Three Conceptions of Law in Democratic Theory

Abstract Democratic theory tends to proceed on the assumption that law requires democratic legitimation because it is coercive. However, the claim that law requires democratic legitimation is distinct from claims about the nature of law. This paper takes issue with the notion that law is coercive by an exploration of three distinct understandings of the nature of law: the state-based conception of law, law as the rules of institutionalized normative systems, and law as social norms. Drawing on insights from legal and democratic theory, the paper defends the view that the ‘law’ to which democratic claims apply are the rules of conduct of institutionalized normative systems. Since rules that belong to such systems are found in associations beyond or below the level of the state, the scope of democratic participation is significantly wider than is usually recognized.


Introduction
It is widely believed that democratic participation is among the conditions for legitimate subjection to the law. A prominent view in support of this contention is that the subjects of legal systems are invariably subject to coercion. Hence, the idea is that legal subjects are presumptively entitled to democratic participation because they are subject to "a coercive order of public rules." 1 The democratic conception of legitimate law is closely related to the image that law is necessarily coercive. 2 The notion that law is coercive is often accepted without closer inspection by democratic theorists who remain largely oblivious of extant debates on the nature of law in legal theory. One reason why the literature on democratic legitimacy tends to ignore controversies on the nature of law might be that the context of the state is taken for granted. The claim that law is coercive gets credibility from the widely accepted image of the state as a coercive organization. Accordingly, the predominant assumption in democratic theory is not just that law is coercive but also that law is state-based. The familiar democratic principle that the 'takers' of the law should be the 'makers' of the law is transmuted into the precept that democratic rights are privileges of subjects to the laws of the state. 3 This paper takes issue with the predominant image of the law in democratic theory. To this end, three influential conceptions of 'law' are examined: law as the coercive order of the state, law as an institutionalized normative system, and law as social norms. These are familiar in legal theory and my aim is not to contribute to theories about the nature of law as such. But, as noted in other contexts, legal theory is bound to have significant consequences for political theory that are often overlooked, perhaps due to disciplinary boundaries. 4 Hence, the purpose of this paper is to exploit legal theory for the benefit of democratic theory. Drawing on insights in debates on the nature of the law, my aim is to advance our understanding of the relationship between 'law' and democratic legitimacy. What is that thing called 'law' that figures in claims to the effect that the subjects of law should be the makers of law?
The first argument advanced here is that the state-based conception of law runs counter to past insights about the relationship between coercion and the law. The law is neither necessarily coercive nor necessarily associated with the state. From the point of view of legal theory, law is a category that applies to a wider range of phenomena, either to social norms or to institutionalized systems of norms. However, these rival conceptions of law are not equally plausible candidates for law in claims of democratic legitimacy. The second argument advanced here is consequently that principles of democratic law-making generate boundary conditions for the relevant understanding of the law. Democratic participation is concerned, exclusively, with laws that are made in accordance with rules. Hence, the legitimation of law by democratic participation does not apply to law understood as norms sustained by normative attitudes (social norms). The concept of law that should take the centre stage in democratic theory is that of norms that depend on institutionalized normative systems: the claim for democratic legitimacy pertains only to the subjects of norms that are regulated by rules of higher-order and that include bodies tasked with their determination.
The position defended here has important consequences for the relevant domain of democratic participation and inclusion. Institutionalized normative systems are present in a variety of contexts above and below the level of the state. Indeed, it is not unusual to talk about democratic participation in civil society, at the supra-national level, or in the context of privately owned corporations. These contexts represent familiar sites of democratic participation and contestation. But to make sense of them as exercises of democratic law-making we should reject both the state-based notion of the law and the notion that law is necessarily coercive. On a state-based understanding of the law, the claim that the subjects of law should participate in their making either does not apply in non-state contexts or requires a very different justification. By contrast, on the updated understanding of the law as institutionalized normative systems, we can readily recognize the potential validity of claims to democratic participation in relation to a variety of non-state associations.

