Why Legal Conventionalism Fails

Other people wrote to me saying, “Isn’t what you call the Moral Law just a social convention, something that is put into us by education?” I think there is a misunderstanding here. The people who ask that question are usually taking it for granted that if we have learned a thing from parents and teachers, then that thing must be merely a human invention. But, of course, that is not so. We all learned the multiplication table at school. A child who grew up alone on a desert island would not know it. But surely it does not follow that the multiplication table is simply a human convention, something human beings have made up for themselves and might have made diffe- rent if they had liked? I fully agree that we learn the Rule of Decent Behavior from parents and teachers, and friends and books, as we learn everything else. But some of the things we learn are mere conventions which might have been different – we learn to keep to the left of the road, but it might just as well have been the rule to keep to the right – and others of them, like mathematics, are real truths. The question is to which class the Law of Human Nature belongs.¹

1. The Story Told by Two Lewises – An Introduction

C.S. Lewis, quoted above, puts it this way: giving an adequate account of the Law of Human Beings including, in his opinion, the chief Moral Law – the Rule of Decent Behaviour, as well as other institutional arrangements that are dependent on moral/ political decisions is not as easy as many may think. It is to say that there is something in morality that is universal for all human beings, across different lands and cultures.

Lewis, being a realist and a cognitivist with respect to moral principles commonly shared within different cultures, states that there is something that just could not be different about morality – the Law of Human Beings – even in radically changed circumstances. Like the table of multiplication is an example of mathematical (analytical) truth, there are some real truths about morality (genuine, substantial principles) which can be qualified as the Moral Law. Such an approach suggests, however, that besides the Universal Law, being in fact a kind of ius-naturalist presupposition, there also exist laws of a purely conventional nature that do not resolve moral problems of any kind, but only serve as “instrumental” facilitations in achieving certain societal ends (either moral or immoral, but it does not matter here). Those laws are accidental and arbitrary in the sense that they could have been different, as they always have alternative equivalents or substitutes. To keep left or right on a road – it does not really matter, so long as most people actually drive on the same side. Moreover, even if certain laws are introduced to solve certain substantial, moral problems, they serve only as instruments, which, analysed in isolation, do not possess intrinsic value of any kind. In fact, certain conventional norms are usually considered morally irrelevant, and the view of the author of The Chronicles of Narnia supports that intuition. C.S. Lewis (whom, for the purpose of this text, I shall call “the first Lewis”) seems to suggest that the structure of the institutions of social order is twofold: consisting both of the invariable principles, as well as of the variable arrangements, being mere conventions, which can, but not necessarily must, serve as a means to certain ends. This view may be assessed as naive or too optimistic. So far, is it not a fact that monstrous and degenerate regimes in history (Nazism, Stalinism, other forms of tyranny) received sometimes, at least temporarily, vast social support? In those cases, had the basic principles of Moral Law been somehow forgotten? Furthermore, assuming that huge social groups, as well as individual people, simply betrayed the “objective” common moral principles – can we say that the core of law in that society was reduced to conventional and instrumental forms? A positive answer to the latter question may be ascribed to many radical thinkers who, as C.S. Lewis, were writing at the same time – namely during the Second World War – including especially T.W. Adorno and M. Horkheimer, whose dark philosophy reveals the “dialectic of enlightenment”. According to their view, the Substantial (metaphysical and teleological) Reason (which could be equated with Lewisian “Law of Human Beings”) is being inevitably displaced by Instrumental Reason.² In effect, a particle of instrumentality, being inherent in every culture, grows gradually within its development and absorbs more and more molecules of teleologically directed substantial rationality. Consequently, the picture of contemporary culture presents a languishing flame of Substantial Reason overwhelmed by different forms of technical rationality, backed by Instrumental Reason, and thus devoid of any further goal or deeper meaning. The radical vision espouses the claim that when Instrumental Reason eradicates its older and wiser, substantial brother, the whole deeper meaning is lost, and thus all prima facie rational actions not only turn out to be immoral, but, most of all, thoroughly irrational. In this case, as in many others, a radical seclusion leads to insanity. The parallel between notions of “conventionality” and “instrumentality” is obvious, since in both the above-cited views, the naive view of C.S. Lewis and the radical view of

