Why anarchy still matters for International Relations: On theories and things

Introduction

The category of anarchy has been conventionally associated with the emergence of an autonomous discipline of International Relations (IR) (Schmidt, 1998). Unlike social and political theory, IR theory studies social systems that are anarchical in nature. And while in general social theory the term anarchy stands for disorder, in IR it has two parallel meanings: (1) an interaction domain among units not governed by a common superior such as world government (Baldwin, 1993b: 14; Milner, 1991: 69–70), or (2) a horizontal order between formal equals to be distinguished from a hierarchical order between subordinate and superordinate units (Bull, 2002 [1977]: 17; Lake, 2001: 130; Waltz, 1979: 114–116). Recently, the conventional primacy of anarchy as central to IR has been challenged by Jack Donnelly (2015). Except for a period after 1979, when Kenneth Waltz published Theory of International Politics inaugurating a structural realist theory of anarchy in IR, Donnelly argues, anarchy has not been formative for IR discourse – if anything, hierarchy is more weighty (see also Hobson, 2014; Lake, 2001, 2009). Notably, this criticism targets not concepts of anarchy but theories of anarchy such as Waltz’s. It is symptomatic of a wider ethos tacitly accepted in the IR discipline: it may be termed the ethos of anti-theory.

This article pursues a twofold task. First, to disclose the defects of the anti-theory ethos, a form of conceptual atomism which is hostile to structuralist and normative theories. ¹ The proposed alternative is theoretical holism (Dreyfus, 1980; Duhem, 1991 [1906]; Quine, 1951; Quine and Ullian, 1970). On a holist view, theory is a framework: a conceptual whole which constitutes the meaning of a set of interrelated concepts and determines their relation, as a set, to the world. ² Building on the insights of theoretical holism, the second task is to defend the continuing relevance of theories of international anarchy for IR, and specifically, of structuralist theories. In what follows they are represented by Waltz’s structural realism (Waltz, 1979) and Bull’s theory of the anarchical society (Bull, 1966a, 1966b, 2002 [1977]) in lieu of their landmark status among American, and, respectively, European scholars of IR. Critics usually treat Waltz’s structural realism, a scientific theory also known as neorealism, as a paradigm of international anarchy. But Bull’s account of international anarchy – a species of normative structuralism – is equally paradigmatic once we move from science to normative analysis.

The IR reader is likely to be familiar with English School theorising (an overview is Buzan, 2014) where Bull was a towering figure, and structural realism and kindred approaches such as neoliberal institutionalism (Baldwin, 1993a; Keohane, 1986; Powell, 1994). One basic problem that remains underexplored in these literatures is theoretical holism. It is important to take it seriously given the rise of theory scepticism in the field (Wight et al., 2013). And yet, the problem has not received attention in the recent anarchy-versus-hierarchy IR debate (Donnelly, 2006, 2009, 2015; Hobson, 2014; Hobson and Sharman, 2005; Lake, 2001, 2009; Mattern and Zarakol, 2016). The aim here is not to review this extensive body of literatures but to make a critical contribution to it by emphasising the role of holism and second-order, philosophical considerations. Such an undertaking seeks to render explicit the underlying premises of theories of international anarchy. ³ Concretely, my central thesis is that although Waltz and Bull disagree over the priority of science and normativity, they share a holist commitment to theory as a set of concepts (including ‘anarchy’) organised into a coherent whole. What is more, both theorise an object that is holist in character – an international system whose structure is anarchical and whose key players are states.

The exposition is developed in five parts. The first introduces theoretical holism. It suggests that theory is a holist construct and that the object it investigates can be a ‘whole’ such as a system or structure. A detailed evaluation of Waltz’s and Bull’s structuralist theories of international anarchy in the next two sections shows that, despite defects, these theories represent overall coherent conceptual ‘wholes’. It is then shown that Donnelly’s proposal for abandoning Waltzian international anarchy for global hierarchy fails to register the holist underpinnings of Waltz’s, and by extension, Bull’s theory. Not only does the structure of international anarchy outlined by Bull and Waltz contain elements of hierarchy, another point misunderstood by critics, but it also has normative significance: the value of ‘anarchical freedom’ is fundamental to it (this form of freedom will be referred to as anarchy in the third sense). Anarchical freedom presupposes the co-existence of mutually independent, formally equal units: states. The conclusion ponders how this value can be best realised in a globalising world where the agency of states is eroding.

