Why we should see international law as a structure: Unpicking international law’s ontology and agency

This article identifies how three dominant ideas of international law (as a process, an institution and a practice) see its agency, concluding that all three share a reluctance to see international law as doing anything more than enabling the operation of other actors, forces or structures. This article argues that we should see international law as a structure because it possesses both the surface structure of rules, principles, processes, personnel and material elements of the international legal system and a deep structure of values that sits deep within our subconscious. As Shklar’s idea of legalism shows us, legalism plays a powerful role in shaping all our understandings of ourselves and the world that surrounds us. Seeing international law as a structure enables us to see how it locates actors within a social hierarchy and how it behaves in similar ways to recognised structures like capitalism and racism.


Introduction
The debate over the power of international law has been a defining feature of international legal theory and International Relations (IR) theory accounts of international law for over two decades. 1 The two disciplines' very different conceptions of power have been a source of conflict and confusion. As Bianchi notes political scientists accuse international legal theorists and lawyers of being 'inattentive to the element of power.
Lawyers, in turn, would retort that the authority of law must be distinguished from the force of power. It would perhaps help both sides to realize that law is power'. 2 This article argues that before this is possible, we must first work out what international law's agency is: what types of agency does it have? Is it an actor? Can it have an independent effect on events? First of all, however, agency itself must be defined. For the purpose of this article, agency is defined as the ability of an actor to exert influence, affect the outcome of a process or event, to make happen or prevent an event or action. 3 'Actor' here is not restricted to humans but can include non-humans. Drawing upon Actor-Network Theory (ANT), I speak of 'actants'. An actant is anything that is 'granted to be the source of action' 4 and there is no assumption that an actant must be human and so the notion of what it means to 'act' is broadened. Seeing actants as possessing agency is central to the argument that international law has agency and will be expanded upon below.
This article outlines three of the most common ways of understanding international law that are common in the IR/international law (IL) literature and teases out their, often implicit, ideas of international law's agency. The first sees international law as a channel, the second as an institution and the third as a practice. 5 The article then goes on to argue that these different ways of understanding international law fail to see the true extent of its causal power. Despite the negative perception of structures held by most mainstream thinkers in IR-IL, there are important benefits to be gained from seeing international law as a structure. This is more than just a heuristic device, however. Viewing international law as a structure enables us to see the full extent of international law's agency and thereby offers a more accurate and nuanced account of what international law can and does do. Bianchi, while entering the usual caveats about the diversity and complexity of international legal theory, offers a definition of a 'classical international legal positivism': international law is a system of objective principles and neutral rules that emanate from States' will, either directly through treaty or indirectly through custom, and ... States are the primary subjects of the international legal order ... a strict test of legal validity must be passed for a rule to be conceived as a binding rule of law, and ... extra-legal considerations (economic, social, moral, or political) are alien to legal analysis. 6 The 20th century's rejection of such black-letter views of law saw a gradual widening of law's ontology, in two respects. First, it widened the scope of international law beyond the state and beyond rules and principles. Second, it saw extra-legal considerations as playing an important role. As the ontology of law widened, so too did the assessment of law's agency.

International law as a process: expanding law's agency
For process scholars, the traditional understanding of international law as rules and principles is too narrow and misses out important questions about how those rules come to be and how they function in practice. Shifting international law's ontology from rules to process has significant repercussions. Most significantly, analysis must start much earlier by looking at how and why particular legal rules emerge in the form they do.
Analysis must also be much broader, seeing legal rules as just one of several elements which can explain state behaviour. 7 It is not simply the legal rules that we should study, but the process of how those rules come to be and the process of their application and interpretation. Chayes' seminal text on the Cuban Missile Crisis rewrote how both international legal thinkers and IR theorists understood international law. It also changed how international law's agency was and continues to be understood. Instead of seeing international law as acting as a constraint on action -a very high bar which is rarely cleared, Chayes argues that international law's power operates in less obvious ways. First, it functions as a basis for justification or legitimation for action. For Chayes, legal justification is a particularly persuasive form of justification because it must proceed in terms of 'more or less universal and generalised criteria'. 8 This means that the public find it more persuasive than non-legal or purely political justification. Second, it provides the organisational structures, procedures and fora in which decisions are made. 9 This is particularly true of international politics where international organisations are, in Chayes' words, 'dominated by legalistic modes of procedure' ... [and thus are] 'responsive to legal argumentation and acting by law-created processes'. 10 Third, it is seen as providing accountability -a much prized political value for Chayes. This ability to provide accountability, and to be the arbiter, allows international law to occupy a powerful place in international politics by being able to stand apart from (and above) it. Finally, much of law's agency comes from the omnipresence of law and the legal mindset. International fora are legally constituted bodies, often with an explicit legal dimension or purpose. Moreover, the presence of lawyers in international politics, as well as the prevalence of legal education among decision-makers and staff at nongovernmental organisations (NGOs) and international institutions, brings the legal mindset into international politics further still. It is not only that so many actors are legally trained, but that the method itself is also seen as a highly persuasive (and hence appealing) form of justification, both for elite political actors and their audiences. Law's agency comes from its omnipresence in the institutional structure of international politics, in the personnel of international politics, and in the perception of legal reasoning and the law as particularly persuasive forms of justification.
