1. Coordination Conventions and the Law

A coordination convention purports to solve a recurrent coordination problem which is a special case of strategic interaction.⁶ Suppose that two or more people wish to coordinate themselves in order to achieve some purpose, but there are at least two different scenarios that lead to this outcome. No previous agreement between the agents has been made. So they must choose the solution based on other elements, for instance: (1) their mutual expectations (what do A expect from B and vice versa); (2) their mutually conditional preferences (a solution is preferred by A but only if its also preferred by B), and (3) some salient feature about one of the solutions that makes the actions of A and B converge. Sometimes the need for coordination is persistent; the coordination problem the participants face may be recurrent. Whenever this happens, a coordination convention among them may arise. Coordination conventions are defined as 'regularities of behavior in a community in recurring situations calling for coordinated activity, where the need for coordination and the fact of general conformity are common knowledge'.⁷

Conventions have a special characteristic. They are binding by nature. They not merely provide reasons to follow them but create genuine obligations to do so. In other words, when a group has a convention it seems right to say its members are under an obligation to conform. In the case of coordination conventions the obligatory character they have is based on the mutually dependent expectations of the participants (i.e. the fact that each one expects the others to conform and others have a reason to expect everyone else to conform, being all this of common knowledge). Besides, as Postema has argued, coordination conventions define a pattern of joint activity for mutual benefit in which the coordination depends on each of the parties doing what others expect of him. Thus every party has an obligation to conform.

Postema used this scheme to explain the social practice in which the law is founded. His claim was that 'Law exists only insofar as it is realized in the actions, beliefs, and attitudes of members of the community'.⁸ What the law is in a social community is a matter of social facts: it depends on what its members take the law to be which itself is a matter of mutual understanding. Moreover, law is a complex practice that involves judges and other public officials charged with the task of identifying, interpreting and applying legal norms. But it also involves private citizens who choose to behave depending on two things: 1) how they understand what the law expects of them and 2) their own expectations regarding how legal authorities and judges interpret the legal framework. So the way officials and private citizens understand the law is interdependent and this form of social interaction (plus their mutually conditional preferences to coordinate) is what defines the social practice of the law.

Postema claims that social interaction between officials and citizens resembles a complex coordination situation. And the regularities in identifying, interpreting and applying the law are properly understood as conventions that impose public officials an obligation to conform. He argues that a private citizen, in order to coordinate his actions with other citizens and officials, may have a reason to follow the convention. This also applies to the citizen that assumes the position of 'the bad man', i.e. one who obeys the law for prudential motives. He also has a reason to coordinate his conduct with the decisions of law-applying organs even if his only motive is to escape from legal punishment. However, it would be inappropriate to say that the citizen is obligated to comply with the convention of identifying, interpreting and applying the law. By the contrary, judges do have such duty based on two powerful reasons. First of all, their preferences to coordinate are determined by the official function they accomplish, that is, mediate between the law and the behavior of private citizens. If law is to be effective, judges must seek coordination in their activities and comply with the conventions that aim such purpose. Second, judges must exercise their power with consistency if they wish to accomplish two goals: 1) to provide the whole institution of law with some reasonable degree of coherence, and 2) to respect the citizens' expectations regarding their decision-making activities (in fact, judges induce such expectations and reliance on them which is an essential feature of their role in the legal system).

Coordination is essential for judges and law-applying organs. As Hart pointed out, one of the conditions for the existence of a legal system is that judges ought to use the same criteria of legal validity, that is, they ought to use and accept the same rule of recognition.⁹ They must seek to coordinate their activity in order to achieve a reasonably coherent pattern in the identification, interpretation and application of the law. So each must consider his own expectations of how other judges identify, interpret and apply the legal norms of the community.

Albeit this account of the social practice of law in terms of a coordination convention seems appealing, it has to be rejected for several reasons. The most important is that law does not purport to solve a recurrent coordination problem. As Andrei Marmor has claimed: the main function of many types of conventional rules is rather to constitute a social practice. These are social conventions which 'evolve in response to a wide variety of social needs, and the need for coordination is just one of them'.¹⁰ Marmor argues that social conventions have a constitutive function: i.e. they create autonomous social practices such as artistic genres and games. They are called 'constitutive conventions'.¹¹ I shall explain them further in the next section.

2. Rules of Recognition as Constitutive Conventions

Lets begin by considering the game of chess. Conventional rules constitute the practice of playing chess which is viewed as a complex social interaction that embodies all kinds of elements: conceptions about winning and losing, values related as to what counts as a good game and a bad one, etcetera.