Theoretical holism in brief

Before we explore the holist facets of the two leading structuralist theories of anarchy in IR, it is useful to clarify the notion of theoretical holism. Here, it is contrasted with practical holism and with object holism. Object holism implies a distinction between two kinds of objects – between a system (a whole) made up of elements and an aggregate of elements. A class of students is a whole, whereas a list of students is an aggregate. Compared to an aggregate, a system has an identity of its own which is logically separate from that of its individual members. If we say ‘The class is misbehaving today’, we are not saying that certain students, John, Jill and Mark, are misbehaving as isolated individuals but that the unit composed of these students does so. The atomism-holism distinction applies to objects as well as to the conceptual standpoint employed in studying objects. This standpoint is atomist if an object is conceptualised under a concept (the concept functions as a general rule for subsuming particular objects; see Kant, 1996 [1771/1778]: A141–B180) and it is holist if it is conceptualised within an entire theory (Duhem, 1991 [1906]; Quine, 1951, 1981). Theory can be defined as a conceptual system inside which certain elementary concepts are glued together – by the theory’s assumptions – so that they yield hypotheses. Or, more precisely, it can be defined on the analogy with language: as a coherent set of sentences (statements) whose constituent bits resemble individual words. The idea goes back to Gottlob Frege’s Contextual Principle which states ‘Only in the context of a sentence does a word mean anything’ (Frege, 1884: §62).

Atomism and holism deal with a common puzzle of how multiple elements can be combined. The difference is in the starting premises: atomists begin with the premise of separate pieces which can then be aggregated, while holists first postulate a whole (a system) which can subsequently be disaggregated into parts. The reason why a system is logically prior to the sum of its parts is that it has structure – the structure is the common relation holding the parts together. (Aggregates are summations of atomist objects not interrelated by any common structure or relation.) The same relationist logic applies to conceptual systems or theories. As a result, the interpretation of (lists of) free-standing concepts differs principally from that of concepts inside theories. A list enumerates disparate analytical elements, so it is fully legitimate to interpret one element from the list by assuming away the rest. A theory however involves a relation between multiple concepts. From a proper theoretical perspective – a perspective that is holist – individual concepts are literally meaningless outside the framework of their host theory. To understand what Bull means by the concept of anarchy, for example, we have to assess his statements about anarchy, and this requires us to attend to his entire theory of international society. Bull’s theory (elaborated below) contains other basic concepts such as ‘society’, ‘system’, ‘order’, ‘institutions’ and ‘rules’ as well as statements that interlink such concepts (e.g. that even under anarchy, states recognise as binding common rules and institutions in their mutual dealings).

The second way to explicate theoretical holism is by contrasting it with practical holism. In a classic paper, ‘Holism and Hermeneutics’, Hubert Dreyfus (1980) differentiates the theoretical holism of philosophers with attachment to science such as Willard Van Orman Quine, from the practical holism of hermeneutic philosophers such as Martin Heidegger and the late Ludwig Wittgenstein. ⁴ Understanding in science is measured against a system of beliefs, whereas understanding in hermeneutics is measured against a social practice. What interested Quine, in Dreyfus’ view, was the problem of scientific theories as open-ended, quasi-conceptual quasi-empirical systems or ‘webs of beliefs’ (Quine, 1951: 38–40; Quine and Ullian, 1970). Heidegger and Wittgenstein were occupied with the question of how agents act within social practices – systems of action that are actual or non-abstract. But casting action as the antipode of theory creates the false impression that action-based (‘practical’) social systems do not lend themselves to theoretical examination. Besides, a system of action can be theorised in concrete terms or in abstract terms, as Waltz does in presenting the international system as an automatic field of forces. It is therefore better to distinguish the standpoint for analysing objects – atomist (single concept) or holist (entire theory) – from the object analysed. The object can be abstract or concrete, it can be atomist or holist and indeed it can have further features.

Why should we favour theoretical holism and what is its bearing on structuralist theories of international anarchy? To anticipate the argument of this article, the principal import of Bull’s and of Waltz’s approach to international anarchy is that it is doubly holist. Its holism is palpable in that it prioritises theory over atomistic concepts and that the object it studies is not an isolated event or process of international politics but the anarchical structure of the international system as a whole. Such an approach runs counter those of more recent IR scholars (Donnelly, 2009, 2015) who tend to zoom into a single concept of anarchy, the benchmark of conceptual atomism (discussed in section ‘Moving beyond anarchy in IR?’).