Democracy and Subjection to Law
The term 'law' figures in a variety of contexts, not all of which are relevant in the context of democratic participation. 5 Rights to democratic participation do not extend either to the laws of physics or to the laws of economics: subjection to either the laws of gravity or the laws of diminishing returns does not trigger a concern with democratic legitimacy. Democratic participation is concerned with decisions about rules that do-or purport to-set standards for behavior that ought to be complied with. A characteristic of the laws to which democratic principles apply is consequently that they are rules of conduct, or norms, not mere regularities of the natural or social world. This is of course consistent with a variety of claims about the additional features necessary and together sufficient for rules of conduct to be considered as 'law.' The laws of the state are evidently among the most important rules of conduct. The laws of the state apply in the jurisdiction of the state and are typically enforced by coercive sanctions. But the state is not unique in making rules of conduct that claim to be normative. A diversity of institutions are engaged in the regulation of behavior by complex systems of rules. There is canonical law, Jewish law, Islamic law, Hindu law, and other religious systems of rules that claim to be regulative. 6 There is international law as recognized and practiced by international 5. Raz points out that the general meaning of 'law' is "rules of some permanence and generality, giving rise to one kind of necessity or another." Joseph Raz, "Can There Be organizations and states. 7 And there is Indigenous law as practiced by a diversity of peoples independently from the state since time immemorial. 8 In addition, a myriad of voluntary associations is engaged in the creation of rules that seek to regulate the behavior of members. Housing associations, political parties, and sport clubs are in the business of making people conform to rules. Of particular interest are the rule-making activities of corporations. Business corporations are not merely 'firms,' economic organizations with the capacity to sell or produce goods and that structure the incentives of employees accordingly. The corporation is a "norm-governed" entity that regulates conduct by means of rules and institutions tasked with their application. 9 In sum, there is a broad range of entities that create and apply rules of conduct in ways that are closely analogous to that of states.
But there are rules of conduct that are neither created nor applied by associations. Rules that regulate behavior can emerge from more or less stable patterns of social interaction or what is sometimes described as social practices. Following the influential argument introduced by Lon Fuller some time ago, the law is basically a set of "stable interactional expectancies" that constitutes a "program for living together." 10 In effect, 'the law' is sometimes just the social norms of a particular community. These norms may represent the custom or traditions of specific communities that are recognized as authoritative for members. Social norms form the basis for traditional legal systems, as practiced by Indigenous peoples and in Anglo-Saxon legal traditions in the form of the 'common law.' 11 A characteristic of legal norms that are social norms is that they are not enacted by procedures designed for that purpose.
The variety of rules of conduct that are recognized as 'law' in different contexts is rarely reflected upon by democratic theorists. Democratic theory appears largely infected by the assumption that only the laws of the state are laws in the relevant sense. Robert Dahl frequently, though not always, refers to the subject of the "government and its laws." 12 This confirms that "the subjects of law" is primarily intended to capture the relationship that obtains between individuals and the laws of the state. 13 The view that democratic participation is conditioned by subjection to the laws of the state figures also in the work of Hans Kelsen. Kelsen argued that a person is "politically free" only if he is "subject to a legal order in the creation of which he participates." 14 By the 'legal order,' Kelsen invariably meant the laws of the state. No wonder then that democratic participation in the process of law-making is widely understood as a precept that applies exclusively to the coercive laws of the state. However, as already indicated, the notion that 'law' exists in contexts that are distinct from that of the state is not uncommon in legal theory. If this view is accepted, 'law' is a category that applies to phenomena that are independent of the existence of states. The claim that principles of democratic legitimacy apply to the subjects of law could accordingly be valid for the subjects of non-state associations too. The domain to which democratic principles apply would be considerably wider than usually acknowledged. Yet, any conception of law depends on theoretical considerations that differentiate the law from other rules and norms. 15 An account of law that is plausible from both the vantage point of legal theory and democratic theory must be responsive to insights from both disciplines.