M. Horkheimer and T.W. Adorno, a convention and an instrument generally serve as a means to achieve certain ends. (However, C.S. Lewis seems to believe optimistically that a mere conventionality or instrumentality cannot, in any case, exhaust the actual institutional order, and M. Horkheimer and T.W. Adorno, being pessimistic, believe that the totalistic expansion of instrumentality into the whole culture is an actual fact. At any rate, the latter view allows for means, through a specific process of alienation, ultimately becoming an end, what the former view rather seems to disavow. Therefore, the notion of “conventionality” has much in common with the notion of “instrumentality”. On the one hand, a convention can serve as an instrument to achieve certain societal ends; on the other hand, the strength of Instrumental Reason depends on whether it can adapt itself to any existing circumstances by taking the form of various conventions. Thus, a convention can be perceived as an instrument, and in this case both the intuition of C.S. Lewis and the theory of M. Horkheimer and T.W. Adorno coincide. Such intuition does not seem to raise many doubts. It may even be considered a platitude. However, as far as every platitude remains uncontroversial, it also remains ambiguous for the more careful researcher and thus any detailed account of the nature of a convention will trigger controversy. The juxtaposition of “conventionality” and “instrumentality” seems to be quite accurate, but doubts arise when we refer to other linguistic intuitions and begin to ask certain questions: if a convention is “up to us” and we – Humans – choose it as we usually select proper means leading to particular ends, can we say that government, law, social institutions, money, morality, or property are conventional? Or, how – in particular in mentioned cases – to differentiate between conventional means or instruments and unconventional goals? This question is especially difficult to answer if we understand that the human, reflexive and rational choice concerns not only human goals but also means to achieve them. Therefore, how can we be certain that our goals are unconventional, non-instrumental and non-arbitrary, while means to achieve them certainly are? Where is the vantage point, allowing to differentiate between “a mere convention” and “unconventional end/value”? Such differentiation could have been possible in ancient times when all people believed in certain metaphysically determined goals. Such belief is to some extent inherent in C.S. Lewis’s writings about the Moral Law. But such a view is easily questioned in the age of post-metaphysics and worldwide pluralism (or even – relativism). Now, a more detailed and – inherently more controversial – theory of conventionality is required. Otherwise, probably each element of our institutional world could be characterized as “conventional”, as dependent on human decisions to some extent. A fleeting shift of the discussion to a more definite ground might reveal one of the most important problems connected with the notion of “conventionality”. The “standard story of the Carnap-Quine debate”³ on the foundations of logic is a classic example of a dispute about whether the truth of logical statements can be founded on a mere convention. According to the popular reading of R. Carnap, he himself held the view that the adoption of a system of logic is fundamentally a matter of linguistic convention, so that a logical truth is a “truth by convention”. Any logical truth could namely have been different if we had decided it to be. W.V. Quine opposed this view and deprived the idea of convention of such an ultimate, constructive function in logic. He claimed that a convention cannot be identified with any kind of “ultimate decision”.

The lore of our fathers is a fabric of sentences. […] It is a pale grey lore, black with fact and white with convention. But I have found no substantial reasons for concluding that there are any quite black threads in it, or any white ones.⁴

Consequently, the idea of the conventional foundations of logic, being in this dispute a paradigm for other sciences, has been disqualified, but at the same time no detailed account of what convention really is has been proposed. We might say that the basic but hazy intuition about conventionality described above was still thought to be valid, but no deeper understanding of it has been developed so far. Quine questions what a convention is “when there can be no thought of convening”, and remains sceptical about the possibility of answering it. The sign of the modern times is that all old platitudes are denounced as real controversies. The major task of the analytic philosopher in such circumstances is to restore the primordial lucidity of the platitude by presenting a detailed theory of its subject and by contrasting it with other theories. Such a task had been undertaken by D. Lewis (whom, for the purpose of this text, I shall call “the second Lewis”), who presented a detailed theory of convention, developing the argument of D. Hume by the use of game-theoretical scaffolding. He was a pioneer in these matters and raised the bar high enough for other philosophers who nowadays cannot propose any theory of convention without either criticizing or accepting his manifold contribution to the problem discussed. D. Lewis answers the Quinean problem without a necessary reference to any kind of pre-existing historical fact of agreement or “a fiction of convening” of any kind that would certainly produce a vicious regress. Rather, he tries to anchor the notion of convention in mutually recognized shared expectations of individuals (which he calls “common knowledge”), built upon the existence of Humean “common interest” in reaching certain goals. As he writes:

Convention turns out to be a general sense of common interest: which sense all the members of the society express to one another, and which induces them to regulate their conduct by certain rules. I observe that it will be to my interest [e.g.] to leave another in possession of his goods, provided he will act in the same manner with regard to me. When this common sense of interest is mutually expressed and is known to both, it produces a suitable resolution and behaviour.⁵

According to D. Lewis, a convention can be described as a solution to a coordination problem (which arises in non-zero-sum coordination games), which is a situation of interdependent decision by two or more agents in which coincidence of interests predominates and in which there are two or more proper coordination equilibria.⁶ In such situations, agents have a common interest in all undertaking the same one of all alternative actions. Moreover, agents can succeed in uptaking one and the same

coordination equilibrium either by chance, by making an explicit or tacit agreement, or – what is the most important to overcome the threat of regress – by reasoning based upon “suitably concordant mutual expectations”.⁷ Such reasoning, characteristic for each individual that has to act in circumstances of relative social uncertainty, leads to the creation of a system of self-perpetuating preferences, expectations, and actions, due to which certain regularity in behaviour agents arise. Thus, a convention can be technically referred to as a certain regularity R in the behaviour of persons in a population P in a recurring situation S if and only if in any instance of S: (1) it is common knowledge in P that (a) there is in P general conformity to R; (b) most members of P expect most members of P to conform to R; (c) almost every member of P prefers that any individual conform rather than not conform to some regularity of behaviour in S, given general conformity to that regularity; (d) almost every member of P prefers general conformity to some regularity rather than general non-conformity (i.e. general conformity to no regularity); (2) part of [the] reason why most members of P conform to R in S is that 1a-1d obtain.⁸