So, what conclusions can we draw about international law's agency? Importantly, international law is not seen as a 'superego setting outer limits to permissible action'; 11 its influence does not operate in terms of 'yes or no', but lies in its role in defining and shaping possible actions. 12 It is, in other words, a channel, making certain outcomes more or less likely but not definitively causing them. International law here lacks direct causal power. As Chayes is keen to point out, international law does not definitely decide what a state can and cannot do. Instead, it frames the options, making some more appealing than others.

Law as an institution
Institutional accounts of international law fall into two camps. First, there are accounts which focus on formal institutions which are specific, concrete and usually formally designed. These are narrowly focused on formal, treaty-based law and see international law's agency as minimal. This article will instead focus on accounts that see international law as a 'broad social phenomenon deeply embedded in the practices, beliefs, and traditions of societies, and shaped by interaction among societies'. 13 For the English School, international law is a social institution, rather than a formal one: 'an established custom, law, or relationship in a society or community', 14 or, in Bull's words, 'a set of habits and practices shaped towards the realisation of common goals ... an expression of the element of collaboration among states in discharging their political functions -and at the same time, a means of sustaining this collaboration'. 15 This is not to say that the English School is uninterested in formal institutions. Rather, it sees formal institutions as secondary institutions which are the manifestation of (arguably, more interesting and powerful) primary institutions which 'represent fundamental underlying norms, and are more evolved than designed'. 16 Secondary institutions are 'relatively specific, concrete, and usually designed (mainly intergovernmental organizations and regimes)'. 17 It is primary institutions which exert more power because they constitute the foundations of international society, including secondary institutions.
Crucially, international law is seen by English School thinkers as one of international society's primary institutions. 18 However, not all primary institutions are equal: certain institutions determine others. For James, Holsti and Jackson, this is sovereignty, and international law is derivative of it. Buzan's notion of nesting is helpful in explaining how different types of primary institution relate to each other. He argues that some primary institutions: Contain ..., or generat[e] ... others. International law, for example, can be seen as a general institution, a set of fundamental principles, and also as the container of the potentially endless particular laws about a wide variety of specific issues that can be built up within it, and which mostly fall under what I have labelled here as secondary institutions. 19 But for Buzan, 'the trick is to find primary institutions that stand alone', 20 deriving from no other institution and international law is not one of them. For Buzan, international law derives from sovereignty. This means that while international law has some agency, it is not a master institution (Buzan's hierarchy being: master, primary and secondary institutions). The master institutions are sovereignty, territoriality, diplomacy, balance of power, equality of people, inequality of people, trade and nationalism. 21 This means that international law for Buzan is a derivative primary institution, deriving from sovereignty. 22 International law in turn produces secondary institutions for Buzan, such as regimes and international courts. 23 So, we see a reticence to see international law as possessing a particular type of agency: the normative power to shape values, a kind of master value from which others derive despite a recognition of international law's significance. This is combined with the belief that independence is a key marker of agency and that international law falls short here because of the belief that it derives from sovereignty. So, while the English School expands international law's agency by recognising the role it plays, we are still a long way from seeing international law as possessing significant agency or as being an actor in its own right.
Reus-Smit, a constructivist who draws heavily on the English School, again sees a hierarchy of different kinds of institution. 24 At the top there are superficial institutions such as specific regimes like the General Agreement on Tariffs and Trade (GATT); beneath these lie fundamental institutions, including international law, and beneath these lie international society's normative foundation which comprises three constitutive values: a shared belief in the moral purpose of centralised political organisation, an organising principle of sovereignty and a norm of pure procedural justice. 25 But it is the shared belief in the moral purpose of the centralised political organisation that is the core of this normative complex and it 'provides the foundations for' the other two. 26 Again, international law's agency is limited.
