I. Introduction

How can law succeed in “mak[ing] us do things we do not want to do” (Schauer Reference Schauer2015, 1)? When legal compliance conflicts with our all-things-considered judgments about how best to behave, why would we ever choose to obey the law? And if law frequently does conflict with our law-independent interests, how is any kind of legal order possible? These are the rather fundamental questions raised by Schauer (Reference Schauer2015) and McAdams (Reference McAdams2015) in their well-matched books.

One approach to answering these questions is to posit an intrinsic preference for legal compliance. Perhaps people feel a moral compulsion to follow legal rules or to comply with the dictates of procedurally or substantively legitimate legal decision makers (see Schauer Reference Schauer2015, Ch. 5; McAdams Reference McAdams2015, 3–4). No doubt there is some truth to this view, but there is also reason for skepticism. The problem is not just that lots of human behavior appears to be motivated by amoral, narrow self-interest. The larger problem, emphasized by Schauer, is that even the most morally motivated, other-regarding types may have little interest in legal rectitude. Those who are inclined to do the right thing will not necessarily prioritize the rightness of legal compliance over other moral imperatives. Even, or perhaps especially, the most virtuous among us may well approach the law like Holmesian “bad men.”

The project that both Schauer and McAdams take on is to explain how people who do not place a particularly high priority on legal compliance for its own sake might nevertheless mange to sustain some tolerable level of legal compliance and social order. Schauer and McAdams develop two different explanations, each of which seeks to reconcile legal compliance with individual interest. Schauer emphasizes the vertical threat of state-imposed sanctions for legal noncompliance, or state coercion. McAdams emphasizes the benefits of coordination with other citizens and the horizontal threat of socially imposed sanctions for failing to coordinate.Footnote ¹

II. Inevitability

For some purposes, social coordination and state coercion might be usefully understood as substitutable mechanisms for encouraging legal compliance. Either might work, for example, to induce drivers entering a traffic roundabout to yield—in order to avoid collisions with other drivers who have coordinated on the yield rule or in order to avoid a traffic ticket or prison sentence.

From the systemic perspective of establishing legal order, however, the two mechanisms do not operate on a par. Simply put, coordination is more fundamental than coercion. This is because state-imposed coercion cannot exist without the prior success of coordination. After all, the possibility of state-imposed coercion presupposes an organized state with the capacity to promulgate legal norms and back them with force. As Schauer recognizes, mobilizing coercion requires organized collective action that will typically rest on a foundational “noncoerced cooperative compact” (2015, 85).Footnote ² How does such a compact come into being when there is no one to “coerce the coercers” (77)?

One way of appreciating the force of this question is to consider legal regimes that cast the state as the subject rather than the source of law. International law and constitutional law are two such regimes that create “law for states” (see Goldsmith and Levinson Reference Goldsmith and Levinson2009). To the extent these legal regimes effectively control and constrain the behavior of states, they do so despite the absence of any super-state enforcement authority capable of bringing to bear coercive sanctions. If ordinary domestic law, administered and enforced by states, can easily be conceived as “the gunman writ large” (see Schauer Reference Schauer2015, 159, quoting H. L. A. Hart), international and constitutional law, imposed upon states, raise the question of why “people with guns obey people without guns” (see Goldsmith and Levinson Reference Goldsmith and Levinson2009, 1796, quoting Stephen Holmes).Footnote ³

Of course, one possibility is that the people with guns do not obey. There is a long tradition of skepticism about the force of force-less law, dating back at least to Hobbes's dismissal of “[c]ovenants without the sword” as “but words” (Schauer Reference Schauer2015, 81, quoting Hobbes [1651] Reference Hobbes and Waller1935, 115). In the context of the US Constitution, James Madison similarly dismissed rights and other legalistic limitations on powerful political actors as merely “parchment barriers” (Rossiter Reference Rossiter1999, 305). Hobbesian and Madisonian skepticism remain robust in both the international and constitutional arenas. Thus, generations of international relations realists have questioned whether international law is a meaningful constraint on state behavior as opposed to an instrument through which powerful states get their way. Political scientists and constitutional theorists have likewise, albeit less commonly, questioned the efficacy of constitutional constraints, pointing to examples such as President Lincoln's arguably illegal actions during the Civil War or President Roosevelt's defiance and defeat of the old constitutional order during the New Deal; or, more broadly, to the fact that the content of judicially interpreted constitutional law seldom departs from the preferences of national majorities or dominant political coalitions. Schauer himself casts some doubt on the willingness of government officials “whose constitutional obligations seem typically to be unsupported by sanctions in the usual sense” to be bound by law, pointing to a number of cases in which officials have ignored or finessed legal rules to take actions that were politically popular with their constituents (Schauer Reference Schauer2015, 87, 89–92).