This definition shows how any basic convention could develop through a process of practical reasoning of agents who share a certain interest (common interest) in achieving certain ends. Agents are forced to act in circumstances where only concordant actions, constitutive of a particular equilibrium, can make it possible to achieve a shared end. Moreover, theoretically each of the coordination equilibria is equally available (and thus – arbitrary), and therefore agents need to construe their common knowledge on the basis of some background empirical factors or premises, being salient facts indicating the most probable coordination-game solution (the fact that there actually exists some kind of regularity in behaviour; the fact that there was some regularity in behaviour a long time ago; the fact that one of the alternatives is to some extent epistemically salient and so on). In this case we can say that agents believe that there exists some kind of precedent to obey precedents in cases of uncertainty and lack of transparency concerning intentions and expectation of other agents (and this is one of the main Humean assumptions, developed by T. Schelling and adapted by D. Lewis). The detailed story of how a convention (linguistic, social, etc.) emerges without any explicit or tacit agreement is one of the most important contributions of “the second Lewis” to modern philosophy. The basic intuition that convention serves as an instrument, one of many available, to achieve shared ends by certain groups of individuals in this case is backed by a detailed account of common knowledge and perfectionist rationality.⁹ However, the extent to which convention serves as an instrument to resolve a coordination problem is per definitionem arbitrary,¹⁰ but the circumstances in which the very problem arises might not be so. Consequently, however, a convention itself is

of no intrinsic value. It might gain moral or social value in case it enables the solving of certain moral or social problems. Thus, we can still differentiate between the inner and relative value of conventional (resp. instrumental) rationality, and the external or at least inter-subjective value of ends a convention (resp. instrument) serves to achieve.¹¹ The arbitrariness, as I. Dąmbska indicates, is an intrinsic but relationary feature of every convention,¹² conceived either as an ultimate decision or agreement (as classical conventionalists do), as a solution to a coordination problem (as D. Lewis does), or in a different way (as other authors, mentioned below, do). If it is so, conventionalism does not imply relativism and conventional arbitrariness should not be confused with indifference.¹³

2. Some Problems with Coordinative Account – A Short Remark

The account of convention proposed by the second Lewis gives a detailed explanation of how it is possible to share something (language, social institutions, justice, government, law) without the need to reach an agreement of any kind. However, some critics¹⁴ show that this account is not as general as it is declared by Lewis to be. In particular, as A. Marmor indicates, there are certain types of social conventions that do not fit this analysis:

Generally, the problem is this: Lewis’s analysis assumes that first there is a recurrent coordination problem in a given set of circumstances, and then a social rule evolves that solves the problem for the relevant agents. For many types of conventions, however, this story [does] not make sense. Antecedently to the emergence of the convention, there is no coordination problem that we can identify, at least without already assuming that the conventional practice is in place.¹⁵

It seems plausible to recognize that the main rationale of different institutional forms of socio-cultural actions is not to coordinate behaviour. A. Marmor gives examples of chess and theatre. Chess is a competitive game where coordinating actions play a secondary role and coordination cannot be considered its main aspect. It is not even the case that chess emerged as a solution to a coordination problem. The point is that from the functional point of view rules of chess do not solve any coordination problem. Rather, chess as a “competitive game” is, by definition, troubling engaged players with different problems, some of which are certainly coordinative by nature. But it would hardly be acceptable to say that the main rationale of chess is to solve a coordination problem. As A. Marmor soberly notices, you can say, of course, that there is a general coordination problem between two agents, consisting in their desires to play some

intellectual board game, and thus the game of chess appears to be a good solution, but the obvious difficulty is that such a coordination problem would be too abstract and underspecified. To make an abstraction in this case could save the construction of the argument, but under pain of losing its explanatory power. Abstract coordination problems are not interesting problems at all. In the case of theatre, the argument is analogous. To quote A. Marmor:

Does it make sense to assume that before theater evolved as a specific form of art, there was some recurrent coordination problem that needed to be solved, and then the conventions of theater evolved to solve it? What would that coordination problem be, and for whom?¹⁶

Now, I think that it is plausible to accept A. Mamor’s objection and agree that social conventions evolve as responses to numerous kinds of social needs. They serve a wide variety of social functions and there is no reason to assume that all those needs are reducible to coordination problems. Theoretically, there can be different types of conventions, which are even more fundamental for the existence of social institutions (but we should leave this case aside for a while). The problem I want to discuss in the following sections is whether we can say that law in general is built upon conventions. If we suppose that it is, what kind of conventions is fundamental for law to exist? Is law in general a solution to a certain coordination problem? If not, is there another plausible conventionalist explanation for the existence of legal institutions? Giving an answer to those questions could not be as easy as one might suppose. Law stands out among other social institutions because it claims to have universal authority. That is why the first thing to do is to discuss the relation between two theses: the thesis that asserts that the existence of law is a matter of social fact (the social fact thesis), which can be further developed into the conventionality thesis, and the thesis that the authority of law depends on the role played by legal rules in the practical reasoning of its subjects (the normativity thesis). These two theses have been feverishly discussed for more than a quarter of a century. Classical positivists reject the latter thesis and proclaim the former; the ius-naturalist philosophers and some nonpositivists reject the former and proclaim the latter. But the theory on which I would like to focus tries to find a middle way and reconcile these two theses, being potentially in conflict, in the name of conventionalist legal positivism.