While a norm of pure procedural justice is derivative of shared ideas about the moral purpose of the political organisation, it still has considerable agency. For Reus-Smit, its agency operates via three mechanisms. First, norms of pure procedural justice define the 'cognitive horizons of institutional architects; they define what is imaginable'. 27 And so when creating ways of coordinating and cooperating between political units, international law colours what decision-makers believe is possible and desirable. The second mechanism by which norms of pure procedural justice shape the fundamental institutions of international society is that they operate as the metanorms that structure the process of communicative action that surrounds the production and reproduction of fundamental institutions. Fundamental institutions are sets of prescriptive norms, rules, and principles that specify how states 'ought' to resolve their conflicts, coordinate their relations, and facilitate coexistence. 28 Norms of pure procedural justice represent that 'higher order values' that structure states' moral dialogue about what the norms, rules and procedures should be that order their international society. The third mechanism operates once the majority of states reach an ideological consensus about the primacy of prevailing systemic norm of pure procedural justice. When this happens, other states who do not have 'a deep cognitive or moral attachment to them' are still constrained because they are part of a society of states. 29 When these states wish to justify their behaviour, they have to do so in reference to the system's primary norms and so have to accept them. In other words, if these other states want to play the game, they have to abide by the rules that the original players set. We therefore see a slightly different, arguably greater agency here for law. There is a recognition of law's imbrication in the international system as well as the impossibility of stepping outside of the law. Here, we see law as constraining behaviour or acting to locate actors in certain places in the hierarchy and compelling certain behaviour (because powerful actors high up in the hierarchy drive the ideological consensus). Reus-Smit's recognition of states' inability to step outside of law is an important move towards a greater idea of international law's agency, one where international law affects actors' material circumstances in a compelling manner.

Law as a practice: the minimalist idea of international law's agency
The idea of law as a practice has proved popular with theorists of international law and IR alike with both drawing upon social theory and sociology. 30 Several approaches use the term 'practice' -'repeated interactional patterns', 31 to describe law but this article argues that there is a significant divide in these approaches, with one side seeing law's agency as reduced by being a practice, while the other sees it as radically expanded. The first, minimalist interpretation of international law's agency includes work by Brunnée and Toope, and Rajkovic, Aalberts and Gammeltoft-Hansen. 32 The second, maximalist interpretation is found in Gramscian accounts of international law, which see law as having a far greater role in the operation and ordering of society.
For Brunnée and Toope, it is the process of interacting with international law that creates an actor's sense of obligation, and gives law its agency. 33 They argue that three elements must be present for legitimate legal norms which exert a sense of legal obligation to be created. First, legal norms can only emerge from a context of social norms. Second, these norms must meet the criteria of legality which comprises eight internal criteria: 1. Legal norms must be general, prohibiting, requiring or permitting certain conduct; 2. They must be promulgated, and therefore accessible to the public, enabling citizens to know what the law requires; 3. Law should not be retroactive, but prospective, enabling citizens to take law into account in their decision-making; 4. Citizens must be able to understand what is permitted, prohibited or required by law -the law must be clear; 5. Law should avoid contradiction, not requiring or permitting and prohibiting at the same time; 6. Law must be realistic and not demand the impossible; 7. Its requirements of citizens must remain relatively constant; 8. There should be congruence between legal norms and the actions of officials operating under the law. 34 The third and final element is a continuing practice of legality whereby legal norms are created, maintained and altered over time. The social and legal must interact and that interaction must be ongoing. The practice of legality is insufficient on its own: it must conform to the criteria of legality. 35 If not, it is not law and exerts no specifically legal compliance pull (though it may have a social compliance pull). As they argue: 'The interactional theory shows that a social norm, reflecting a shared understanding[,] that meets the criteria of legality is upheld through practice that is congruent with the norm'. 36 This constitutes a 'practice of legality' and 'it is practice itself that grounds continuing obligation, but practice rooted in the criteria of legality'. 37 Similarly, Rajkovic, Aalberts and Gammeltoft-Hansen see international law as a practice and a series of 'social performances'. 38 They use Adler and Pouliot's definition of practice as 'patterned activity' 39 that is attributed social meaning; it is the '"stuff" that state agents and other international actors do, on an everyday basis'. 40 In this way, the number of factors which shape (and are shaped by) the practice of legality is expanded beyond the 'formalist box of doctrinal international law ... to ... a broader plurality of actors and scholarships, and also over a multitude of formal and informal international settings'. 41 However, where Brunnée and Toope see all of international society as relevant, Rajkovic, Aalberts and Gammeltoft-Hansen focus on three communities: scholarship, institutions and policymaking. 42 Moreover, as Schindler and Wille argue, Adler and Pouliot's approach overestimates continuity and underestimates uncertainty. 43 It is also very focused on human activity and social meaning, exacerbating Neumann's criticism that discursively based approaches have focused too much on language and have wrongly downplayed material conditions. 44 Seeing international law as a practice emphasises the way in which law is socially embedded. The legal and the social must interact and this produces a more complicated sense of law's agency because it is no longer separated from society and acting upon it. Rather, it is embedded within it and loses the freedom that comes from separation. 45 Consequently, the minimalist idea of international law's agency sees law's imbrication in society as a lessening of its agency from the traditional black-letter view of law that Bianchi outlines. Its ontology has expanded, and more actors are involved, but this also produces more brakes on action.