Conceptual atomism is problematic on two major counts. To begin with, each isolated concept, taken by itself, is inert. Its productive power is harnessed when it is connected to other concepts, inside hypotheses (statements). But since no individual hypothesis supplies its own interpretation, to be able to interpret it we would have to first determine its context and for holists as back as Pierre Duhem (1991 [1906]: 187) this is theory as a whole. Inspired by Duhem, Quine (1951) argued that hypotheses about the world are evaluated not individually but in conjunction with theories of which they are a part (Block, 2000: 360; Quine, 1951: 38). Translated into IR discourse, this suggests that a full-blooded theory of anarchy cannot be compressed into a concept of anarchy. Only after we have established whether a particular theory, as a system of assumptions, concepts and hypotheses, fares ill or well, can we tell whether a given concept (such as ‘anarchy’) as embedded in this theory, is coherent or not.

The second major drawback of conceptual atomism pertains to the type of objects concepts are about. Atomism equates objects to discrete things in the world. This may be appropriate for concepts like ‘chair’ or ‘tree’ as these, arguably (cf. Quine, 1981: 20), have things in the world as empirical referents. But ‘anarchy’ is a different order of concept: no discrete empirical referent corresponds to it. The thrust of Bull’s and of Waltz’s position, as we shall see, is that international anarchy is an unobservable, structural feature of the international system – at most we observe its effects. We infer the presence of structure and this inference is a theoretical enterprise.

In short, in order to properly make sense of objects that constitute wholes (structures and systems), we need conceptual wholes (theories). Once this crucial holist insight is appreciated, we can understand why structuralist theories of an anarchical international system still matter.

Kenneth Waltz on international anarchy

Since its appearance, Waltz’s Theory of International Politics (1979) has attracted a plethora of appraisals covering various aspects of structural realism such as security dilemma (Schweller, 1996), foreign policy (Tebami, 2002) as well as questions of methodology and theory building (Dessler, 1989: 448–451; Guzzini, 1998: 128–129; Wendt, 1999). Few, early contributions by Ashley (1984), Buzan et al. (1993) and Ruggie (1983) offered an in-depth analysis of the premise at the heart of Waltz’s theory – that the international system is anarchical. Waltz employs the two standard in IR concepts of anarchy – anarchy as a lack of common superior and anarchy as a horizontal order between equals – to which we shall return in a moment. The point to flesh out is that theories are not on par with concepts. When commentators like Donnelly (2015) pronounce the end of international anarchy, invoking Waltz’s scheme as a prototype, they do not merely claim that certain concepts of anarchy are defective but, rather, that theories of anarchy are. However, if concepts, as holism assumes, are the constituent elements of a theory, it would be a mistake to analyse Waltz’s concepts of anarchy by abstracting away the background theory. Does Waltz himself accept theoretical holism? The short answer is affirmative, even though we shall have to integrate Waltz’s position, clearing up ambiguities and infelicities.

Waltz (1990) sets out to develop a theory of international politics that goes beyond mere reflections on political thought. Theory is a systematic conceptual framework designed to explain laws or law-like regularities (Waltz, 1979: 6, 118). Structural realism explains the laws of state interaction in the sphere of ‘international politics’, a term of art for Waltz. Two mutually exclusive structures of system organisation or ‘ordering principles’ (Waltz, 1979: 81) are posited – hierarchy versus anarchy (Waltz, 1979: 88, 114–116). Domestic politics is organised hierarchically, under common government; international politics is organised anarchically, in the absence of common government:

Formally, each [state] is the equal of all the others. None is entitled to command; none is required to obey. International systems are decentralized and anarchic. The ordering principles of the two structures are distinctively different, indeed contrary to each other. Domestic political structures have governmental institutions and offices as their concrete counterparts. International politics, in contrast, has been called ‘politics in the absence of government’. (Waltz, 1979: 88)

In this passage, Waltz relies on two separate concepts of anarchy. The first, let us call it demarcational, stands for a lack of world government (it corresponds to anarchy in the first basic sense identified above). Parallel to it, Waltz uses anarchy as a structural concept to identify the ‘ordering principle’ or structure of the international system. It is equivalent to anarchy in the (second) basic sense of a horizontal order between formally equal states. Because these two concepts differ in scope, they are not mutually exclusive. It is possible to combine the broad idea of an absence of world government with the narrower idea of a horizontal order of equals. Waltz’s proposition that among states ‘none is entitled to command; none is required to obey’ is ambiguous because it may be taken to refer either to the broad concept of anarchy alone or to a combination of the broad and the narrower one – and Waltz often runs them together. Nevertheless, the two concepts are distinctive and to establish how, and indeed whether, they fit together inside a theoretical whole, we have to look more closely into the framework of structural realism.