3. The Quest for Normativity17

As G. Postema underlines, the quest for determining the proper relationship between the social fact thesis and the normativity thesis “set the agenda for much of philosophical jurisprudence”.¹⁸ The prospected artefact by theorists in this case is a “fully fledged theory of law”, which not only gives an acceptable picture of legal practice as a social fact, but also aims to convincingly explain the very nature of law’s characteristic normativity. The two theses that should be reconciled are, according to G. Postema,¹⁹ as follows:

The normativity thesis: “Law is a form of practical reasoning; like morality and prudence, it defines a general framework for practical reasoning. We understand law only if we understand how it is that laws give members of a community, officials and law-subjects alike, reasons for acting. Thus any adequate general theory of law must give a satisfactory account of the normative (reason-giving) character of law and must relate the framework of practical reasoning defined by law to the framework of morality and prudence.”

The social fact thesis: “Law is a social fact; what is and what is not to count as law is a matter of fact about human social behavior and institutions which can be described in terms which do not entail any evaluation of the behavior of the institutions. We understand law only if we understand it as a kind of social institution which can be said to exist only if it is actually in force and directs human behavior in the community. Any adequate general theory of law must give a satisfactory account of law as a social phenomenon.” Some contemporary legal positivists, who perceive themselves as heirs and developers of H.L.A. Hart’s upgraded version of positivism, have a strong incentive to apply the concept of convention to fill the gap between these two theses. They usually point to The Concept of Law, where Hart openly expresses the idea that law rests, at its foundations, on a special and complex custom or convention.²⁰ The widely cited passage concerning the conventional nature of the rule of recognition reads as follows:

Certainly the rule of recognition is treated in my book as resting on a conventional form of judicial custom. That it does so rest seems quite clear at least in English and American law for surely an English judge’s reason for treating Parliament’s legislation (or an American judge’s reason for treating the Constitution) as a source of law having supremacy over the sources includes the fact that his judicial colleagues concur in this as their predecessors have done.²¹

It is characteristic of contemporary legal positivists that, as J. Coleman (and J. Dickson after him) notices, all of them accept that the criteria of legality are conventional.²² Such a view about the conventional nature of the ultimate rule of recognition is strictly connected with the Hartian account of social rules, namely the “practice theory of rules”. When we interpret Hart’s enterprise, we find that his aim was to present a version of legal positivism that does not inherit the “mistakes” of J. Austin’s theory (who was a naturalist, very suspicious of normativity). To achieve this, Hart had to propose a theory of law that plausibly explains the nature of legal obligation. Hart was aware that the existence of a social rule is a necessary but not sufficient condition for a determined behaviour to be configured in terms of obligation: if a person has an obligation to do something, then it will always be possible to trace a social rule at the basis of this obligation.²³ The practice theory of rules has been discussed in detail by P.M.S. Hacker, who ascertained that there are eight conditions by which Hart tries to elaborate an adequate theory of legal obligation.²⁴ First, the social rule requires of those

people that are subject to it that they behave (or abstain from behaving) in a certain way in given circumstances. Second, the majority of the members of the group believe that the social rule at issue is important for the maintenance of social life or some highly appreciated characteristic of the latter. Third, a potential conflict exists between the behaviour required by the social rule and the desires of those people that are subject to the rule. Fourth, a generalized conformity exists on the part of the members of the group to what is prescribed by the rule (i.e. the rule is potentially effective). Fifth, deviations from the rule are followed by serious critical reactions such as to make the deviant behaviour less advantageous. Sixth, possible deviations from the rule are being considered a good reason for a critical reaction. Seventh, this kind of critical reaction is generally considered to be legitimate (i.e. criticism of the deviation from the rule is not usually followed by counter-criticism). And last, but not least, in criticizing the deviant behaviours, major use is made of normative language. Although the Hartian theory of obligation deserves to be discussed in greater detail, I will confine my presentation to these eight conditions, because I share the belief that these conditions should be considered central to Hartian positivism. However, we notice that there are some interpreters of Hart who claim that it was not Hart’s main aim to give any theory of legal obligation and, what follows, any detailed account of law’s normative force. Rather, they say, Hart’s project was profoundly descriptivist and any concerns whether we should treat legal provisions as actually justified and reason-giving, binding norms are a matter of normative, political philosophy justifying legal order. In this interpretation, Hart is supposed only to propose a general sketch of relations between central legal concepts (obligation, sanction, legal practice, rules, etc.). Moreover, on this account to say that the criteria of legality are conventional is simply to say that there is a widely-shared agreement between officials who apply the same criteria of identifying law (i.e. the same rule of recognition) – and nothing more. This is a social fact that can be intuitively named “convention”. Such a view recognizes, what is obviously true, that the accent in Hartian jurisprudence is put on descriptive matters (“descriptive sociology”), but incorrectly assumes that Hart himself cut off political and normative deliberation in the context of law by simply evoking a concept of the internal point of view connected with the notion of “serious social pressure” that does not need any further explanation. Contrary to what those theorists might say, probably the most important problem in Hartian jurisprudence is to give such a general interpretation of his theory that would both save its descriptive value and present a coherent and plausible understanding of the internal point of view (on which almost the whole burden of law’s normativity rests). Such a task would reveal some internal contradictions in Hart’s own theory and would force one to choose those solutions that best fit the basic assumptions of the interpreter. My basic assumption, to make it clear, is that Hartian positivism aspires to be a fully fledged theory of law, as otherwise it cannot compete with other theories that have such ambition. The Hartian account of rules, namely the practice theory, maintains that a social rule (including the rule of recognition) is a certain kind of complex social practice that consists of the existence of a general and regular pattern of behaviour among some group of persons, together with a widely shared attitude within the group that this pattern is a common standard required to conform with.²⁵ This is thought to discriminate between