Practice: the maximalist idea of international law's agency
The alternative is to see law's imbrication in society as considerably expanding law's agency by seeing law as deeply involved in every element of government, such that, Rajkovic argues, law is a 'practice of government aimed at directing human conduct'. 46 This Foucauldian approach shows a clear sense of a driver behind the practice. As such, there is no clear dividing line between law and politics. Indeed, for Rajkovic, law is a 'powerful ordering rationality and hence means for government' in the Foucauldian sense. 47 The traditional belief in the separation of law and society masks what Rajkovic terms as law's 'conflicted nature'. Law has an 'image, lore and reverence', a kind of 'juridical and legal idolatry' 48 which empowers it but reflects its capacity to uphold its own legal values even when they conflict with those of powerful actors. 49 At the same time, law is also an integral participant in rule and, as Rajkovic argues, a better understanding of law lies in blending and relating it to extra-legal processes and practices. 50 Rajkovic's willingness to explicitly address the question of international law's agency marks him out. Here, he turns to Barnett and Duvall's taxonomy of power and their category of productive power. To re-state, productive power is 'the socially diffuse production of subjectivity in systems of meaning and signification'. 51 It works through social relations of constitution but does not do so directly. Productive power is important in assigning asymmetric categories of social relations: civilised versus uncivilised, for example, as well as legal categories of civilian versus combatant. 52 These categories privilege some, and disadvantage and subordinate others. Barnett, writing with others elsewhere, argues that law possesses productive power. 53 So, for Rajkovic, the Foucauldian idea of law as productive is the most accurate because it can show us how international law is a practice and how that practice functions as a tool of government or power. The Foucauldian reading of law gives it considerable agency, but it remains driven by government, or, presumably, at the international level, powerful actors. It is worth unpicking this complex set of ideas about how international law works. We see two ideas about law's agency here. First, law is seen as masking both the operation of power, typically understood in terms of maintaining an unjust social hierarchy, and as masking its own involvement in power. International law is able to perform this masking role because it is central to the knowledge/power nexus and it produces discourses of truth by mediating conflicting claims. 54 As Cutler argues, laws of private ownership and property 'neutralise the commodity system by presenting the communal protection of private property rights and entitlements as natural incidents of the rationalised commodity form'. 55 This means that a system which is 'inherently inequitable' 56 is seen as the exact opposite. 57 Moreover, the relegation of economic activity to the private sphere of civil, economic law minimises and masks its political function. 58 Second, law is seen as mediating between power and its subjects, partly by making rule palatable but also by providing a method of (limited) redress. Law is increasingly called upon to mediate between political, economic and social relations among competing actors. 59 The ultimate source of agency is not found in international law but -in Foucault's depiction, in government and the state, and in accounts which apply Foucault to international law, in global governance or powerful global actors.
What is clear is that international law's agency is, once again, limited because it is seen as being possessed and deployed by other actors to serve their interests.