Law as Coercive Order
The claim that 'law' necessarily refers to the 'laws of the state' sits comfortably with the claim that law is necessarily coercive. The laws of the state are coercive and differ from the rules of other associations by the fact that they are coercively enforced; the subjects of the laws of the state are "intermittently subject to coercion." 16 However, it is doubtful that the coercive powers of the state are enough to explain why only the state is able to make and enforce law. Though states are invariably coercive, so are a host of other associations. 17 An employer exercises coercion when employees are subject to sanctions for failure to comply with the rules of the workplace. Social clubs and associations engage in coercion when they exclude individuals from membership. International law is enforced by economic sanctions and, in the extreme case, by military intervention. Though the state has the capacity to enforce rules by coercive means, the state-based view of law inevitably depends on additional premises to justify the claim that only the laws of the state are 'law.' There must be more to the laws of the state than the fact that they are coercive. The claim that a concern with democratic legitimacy is premised on subjection to coercive laws is not sufficient to justify the claim that only the laws of the state raise a concern with democratic legitimacy. Now, the view that law is necessarily coercive has a long pedigree. In the following sections, two of the most influential versions of this claim are to be considered. Both take law to be coercive and insist that only the state can establish legal rules. 18 The first holds that legal rules summarize the commands of the sovereign. The sovereign is the body that controls the state, and the state has the capacity to maintain general obedience by means of coercive sanctions. Hence, the laws of the state necessarily depend on coercion and the capacity to establish society-wide compliance. Following the second view, law is a 'technology' for the regulation of coercion. The rules enforced by the state are unique on this view because only the rules enforced by the state are designed to regulate its coercive institutions. Though law is necessarily coercive on both views, they explain the basis of this claim in radically different terms. As noted by Norberto Bobbio, the first view identifies law and coercion based on the means employed. By contrast, the second view proposes that instructions for coercion sanctions are a defining content of legal norms. 19

Law as Orders Backed by Coercion
Following the legal theory developed by John Austin, the uniqueness of state law is due to the presence of two distinct features: coercive sanctions and sovereignty. 20 In line with predecessors like Hobbes and Bentham, Austin conceived of the law as 'commands' backed by threats of coercion. 21 The core claim defended by Austin is that only the commands made by the sovereign count as law. Though others may make 'commands,' only the sovereign is able to induce habitual obedience in the 'bulk' of the population. 22 The implication is that the laws of the state are distinct not just because they are coercive but also because only the sovereign has the power to establish wide-scale 'habitual obedience.' This is in the end why Austin thought that the laws of the state are distinctive.
18. Lamond distinguishes between three views of the relationship between law and coercion: i) that coercion is a defining element of law; ii) that coercion is the most prominent feature of law; iii) that coercion is one possible feature of law. The two accounts identified in this

Beckman
From this vantage point, the state-based reading of the democratic principle of inclusion applies only to the subjects of the state because only the state is able to establish law. However, there are well-known issues with Austin's conception of law, the most serious being that it fails to recognize the law as rules for conduct. Following Austin, laws are but commands to which subjects comply in order to avoid the "evil" of sanctions. 23 The law is little different from the raw force exercised by bandits, except that the state is able to wield coercive force on a larger scale.
To see this, consider a paradigmatic case of coercion-the highway robber that uses threats of physical coercion to deprive victims of their belongings. 24 The victim of the 'highway robber' is evidently subject to coercion. But the command 'give me your money' is not a rule. A defining attribute of rules is that they intend to regulate not just actual behavior but also hypothetical situations. 25 Hence, the highway robber does not subject victims to rules of conduct. The point is that subjection to coercion is insufficient to explain either why subjects of the state are subjected to the laws of the state, or why subjects of the state are not just subjected to the 'highwayman writ large.' 26 Austin is unable to differentiate between the victims of the highway robber and the subjects of the laws of the state. The only distinction between them, from Austin's viewpoint, is that the sovereign can establish obedience that is habitual.
Legal theorists from Kelsen to Hart and beyond have virtually unanimously rejected Austin's conception of the sovereign. However, they do not unanimously reject Austin's insistence that the laws of the state are coercive. The raw force of the laws of the state is particularly emphasised by Frederic Schauer. The capacity of the state to impose 'evil' in response to non-compliance significantly surpasses the capacities of other agents due to the fact that subjection to the state is inescapable. For most people, subjection to the coercive sanctions of the state is "nonoptional." 27 The point that the state is coercive and non-optional is of course reason to conclude that the state is of normative significance. The state's claim to a monopoly on legitimate coercion places a heavy burden on the justification of the state. But even if the laws of the state are both coercive and non-optional, it does not follow that coercion is a necessary feature of 'subjection to the law.' If the law is a set of rules for conduct, there must be more to the law than the fact that it is coercively enforced. And if there is more to the law than coercive sanctions, it is conceivable that entities that are neither states nor coercive can make and apply laws.