It must be noticed that rules of chess constitute the practice of playing the game, but they don't exhaust it. The relation between the conventional rules and the social practice is not of identity. There is no social practice without the rules, but the practice is not exhausted by following them. Something similar is true about the law. Rules of recognition constitute the law as a social practice. They are a special type of constitutive conventions. All constitutive conventions have these features in common: 1) they are systematic; 2) they are prone to change; 3) they are arbitrary; 4) they are partially autonomous; 5) we have limited knowledge of them, and 6) the reasons one has to follow them are strongly connected with the fact that others do. Two of these features have special importance: the arbitrary character of conventions and the fact they must be efficacious. A conventional rule is arbitrary in the sense that there must be at least one other conventional rule that could have been followed instead, achieving the same purpose.¹² On the other hand, a conventional rule must be actually followed by the majority of the group in order to exist. There is no point in following a conventional rule which is not actually followed by a significant number of people.

Marmor has emphasized this last condition: conventional rules can constitute a social practice only if they are actually followed. Hence -he concludes- only practiced conventions are conventions. But most important of all is that the reasons for following the rules are strongly connected with the fact that other members of the group follow them as well. These reasons to act in accordance with the convention are called 'compliance dependent reasons':

The reasons for participating in a conventional practice crucially depend on the fact that the practice is there and its rules are actually followed by the relevant population. Without some level of general compliance (in the relevant population), there is no social practice. To the extent that anyone has a reason to participate, the reason must partly depend on the fact that the practice actually exists.¹³

The efficacy of conventional rules is a condition on which Hart also insisted. In the Postscript Hart established a difference between conventional rules and merely concurrent practices: 'Rules are conventional social practices if the general conformity of a group to them is part of the reasons which its individual members have for acceptance'.¹⁴ In other words, a conventional rule must be actually followed in order to be a conventional rule.

In more recent works, Marmor has recognized another type of social conventions which are also arbitrary and efficacious. He calls them 'deep conventions'.¹⁵ They are responses to human needs deeply rooted in the world we live in. Whereas constitutive conventions are rather superficial and come into existence through the conventional practice, deep conventions are independent.¹⁶

There are many examples of social practices founded on deep conventions, including law. Law is a response to some of the most important social needs we have: social order and security for instance. We may all agree that these are natural purposes of the law and that a legal system that doesn't pursue such goals would be seriously defective. I think Marmor is quite right at this point and I also believe his analysis is original and interesting. But there is a small problem in it. Conventions do not merely have a constitutive function. They also give rise to obligations. We may say they have a normative character that needs to be properly explain and justified. These are matters that are not quite clear in Marmor's account. He accepts that law is an authoritative institution imposing demands on its subjects. Its part of law's nature to make such demands. The question is how to reconcile this fundamental feature of law with its conventional character.

At this point, Marmor makes a distinction between primary and auxiliary reasons for action concerning the normativity of conventional practices.¹⁷ Primary reasons are not created by the existence of the conventions but derive from values that are either external to the practice or inherent to it. Constitutive conventions partly constitute those values. They render the practice desirable or intelligible and determine the reasons one may have to participate in the practice. On the other hand, auxiliary reasons are engendered by the conventional rules. They are second-order reasons for they depend on the primary reasons we may have to follow the convention. For example: the greeting convention that prescribes that one ought to say such things as "hello" or "good morning" to an acquaintance, presupposes that there are values like politeness and civility which people cherish. These values explain why should a person follow the convention in the first place or why is important for her to do so. But the convention does not provide an answer as to what reasons should a person have for participating in the practice. The convention merely defines what the practice is and what it is to be done to participate.

According to Marmor, conventions can never constitute a complete reason for action. They only engender auxiliary reasons. Nonetheless they are characterized as mandatory norms (i.e. obligatory) in a conditional way. They prescribe actions that ought to be done if one is willing to participate in the conventional practice. In chess -for instance- we often find such requests as 'Move the bishop diagonally' that shall be interpreted as follows: 'If you want to play chess you ought to move the bishop diagonally'. Social conventions are conditionally mandatory: 'the "ought" such rules prescribe engenders reasons for action only on the condition that one is initially committed to participating in the practice constituted by these same conventions'.¹⁸ In other words, the convention is obligatory on condition there is a previous commitment to participate in the practice. This commitment is like a decision already made. My decision is a reason to act in accordance to the convention. It is the decision I took what makes the convention obligatory (at least for me). Had I not decided to participate in the practice, there would be no point in following the convention. Therefore the normativity of conventions is always conditional:

Constitutive conventions could not in themselves amount to mandatory norms, but for one who ought to be a committed participant in the practice which is constituted by those conventions.¹⁹

This is crucially important for the normativity of law: if rules of recognition are constitutive conventions, then their obligatory nature is also conditional. It depends on the underlying reasons for participating in the practice of law. The consequence is that rules of recognition do not provide the answer as to why should we obey the law or why should we endorse its demands. They only determine what counts as law in a given community.