Both the minimalist and maximalist ideas of law as a practice focus on individual practices and laws. This is because of practice theory's origins in sociology and its focus on 'following the actors', and the micro level. As Adler and Pouliot argue, it is by 'zoom[ing] in on the quotidian unfolding of international life' that we can understand how the 'big picture' is constituted. 60 As a consequence, this makes it difficult to identify overarching values or patterns and, as yet, the application of practice theory to international law has yet to attempt this. When practice theorists theorise agglomerations of practices, 61 we move towards what IR and political theorists would call a structure but which they theorise differently: 'the principles that pattern social practices that make up social systems'. 62 Schatzki's idea of rules is a good description of law: 'explicit formulations, principles, precepts, and instructions that enjoin, direct or remonstrate people to perform specific functions'. 63 They are also enforced by those in authority. But rules are separate from practice and help to bind practices together. This means that they fade into the background because practices are the focus of attention. This idea of rules explains part of law but not all. Schatzki also offers teleoaffective structures: 'a range of normativized and hierarchically ordered ends, projects, and task, to varying degrees allied with normativized emotions and even moods'. 64 This captures the deeper values of law but misses out on the ways in which law sanctions behaviour. Latour and McGee both see law as a network, mainly because 'there is nothing but networks' and law is no different. 65 Leaving aside the agent-structure debate and the reasons for ANT's rejection of the idea of structures, what is important here is how seeing law as a network affects our perception of its agency. Without a doubt, this increases law's agency beyond existing accounts which struggle to see international law as doing more than merely enabling or channelling. Networks do not just enable but they situate actors and constitute identities (as well as being shaped and constituted in turn by actors). The limitation of ANT when applied to international law is that law as a network is embroiled in countless other networks and is seen as no different to them. This jars with a seminal assumption common to both international legal theory, jurisprudence and IR that law is distinct from other social phenomena. 66 Certainly, law and its deeply embedded values do seem further removed from critique and challenge than other social phenomena like science. 67 This may explain, at least in part, why the idea of international law as a network has yet to make any real headway in the field of IR-IL.
What practice approaches do is help us to see the ways in which law helps to create meaning and to privilege and silence certain actors and actions by bestowing competence. This is law's power as productive, familiar to approaches which focus on the language of law and which reinforce the idea of international law's agency as 'enabling' or 'channelling'. Practice approaches also emphasise performativity, drawing the eye to humans and away from non-human factors, despite their argument that social and the material are equally 'wrapped up' in practices.
What practice theory, specifically ANT, can offer is the idea of international law as an actant. Seeing law as an actant enables us to accord it a role beyond enabling or channelling. This is important because the role that a legal rule plays is important and attributing agency to either the actors who apply or interpret it, or to the actors who created it, misses something important. The rule is not causally irrelevant but existing IR ideas of international law's agency have difficulty in attributing either agency or any particular significance to the legal rule itself. Rather explanation and interest lie in the wider political battles in which legal rules are little more than tools.
When we look at international law's agency, we see multiple forms: the law itself imposes limitations (like a channel), it colours how we see and institutionalise cooperative coexistence (i.e. institutions), it helps to produce our identity and our understanding of how to 'go on' (like a practice), but it also imposes concrete, sometimes physical, restrictions on what we can and cannot do. We must recognise law's composite nature: spoken and written words, beliefs about justice, its physical performance but also 'weaponized city benches, surveillance cameras, speed-bumps, and steel barriers'. 68 In other words, the 'law-scape' where law affects every aspect of law and is embedded not just in its social fabric but its spatial fabric, too. 69 In trying to capture the agency of law as a whole, we must see law as a whole and as comprising different elements with different agencies. Gramscian understandings of law get us closer.

Gramscian international law
Gramscian accounts of international law move our estimation of international law's agency on further still, although they need some adaption before doing so. 70 There are two possible readings of the Gramscian idea of law. On the one hand, we might read the Gramscian idea of law as limiting law's agency to being an institution which is used by the state and dominant class to maintain their dominance. It is a method of maintaining hegemony. This does not differ from the more Foucauldian end of practice accounts of law seen in Rajkovic's work which see international law as a tool to be controlled by other actors. On the other hand, however, there are elements in Gramscian accounts which point to a more substantial idea of law's agency and which hint at the possibility of the law as an actor in its own right. Gramsci's praxis conception of law certainly comes very close to depicting international law as an actant and it is the account which affords international law the most agency. It sees legal institutions and ideologies as central to the constitution of the hegemonic class. International law is part of the process by which a class or group establishes the conditions necessary to establish control, either through force or by ideologically capturing popular support as the articulator and protector of the public interest or common sense. 71 Contrary to the depiction of international law as an enabler, both law and capitalism work to order society and the economy and both benefit. As Cain notes: Insofar as laws such as constitutional and administrative laws then secure the authority of state, the influence of law and state is reciprocal. So too is the influence between law and capital reciprocal, for just as law is given its form by capital, capital is given its form by laws that enable its accumulation, disposition, alienation, and so on. 72 It is not a passive or disinterested actor that has been captured by capitalism. It actively participates and, crucially, benefits. When we see international law as benefitting from being central to the form society takes and shaping its operation, it becomes harder to see it as passive. Thus, for Gramsci, capitalism and law developed together, and it is impossible to have capitalism without law. As Cutler argues: the global political economy is undergoing a process of juridification in which a commodified legal form provides the template for economic and social regulation. Domestic and global orders are increasingly linked to and disciplined through law by the logic of capital. 73 Moreover, we start to see law as a specific type -the specific law of capitalist systems which dominates the globe and international law. It has a particular set of values and property is central to it, just as it is to capitalism. Moreover, this is not the first time that international law has reflected and spread a particular set of values: European imperialism was made possible by a particular understanding of international law, territory, possession and sovereignty. 74 International law has a considerable stake in maintaining and expanding the current reach of global governance and ensuing it takes a particular legalistic form. But international law's agency here is more than channelling political actions into particular legal forms, or of enabling the operation and spread of capitalism. There is a synergy between economic and legal values that should lead us to question why we see international law as being neutral or value-free. It is not simply that international law has been captured by capitalism or that economic values have percolated into international law but that international law's own values are inherently capitalist and capitalism is inherently legalistic. It would make no more sense to say that capitalism has been captured by legalism than vice versa. If we recognise international law as an equal, and equally involved and active, partner in maintaining and creating global governance with economics and politics; if we see international law as an equal partner to recognised structures, then shouldn't we see international law as a structure?