But Schauer does not believe that all international and constitutional law rules amount to nothing more than parchment barriers or mere words. To the contrary, he recognizes the efficacy of international institutions such as the WTO (2015, 136). He also appreciates the pervasive efficacy of constitutional law in “establishing the very structure of government” and in “regulating and constraining” the behavior of public officials (88). Certainly in the US constitutional context, at least some degree of efficacy is difficult to deny. It is one thing to raise questions about, for instance, whether the president is meaningfully constrained by constitutional limitations on his war powers.Footnote ⁴ But asking that question presupposes the existence of a constitutionally recognizable president, elected in the constitutionally prescribed manner and possessing widely agreed-upon institutional structures and powers and operating within a widely agreed-upon system of political organization and decision making. We do not typically worry about the president canceling elections, ordering the military to take over Congress, and ruling by decree; even the most popular presidents do not run for third terms or directly defy the Supreme Court. When we foreground particular constitutional rules to question their constraining force, we do so against a background of unquestioned constitutional constraint.

What accounts for these constraints? Here again, we might explain compliance with noncoercive law by hypothesizing some sort of intrinsic preference for legal compliance. This is the approach taken by international lawyers and international relations theorists of a “constructivist” bent (Ruggie Reference Ruggie1998; Wendt Reference Wendt1999), who hypothesize that international law “socializes” states into compliance (Goodman and Jinks Reference Goodman and Jinks2013), or “exerts a pull toward compliance” stemming from its perceived legitimacy (Franck Reference Franck1990, 14), or becomes “internalized” by states through their domestic legal processes (Koh Reference Koh1997, 2602). Many constitutional theorists similarly assume that the hyper-democratic pedigree of the US Constitution or the principled decision making of justices and other constitutional interpreters will generate a kind of legitimacy that fosters compliance (see Goldsmith and Levinson Reference Goldsmith and Levinson2009, 1839). Whatever the merits of these explanations, all of them jettison the bad man assumption that Schauer and McAdams want to hold constant.

Theorists seeking bad-man-compatible explanations for legal compliance at the international and constitutional levels inevitably turn to explanations based on broad or strategic self-interest. At the international level, these explanations include institutionalist theories of international relations and rational choice theories of international law that make use of game-theoretic models—the familiar array of prisoners' dilemma, hawk-dove, and other mixed-motive game forms surveyed by McAdams—to account for legal compliance and other kinds of cooperative behavior by states (see Goldsmith and Posner Reference Goldsmith and Posner2005; Goldsmith and Levinson Reference Goldsmith and Levinson2009, 1826–28). Political scientists and legal theorists have applied a similar set of models to explain constitutional compliance, showing how powerful political actors might be willing to sacrifice their short-term interests in exchange for the broader or longer-term benefits of coordinating or cooperating around constitutional rules and institutions (see Goldsmith and Levinson Reference Goldsmith and Levinson2009, 1835–36).

While this is not his primary focus, Schauer seems content to embrace game-theoretic accounts along these lines as foundational to the creation of a legal system. In discussing the “foundations of law,” he recognizes that legal systems “typically rest on the decisions of multiple individuals to create a common enterprise for what they perceive as their own collective good,” relying not “on sanctions but on the more complex dynamics of human interaction that lead to cooperative or coordinated behavior” (2015, 82, 85, 81). Once a system of government has been coordinated into place, “those who are at the apex” can then mobilize coercion as necessary to bring the rest of the population into line (85). Schauer's emphatic point is that these top-down sanctions may be necessary and important to the successful operation of a legal system. As he clearly recognizes, however, sanctions can only come into play after, and as a result of, prior game-theoretic coordination or cooperation.

On this fundamental point, Schauer and McAdams seem entirely in accord. For his own part, McAdams offers a stylized, Hobbesian account of the game-theoretic origins for foundational legal order, in which the benefits of cooperating to produce public goods—“[b]uilding roads, fighting crime, regulating pollution, creating currency, waging war”—induce people to coordinate on leaders and governments who are empowered to make and enforce law (2015, 61). McAdams proceeds to describe how compliance with constitutional rules and rights might be sustained through the repeat-play interactions of groups of citizens who use the constitution to coordinate the policing of government officials pursuant to a mutually beneficial cooperative equilibrium (72–73).Footnote ⁵ “Once we have robust legal institutions,” McAdams explains, “the state can usually credibly threaten individuals with sanctions for violating the law” (62). But McAdams fully appreciates that the formation of a state or government or legal system capable of mobilizing sanctions rests on a foundation of game-theoretic coordination or cooperation (71).