Law as Instructions for Coercion
Hans Kelsen offers a distinctive argument for the claim that "all law is state law." 28 Kelsen depicts a legal system as a hierarchy of norms where the Grundnorm has replaced the sovereign as the ultimate source of legal validity. Sovereignty is not external to the legal system and founded in the capacity to secure compliance, as Austin thought. Instead, sovereignty is a legal capacity that depends on and is therefore construed by the legal system. 29 Kelsen nonetheless saw coercion as a defining element of legal systems and one that separates them from other social systems. Coercion is a characteristic of law because the nature of law is to regulate the coercive acts of public officials. The law is a particular technology for the regulation of coercion. 30 Kelsen thus recognizes that subjection to the laws of the state is different from mere subjection to coercion. Public officials are subjected to norms of the legal system, whereas the general population is subjected to the coercive acts of public officials that are regulated by these norms. 31 This account is much more sophisticated than that offered by Austin, as it avoids the reductive tendencies of the latter and is able to explain the difference between legal systems and the 'highway robber.' In contrast to the bandits and robbers, the state secures compliance only by coercive acts that are regulated by a hierarchy of norms.
Yet, Kelsen shares with Austin the conviction that law is essentially dutyimposing. The law is a system of legal 'oughts.' The law requires that subjects avoid behaviours that qualify as conditions for the imposition of coercive sanctions. But as noted by Hart and Raz, this is to ignore the fact that the legal norms are not all duty-imposing. 32 Legal systems do not merely include norms that define duties but also norms that grant permissions, confer powers, and identify immunities-none of which are duty-imposing. duties for subjects to comply with. The laws of marriage define legal powers by which subjects are able to establish a particular legal relationship. 33 For Kelsen, the legal duties of the subject population are necessary conditions for coercive sanctions. But laws that define legal duties do not always conform to the explanation offered by Kelsen. It seems clear that a person can be subject to legal duties that do not depend on instructions for coercive sanctions. Raz argues convincingly that such legal duties do exist. 34 For example, the law regularly defines legal duties to comply with the legislative procedures for members of the parliament. Yet, members of parliament are not liable to either prosecution or coercive sanctions for failure to comply with these duties.
In sum, Kelsen's claim that subjection to the laws of the state can be fully accounted for by reference to the conditions for legally authorized coercion is a distortion of the nature of legal systems. The claim that the laws of the state are legally regulated acts of coercion and that the laws of the state are therefore distinct from other rules of conduct should be rejected. That coercion is not a defining element of the law is arguably well recognised in legal theory today, while less so in democratic theory, which instead remains under the spell of either Kelsenian or Austinian conceptions.

Law as an Institutionalized Normative System
How then should 'law' be understood if coercion is not among its defining attributes? Perhaps the law is a phenomenon that does not depend for its existence on any of the defining attributes of the state. In that case, we should expect that subjects of the law are to be found in domains that are distinct from the state. But even so, the relevant account of the law must differentiate the law from rule-governed relations that are not 'legal.' An influential view is that 'the law' is referring to rules of conduct that belong to a system of rules-a normative system. Rules that regulate conduct are part of a normative system if and only if they are themselves regulated by rules. The rules that regulate rules of conduct have a distinctive purpose: they determine how to create, revise, and abolish rules of conduct. Hart named the former "primary rules" and the latter "secondary rules" and argued that a "distinctive feature" of law is that it represents a system of rules in the precise sense of providing for a "union" of primary and secondary rules. 35 Laws are thus rules of conduct that are themselves governed by rules of a particular kind. By this account, rules that are mere habits, conventions, or unilaterally declared instructions, do not qualify as law. They are not legal rules because they are not the product of normative systems.
The notion of a normative system is unlikely to be sufficient for the differentiation of law from other rules of conduct, however. Only systems of rules that are

Three Conceptions of Law in Democratic Theory
institutionalized should be recognized as legal. A system of rules is institutionalized only if some agent is tasked with "ensuring conformity" and "dealing with deviations" from the rules. 36 According to MacCormick, the mark of an institutionalized normative order is procedures for "settling and finalizing disputes" about rules. 37 Why must normative systems be institutionalized in order to count as 'law'? The answer is that the law exclusively refers to rules that are practiced and that institutionalization separates normative systems that are practiced from normative systems that are not. Norms that are not practiced might be part of normative systems. It is conceivable that an extinct legal system can be fully reconstructed by students of the law that provide a full account of the primary and secondary norms of that system. Yet, if that legal system is dead, the norms that belong to it are no longer 'laws.' The reason why is that no institution exists with the authority to determine the content of valid norms of that system.