Defining structure
Structure is a much-used and rarely defined term in IR theory. My idea of structure is neither the Marxist notion of an economic structure nor the linguistic notion of language as a structure. 75 Instead, I draw upon Sewell's clarification of Giddens' idea of structures as comprised of 'rules and resources'. 76 Sewell argues that we should think of rules as 'informal and not always conscious schemas, metaphors, or assumptions' 77 presupposed by formal rules. These are, in other worlds, social rules. Resources, in Sewell's reworking of Giddens, are anything that 'can serve as a source of power in social interactions'. 78 These can be human or non-human (what is often termed 'material'). They must always be interpreted within context and are never self-evident. Nor are they evenly distributed As Barnett and Duvall argue, structures define the kinds of social beings actors are and an individual's social capacities, subjectivities and interests are directly shaped by the social positions they occupy. 79 Moreover, these relations are always hierarchical, and structures serve to allocate places in a hierarchy. Structural power affects actors in two ways: first, 'structures allocates differential capacities, and typically differential advantages, to different positions'. 80 Second, it shapes the actor's self-understanding and subjective interests. 81 It can operate both overtly but also covertly when it generates 'the social powers, values, and interpretations of reality that deeply structure internal control'. 82 Structures, then, are 'the principles that pattern social practices that make up social systems'. 83 Structural power, therefore, differs from the productive power which characterises practice theory. Practice theory focuses on the social practices, not the principles which pattern them. Finally, a structure is always structuring social roles and relations: class structuring politics, gender structuring employment opportunities and pay, race structuring development and so on. Where these structures exercise considerable causal power for structuralists, in practice theory, they are mostly absent. And so, while constructivism, practice theory and structuralism all share the epistemological assumption of mutual constitution, constructivism and practice theory's narrower ontology denies structures any serious explanatory potential.
Significantly, structuralism typically distinguishes between surface and deep structures, with deep structures generating surface ones. For example, patriarchy is a deep structure which generates and structures the surface structure of appropriate dress for women in a particular context. The deep structure is 'deep' because it is more deeply rooted in our subconscious and harder to change: the appropriate skirt length has changed over time in Western society but the deeper patriarchal right to judge women by their dress, and to discipline them, remains. It is the deployment of men's superior social, economic and political position (and the material consequences that flow from this) that makes this possible. Systemic inequality both makes patriarchy possible and is a consequence of it. It is not just ideas but hard material and often financial and physical disciplining of women and other subordinated groups that allow this to happen. 84 And the deeper into our subconsciousness a structure is, the more powerful it is.
This article argues that we should see law, and international law, in similar terms, that is, as possessing both a surface and deep structure. Law's surface structure comprises legal rules, precedents and principles. But it also includes law's personnel: everyone who works in the justice system from clerks and prison guards to judges and prison governors. It also includes 'non-professionals' -ordinary people who get caught up in the law as jury members, witnesses, complainants and defendants. Third, it includes the law's procedures: when people enter and leave a courtroom, how inmates are processed through the prison system, who may speak, when and how. This also has a strong material component: buildings, furniture, clothing but also the physical ordering of people within these spaces, like who sits where and when.