rule-following behaviour from mere coincidence of habit. So, in fact the concept of a rule includes those very facts of acceptance and use, and therefore a social rule does not even exist unless a sufficiently large number of people within the requisite group adopt the internal point of view with respect to some regular pattern of behaviour.²⁶ The notion of rule acceptance, being central to the internal point of view, is present, at least implicitly, in conditions two, four, five, six, seven, and eight on Hacker’s list referred to above. Moreover, the general acceptance of the foundational rule of recognition makes it a duty-imposing rule²⁷ (but, with respect to its foundational function, one should also appreciate its power-conferring function). So, the key to explaining law’s normative force together with the concept of legal obligation is to present in detail the notion of having the internal point of view on the social rule that avoids vicious regress in tracing the very foundations of legality. According to S. Perry, the internal point of view serves two particularly important and related roles in Hart’s theory of law:

The first is […] to specify one of the constitutive elements of the complex social practice that comprises a legal system, and, more particularly, to specify that element which permits us to say that law is not just a social practice, but a normative social practice. The second role is to explain the normative dimension of the meaning of such statements as ‘It is the law of Pennsylvania that everyone has an obligation to do X’.²⁸

The question is whether such “a non-cognitivist account of the internal point of view” (as S. Perry calls it) allows Hart to escape the difficulties of the Austinian theory, which Hart himself revealed. The answer to this question seems to be negative. Below I present in a nutshell an argument that is partially based on considerations of N. MacCormick, S. Perry, and A. Schiavello. Hart’s theory of legal obligation, as well as – in the original edition of the Concept of Law, before Dworkin’s substantial critique – moral obligation, is based upon his account of social rules that are simply accepted by major groups of individuals. Hart offers no deeper investigation into the nature of this attitude of acceptance (that is why S. Perry calls it a “non-cognitivist account of the meaning of legal statements”). Hart simply focuses on an “empirical” assumption about the notion of law’s acceptance: a general endorsement of the standard of conduct. As S. Perry notices:

The internal point of view, properly understood, is the perspective both of the authorities who make this claim and of the subjects of law who accept it. To accept the legitimacy of the law’s claim to authority is to believe that the law has such authority, and not simply to adopt an attitude of endorsement towards the law’s requirements.²⁹

This remark shows that Hartian positivism, in order to fulfil its main task, needs to present a deeper understanding of the internal point of view in terms of reason for action. S. Perry thinks that adopting a “liberated”, cognitivist understanding of the internal point of view, and of the meaning of normative statements generally, will naturally lead to the recognition that the meaning of normative expressions in legal, as well as in moral contexts is the same. However, such a consequence of “liberation” seems to be contrary

to Hart’s basic commitment, namely the basic distinction between legal and moral obligation. This can be clearly read as a reason why Hart rejects the assimilation of the internal point of view and the moral point of view. The non-cognitivist understanding of the internal point of view means that acceptance of law can come about for different reasons that are all on the same level,³⁰ none of which seems to be a necessary reason (this can be called “a weak-dependence condition”).³¹ As A. Schiavello points out, this thesis of Hart is questionable to a large extent. For instance, the “conformist” – who is law-abiding because others are – though being, Hart says, perfectly referable to the perspective of the participant, in effect shows many analogies with the perspective of the bad man. The only difference between these two situations is that the bad man “goes straight” out of fear of punishment, and the conformist out of fear of social reproof. Social reproof, however, is nothing but a sanction that is not institutionalized. Characterization of the internal point of view in a weak sense thus reduces the distance between Hart’s conception of legal obligation and the model of bandit. If the reasons for accepting the law are on the same level, a case can be hypothesized in which all the participants accept the law out of conformism, and where this happens the difference between “having an obligation” and “being forced” ceases to be substantial.³²