Thus, legal normative systems do have institutions with the capacity to make decisions on the norms of conduct that apply to subjects. Such institutions are of course present in well-functioning states. However, we should carefully avoid the assumption that all institutions of states are necessary for the institutionalization of systems of norms. Following Raz, only "primary institutions" tasked with the "authoritative determination of norms" are necessary. 38 The point is that the capacity to authoritatively determine norms is distinct from the capacity to enforce norms. 'Norm-enforcing institutions' are not necessary for the institutionalization of a normative system.
In the context of the state, courts of law represent the primary institution par excellence. This is not to deny that a variety of other public authorities are also involved in the determination of the law. Taxes are determined by tax authorities, social benefits are decided by various welfare authorities, and so on. Similarly, non-state systems of norms provide their own specific assortment of mechanisms for the determination of norms. The board of the university department decides about the internal rules of the department; the CEO determines the rules that apply to the corporation; the board and ultimately the annual meeting is the final arbitrarer of the rules that apply to the housing association, and so on. They are all 'primary institutions' of normative systems that are distinct from the state. Moreover, though some institutionalized normative systems include 'norm-enforcing institutions,' it is not necessary that they do. This is yet another illustration of the point that institutions with the capacity to enforce norms are not necessary for the existence of an institutionalized normative system. 'Law' is possible even in the absence of bodies that enforce rules by coercive means.
The implication is that the legal system of the state is not the only association that is governed by law. There are also institutionalized systems of law in sport 74 Beckman associations, social clubs, educational institutions, trade unions, and so on. As observed by Raz, "the features of legal systems : : : are not peculiar to legal systems." 39 There's but a minor terminological difference between Raz's conclusion and the claim defended by Brennan and colleagues according to which clubs and associations are "non-legal formal systems of rules." 40 Though they prefer 'non-legal,' Brennan and colleagues are keen to identify corporations, voluntary associations, and other associations as formalized normative systems that make rules applicable to their members.
In fact, it may be that institutionalized normative systems are also to be found in associations that are not part of 'civil' society, for instance, pirates and other outlaws. 41 The rules adopted by a gang of pirates correspond to a normative system to the extent that they identify normative relationships that apply to the members of the group while also including rules for the making and revision of these rules. The normative systems of pirates are institutionalized to the extent that they include mechanisms for the authoritative determination of the rules that apply to them. 42 The subjects of law are on this view to be found in a variety of contexts beyond the state. Given that a precept of democratic legitimacy is that the subjects of law should presumptively be entitled to participate in the making of law, the domain to which that precept applies must be expanded beyond the laws of the state. Democratic participation in the creation of legal norms should no longer be thought of as a possibility and ideal only among the subjects of the state.
There are of course objections against this broader and more inclusive understanding of the law. One is that it ignores the special normative significance of the laws of the state. The laws of the state are different from the norms of other institutionalized normative systems, as the legal system of the state claims for itself the right to regulate all normative systems within its domain. In the words of Raz, the legal systems of the state do not "acknowledge any limitation of the spheres of behaviour which they claim authority to regulate" and consequently insist on "comprehensive" legal authority. 43 The laws of the state are of special normative significance because their claims to authority are virtually inescapable. The consequent argument would be that democratic legitimacy should primarily be considered as a normative standard for the laws of the state. Though other institutionalized normative systems may exist and though we might call them 'legal,' only the laws of the state are of sufficient importance to conclusively demonstrate the need for democratic legitimacy.
The objection is premised on the claim that standards of democratic legitimacy apply only to entities of sufficient normative significance. Democratic participation in the process of law-making is a requirement for the legitimacy of law that applies to the laws of the state due to the overwhelming importance of the interests at stake. Stronger and weaker versions of this precept are clearly imaginable. On a weaker version, democratic legitimacy is not exclusively applicable to the laws of the state but in proportion to the normative significance of an institutionalized normative system. However, both stronger and weaker versions are premised on the claim that normative considerations decide the extent to which standards of democratic legitimacy apply. This premise is likely to be mistaken for reasons to follow.
