Statman do provide reasons for thinking that we cannot count on the refusal of morally conscientious combatants to obey unjust orders, because of the presence of propaganda and misinformation exploited by morally unscrupulous regimes. But the contractarian cast of their argument is to some extent protected against these contingencies, because, as contractarians, they are not interested in securing optimal outcomes. Real world outcomes only have to be good enough. The non-consequentialist framework they adopt helps to shield them against these theoretical pressures towards optimality.
It will be instructive, in this connection, to examine Benbaji's and Statman's reasons for distancing their war contract from a more explicitly consequentialist theory of war. Of course, my argument 22 Indeed, that no wars will be fought, since wars are always unjust on at least one side of them. Even if a global refusal to fight wars remains a remote possibility, there may at least be fewer unjust wars if refusals to fight are deeply embedded in soldiers' professional self-conception. that their war contract can exist only in an 'as if' form is not one that they endorse or have anticipated. As it happens, however, they also provide a number of substantive reasons for maintaining their theoretical distance from rule-consequentialism, and it seems to me that these reasons prove unreliable upon further examination. 23 Benbaji and Statman enumerate four differences between contractarianism and rule-consequentialism. 24 First, contractarianism, unlike rule-consequentialism, does not offer a complete theory of morality, and assumes a background picture of pre-contractual rights and duties. Second, and relatedly, contractarianism places fundamental importance on rights, unlike rule-consequentialism. Third, contractarianism is not a maximizing theory. Benbaji and Statman explain that 'a set of social rules is mutually beneficial if the state of affairs where the rules are followed is Pareto superior to a state of affairs in which the relevant parties tried to follow pre-contractual morality'. 25 That is a good enough target for contractarianism. Contractarianism aims at improved outcomes but not optimal outcomes. It is therefore under no special pressure to make the rules as good as they can possibly be. Fourth, contractarianism has a different normative foundation, according to which 'the fact that rules are mutually beneficial validates the presumption that they are freely (as well as tacitly) accepted, and the fact that they are tacitly accepted entails a waiver of rights'. 26 The fourth of these features need not be questioned, but I have already argued that it does not apply to the 'as if' contract. The first and second of these features emphasize the point that the contractarian account is not consequentialist all the way down, and that therefore it does not count as fully rule-consequentialist, since it builds on the pre-contractual common-sense picture and is applicable only to war. 27 These points do not eject rights-consequentialism 23 One additional relevant consideration about rule-consequentialism, in particular, that goes unmentioned by Benbaji and Statman is that the standards of rightness and wrongness in rule-consequentialism standardly assume full or at least nearly full compliance, which means that the favoured body of rules may be considerably sub-optimal in conditions of partial compliance. Benbaji's and Statman's reliance on Actuality therefore suggests that an adjusted quasi-consequentialist interpretation of their war convention should not take a specifically rule-consequentialist form. 24  from consideration as a normative basis of the 'as if' contract, however, for rights-consequentialism is similarly hybrid: it wishes to build a consequentialist structure around the protection of rights.
That leaves us with the third feature, concerning sub-optimality. But if, at this stage, it is only the sub-optimality that is keeping the theories apart, then it is natural to suspect that this may be little more than a self-serving stipulation that does not create genuine critical distance between the revised understanding of the Benbaji-Statman account and consequentialism. We should be suspicious of reasoning that unfolds in the following way: (a) contractarian theories are not optimizing theories; (b) the revised Benbaji-Statman account still counts, in virtue of its sub-optimality, as a residually contractarian theory; and thus, (c) as a theory of this sort, it is under no theoretical pressure to aim at optimality.
My quarrel here lies with (b), rather than (a). Even if sub-optimality is one of the features that can differentiate contractarian theories from consequentialist theories, it is a less reliable basis for classifying a theory as contractarian rather than consequentialist when the other theoretical features of contractarianism are no longer in play. In Benbaji's and Statman's case, if the reasons for acting as if the war contract still holds are sensitive to the demand that rights violations be reduced, and if at this point nothing is competing in justificatory space for the reduction of rights violations, then it is not clear why they should be entitled to refuse the further burden of attempting to minimize them, as opposed to simply reducing them. Why should 'good enough' be good enough in this context?
Once the relevant justificatory considerations are detached from the normative force of contractarianism, it seems to be consequentialism, as encapsulated in Mutual Benefit, which is taking the normative strain. The reasons Benbaji and Statman advance for making contractarianism critically distinct from consequentialism dissolve under further scrutiny. When they squarely acknowledge that consequentialist grounding, the reassuring inflexibility of these rules will seem less secure. The high stakes character of war will make it more problematic for combatants to accept their obedient roles in an unquestioning way. Benbaji and Statman may still be protected, of course, by the substantive arguments they offer for moral and legal symmetry at the level of in bello. However, they can no longer claim protection from the very structure of the theory they advance.