Conventions should not be invoked in order to answer the question of why we should have law and legal systems, but only to answer the question of what counts as law in any given society. Social conventions cannot explain the point of law and its functions in our society. Law is conventional only in the sense that its rules of recognition are social conventions, and the function of such conventions is to determine the recognized criteria of legal validity, that is, to answer the question of what counts as law as opposed to other normative domains.²⁰

In my opinion, the problem with Marmor's account is that it focuses exclusively on the normativity of law issue from a descriptive point of view. He gives an explanatory account on how social conventions engender reasons for action and what kinds of reasons are involved. Conventionalism is important but only with regard to the explanatory aspect of the normativity of law.

I must say I differ from Marmor at this point. The normativity of law cannot be fully grasped unless we take in consideration the justificatory aspect as well: why should people (or at least some of the them) regard legal duties as legitimate or rightful demands? This is one aspect that any complete philosophical account of normativity of law must acknowledge. Against the idea that rules of recognition only define what the practice of law is, I will argue that this rules actually impose duties upon judges and other public officials. However, before I take care of this matter, I shall resume the conventionalist approach.

3. The Conventionalist Thesis

Coordination and constitutive conventions fall within the conventionalist approach. As I mentioned earlier, the conventionalist approach maintains that law as a social practice is founded on a social convention. There is an important connection between the conventionalist approach and the Social Thesis. While the Social Thesis claims that what the law is in a certain community depends on social facts, the conventionalist approach maintains that these social facts constitute a convention. Therefore the conventionalist approach may be seen as a specification of the Social Thesis.²¹ In accordance with the Social Thesis Hart claimed that in every community there is at least one rule of recognition that sets out the criteria of legal validity. If someone needs to know what the law in his community is, he must appeal to the rule of recognition that is actually followed by judges, public officials, jurists, etcetera. For Hart, the rule of recognition is a social rule, which means that its existence is a social fact. The rule of recognition finds expression in the judicial practice of identifying and applying certain standards as legal norms. To state the existence of that practice is to formulate an external statement, that is, a pure descriptive statement on facts.

In the Postscript, Hart describes the practice of identifying and applying the law as a conventional practice. There is an important distinction between conventional and concurrent practices. 'Rules are conventional social practices if the general conformity of a group to them is part of the reasons which its individuals have for acceptance'.²² By the contrary, simple concurrent practices such as positive morality are not constituted by convention 'but by the fact that members of the group have and generally act on the same but independent reasons for behaving in certain specific ways'.²³ The rule of recognition is a conventional social rule. Hart defines it as a 'form of judicial customary rule existing only if it is accepted and practiced in the law-identifying and law-applying operations of the courts'.²⁴ As a conventional rule, the rule of recognition must be actually practiced by the relevant group of people (courts and legislatures fundamentally), and this fact is part of the reasons a group member has for following and accepting it.

The claim that rules of recognition are conventional rules is part of the conventionalist approach. Jules Coleman refers to this approach as the 'Conventionalist Thesis':

Roughly -he says-, the Conventionality Thesis is the claim that law is made possible by an interdependent convergence of behavior and attitude: what we might think of as an 'agreement' among individuals expressed in a social or conventional rule. For Hart, this is the rule of recognition.²⁵

The Conventionalist Thesis claims that law ultimately rests on a social convention. It explains the very existence of the law and tries to figure out how is the legal phenomenon possible. As I said before, legal positivism is committed to the idea that law depends on some social facts. But there is no settled agreement among positivists about the nature of these facts. The Conventionalist Thesis holds that law is based on a set of social facts that constitutes a convention.

In an early stage of his thought, Coleman supported the idea that the rule of recognition resembles a coordination convention and that the rule creates reasons for acting in the way in which this type of conventions normally do, that is, by establishing a system of mutual expectations.²⁶ However, in more recent writings, Coleman criticized that view and adopted a different one.²⁷ As he explicitly admits:

In my earlier work I claimed that the practice of officials in regards to the rule of recognition constituted a Lewis-like coordination convention. Law exists when there is a rule of recognition that sets out criteria for legality that are adopted by officials. Their adoption, which Hart referred to as acceptance from an internal point of view, was best thought of as constituting a coordination convention. This view expressed the sense in which law ultimately rested on social facts, and did so in a way that explained the normative structure of the relationships of officials to the criteria of legality and to one another... [However] Shapiro's objection to my view put an end to the coordination convention version of the social fact thesis. If the practice of officials is not a coordination convention, what kind of social practice is it? Borrowing from some of the work of Michael Bratman on shared agency, I suggest... that the social practice among officials might plausibly be analyzed along the lines of a shared cooperative activity (SCA). I believed that conceiving of the practice of officials as an SCA might help us better understand the role of the rule of recognition as a source of reasons and reasoning in the lives of officials.²⁸