Waltz’s structural realism is a scientific theory influenced by Karl Popper’s famous methodology of falsification (Popper, 1959: 84–92; Waltz, 1979: 123–125). But a more technical aspect of Popper’s analysis – demarcation (Popper, 1959: 34–44, 1963: 253–292) – elucidates Waltz’s take on anarchy. Popper demarcates science from pseudo-science, Waltz demarcates anarchical from non-anarchical systems of action. The demarcational concept of anarchy sorts out anarchical systems (with no common superior) on one side (that of ‘international politics’) and hierarchical systems (with a common superior) to the other (that of ‘domestic politics’). But it cannot explain what states do, with what consequences, once they find themselves inside the domain of international politics. This explanation requires the second, structural concept of Waltzian anarchy.

This second, structural concept of anarchy performs most of the analytical work in Waltz’s scheme. Structural realism is a systemic theory, and integral to it are notions of structure and system. As mentioned, the term ‘system’ stands for a whole over and above a sum of elements, where the principle of systemic organisation is structure. For example, a brick house has structure (the connection holding the bricks together) which a mere heap of bricks lacks. For Waltz, the identity of a system is determined by its structure and its units. Because he studies systems of action, he defines structure as action constraint (Waltz, 1979: 73, 90, 122). The proposition is that the units would have acted differently were it not for the constraints imposed on them by structure.

Waltz is primarily concerned with explanatory theory. This still leaves open the question of what counts as theory. On the standard definition, it is a framework of interrelated concepts which, when mixed with key premises, generate hypotheses. Waltz’s belief that theory must be scientific theory supposes a particular way of hypothesis testing. The hypotheses of a scientific theory either purport to explain empirical facts (if the theory is empirical) or (if the theory is non-empirical) to offer compelling explanations by route of analytical models (Nagel, 1961: 79–105). Models may represent reality at one or two removes, but Waltz treats models as representations of theory (Waltz, 1979: 7). Apart from ‘system’ and ‘structure’, concepts central to Waltz’s neorealism include ‘anarchy’, ‘hierarchy’, ‘power’ and ‘balance of power’. These concepts demand definitions. But after being defined, they would stay inert without the driving momentum of the theory’s basic assumptions such as Waltz’s assumption that the international system is anarchical, or that states seek survival and power.

Let us see what hypotheses are entailed by Waltz’s structural realist theory. Waltz’s concept of anarchy with its duo of meanings was noted previously. His concept of structure has three analytical layers: unit arrangement (ordering principle of the international system), functional differentiation of the units and distribution of capabilities (Waltz, 1979: 88, 100–101). The ordering principle of the international system is captured by Waltz’s structural concept of anarchy. Waltz assumes that all states seek to survive as a minimum, and beyond this, to amass more power than they already have (Waltz, 1979: 91, 118, 126). When this assumption is coupled with the structural concept of anarchy, it generates the hypothesis that due to the constraints inherent in international structure, states may not always get what they want (Waltz, 1979: 119). A state may aspire to become a global hegemon but, because of the constraining effect of the international structure, the likely outcome of its interaction with other states would be something else than global hegemony – a system of balance of power.

Actually Waltz puts forward two hypotheses about international structure. The first holds that the international structure exerts equalisation effects across the international system – it limits the feasible options for action, so that units with different capabilities – great powers as opposed to rank-and-file states – will end up reproducing the same pattern: a balance of power (Waltz, 1979: 72, 77–78, 123). The second hypothesis stipulates that no balance-of-power configuration, not even a unipolarity, would count as hierarchy unless states cease to see each other as formally equal (breaking with anarchy as structure). But, and here Waltz is obscure, transformation into a hierarchical system requires also a break with anarchy as a demarcation condition. For this to happen, the system must transform into world government (Waltz, 1979: 199).

These two hypotheses present differently calibrated explanations in Waltz’s theory. The first explains state action by reference to general structure, the second explains it by reference to a particular, anarchical structure. The distinction is non-trivial because Waltz does not treat structure as equivalent to anarchy. Anarchy comprises only the first layer of structure. The second layer, functional differentiation, is bracketed from the analysis since, as actors in international politics, states are functionally undifferentiated units – all perform an identical function (provision of security). The first layer corresponds to ‘deep structure’ (Ruggie, 1983: 266), the second layer is shallower and the third is ‘surface’ structure (Buzan et al., 1993: 87). The surface serves as a measure of ‘polarity’ or the distribution of power (capabilities) across all the units in the international system (Waltz, 1979: 98, 131).