Schauer and McAdams thus agree that a society of bad men (or, in the international arena, bad states) who have no intrinsic interest in complying with or enforcing the law might nonetheless decide to do so on account of the instrumental benefits of reciprocal cooperation and coordination on a system of law and government. The simple and seemingly inevitable intuition is that, at bottom, legal regimes are capable of constraining powerful and self-interested actors only because, and to the extent that, the costs of constraint are outweighed by the benefits of legal order accruing to these actors. To the extent the bad man view of human (and state) motivation is correct, game-theoretic explanations of legal order may be the only game in town.

III. Indeterminacy

Once we embrace game-theoretic explanations of foundational legal order, however, it becomes less clear why we need the kinds of mechanisms for securing legal compliance emphasized by Schauer and McAdams, or what role those mechanisms play in building or sustaining a legal system. Again, neither state-imposed sanctions nor localized coordination around specific legal rules would exist at all without the prior success of coordination or cooperation in creating a political and legal system capable of promulgating and enforcing authoritative rules. But once we suppose successful coordination at this foundational—or constitutional—level, why is that not sufficient to explain compliance all the way down, eliminating any need for coercive sanctions or localized coordination? Why is compliance with a legal system not a package deal, one in which actors make a single, all-inclusive calculation to comply with both constitutional law and with whatever subconstitutional laws are promulgated by the constitutionally specified authorities?

As McAdams puts part of the point, coordination at the “foundational” level of a legal system might obviate the need for coordination at the localized, or “marginal” level: “A skeptic might therefore say that … the theory [of coordination] explains how people create a basic order and how that order imposes sanctions, but that, once those foundations exist, the rest of the work is done by sanctions without the further influence of focal points” (2015, 76). But—and here is the other part of the point—a skeptic might also wonder why sanctions are necessary after foundational coordination has been established. Once a critical mass of people in society makes the wholesale calculation that the benefits of legal order outweigh the costs of sacrificing law-deviant interests, why is this not sufficient for a stable, cooperative equilibrium of legal compliance to take hold systemically, from top to bottom?

In fact, Schauer and McAdams each point to a number of contexts in which game-theoretic coordination and cooperation seems to work in just this way—operating at the wholesale level of an entire system of rules, and eliminating any need for sanctions or localized coordination to sustain compliance with individual rules within that system. Schauer's chess (2015, 34) and cricket (160) players, for example, seem to coordinate on the rules of their games as a complete package. Chess players do not make game-theoretic calculations about compliance at the level of specific rules, such as whether a pawn can be moved two squares or one. Choosing to play the game means playing by all the rules; break a single one and the entire game falls apart. Or consider McAdams's example of parliamentary procedure manuals like Robert's Rules of Order, which even obstreperous law faculties have been known to embrace (2015, 113). It is doubtful that a law faculty could sustain game-theoretic coordination or cooperation on each individual Robert's rule assessed in isolation. Just to give one example, loud and obnoxious faculty members would find no reason to comply with the rule against interrupting speakers who have the floor. Nonetheless, those same faculty members, and all the rest, have sometimes proven themselves willing to comply with the anti-interruption rule as part of an aggregated bundle—evidently making game-theoretic calculations not on a rule-by-rule basis but at the level of a systemic package of rules.

On the other hand, Schauer and McAdams also give a number of examples of cooperation or coordination around discrete rules, considered in isolation from any larger systemic package. For instance, Schauer points to Robert Ellickson's (Reference Ellickson1991) game-theoretic accounts of social norms among neighbors governing specific interactions, such as building fences or compensating for cattle-trespass harms (Schauer Reference Schauer2015, 142). Ellickson's neighbors do not attempt to coordinate on broader codes of conduct dealing with other kinds of interactions. (Then again, Schauer also invokes Lisa Bernstein's [Reference Bernstein1992,Reference Bernstein2001] examples of the comprehensive and institutionalized systems of norm-based governance adopted by the diamond and cotton industries, where the cooperative benefits of compliance are evidently assessed at the level of the entire, systemic regime rather than at the level of specific, individuated rules [Schauer Reference Schauer2015, 142, 224 n7].) McAdams's paradigmatic examples of coordination also involve discrete rules, such as those telling people where they can smoke or when to yield to other drivers.