The notion that an institutionalized normative system should be accountable to standards of democratic legitimacy depends on the possibility that such standards apply to that entity. The claim that an entity ought to be democratic makes no sense unless it is possible for that entity to be democratic. The possibility of an entity being democratic does not depend on normative reasons but on conceptual considerations. Hence, any claim to the effect that an entity should be democratic is premised on a prior conceptual claim to the effect that the relevant entity is one that satisfies the conceptual criteria according to which the application of democratic standards is possible. The point is that normative assessments inevitably depend on prior conceptual judgments on the domain to which normative standards apply. This then is why we must reject the claim that the normative significance of the state is sufficient to determine the domain to which standards of democratic legitimacy apply. That view is parasitic on conceptual reasons to believe that the state is an entity to which democratic standards apply.
The notion that democratic standards apply to the subjects of law represents a claim about the relevant conceptual preconditions of democratic participation. The content of that conceptual claim depends on an account of 'law.' One such account is the theory according to which 'laws' are but rules of conduct embodied by normative systems that are institutionalized. The precept that the subjects of law should be able to participate in the making of the law should on this account be understood to mean that democratic participation is conceivable for the subjects of institutionalized normative systems. Thus, the conceptual universe of democratic legitimacy is not limited to the laws of the state but extends to all associations that provide for systems of norms and the institutions required for their determination.

Law as Social Norms
The previous section introduced a wider account of law than the traditional state-based conception. Yet, even the wider account might be considered overly narrow. Law can be construed to refer to rules of conduct grounded in conventions or mutual expectations. So understood, the law permeates all social contexts 76 Beckman where human behavior is ruled by normative standards and comprises rules that regulate everyday practices such as eating, saluting, and dressing. In fact, social norms do not merely regulate everyday practices but also behavior in business and politics. Social norms are normative standards of behavior reflected in more or less stable and shared attitudes within some particular social cluster. Following writers in the tradition of 'legal pluralism,' there is no reason why social norms should not also be considered 'law.' The term 'law' extends to "any set of observed social norms." 44 This view traces back to Eugen Ehlrich, who famously called for the study of the 'living law,' by which he referred to rules of conduct that participants in any social context recognized as binding for them. 45 Social norms are 'laws' that are embodied in 'shared practices' that are not written down in law books or enacted by formal decisions. Refusal to recognize a body of norms as 'law' for no other reason than they are not codified is to mistake a particular technique of law-making with the object itself. 46 The claim that the concept of law extends to social norms is of course controversial. The immediate objection is that social norms are not enforced. The separateness of law is reflected in the distinction between "enforced norms" and "lived norms." 47 Social norms are 'lived norms,' and though they are standards of conduct, non-compliance is not subject to sanctions.
It is unclear whether social norms are necessarily un-enforced, however. Behavior that violates established social norms is typically subject to negative attitudes. Someone who, for example, ignores the rules of etiquette in a restaurant is likely to face rebukes from others, or at least frowning eyebrows. In the case of gross violations of etiquette in public spaces, the agent may be expelled and denied access.
However, the argument that law is necessarily enforced and that social norms should therefore not be considered as laws can be restated in terms that make this reply moot. Conceding that social norms are occasionally enforced by negative attitudes, the point is that social norms are not correlated with 'tangible' sanctions. 48 The argument is that sanctions following violations of social norms are not sufficiently substantial for them to qualify as law.
Evidently, the correctness of this claim depends on what counts as 'tangible.' On one reading, sanctions are 'tangible' only if they would inflict a substantial cost on the victim. But on that reading, non-compliance with social norms could conceivably be subject to tangible sanctions. Harsh punishment directed against "socially deviant" behavior is well-known. 49 The sanctions delivered against violations of social norms may include 'naming and shaming' and forms of social ostracism intended to deny access to the benefits of social community. Surely, these sanctions can be more costly to the victim than pecuniary or even physical punishments. Adam Smith is known to have said that "[c]ompared with the contempt of mankind, all other external evils are easily supported." 50 The merits of the argument that social norms do not qualify as 'law' because they are not associated with tangible sanctions is in other words dubious.
The final objection against the notion that social norms are law is that the sanctions associated with law are necessarily regulated by rules. Legal sanctions are governed by rules and executed by "specialized enforcers." 51 By contrast, the sanctions imposed in cases of non-compliance with social norms remain unstandardized. The reason why social norms are not 'legal' is not that sanctions are insufficiently severe but that sanctions are insufficiently standardized.