This tripartite division of structure has implications for the problem of system change. The international system can change its internal parameters in two ways: either because some units might acquire new, specialised functions – a group of states might become security providers for all the rest – or because the system-wide distribution of power might shift. Such contingent change pertains to functions and capabilities and is located in the second and third layers of structure. Systemic change, where the system as a whole transforms its form, depends on the first layer of structural anarchy (Waltz, 1979: 100). For a system of international anarchy to materialise, it is not enough that states are formally equal, as presupposed by structural anarchy. It is also necessary that no world government is present (meeting the demarcational condition of anarchy). As long as no common world government or comparable global authority has been recognised by states, the international system will retain its anarchical form and there will be no change of the system into global hierarchy even if one superpower has gained total preponderance in capabilities. Waltz concludes, ‘no authorized manager of the affairs of nations will emerge in the nearest future’ (Waltz, 1979: 199, emphasis added).

If we are to summarise this section from the vantage point of theoretical holism, it is that Waltz is not testing isolated hypotheses about international politics, but a set of hypotheses, concepts and premises fused into a whole – or theory. As he recognises, ‘hypotheses about the association of this with that, no matter how well confirmed, do not give birth to theories’ (Waltz, 1979: 8; see also Mearsheimer and Walt, 2013). Waltz’s structural realist theory, we might say, takes theoretical holism seriously. What remains dissatisfactory nonetheless is that Waltz never clarifies the relation between power and authority. If the accumulation of power by a single superpower is insufficient to transform international anarchy into global hierarchy but demands global authority, we have left the realm of a power-based rule and entered that of a rights-based rule or authority (on authority, see Raz, 1990). The trouble is that authority and rights are normative categories and that Waltz discounts normativity. This brings us to the normative perspective of international anarchy.

Hedley Bull and the anarchical society

The most popular normative theory of international anarchy in IR is Bull’s theory of the ‘anarchical society’– a society whose members are states, bound by common rules (Bull, 2002 [1977]). Bull (1966b, 2002 [1977]: 25–30) calls his approach ‘Grotian’ to acknowledge its origin with Hugo Grotius, and the likes of Samuel Pufendorf, Emmerich de Vattel and Christian Wolff. At first blush, Bull’s normatively laden talk of society may appear to clash with Waltz’s scientific vocabulary of falsification, general laws and explanation, but both theorists are of one mind in viewing anarchy as the differentiating tenet of the international system (Waltz) or international society (Bull). Both moreover resort to a holist notion of theory (conceptual whole) as a framework for explaining the workings of the international system (an object whole).

Bull is often read as a thinker who drew a sharp distinction between an international system and an international society (Bull, 2002 [1977]: 10, 13; Watson, 1987). The former, the idea goes, is mechanical, the latter is imbued by meanings, normativity and rules. Yet putting the system-society distinction in mutually exclusive terms might be misleading since, logically construed, society is a type of system. Following HLA Hart (1961), Bull defines society as a system of rules (Bull, 2002 [1977]: 65, 122–123). It is a normative system whose rules and norms provide common standards of action by assigning rights and responsibilities to societal members. States members of the anarchical society hold rights of territorial integrity and sovereign independence under the common rules of international law (Bull, 2002 [1977]: 130, 146). Rules of balance of power give rights to member states to thwart the bid of any state for global hegemony, by means of war if necessary (Bull, 2002 [1977]: 103). There are correlative rules of war and diplomacy, and finally, rules of great powers which accord privileged status to a club of eminent states. These core rules constitute the ‘fundamental institutions’ of international society (Bull, 1966a: 48, 2002 [1977]: 65; Buzan, 2004: 167–176).