Coleman developed this particular conception in his book The Practice of Principle.²⁹ He begins by considering the rule of recognition as a duty- imposing rule. This rule is treated as the source of judicial duty (i.e. the duty judges have to apply certain criteria of validity). But how is this possible? How can the rule of recognition be a duty-imposing rule? Coleman believes its not enough that judges and legal officials adopt the internal point of view. This one is understood as the psychological capacity of human beings to consider a practice or pattern of behavior as a norm. The internal point of view creates a reason to conform to the practice for those who adopt it. But not every reason is a duty. Suppose I have the habit of reading sixty minutes every day. However, I decide to make this pattern of my behavior a norm by considering it from the internal point of view. I have created thus a reason for acting. Still, I do not regard myself as being under any obligation.

If I can create a reason by adopting a pattern of behavior as a norm, then it would seem that I can subsequently extinguish the reason that norm provides simply by withdrawing my commitment to it. Yet it is the nature of duties that those bound by them cannot voluntarily extinguish them as reasons.³⁰

In other words, there aren't self-imposed duties. It is in the nature of duties that one cannot autonomously extinguish them. This is the difference between reasons for acting and duties. That being so, how can the rule of recognition be a duty-imposing rule if the internal point of view does not create genuine obligations? Coleman thinks we must take a closer look at the normative structure of the social practice itself.

...When judges adopt the practice of applying the rule of recognition, the actions and intentions of other judges are reasons for each; it is as though they are going for a walk together, rather than simply walking alongside one another. It is this feature of the normativity of the rule of recognition that is left unexplained by the internal point of view. For it is not just that different judges decide individually and separately to evaluate conduct in the light of standards that satisfy certain criteria, thereby creating reasons for themselves that they can unilaterally extinguish; rather, they are engaged in a practice that has a certain normative structure -where, among other things, the fact that some judges apply criteria of legality is a reason for others to do so. It is this fact about legal practice, and not the internal point of view that judges take toward it, that is the key to understanding how the conventional practice of applying criteria of legality can be a source of a duty to do so.³¹

To consider the rule of recognition as a duty-imposing rule, Coleman claims that we must think of it as consisting in a conventional social practice. But not as a convention that purports to coordinate the conduct of judges regarding the use of the same criteria of legal validity. According to him, the coordination convention account has proven to be inadequate to elucidate the rule of recognition's normativity. The account is too weak since it cannot describe the kinds of reasons judges have for following the practice. The system of interdependent and reciprocal expectations that the convention creates falls short of explaining this point.³²

Coleman gives a philosophical account of the legal practice following Michael Bratman's social theory. There is a class of practices that Bratman identifies as 'shared cooperative activities' (SCA). SCA are things we do together -like taking a walk, building a house, singing a duet, etcetera. Coleman believes the practice that constitutes the rule of recognition is an instance of SCA. All SCA have three characteristic features: 1) mutual responsiveness; 2) commitment to the joint activity and 3) commitment to mutual support. The first condition means that each participant is responsive to the intentions and actions of the others and seeks to guide his behavior depending on what other participants do. This entails the existence of shared intentions between members that converge on a common goal (regardless of the reasons for doing so). Commitment to the joint activity and commitment to mutual support are also entailed by this first condition. They mean that each member of the group must commit to the joint activity (playing in a musical band for instance) and support the efforts of other participants to perform the joint activity successfully. Coleman maintains that the social practice in identifying and applying legal norms by using the same criteria of validity exhibits these three features. Judges coordinate their conduct among themselves and are responsive to the intentions of one another. And the best proof of their responsiveness is their commitment to the goal of making possible the existence of a durable legal practice. I myself will defend later a similar approach of the legal practice using the idea of a joint commitment as the basic element of the convention and the source of judicial duty. However, right now, I shall finish exposing Coleman's line of argument.

To sum up this far: in his 2001 book The Practice of Principle, Coleman thought that SCA account was the key to explain the rule of recognition as a duty-imposing rule. Yet, Coleman did not remain faithful to this approach. Some years later, in 2005, he modified his theory once again. He then affirmed:

I no longer believe it follows from the fact, if it is one, that the behavior of judges is an SCA that judges have obligations to one another or to comply with the rule of recognition. Nor do I believe that one can derive reasons from the social facts that constitute SCAs, including those among officials.³³

He proposed a different point of view according to which legal practice is best understood as a set of jointly intentional actions. But he explicitly admitted that he hadn't developed an argument of that kind, not even close (as a matter of fact, he confessed his efforts on this issue were fruitless). The truth is Coleman has moved from one position to another over the years. Though these positions exhibit important differences from one another, all of them might be seen as instances of the Conventionality Thesis. All of them are expressions about the conventionality of law. I shall now focus on one last form of legal conventionalism.