McAdams takes a similar, one-rule-at-a-time view of international law. He explains treaties that set territorial boundaries or standards as supported by issue-specific coordination based on the localized benefits of conflict avoidance, communication, and trade, without regard to any broader, systemic commitment to international law (2015, 68–70). He presents evidence along the same lines that states assess International Court of Justice (ICJ) decisions on a case-by-case basis, choosing to comply where there are coordination benefits to treating the court's decision as focal but otherwise not (120). Here again, the unit of analysis for game-theoretic calculation is a single rule or ruling, assessed independently of any systemic package.

Between “top-down” coordination at the systemic level and “bottom-up” coordination at the level of discrete rules, many intermediate possibilities exist. For example, in contrast to McAdams's finding that state compliance with ICJ decisions is assessed on a decision-by-decision basis, other theorists of international law and relations deploy the very same game-theoretic logic to explain the willingness of states to abide by all the decisions of one or another international tribunal, or with a disparate set of rules that have been linked together in a treaty (Goldsmith and Posner Reference Goldsmith and Posner2005, 153–54). In this view, equilibria of coordination or cooperation are established not at the level of specific rules but at the level of larger bundles. Indeed, larger bundles may be necessary to make mutually self-interested cooperation or coordination possible. Developing states that are disadvantaged by strong intellectual property rights may nonetheless agree to comply with a treaty or governance regime that links intellectual property to free trade (see Guzman Reference Guzman2008, 165–66). Warring states may be able to sustain a cooperative equilibrium of reciprocal compliance with the Lieber Code or the international laws of war when no such equilibrium would be possible at the level of any specific rule—much less at the level of compliance with the jus ad bellum rules that would have prevented war in the first place.

Recognizing that game-theoretic calculations of mutual self-interest can be applied to rule packages of various sizes—up to and including legal systems as a whole—raises the question of how analysts and players determine which size will actually prevail. Consider Robert Axelrod's (Reference Axelrod1984, 73–87) classic game-theoretic explanation of “live and let live” norms that developed during World War I between German and British forces faced off in the trenches of the Western Front. These norms specified times and places when shooting and shelling would be suspended. During the informal Christmas Truce of 1914, for instance, soldiers from both sides met in no man's land to exchange gifts. Axelrod's explanation for this odd state of affairs is that British and German soldiers, playing a strategy of tit-for-tat in an iterated prisoners' dilemma game, were able to sustain norms of nonaggression as a cooperative equilibrium. Yet we might wonder why the very same game-theoretic logic failed to generate a peaceful equilibrium on a larger scale. If British and German soldiers could sustain cooperation at the level of the Christmas Truce, why not at the level of laying down arms altogether, or avoiding war in the first place? Or, moving in the other direction, why was the cooperative equilibrium not limited to the Christmas Truce as a one-off arrangement, detached from the more expansive set of live and let live norms that Axelrod takes as his unit of analysis?

Predicting or explaining the scope of cooperative equilibria seems to be a quite general, if little noticed, indeterminacy in game-theoretic explanations of cooperative behavior. It is related to, but importantly distinct from, two more familiar kinds of indeterminacy in game-theoretic analysis. One is determining whether a cooperative equilibrium will prevail at all, given the possibility of multiple equilibria. Even if we can predict or explain the emergence of a cooperative equilibrium, there is the further question of what the scope of that equilibrium will be. That question goes not just to the choice of solutions in a set derived from a predetermined game form, but to the determination of the game form itself, specifying the players, choices, and payoffs. A second well-known indeterminacy in coordination analyses relates to which of many conceivable focal points for coordination (if any) will emerge as salient (see McAdams Reference McAdams2015, 42–48). Indeterminacy of cooperative scope might be understood as a distinct variant of this question, focused on the dimension of size: Will coordination occur around a focal pinpoint, a focal beach ball, or an even larger focal sphere?Footnote ⁶

If cooperative equilibria can come in all shapes and sizes, so, too, can game-theoretic accounts of legal order. Legal compliance can be explained at the level of particular rules or interactions, for example, by modeling a discrete game around the isolated decisions of drivers to yield. But compliance with specific traffic rules can also be explained by modeling a game around the entire legal system, or around some intermediate component, such as the rules of the road or the institution of a traffic court. How do we, or the relevant players, know which of the many possible game forms to choose?