It is certainly true that sanctions pursuant to violations of social norms are unregulated and therefore potentially arbitrary and haphazard. Yet, this observation represents a valid argument only on the premise that law is necessarily associated with sanctions. In the case that the existence of legal norms, or even of legal systems, is not premised on coercive sanctions, there is no basis for the claim that law is necessarily associated with regulated sanctions. That law need not depend on coercion has already been established. Rules of conduct that are recognized as binding by an institutionalized normative system are legal whether or not they are coercively enforced. In the case that law need not be coercive, the fact that social norms are not coercive should not lead us to reject the claim that social norms are laws.
Social norms are of course different from the laws of institutionalized normative systems exactly because social norms are neither regulated by secondary norms nor amenable to authoritative determination. These differences are obvious in the case of rules of etiquette. The norms that regulate how to eat in public spaces are not made according to rules, not interpreted according to rules, and not revised according to rules. Nor are rules of etiquette institutionalized, as is evident from the fact that no agent has the authority to finalize judgements about their content and application.
These remarks are sufficient to re-affirm the distinction between institutionalized normative systems and social norms. Yet, it is unclear that it justifies the conclusion that social norms should not be recognized as 'law.' Advocates of 'legal pluralism' insist that clusters of social norms can legitimately be studied as articulations of law even if they are neither systemic nor institutionalized.

Beckman
A sufficient condition for the existence of legal rules on this account is that there are rules of conduct that are recognized as binding.

A Democratic Conception of Law
Our interim conclusion is that the dominant understanding of law in democratic theory receives scant support from legal theory. Neither the notion of law as necessarily coercive nor the idea of the law as the prerogative of the state can be sustained. The implication is that the democratic notion that the takers of law should be the makers of law is not exclusively applicable to the subjects of state law. The law is a wider phenomenon and democratic law-making is conceivable in contexts that are neither coercive nor associated with the state.
On the other hand, the relevant understanding of 'law' remains ambiguous. There remain two distinct and ultimately incompatible conceptions of the law: the legal pluralist view according to which the law is constituted by social norms, and the legal positivistic view according to which the law is constituted by institutionalized normative systems.
Though it may be that both represent plausible candidates of 'law' in legal theory, it does not follow that both are plausible candidates for democratic law-making. In fact, on the basis of democratic theory, there is reason to accept only institutionalized normative systems as the relevant loci for claims to democratic participation, or so I shall argue in what follows.
The language of democratic participation pertains to procedures for the making of decisions. Procedures are rules, and procedures for decision-making are, accordingly, rules that regulate how decisions are to be made. The notion of democratic participation is for these reasons premised on the existence of a particular regulatory framework that establishes a specific blend of "rule-governed relations." 52 Democratic participation is conditioned by procedures for decision-making that are constituted by rules that serve to regulate the activity of participation.
The notion that democratic participation depends on rules serves an additional purpose in regulating when the activities of participants should count as decisions. An entity (say, a parliament) that recognizes democratic rules for decision-making is unable to make decisions unless it acts in ways that satisfy the procedural requirements of democracy as specified by its own rules. The fact that a majority in the parliament 'likes' a piece of proposed legislation is not sufficient for the proposal to be adopted by the parliament. 53 Democratic decision-making is a rule-governed activity that identifies the necessary and sufficient conditions for outcomes to be recognized as binding. Now, the claim that democratic participation depends on rules that regulate decision-making does not preclude disagreement on the content of these rules. In fact, the notion that democracy is a rule-governed activity is consistent with the view that democratic rules for decision-making may include non-procedural conditions. However, it is safe to say that a necessary precondition for democratic participation is rules that define some procedures for decision-making.
The claim that democracy is a rule-governed activity is acknowledged in otherwise conflicting accounts of democracy. Consider, for example, the well-known distinction between thick and thin conceptions of democracy. On thick accounts, democratic procedures depend on rules that determine the "rights, liberties, and resources sufficient [for a people] to participate fully, as equal citizens, in the making of all the collective decisions by which they are bound." 54 A thick conception of democratic procedures evidently stipulates rather demanding conditions for decisions to be fully democratic. Not so on thin accounts of democratic procedures. On such views, rules for decision-making are democratic if they allow participants to competitively select a winner. 55 These conditions are clearly much more lenient. Notwithstanding the differences between thick and thin accounts, the point is that both are premised on the idea that democratic participation is necessarily regulated by rules and that only decisions that comply with these rules should count as democratic.