4. Law and Plans

During the last years, Scott Shapiro³⁴ has been advancing a particular view about the law that involves the concept of social planning. According to this conception, legal practice is a master social plan that includes other plans as subplans.

Social planning is when members of a group stipulate a plan about how to engage in a shared activity, for instance: playing together in a musical band. Planning is the easiest way to coordinate themselves. In the plan they specify all the provisions concerning the different ways in which the shared activity can be executed. These provisions are subplans which serve as means for the ultimate end and larger plan (i.e. playing together in a musical band). Planning is something a person can do by her own (I plan to cook dinner tonight). But it can also be done by two or more people together (me and my friends plan to cook dinner together tonight). If a group of people have a plan, it thus becomes a shared plan. A plan is shared by a group if the following three conditions are met: first, the plan is designed for the group (i.e. with the group in mind) as a joint activity constituted by each member's actions; second, the group accepts the plan, which means that each participant accepts to do his part and let the others do theirs (actually, accepting a plan entails a commitment on behalf of each member, namely, the commitment of following the plan). And the third condition is that the plan must be publicly accessible (the participants should be able to know the parts of the plan that pertain to them and to others).

Shared plans are constitutive of shared agency, that is, acting together. Acting together means that the group shares a plan; that all the participants act intentionally according to the plan (all of them play their parts) and the joint activity takes place because everyone do so. Additionally, for a group to act together two more conditions are necessary. First, the existence of the shared plan must be common knowledge (a group can act together intentionally only if the participants know that they share the same plan); second, members of the group act together intentionally only if they are capable of solving the conflicts that arise between them in a peaceful and open manner.

Planning in larger groups may be difficult. It requires extensive deliberation and negotiation between all the parties. In this context, it would be highly recommendable to let an individual or several take charge of the planning process in the name of the group. He will be the one making policies that guide and organize the shared activity. His policies will be considered also as plans. By appointing someone in charge of planning, hierarchy is introduced in the planning scheme.

Hierarchy is a planning mechanism for producing shared plans, but it is also the product of shared plans as well. In other words, authority within the group is a result of planning. The source of authority is a shared plan. Those in charge with the planning strategy direct the actions of others if the joint venture is to be successful. The planners may organize the labor of the group members by issuing different kinds of policies (directives and permissions) that constitute subplans of the shared plan. Their function is to guide and organize the actions of the participants, but also to monitorize their behavior and make them responsible for any violation of the plan. In order to make more efficient the joint activity, besides hierarchy, some forms of decentralization are required. The goal is to decentralize the process of planning by appointing supervisors who are authorized to apply the shared plan and to make other plans as well. Decentralization is also done by planning. There are plans called authorizations that grant powers on supervisors and instructions that specify how they ought to exercise their authorized powers. Authorizations and instructions guide the planning process. Shapiro calls them 'plans for planning'. Authorizations specify who is to plan, while instructions determine how to plan.

These remarks are important because legal practice is a form of social planning. The fundamental rules of a legal system are shared plans. Shapiro claims that human beings are planning creatures, and adopting plans is a manifestation of human nature. But how exactly does social planning make law possible?

Shapiro invites us to imagine a group of islanders living together. Following a line of reasoning very close to Hart's, Shapiro describes the life of the community from a primitive stage where we can only find small-scale group planning through negotiation, to more complex levels of shared agency. For instance, as the population of the island becomes larger, the inhabitants decide to adopt a system of private property that replaces the old system of common ownership they had since the beginning. They also introduce policies that protect private property and allow for the transfer of property rights. All these policies are plans which the islanders generate by consensus. However, the move to a system of private property increases the number of transactions and dealings. The economy grows and the population multiplies. Planning via consensus becomes very difficult and costly. At the same time, the new situation creates a huge amount of conflicts between the islanders. Bargaining cease to be an adequate mechanism for solving disputes. So the islanders decide to introduce hierarchy as a remedy. They appoint a small group of them as planners, and another small group as plan appliers. This last group is in charge with dispute resolution. Plan appliers settle disputes between islanders when they disagree about the proper application of prior plans. There's now vertical and horizontal division of labor in the island. Hierarchy and labor division are created by a shared plan, that is, a plan designed for the social planners and accepted by them. This shared plan guides and organizes the behavior of social planners. It regulates the activity of social planning. It is called the master plan of the group.

In accordance to the master plan, the new policies adopted by the planners are deemed binding for the whole community. Similarly, when one of the appliers determines that a policy adopted by the planners has been violated, his or her decision will be considered as obligatory; as long as these plans are adopted according to the master plan, they will be regarded as authoritative, both by planners and by islanders generally. To ensure the continuity of the planning process over the years, even when the first planners are gone, the master plan should authorize people to adopt plans in broad terms, designating offices (for example, the office of 'planner'). The office specifies just general qualifications a person must have for being appointed as planner (for example: 'A planner must be over 60 years old at the time of designation'). The office carries with it several rights and duties that persist over time; they are inherited to whoever happens to occupy the office at the given moment. To guarantee the permanence of the policies dictated by those appointed as officials, the master plan should mandate that policies are to be followed regardless of the person occupying the office. All these provisions in the master plan entail the institutionalization of social planning, which means that the community of islanders has reached a very high level of impersonality. At this point, Shapiro is inclined to think the island has developed a legal system.