To make matters more complicated, when it comes to legal order, it seems likely that multiple games are being played at once. McAdams and Schauer seem to share a view of how coordination and cooperation contribute to legal order in the context of a well-developed legal system like that of the United States. Both envision a two-level game. At the highest, or constitutional, level, a coordination game among officials and citizens generates, in equilibrium, a state with the ability to enact laws and mobilize sanctions. At the same time, McAdams and Schauer assume that the dispersed decisions of individual citizens and officials about whether to comply with discrete legal rules (or at least some kinds of legal rules?) will be independent of the constitutional-level game. Compliance with these rules consequently requires some combination of coercive sanctions (available owing to the independent stability of coordination at the top) and more localized equilibria of coordination and cooperation.

Other game-theoretic accounts of legal order suggest even greater complexity—and indeterminacy. Consider the bewildering variety of explanations for constitutional compliance offered up by political scientists and constitutional theorists. Some theorists focus on cooperation or coordination at the level of specific constitutional rules, such as equal state representation in the Senate, the requirement that presidents be at least thirty-five years old, or the frequency of the census (see Levinson Reference Levinson2011, 707–11; McAdams Reference McAdams2015, 73). Others offer game-theoretic explanations of compliance with particular rights provisions, such as free speech and religious liberty, or compliance with the entire package of constitutional rights (see Levinson Reference Levinson2011, 711). Some accounts focus on the game-theoretic stability of institutional arrangements on a variety of scales, ranging from the relatively micro (e.g., the Senate filibuster [see Levinson Reference Levinson2011, 730] or the antebellum “balance rule” between North and South [see Weingast Reference Weingast1997,Reference Weingast and Bates1998]) to the relatively macro (e.g., federalism [Weingast Reference Weingast1995; de Figueiredo and Weingast, Reference de Figueiredo and Weingast2005] or judicial review [see Ramseyer Reference Ramseyer1994; Stephenson Reference Stephenson2003]). Others take as the unit of the analysis the constitutional system of government writ large—presumably incorporating and superseding all of the above (see Hardin Reference Hardin1999, 82–140; Hadfield and Weingast, Reference Hadfield and Weingast2012,Reference Hadfield, Weingast, Galiani and Sened2014; Weingast and Mittal Reference Weingast and Mittal2013). No one has offered any theory of how these multiple, overlapping, and seemingly duplicative games fit together or the factors that determine which should be relevant to predicting or explaining anything that happens in the real world.

To the (limited) extent patterns emerge, moreover, the causal mechanisms are left unexplained. Consider again Schauer's and McAdams's shared intuition that constitutional-level coordination operates differently and independently from more quotidian legal compliance decisions. As a descriptive matter, this seems entirely plausible: ordinary citizens probably do not obey traffic rules because they believe noncompliance will cause the entire constitutional order to unravel, leaving them in the state of nature. But it would be helpful to know why the legal system works in this peculiarly bifurcated way instead of some other way that is equally consistent with the hypothesized game-theoretic underpinnings.

Or consider the related and recurring intuition among constitutional theorists that structural arrangements—the basic design of democratic decision-making institutions and processes—tend to be more stable and constraining than discrete constitutional rights and rules. Thus, even while denigrating enumerated rights as merely “parchment barriers” against politically powerful majorities, Madison had faith that his preferred structural safeguards for protecting minorities—shifting power to the national government, indirectly electing senators and the president, and the like—would somehow possess greater constraining force. The plausible assumption that contemporary presidents are more likely to violate discrete rules and rights than to unsettle basic structural premises of the constitutional order by suspending elections or ruling by decree suggests that Madison was on to something. Unfortunately, here again, no one seems ever to have explained why cooperation or coordination at the level of constitutional structure is likely to be easier to establish or maintain (see Levinson Reference Levinson2011).

Explaining legal compliance and legal order as equilibrium behavior in some sort of game-theoretic frame turns out to be the easy part. So long as playing by a set of rules can generate some cooperative surplus—which will always be the case if the alternative is war, anarchy, or the Hobbesian state of nature—then legal compliance can be modeled as a potential equilibrium. The real challenge lies in explaining the size, shape, and location of that equilibrium. How much of systemic legal order can be derived from higher-level constitutional coordination or cooperation. Which components of legal order must instead be based on independent game-theoretic interactions and equilibria? When must legal compliance be undergirded by coercive force, and when can it be derived from the same kinds of game-theoretic calculations that make coercive force possible in the first place? Are there predictable pathways for the development legal order—bottom-up, top-down, or more circuitous? Which groups of actors coordinate or cooperate at which levels of the system? What are the variables that predict and explain the levels, locations, and boundaries of coordination and cooperation, as applied to legal compliance and cooperative behavior more generally? These are the kinds of questions that Schauer and McAdams can be profitably read to raise and that game-theoretic accounts of legal order should now be striving to answer.