A. Schiavello shows that certain types of reasons cannot be treated as fundamental for upholding the internal point of view on the rule of recognition as far as we conceive the internal point of view as having genuine, normative significance. Considering the “general conformity” to a conventional rule as part of the reason to conform, the way Hart does, leads only to the acceptance of the weak dependence condition – and such a view cannot be considered an alternative to the Austinian “model of bandit”. It means that from the whole scope of reasons for acceptance by the officials of the rule of recognition, including a bad man’s reasons, conformist reasons, moral reasons and conventionalist reasons, we have to exclude the first two types of reasons if we want to save Hartian positivism from the “mistakes” of Austin, which Hart himself thought to overcome. In other words, “the over-weak” characterization of law’s acceptance (as N. MacCormick puts it)³³ would mean that the whole Hart project is a failure. Moreover, due to Hart’s ardent insistence on having two different and separate explanations for legal and moral normativity (the separability thesis), the cognitivist (stronger but liberated) account of acceptance cannot be reduced to moral acceptance. Consequently, the postulated autonomy of legal obligation implies that moral reasons should also be

excluded. As we can see, the only type of reasons for the acceptance of the rule of recognition that still remains untouched after such short examination is conventional reasons. Hart himself gave some implicit clues to interpret his theory in this way. It seems that Hart originally had thought that the practice theory of rules can be conceived as a general theory of obligation, but the critique presented by R. Dworkin demonstrated that the notion of moral obligation cannot be founded solely on social rules and that the theory of moral obligation should focus on some unconventional forms of morality, namely the critical morality. Nevertheless, although Hart accepted this line of critique with respect to the concept of moral obligation, he announced that he still believed that the practice theory of rules is a plausible theory in relation to conventional rules and may be useful in a legal context. This confession, backed by the citation presented above (about the partial, conventional reasons for having law, see p. 10), had provoked some legal scholars (among them J. Dickson) to speak of a “conventionalist turn” in Hart’s legal theorizing. However, the given interpretation of Hart shows that from the very beginning the only way to interpret Hartian positivism as a fully fledged theory of law was to treat the rule of recognition as a conventional rule not only in a purely descriptive, weak sense (which simply equates the conventionality thesis with the social fact thesis), but also in a stronger sense, in which the whole explanation of law’s normative force depends on conventional reasons that people have to follow law’s requirements³⁴ (the strong conventionality thesis). As we can see, non-cognitivism in relation to the reasons for accepting law cannot be sustained and this leaves us with two possible, cognitivist interpretations of the internal point of view: moral and conventional. Many legal theorists (amongst them S. Perry, J. Raz, L. Green, J. Dickson) argue that the proper way of explaining legal normativity is to show that the legal claim to authority is actually a moral claim. There are also some theorists who make a serious effort to save a conventionalist interpretation of Hartian positivism by developing certain philosophical concepts of convention in a legal context (besides the concept of coordination convention, they refer to the concept of shared cooperative activities or constitutive conventions). They aim to reconcile the two main theses, namely the social fact thesis and the normativity thesis under the aegis of the conventionalist legal positivism, which ascribes to a convention of the foundational role both as providing certain criteria for the recognition of law, as well as by ascribing some normative significance to them. It is to say that conventionalist accounts of law’s foundations are meant to show how appropriately characterized conventional rules on certain conditions can be reason-giving. In my view, “the moral way” of explaining law’s normativity (i.e. that law’s claim to authority should be analysed in moral terms) is correct. But I also think that one of the ways (although not perfectly conclusive) of showing why it is so is by contrast – by the use of some kind of negative test, by which we can exclude all unreliable theories in this matter. Thus, a part of an argument that legal obligation is in fact a kind of moral obligation consists of understanding why legal conventionalism in its contemporary forms fails. Obviously, refutation of a conventionalist account of normativity does not prove per se that the thesis of the moral nature of legal normativity is justified. But if such refutation is successful, we are at least half way in the argument for the moral thesis.

Conventionalism in philosophy is a broad view claiming that conventions have a special, foundational role in explaining certain cultural phenomena.³⁵ In a legal context a convention is supposed to solve two theoretical problems at the same time: to explain the validity of law solely by reference to the ultimate norm without circularity and without reference to a brute fact of any kind.³⁶ Even if restricted to the domain of legal theory, conventionalism still has many faces. But, by referring to the parallel between “conventionality” and “instrumentality” mentioned in the beginning, we can say that the contemporary “fully fledged” legal conventionalists are only the positivistic descendants of Hart, whose views may appear to be “twin-brothers”: the early idea of J. Coleman, G. Postema and E. Lagerspetz that the rule of recognition is a coordination convention, and the more “instrumentalist” view, advanced in recent years by J. Coleman and S. Shapiro, that law is a kind of a general plan, developed and realized through a shared cooperative activity. To some extent these views are similar and both are thought to be a positivist response to an anti-conventionalist critique of legal positivism. Therefore, it seems that, also because of many analogous strengths and weaknesses in both accounts, they should be examined together. Moreover, it must be said that these conceptions are of extraordinary architecture and the whole critique of them relates to a substantial understanding of law’s claim to authority and the very nature of reasons law can give. This means that the critique is clearly external as far as internal relations between concepts used and the internal coherence of these theories are really hard to undermine. The first view, namely the view that the gap between the social fact thesis and the normativity thesis can be overwhelmed by applying the notion of coordination convention, has been introduced to defend Hartian positivism against attacks of Dworkin by G. Postema³⁷ and J. Coleman,³⁸ and further developed in great detail by E. Lagerspetz.³⁹ We know already that as far as coordination conventions are concerned, the fact that there is a convention is a reason (a partial or auxiliary reason) for conforming one’s behaviour to the rule founded on such a convention. But one has a general reason to conform only if one shares a certain interest in achieving certain goals with other people (one has a reason to drive on the right-hand side of the road, as others do, when one has reasons to drive safely, etc.). As L. Green writes:

Coordination conventions promise to explain how law is both positive – a matter of fact – and normative, or action-guiding. The existence of a convention makes salient one of the alternative possible ways of coordinating action; it makes it one to follow. That is not true of every general practice: the fact that there is a general practice of opening a certain way in chess is not itself a reason for opening that way. On the contrary, here the reason is simply the set of strategic considerations that justify the opening itself.⁴⁰

Here the main question arises: is the rule of recognition a coordination convention? Or, is the first, hypothetical constitution a solution to a general coordination problem? S. Shapiro, for example, asks in this context whether it is possible to treat such a constitution as an arbitrary solution that can be replaced by another text at any moment.⁴¹ But I think that Shapiro’s question can be misleading to the extent that it seems to equate arbitrariness with indifference. So, we should rather consider why coordinative convention cannot be treated as foundational in a legal context despite the fact that law seems to coordinate many actions, especially among officials (but also, as G. Postema points out, in other dimensions). L. Green believes that it is true what G. Postema, J. Coleman and especially E. Lagerspetz say when they claim that a) legal systems help coordinate social action, b) their capacity to do this is sometimes valuable, and c) neither (a) nor (b), nor both together, suffice to establish that every legal system is morally legitimate or that there is a prima facie moral obligation to obey the law.⁴² Green’s main point is, however, that neither (a) nor (b), nor both together are sufficient to explain the normative character of law or warrant anyone regarding law as authoritative. His criticism is twofold. First, he shows that not every coordinative convention imposes obligations (i.e. coordinative convention is not sufficient to establish any kind of obligation). He presents a simple comparison that makes these matters immediately clear:

It is a convention in my society that men do not wear skirts, but there is no obligation to refrain from doing so. It is a convention in my society that we speak English, but there is no obligation to do so. Thus, the conditions necessary for the social convention with respect to ø-ing are not sufficient for the existence of an obligation with respect to ø-ing.⁴³

If not every coordination convention is obligatory, and law, ex definitione, certainly is, there must be a fundamental difference between practical reasoning in the case of any convention (that has value of informational restrictions) where the processes of the balancing of reasons take place, and in a legal context, where the obligatory (mandatory) character of law is granted by reference to exclusionary reasons that legal authority purports to give. Conventional reasons are only first-order reasons that could be balanced and over-weighted, whereas authoritative reasons are exclusionary reasons that could not be balanced in this way. Coordinative conventions are conditional and their capacity to create obligations depends on whether one wants them to be created. Such will is usually generalized and referred to by conventionalists as “common interest”. On the other hand, at least from the conceptual point of view, the law’s claim to authority is an unconditional claim that demands conformity in all circumstances. The authority of law seems to be neither necessary nor sufficient to resolve coordination problems. It is not necessary, because there are many extra-legal (deliberate, as well as spontaneous) methods of resolving coordination problems. Thus, conventions are not necessarily authoritative. But one may ask whether general, legal authority is sufficient to solve peculiar coordination problems that may be generally relevant for the existence of a legal system. That would mean that if we have legal authority, it naturally resolves coordination problems that could be unresolvable without it. But again, this claim fails as far as we can see that law by its authority sometimes hinders coordination in social contexts (e.g. against coordination of monopolies, gangs, anarchists…). The general

If coordination is preeminently valuable, and if authority is the only way of establishing a coordination convention, and if the authority of law is the only feasible sort, then it is rational to comply with it, provided it is also rational to expect others to do so as well. But the legal system does not confine its reach to this narrow set of conditions. Law also claims authority when it seeks to prevent coordination, when it replaces coordination equilibria with other kinds of solutions, and when it performs all of the other functions of legal systems.⁴⁴

Thus, the critique presented by L. Green shows that legal conventionalism in its basic form, referring to the concept of coordination convention, has inadequately narrow ambitions. The conditional character of coordination conventions cannot correspond to the whole scope of authoritative actions and therefore grant them normative force. The second, purely “instrumentalist” view on the role of conventions in determining legal authority is S. Shapiro’s “planning theory of law”. Shapiro claims that secondary rules (including the rule of recognition) are elements of a much larger shared plan that sets out the constitutional order of a legal system. The value of the plan is that, as S. Shapiro writes, it does the thinking for us.⁴⁵ The basic plan, shared by everyone (especially by legal officials) is not a plan that is determined in every detail. Plans are general but to some extent vague, and their application demands creating more detailed sub-plans. But if one accepts the overall intention to realize a general plan, one could realize it through more specified sub-plans almost automatically. To that extent, realization of the once accepted plan is clearly instrumental and there is no need to deliberate once again about the general plan’s value – and in fulfilling such an instrumental function plans are pre-emptory in character.⁴⁶ Shapiro stresses that subjects that accept the general plan have “rational authority” to obey it at its realization stage. In this sense each and every plan puts a rational pressure on its subjects and thus becomes a justified standard of conduct, although plans do not have any intrinsic moral value (to paraphrase Austin, “having a plan is one thing, its merit or demerit another”). In this theory the hope to explain law’s normativity rests on the notion of instrumental rationality that forces subjects who have accepted a general plan also to accept its particular effects. In this sense, plans determine certain roles/functions for agents, who accept them as norms. The whole structure of hierarchical and impersonal social planning is embedded within the structure of M. Bratman’s shared cooperative activity. This account, developed recently in great detail in Shapiro’s book Legality, suffers from a certain flaw.⁴⁷ As M. Bratman emphasizes, instrumental rationality cannot itself warrant pro tanto normative pressure that would lead to the unreflexive realization of the once accepted plan, because at each and every stage of the plan its realization