V. REVISIONIST SURFACES
In conclusion, I want to point to certain implications arising from the theoretical classification of the Benbaji-Statman account when it is understood in the way I favour. In revisionist literature, there are usually concessions made to the practice of war. In particular, the asymmetrical 'deep morality of war', in which failures at the ad bellum level then make it morally impermissible to proceed any further, is in some sense supplanted by a more symmetrical depiction of the status of combatants when we move from moral matters to legal matters.
A prominent example of this tendency is Jeff McMahan's distinction between the 'deep morality' of war and the 'laws of war'. 28 McMahan sensibly maintains that the laws of war should 'mitigate and contain the destructive effects of war rather than exacerbating them'. 29 McMahan's primary concerns here are to avoid exercises in 'victor's justice' and perverse incentives against surrender, and to curb tendencies towards total war. To avoid such unhealthy incentives for prolonging the war, and to allow the post-bellum situation to be managed in a more orderly fashion, McMahan is pushed in the direction of the 'legal equality of combatants' doctrine. One cannot fail to notice that these concerns strongly overlap with the substantive considerations Benbaji and Statman press into service under the heading of Mutual Benefit. Now Benbaji and Statman have sensible and penetrating points to make about the stability of this revisionist distinction between 'deep' morality of war and -presumably, to fill out the relevant contrastthe less morally deep legal regulation of war: 30 28  What McMahan really has in mind is a distinction between two types or two levels of moral considerations; those that are in some sense 'deep', probably having to do with fundamental human rights, and those that are not deep -maybe shallow is the appropriate term -mainly involving consequences. He concedes that there are moral reasons to comply with the current war convention, though they are, in a sense, shallow. As far as the deep morality of war is concerned, such compliance is nonetheless problematic. 31 As Benbaji and Statman forcefully argue, the metaphors of 'deep' and 'shallow' seem out of place, since the (merely) 'shallow' considerations to do with the consequences of legal asymmetry are still morally compelling. If the symmetrical legal regulatory regime is one that we ought, morally, to uphold, then it is as deep as it needs to be. It is deep enough. In the war context, moral 'oughts' are coextensive with legal 'oughts': these are the standards that ought, morally, to govern fighting in war.
Does that mean that there can never be any meaningful gap between ('deep') moral standards and morally decisive legal standards? By no means. Examples abound of how these standards might come apart. For example, it is perfectly coherent to think that abortion is morally wrong and yet morally favour, all things considered, a permissive legal framework for regulating abortion. Fairly clearly, the fact that the law makes it permissible to have an abortion and that there are morally compelling reasons for operating with this permissive legal system does not settle the case for the moral permissibility of any particular individual's decision to seek an abortion.
Similarly, one might think, the fact that unjust combatants do not face criminal penalties for fighting does not settle the case for the moral permissibility of their involvement in the first place. But the problem actually goes deeper than that. If unjust combatants do decide to fight, then they are bequeathed with a status, approved by morality, that makes it permissible for them to fight. Symmetrical legal regulation is predicted to have better consequences than asymmetrical moral regulation, but these better expected consequences ultimately ensue, in part, from unjust combatants' entitlement -their morally approved entitlement -to think of themselves, and to be regarded by others, as combatants who are fighting permissibly in a rule-governed framework. (There is no exact analogy in the abortion case.) 31 War By Agreement, The revisionist question of how unjust combatants can succeed at in bello while representing a side that has failed at ad bellum remains a pressing one. But lines are blurred when symmetrical regulation is introduced, because, in deciding whether to fight, unjust combatants will also be in a position to know that, if they do decide to fight, then their fighting cannot be significantly quarrelled with on moral grounds.
These issues run deep. There is much more to say about them. But there is also more to say about the theoretical destination of the adjusted Benbaji-Statman account. For once we dispute the straightforward tenability of the revisionist distinction between ('deep') asymmetrical standards and ('shallow') symmetrical standards, and also dispute the application of genuinely contractarian reasoning to Benbaji's and Statman's picture, then there is at least the possibility that the Benbaji-Statman account and revisionist account are going to meet in the middle. The possibility that the deep structure of the Benbaji and Statman account is actually isomorphic with the application of revisionist theory to the practice of war may seem perplexing, but it is the conclusion to which we seem headed. I do not see how it can be ruled out. 32

OPEN ACCESS
This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from 32 This article started out as a contribution to the Nuffield Political Theory Workshop on Yitzhak Benbaji's and Daniel Statman's War By Agreement in December 2020, organized by Cécile Fabre and Janina Dill. I thank the contributors that day, especially Yitzhak Benbaji and Daniel Statman, for their various helpful comments, as well as Anat Shapira for enlightening subsequent email correspondence. I'm also grateful to Janina Dill and Cécile Fabre for their very helpful, detailed comments on the initial full drafts of this article, and an anonymous reviewer for similarly helpful, detailed comments on a later draft.