The planners are the legal officials; the plan adopters are the legislators; and the plan appliers, the judges. The master plan is the constitution that defines their offices. The plans created and applied by these officials pursuant to the shared plan are the laws of the system: the policy directives are the duty-imposing rules and the authorizing policies are the power-conferring ones. Finally, the islanders all act according to plan. They are law- abiding citizens.³⁵

To summarize: legal practice is a result of planning. Legal institutions are structured by shared plans adopted by officials in accordance with a master plan. The plans are norms that set out the vertical and horizontal divisions of social labor, specifying who has authority to formulate, modify, apply and enforce the adopted plans. These shared plans are called the 'law's plans for planning'. Along with this view, law is just a matter of social facts: the fundamental rules of a legal system constitute shared plans, and shared plans exist precisely as pure social facts. As we know, a shared plan exists when it has been designed with a group in mind so that they can participate in a shared activity, it is publicly accessible and it is accepted by most members of the group in question. Shared plans are thus determined exclusively by what people think, intend, claim and do. The existence of a plan cannot depend on moral facts. If the group had to rely upon morality to ascertain the existence of a shared plan, then it would generate extensive deliberation among participants and thus the plan would not fulfill its purpose: i.e. guide and coordinate behavior by resolving doubts and disagreements about how to act.

According to Shapiro's planning theory, the ability to plan is part of human nature. Planning is an expression of our rationality, and practical rationality is governed by rules. However, the rules of rationality are not plans themselves. They are not created by any authority. They 'exist simply in virtue of being rationally valid principles'.³⁶ Therefore the law as a social phenomenon is ultimately based on this fact of life.

My intuition tells me Shapiro is right in asserting that we are rational creatures and that instrumental rationality is part of what we are. However, I do not think this fundamental truth is able to explain and justify the normativity of law. Shapiro believes that legal authority depends on the acceptance of a master plan. Here we find a parallelism between Shapiro and Hart: both appeal to the idea of acceptance. But there is also an important difference: Shapiro's theory does not require officials to take the internal point of view in order to accept the master plan of a legal system. In other words, there is no need to adopt a normative attitude in relation with the law. The existence of the shared plan is a descriptive fact. So it can be accounted for by means of purely descriptive statements. Anyone can account for the existence of the plan in a purely descriptive way, regardless of his moral commitments. However, the problem with this view is clear: how are we going to justify the adscription of legal rights and duties using as basis a purely descriptive fact? Shapiro actually claims that all legal statements (i.e. 'X has a legal duty to ɸ') are descriptive. So they do not have justifying aims (at least in appearance). But they do.

Legal statements describe the moral perspective of the law. Shapiro calls this perspective 'the legal point of view', and the legal statements formulated from the legal point of view are called 'perspectival legal claims'. These claims have no moral implications. They simply state that from the legal point of view of a legal system those authorized by the master plan to adopt plans have moral legitimacy, and the norms created as subplans in accordance to the master plan are morally legitimate and binding. This is not the same as if the norms were truly legitimate. They are but from the point of view of the master plan. When we say that someone has a legal duty to perform ɸ, we mean that from the legal point of view he is compelled to do ɸ because there is a norm adopted in accordance with the master plan, and consequently the norm is legitimate and binding. Presumably, this is all we need to justify the adscription of legal rights and duties.

I believe Shapiro's theory is flawed in various ways. Here I will mention what I think is one major error. Even if it is right to assume that law is grounded on a master plan that is accepted by most members of the community, and that legal norms adopted in accordance with the plan are actually followed, these facts alone are incapable of justifying legal authority. We need something else. We need to appeal to the commitment that the acceptance of the plan entails. To share and accept a plan implies that the person is committed to it. What's the point of having a plan if the people that accept it do not commit? Shapiro eventually speaks about a commitment between the planners, but he doesn't pay enough attention to this element. So, in section III, I will develop the commitment idea as the basis of my own view about the content of the legal practice. Right now, however, I shall turn to examine one of the most powerful attacks launched against the conventionalist approach.

1. Margaret Gilbert on Social Convention

Gilbert outlined a theory of social conventions as a new point of departure after criticizing David Lewis' account of coordination conventions. I will not summarize all those criticisms for my intention is to concentrate on Gilbert's proposal.⁴⁴ Over the years, her theory on social conventions has been under constantly re-examination. It would seem there are at least three different versions of her theory. In a chronological order, these versions are the following: 1) Conventions as norms of quasi agreements (1983); 2) Conventions as group fiats jointly accepted (1989), and 3) Conventions based on joint commitments (2008). As we shall see more ahead this last version is just a refinement of the second one.