Alongside this surface structure there is a less well-seen deep structure. Shklar's notion of legalism gives us an excellent account of law as a deep structure. Legalism is 'the ethical attitude that holds moral conduct to be a matter of rule following, and moral relationships to consist of duties and rights determined by rules'. 85 'Legalism' is the ideology internal to the legal profession as a social whole but also extends far beyond the legal profession to society as a whole. 86 It is a way of being in and seeing the world. These legalistic values structure both the operation of the legal system and our personal morality. It extends far beyond the courtroom and the outward manifestations of the law. This means that, for Shklar, legalism exists on a spectrum. At times it is openly and clearly articulated and promoted (e.g. promotion of the rule of law by states and bodies like the United Nations). But at others, the deep values which structure most of human relations are hidden and internalised. She argues that legalism is 'both strongly felt and widely shared' 87 and as such is deep and pervasive. It is often not articulated at all. It manifests itself in social institutions, political ideologies, philosophical thought, common social ethos and personal behaviour. It is an internalised and partly hidden set of social rules, which also manifests itself in more obvious, surface sets of social rules. Shklar's notion of legalism therefore fits with the idea of law as both a deep and surface structure. What are the social rules which comprise legalism? For Shklar, it is 'The dislike of vague generalities, the preference for case-by-case treatment of all social issues, the structuring of all possible human relations into the form of claims and counter-claims under established rules, and the belief that the rules are "there"'. 88 The centrality of accountability and the rule of law are two such manifestations. McGee identifies an emphasis on rational-legal reasoning, the privileging of law as a method of conflict resolution, resource distribution and organisation as central. 89 It also changes to fit its context. Writing in the domestic American context, Litowitz identifies the following beliefs as central: 'private ownership of property, employment at will, inheritance, freedom of contract, limited liability for business organizations, patriarchy, and a regime of negative rights that ensures individuals must secure their own health care, day care, and other benefits'. 90 Legalism also works to set our cognitive horizons and is all-encompassing. Litowitz sees it as a grid, forming 'a bounded universe of possibilities ... This grid will filter and frame all legal disputes within its parameters, recasting them in the dominant language of the legal system at the time, thereby extending the system ... law recasts ... events in the conceptual grid of the law', 91 in ways that make it fit into the grid. So social disputes become legal disputes and individuals become witnesses, defendants, complainants and so on. Once a social problem has been fitted into the grid, the grid is able to produce an answer, which legitimates the legal concepts and the grid itself, and its ability to provide solutions. 92 Even if we do not take the dispute to law, legalism still frames our understanding of right and wrong, rights and obligations, grievance and wrong-doing. As Shklar argues, we cannot separate our idea of what is right from legalism. Moreover, one cannot step outside of the grid because it is hard-wired into our consciousness, even if we were capable of 'opting out' of our society's legal system. Thus, even when the law can operate as a site of resistance and counter-hegemony, it can only do so through the forms of law. It must use its language, its concepts, its institutions, its personnel. One can only use the tools that law provides: one cannot, in Litowitz's example, use the law to secure universal health care because (in the United States) there is no legal right to it. One may be able to challenge key elements of the legal system but the price of doing so is acceptance of the system as a whole. Even if successful, such acts of resistance can only happen once the law has been assumed. As Latour argues, law's force comes from its ubiquity: its ability to link all peoples and acts through chains of obligations. 93 Law 'judiciarises all of society'. 94 It 'works on it, kneads it, arranges it, designates it, imputes it, makes it responsible, envelops it'. 95 And the deep structure of law, and international law, is always assumed.
And it is clear that many of our ideas about morality, justice and the right way to behave are a consequence of legalism, rather than any other structure. Instead of seeing law's agency as located elsewhere, we should recognise that law has its own power. It does not derive from sovereignty 96 or the moral purpose of the state. 97 It has its own power and as many have recognised, the legal way of decision-making is seen as 'special'. This can be that the legal principles themselves are considered 'quasi-universal or at least generally accepted', making them 'well adapted' to the needs of public justification. 98 Others have argued that law's particular form of argumentation and reasoning is what explains its uniquely persuasive power. 99 Both contribute to the idea of accountability -a central strand in the Western political tradition -as being accountability before the law. As Shklar points out, legalism has a strong preference for case-by-case analysis, seeing it as the most appropriate way of understanding and identifying bias and injustice. This focus on individual cases, usually involving individual complainants and defendants, is no coincidence; it reflects the individualism that is common to liberalism and Western political thought. It is certainly not the case that liberalism as a structure is shaping or constituting legalism. Rather it is a case of two structures with a shared history, supporting and bolstering each other. Neither is derivative of the other. When we view international law as acting in similar ways to, and possessing similar agency to other, recognised structures, it raises the question of why we do not see international law as an agent. International law certainly seems to fit with the description of a structure offered by Sewell, and by adding Shklarian legalism, we see that international law does in fact have both deep and surface structures. This belies the argument common to process, institutional and minimalist practice accounts of international law, that international law is simply a manifestation of other, more powerful non-legal structures or actors.