needs strong and substantial endorsement.⁴⁸ The only requirement of instrumental rationality is a wide-scope requirement of intention-effect coherence, which, however, may be fulfilled in a conditional way: acceptance of a general plan at the starting point means that either one has an obligation to act according to the basic intention, or – simply – abandon that intention and give up the realization of the plan’s requirements. The internal rationality of law, based upon the requirements of Instrumental Reason, is not sufficient to guarantee endorsement by the subjects taking part in “a legal game”. Instead, such endorsement can be warranted only by principles of substantial, critical morality.

5. Why Legal Conventionalism Fails – A Final Remark

The above description focuses on the main shortcomings of contemporary conventionalist theories of law’s normativity. Some other accounts of a conventionalist reinterpretation of Hart, namely A. Marmor’s idea that the rule of recognition is a constitutive convention, and M. Gilbert’s alternative proposal that the rule of recognition should be treated as a group fiat, were deliberately omitted. To some extent, the former conception accepts the minor role of coordinative conventions in law, but introduces different types of conventions, which are constitutive for institutional order of any kind. This idea is very interesting, but – the basic assumption is that these conventions are only partly autonomous (they are characterized only by “internal normativity”) and to that extent this conventionalist explanation is equated with the relativized, detached judgment about law’s identification and validity. The external, fully fledged reasons for obeying law are not connected with the notion of convention, since Marmor’s aim is to connect the non-cognitivist Hartian picture of the internal point of view with the notion of convention. In this case, although the burden of explaining law’s normativity is on external reasons for taking part in the legal game, it deprives the notion of convention of the meaning that some followers of Hart, through long discussions, have ascribed to it. As J. Dickson notices:

Marmor views Hart as offering an account of the rule of recognition wherein common judicial practices do not supply judges with primary or operative [external – A.D.] reasons for following that rule, but rather identify for them what they have the reason to follow, assuming the existence of underlying primary reasons for following it.⁴⁹

In this case the proper reason-giving character of law seems to be assumed. A. Marmor then tries to justify such an assumption by referring to the notion of deep convention and by applying a slightly amended vision of Razian authority. But still, the notion of convention here plays a different role from the role it has in conventionalist theories. In the case of M. Gilbert’s idea of the rule of recognition as conventional, group fiat, I will limit my commentary to the statement that this conception simply assumes the whole normativity of social rules, being kinds of self-addressed commands, and thus not only lacks any explanatory force in this matter, but also validates all objections against so-called “imperative theories of legal obligation”.

In order to answer the question “why legal conventionalism fails?” I focused on theories that not only refer to the notion of convention as a foundational social fact for legal order, but also give a certain, cognitive understanding of law’s normativity along the lines proposed by Hart. The examination of these conceptions and testing whether they can afford explanation of the law’s claim to general authority showed that they cannot do so (for different, but cognate reasons). Now, as far as it is plausible to adopt any cognitive understanding of the reasons that law claims to give (remember that this is the only way to present a fully fledged theory of law), the only reliable way out is to adopt a service conception of authority (or a related one) according to which the law’s claim to authority is actually a moral claim. In this case it is, however, important to remember that although it may sometimes be difficult to differentiate between the descriptive account of what law is and the political/moral justification of it (political or normative philosophy), the point is that when we explain the normative dimension of law – i.e. the claim to authority in its most abstract and general sense – we should introduce moral terms only to conduct a conceptual analysis, and not to assess the actual value of that claim (in other words: to answer the question whether the law actually does have authority). We should remember that the law’s moral claim is still only a claim which can be (and actually often is) false. The whole effort of evaluating the strong conventionality thesis in legal positivism, namely the thesis that a reason-giving character of law can be explained by reference to a specific kind of social fact, built of mutually recognized intentions and expectations (common knowledge), leads to the conclusion that these accounts are useful only to the extent that they aspire to give a reliable description of the structure of legal order (however questionable the sole description might be). But this description, even if reliable and adequate, cannot afford a theory of genuine, unconditional normativity. That is why we need (as J. Dickson states) at least conceptual theories concerning the justifiability of legal authority, and the character of legal obligation in order to answer questions such as whether and under what conditions judges ought to accept as binding and follow the rule of recognition of their legal systems.⁵⁰ The whole idea that Instrumental Reason, by application of the notion of convention, can replace the Substantial Reason with its normative claims, in the beginning of this article was shown to be intuitively doubtful. Now, we have reached the more specified conclusion in this matter in the context of legal theory. The fully fledged theory of law cannot explain the concept of law’s normativity by reference to any kind of “instrumentality principle” or “convention”. To show why legal conventionalism fails is to show that to substitute substantial claims with instrumental or conventional ones leads, at best, to conceptual confusion, and, at worst, to a theoretical catastrophe.