Since the beginning, Gilbert⁴⁵ defended the idea that conventions engender an ought to act in a certain way. The ought of the convention is neither a moral one nor it is based on the personal preferences of people. The convention might prescribe an action that has no moral relevance at all or that is plainly against morality. Therefore, what kind of duty does the convention create? Gilbert asks whether the ought of social conventions has a correspondence with that of an explicit verbal agreement. It is clear that when two people agree to do something they are bound to act in accordance. Their obligations are founded on their agreement. Perhaps social conventions work with the same logic. It might be natural to think that conventions emerge from explicit verbal agreements. Moreover, they might be characterized as agreements of this type. However, Gilbert claims this is not necessarily true.

...Although there may indeed be a sense of the term convention in which it means something like "verbal agreement"... there is clearly no need for -and often no possibility of- an explicit verbal agreement between the parties to a social convention. Further, it seems implausible to suppose that social conventions must involve something which is actually a species of agreement -to suppose that the convention that men wear ties on formal occasions, for instance, involves what is literally speaking an agreement between the members of the society that men shall wear ties on formal occasions. Let us pursue the idea, then, that social conventions involve an analogue, rather than a species, of agreement.⁴⁶

So Gilbert came up with the idea of a quasi agreement. It happens that people -in certain circumstances- make ought judgments in much the way they would if there had been an explicit agreement between them, except there has been no such agreement and they are perfectly aware of this fact. However, they behave as if an explicit verbal agreement had existed since the beginning. And whenever they are required to justify the ought judgments they use, they describe their situation as one in which 'It is as if we have agreed'. This is called a quasi agreement. Gilbert's proposal is that a quasi agreement supports an ought. When a member of the group with the quasi agreement is required to justify his conduct and the conduct of other members of the group, his response will be something like this: 'We ought to because it is common knowledge that most people think that we ought to', or simply 'That's what we are supposed to do'. According to Gilbert, the ought of social conventions is that of a quasi agreement. The conception of social conventions she stands up for is 'the conception of a generalized, commonly known ought judgment which is itself seen as grounded on a quasi agreement'⁴⁷ (though she prefers not to use this technical term).

...There is a social convention in a group -she says- when and only when it is common knowledge in that group that most people think that one ought to do such-and-such in a certain context, and that one ought to do this because it is common knowledge that most people believe that one ought to do this.⁴⁸

Gilbert's account of social conventions does not require that parties of a convention think in terms of agreements. People may have a convention without having the concept of an agreement. But conventions do involve an analogue of agreement on which the ought of the convention is founded. This is the view of conventions as norms of quasi agreement.

Some years later, Gilbert offered a different account of social conventions in her first major work On Social Facts.⁴⁹ She then put forward the thesis that a social convention is a jointly accepted principle of action. In accordance with her new account a social convention exists when the members of a group accept all together a simple fiat of the form: 'Whenever a member of a population P is under circumstances C, he or she has to perform action A'. It is the fact that someone is a member of P -combined with the fact that the group has issued a fiat with the aforementioned content- what creates for him or her the duty to conform to that principle.

I propose -says Gilbert- that our everyday concept of a social convention is that of a jointly accepted principle of action, a group fiat with respect to how one is to act in certain situations... in my view social conventions on this account are essentially collectivity-involving: a population that develops a convention in this sense becomes by that very fact a collectivity. Further, each party to the convention will accept that each one personally ought to conform, other things being equal, where the ought is understood to be based on the fact that together they jointly accept the principle: "I ought to conform, in so far as I am one of us, because it is our principle".⁵⁰

This scheme was criticized by Marmor in his early essay On Convention.⁵¹ According to his own view, Gilbert's account is defective since it does not consider two important elements of conventions. All social conventions must be arbitrary and they must be actually followed in order to exist. As we know, Marmor claims that arbitrariness and efficacy are essential parts of the concept of convention. But a principle of action jointly accepted does not seem to be arbitrary at all. Furthermore, its validity does not seem to depend on the fact that other people follow it. Thus Gilbert's approach is misleading -Marmor claims. However, Gilbert re-examined her theory once again in an article published a few years ago where the following definition of convention is given:⁵²

A population P has a convention of conformity to some regularity in behavior R in situations of type S if and only if the members of P are jointly committed to accept as a body, with respect to themselves, the fiat: R is to be conformed to.⁵³