This simple observation does have important implications for the conception of law that is relevant to democratic theory. If 'democracy' depends on rules that regulate procedures of decision-making, it follows that 'laws' can be democratic only if they are made by decisions that are regulated by democratic rules. In other words, democratic participation is conceivable only in the making of laws that depend on rules for their existence. Now, the notion that the creation of legal norms depends on a regulatory framework is explicitly rejected by the pluralist theory of law. On this view, legal norms are social norms that emerge from social practices supported by normative attitudes. The law refers to behavioral norms that are internalized among individuals of a particular social setting. The point is that no rule-governed activity is required for the existence of law on the pluralist account. Social norms do not come into existence through procedures but emerge spontaneously through patterns of social interaction. Thus, 'law' that is constituted by social norms is a rule-independent entity. But if democratic participation is conceivable only in the making of laws that depend for their existence on rules, it inevitably follows that democratic participation is not feasible in the creation of 'law' as defined by the pluralist theory.
To illustrate this point, consider again rules of etiquette. Rules of etiquette are typically though not necessarily social norms that purport to be normative standards of conduct in everyday settings. Social norms of etiquette depend on social practices supported by relevant normative attitudes. Now, imagine a person that is either unaware of the norms of etiquette or that deliberatively ignores them. That person is still subject to the normative attitudes that constitute the social norms of etiquette. Where social norms are recognized as 'laws,' we should consequently acknowledge that a person that fails to comply may still be subject to the 'law' of etiquette. On the dictum that laws are democratic only if the subjects of law are able to participate in the making of law, it is tempting to conclude that the laws of etiquette are democratic only if everyone subjected to them is able to participate in their making. But this principle of democratic legitimacy arguably makes no sense. Participation in the making of the law is premised on a rule-dependent conception of the law. Only laws that are created by decisions regulated by rules can be the object of a democratic decision. Since the 'laws' of etiquette are social norms, they are rule-independent entities. No democratic decision is therefore possible on the content of the laws of etiquette. Against this conclusion, we might object that it is possible to make decisions on rules of etiquette. There can be rules that regulate decisions on etiquette just as there can be rules that regulate decisions on other issues. Imagine the following example: the guests at a dinner party discover that they all behave according to different and conflicting standards of etiquette. They wish to reduce confusion and coordinate themselves by making a decision on what norms of etiquette to comply with. In order to make the decision, the guests need to agree on rules that regulate the procedure to follow in making the decision. Once these rules are agreed on, they can make a decision on the rules of etiquette at the dinner. Arguably, this sequence of events serves to illustrate that social norms can be the object of democratic participation. Once the guests are subject to rules that are rule-dependent, they are subjected to 'laws' that can be decided by democratic procedures. The subjects of social norms are in other words able to decide them by recourse to democratic procedures.
In reply, we should point out that 'laws' established by a decision-making procedure are no longer social norms. When the guests agree on rules that regulate the procedure to follow in deciding the rules of etiquette, they are effectively introducing a normative system. The 'laws' produced by normative systems are no longer 'laws' in the sense defined by the pluralist theory. Instead, they are 'laws' in the positivistic sense of being primary norms of conduct in a system of norms that depend on secondary norms regulating the creation and revision of primary norms. The point is that norms of etiquette that are decided by democratic procedures are no longer social norms but the norms of a normative system.
It is tempting to conclude that the democratic conception of 'law' is equal to what is commonly referred to as positive law. Democratic claims apply to rules of conduct that are 'positive' in the sense of being part of institutionalized normative systems. That reading should be treated with caution, however, as the idea of 'positive' law is multi-faceted. Positive law denotes both rules that are made Three Conceptions of Law in Democratic Theory and rules that are arbitrary. 56 The notion that positive law is necessarily 'made' is uncontroversial. Laws that are 'laid down' in accordance with rules are part of the conception of law that is relevant for democratic theory. But the extent to which laws decided by procedures must necessarily be 'arbitrary' is controversial. Though a decision that is regulated by rules does not preclude outcomes that are either morally deficient or illegitimate, it is controversial that morality imposes no limit on what may properly be termed as 'law' governing human conduct. 57 It consequently remains contested if arbitrariness should be part of the meaning of positive law. The conclusion then is that the conception of law in democratic theory is 'positive' only in the sense that it refers to rules of conduct that are made in accordance with the rules of institutionalized normative systems.