As a consequence, applying our understanding of social norms and the social world to the law is unsatisfactory and incomplete. We need a composite approach to theorise a composite entity. Existing, implicit ideas of law's agency see parts but not the whole. Process accounts of international law see the power that comes from providing the rules of the game; the organisational structures, procedures and fora, the language of redress and accountability, and the omnipresent legal mindset. Seeing international law as a social institution enables us to see the ways in which legal values set cognitive horizons. Seeing international law as a practice (or maybe it would be accurate to say a myriad of practices) sees how repetition of action and language work in millions of tiny ways to create behaviour. But they struggle to see law as an actant.
More significantly, if we see law in parts, we cannot see the whole. The composite picture of law comprises rules, principles and precedents; material elements like courtrooms and jails; procedures like jury selection or testimony; the personnel of law and the 'ordinary' people it sweeps up. This is the surface structure of law. But we must also see law's deep structure and the powerful ways in which it shapes all our cognitive horizons, not just those of institution-builders (à la Reus-Smit). Trying to imagine an alternative value set to legalism highlights just how central it is to our human experience. But law is more than the sum of its parts. When we see law in parts, we cannot see how law operates to locate us in our social hierarchy and to construct our subjectivity. We may see law as the omnipresent rules of the game, maybe even as setting cognitive horizons but we do not see it as placing us in particular positions or of structuring other powerful forces.

Conclusion
By seeking to unearth and unpick the ideas of agency that underpin the most common readings of international law, this article seeks to clarify what we really think international law does and is capable of doing. Among the first three accounts of international law: international law as a process, an institution and the minimalist idea of practice theory, we can see three overlapping ideas of international law. First, law can be seen as reflecting or possessing particular values. Process accounts like Chayes' argue that international law provides accountability, something which politics cannot do, and so international law (and law in general) is seen as providing a public good which politics cannot. It is one of the actors that holds politics and political actors to account. This makes international law a persuasive form of justification. These are desirable values and, as a consequence, they provide power. They also create the second set of ideas about international law in that international law provides a particular way of doing things. It is a question of the practical organisation of international politics. So, as both process and institution accounts argue, international law provides the organisational structures, procedures and fora in which decisions are made. Practice theory's minimalist idea of international law's agency agrees: the daily practice of global politics is shaped by international law's form. Moreover, for practice theorists, entrenched social practices also perform a similar function. We also see a harder edge to international law's agency here in Reus-Smit's argument that once a consensus is established, those states who do not agree are essentially compelled.
But particular ways of doing things do not simply fall from the sky. They are manifestations and reflections of a particular way of seeing the world: a Shklarian legalism. As Reus-Smit argues, the norm of pure procedural justice defines the cognitive horizons of institutional architects and it structures the process of communicative action that surrounds the (re)production of fundamental institutions in international politics. It is the omnipresence of law and the legal mindset, seen by process, institution and practice understandings of international law's agency that gives international law its ability to enable outcomes. International law's agency is productive. It does not directly cause actions to happen (or not). Rather it makes certain outcomes more or less likely. It provides reasons for action, rather than causes. And so, we see that the three sources of international law's agency: that it has particular values, a particular way of doing things and a particular way of seeing the world, collapse into one, all united by the same sense of international law's agency. While each has worked to expand international law's ontology and agency, all three stop at the same point: international law cannot do more than enable. It cannot make something happen by itself. It has no independent agency, nor is it an independent actor.
We have underestimated the agency of law, and international law. We see the tip of the iceberg. We see the daily practices, the particular court cases or pieces of law. We see how legal education is common among decision-making elites. We see how convincing legal reasoning is. But we do not see its true reach. We do not see how we are all, even those of us who have no legal education, socialised into the legal mindset. The idea of what justice is, and that we are rights-bearing individuals, is pervasive, so much so that it is hard to imagine how to be in a world where there is no talk of rights, duties and justice. It is one of the most deeply embedded and unconscious sets of belief we hold. But law's power moves beyond shaping our beliefs. It limits our range of actions by requiring that our social battles are put into legal form. This translation inevitably skews our understanding, silencing some actors and narratives, altering others and producing particular legalistic emphases. Moreover, it intersects powerfully with and in similar ways to other structures and indices of control like gender, race and class. By recognising international law as an actant, we expand its agency and start to capture some of the powerful effects it has on global politics. The essence of structures is that they locate us within a social hierarchy and structure our subjectivity. When we hold international law up to this mirror, we see it reflected perfectly.

Funding
The author received no financial support for the research, authorship and/or publication of this article.