There are several changes in Gilbert's new improved account of social conventions (the joint acceptance account of convention). For instance, she now claims that conventions are a type of social rules. All such rules involve the joint acceptance of a fiat of some kind. The fiat 'R is to be conformed to because morality requires this', would imply the group has a moral principle. The fiat 'R is to be conformed to on account of our generally conforming to it', would indicate that the group has a custom. And a fiat such as 'R is to be conformed to because it has been conformed in the past', would mean the group has jointly accepted a tradition. Moral principles, customs, and traditions are social rules as well as conventions. But conventions make reference to simple fiats jointly accepted. A fiat is simple in the sense that it does not require a particular rationale. Therefore, conventions are different from other types of social rule because of the simplicity of the involved fiat. On the other hand, inasmuch as the fiat is simple in the sense explained, it is also conceived by the group as arbitrary (because it is the result of an arbitrary decision). Notice that Gilbert has incorporated -under this latest account- the two key elements of social conventions that Marmor pointed out.

Finally, she thinks a convention does not require an agreement in order to exist. Perhaps a convention can emerge from an explicit agreement, but this is just one possibility among others. Thus an agreement is in no way a conditio sine qua non for the emergence of a social convention. Gilbert sees an agreement as a joint decision to do something as a single body, based on a joint commitment as well. In this respect we encounter an analogy between a convention and an agreement. Both are based on the joint commitment established between the parties. If that is correct, then the most basic element of a whole range of social phenomena including conventions and agreements is the joint commitment. I turn to this idea next.

2. Joint Commitment

The following commentaries about the concept of joint commitment are based on some of Gilbert's essays published during the last two decades.⁵⁴ I will follow her general ideas insofar as they do not oppose my own. As a couple of preliminary points it should be mention, first, that the concept of joint commitment has been used in several disciplines and by different authors, including social psychologists and cognitive scientists.⁵⁵ Second, Gilbert has often claimed that the concept of joint commitment underlies many of our central collectivity concepts, such as the concept of social convention. But what exactly is a joint commitment?

A joint commitment is something different from a personal commitment which is like a personal decision. For instance, I could take the decision to follow a diet for the next months or to do exercise every morning. In both cases, my decision constitutes a reason to act in a certain way. But my decision is not a joint commitment. For a joint commitment to exist its necessary that two persons (at least) express their willingness or readiness to commit themselves jointly to do something together as a single body. The joint commitment is formed by the personal commitments of the parties, though it is not the simple sum of them. It is rather the commitment of all the participants. This last point is related with the fact that the joint commitment creates a plural subject different from each individual. Because of the joint commitment, two or more people commit themselves to act together (as one single subject). Thus acting together is possible due to the joint commitment. Some examples of plural subjects acting together are a musical band, a football team, a political party, etcetera.

To constitute a joint commitment there are two conditions that must be satisfied. I have already mention the first one: each party must express his or her willingness to commit with all the others to act together. The second condition is that these expressions ought to be known by all the participants, that is, they must be common knowledge. Let's imagine that two people A and B want to go for a walk together. To accomplish their goal it's not enough that A walk side by side with B. What's necessary is that A and B commit jointly to walk together. But for this purpose it is required that each one express the other his or her willingness to do it and that both know such fact. Something alike happens in larger groups but with an important difference. In large social groups there is no need that members know each other face to face. Nor is necessary that one by one express the others his or her willingness to commit. It suffices that the person is a member of the group (for example, a member of a football team) to presume that he or she is willing to commit with all the others.

Once the joint commitment is formed, what are its consequences? One of them is that the joint commitment acquires autonomy. The commitment is autonomous with regard the personal commitments of the parties. The joint commitment is neither yours nor mine, and therefore it cannot be overridden by one of us. Since the commitment is ours we all must end it together. Gilbert calls this condition the joint abrogation constraint of the joint commitment.

Another consequence -far more important for my own purposes- is that once the commitment has been formed a whole set of rights and obligations come into existence simultaneously between the parties. The joint commitment creates a bond between them. We might say that each one is bound to the others in virtue of the joint commitment, and that each has the right to demand the fulfillment of that obligation. What we have here is an interdependence of obligations: A is obligated to B only on condition that B is also obligated to A. In the previous example of two people that commit jointly to go for a walk together, each one has the obligation to keep the pace. If someone starts to walk faster without justification the other has the right to formulate some kind of reproof and demand him or her an apology. From this point of view, being part of a joint commitment not only means to have a reason to conform, but to be under the obligation to do so. This binding nature of the joint commitment has an important implication: inasmuch as the social convention is founded on the joint commitment, the convention itself is obligatory. The binding nature of the convention is the same as the one of the joint commitment. If in a social community there is a conventional rule, then its members are in an important sense bound to conform. The lack of conformity makes the transgressor subject to criticisms or reproofs. It could be said that the simple threat of failing to comply might cause among the others some demand of conformity. This last point reveals the obligatory nature of every social convention.