The concept of authority is central to Bull’s argument. What binds states into an international society is the authority of common rules (Bull, 2002 [1977]: 51–73). This society is ‘anarchical’ because states accept common rules without accepting a rule by a common sovereign. Like Waltz, Bull uses the term anarchy in the two standard IR senses: for a society without a common superior (Bull, 1966a: 38, 48, 2002 [1977]: 59, 124), and for a horizontal order between formal equals (Bull, 2002 [1977]: 17). Unlike Waltz, Bull theorises this horizontal order in the idiom of international law. International law is a defining institution of international society, which accounts for its anarchical form. One of the peculiar features of the international legal system is that it is decentralised. Another is that states – even more so than citizens subject to municipal law – keep the rules of international law not because they fear a global sovereign but because they have internalised these rules as norms (Bull, 2002 [1977]: 128–130). In other words, states need not be coerced into keeping the rules for they already accept the rules – this attitude of reverence towards the rules is what Hart (1961) called ‘the internal point of view’ (1961: 82–88).

In articulating the principles of international anarchy, Bull contextualises it inside a normative theory and it is helpful to highlight how this type of theory differs from Waltz’s scientific outlook. Norms or ‘prescriptions’ are rules which prescribe conduct: they tell the agents what they ought to do (Hare, 1952). Normative theory differs from its scientific counterpart in the mode of hypothesis disconfirmation. In scientific theory of a deductive sort such as Popper’s observed behaviour that runs against the hypothesis being tested compromises the theory’s key assumptions and thus the entire theory. One typical objection to Waltz’s structural realism is that after the end of the Cold War, the formerly bipolar international system has turned into a unipolar American empire. Waltz (1993: 59) responded that unipolarity (a recent assessment of unipolarity is Ikenberry et al., 2011) on the surface has not changed the deep structure of international anarchy, a response which leaves us wondering whether Waltz adheres in earnest to Popperian falsification. The distinguishing mark of falsification is that a single instance of an empirical counter-observation suffices to render invalid the hypothesis under examination. ⁵ Contrariwise, Bull’s prescriptive hypothesis that states in international society ought to abide by its basic rules – international law, the balance of power, war, diplomacy, great powers – is not invalidated by one instance of rule breaking by a single participant (it would be invalidated if the majority of participant states began to violate the rules persistently) (Bull, 2002 [1977]: 131; Kratochwil, 1989: 63).

The difference between Bull’s and Waltz’s normative and, respectively, scientific account is obviated in their treatment of the balance of power. According to Waltz’s ‘invisible hand’ market model (Waltz, 1979: 89, 91), the balance of power is the unintended outcome of the interactions of multiple units, states (1979: 119). Here, states are equated to automatically interacting units and their behaviour is explained by scientific (non-normative) theory. For Bull, in stark contrast, the balance of power is an institution based on common rules, and rules cannot be followed automatically – rule-following implies intention (Bull, 2002 [1977]: 31, 35). If a balance of power exists within international society, this is because its member states want to reproduce it: the institution of balancing exerts a normative pull on their conduct.

But why do states want to preserve a balance? Bull’s answer is, because they attach value to international anarchy. Above this was referred to as the third, distinctively normative sense of anarchy. It designates the value of freedom symbolised by the mutual independence of sovereign states. The balance of power, Bull maintains, preserves the independence (freedom) of all states by preventing global hegemony (Bull, 2002 [1977]: 102–104). Together with international law, it is a cardinal institution in an anarchical society of freedom-valuing states.

The upshot is that anarchy has acquired an axiological meaning inside Bull’s normative theory which it normally would not have in Waltz’s scientific theory. Curiously, in an essay on Kant which predates Theory, Waltz himself recognises the value of freedom. There, Waltz (1962) reads Kant as a liberal who understands that freedom and war go hand in hand. As Waltz emphasises, Kant’s pacific league of free states (renouncing war in their relations) may regulate interstate war but it cannot eradicate it (outside the league): war-proneness is a perennial feature of international politics. Bull’s message is similarly sombre. The anarchical society is directed at preventing (negatively) states from annihilating each other rather than at promoting (positively) common goals such as economic prosperity or political development. The freedom of states in international society is the negative freedom of peaceful co-existence (Bull, 2002 [1977]: 35–36, 66–67, 150).

As significant as the value of freedom associated with anarchy may be, Bull’s defence of it is strained. Recall his claim that states in international society have a right to wage war on a rising hegemon to preserve the balance of power because the balance guarantees their freedom and independence. For Bull, this right can be invoked regardless of whether the hegemonic state has committed a wrong under international law or not. Such an anti-hegemonic war is motivated by power differentials – it violates the rules of international law which permit self-defence or punitive action only in response to aggression or injury. ‘It is a paradox of the principle of the balance of power, ’ Bull admits, ‘that while the existence of the balance of power is an essential condition of the operation of international law, the steps necessary to maintain the balance often involve violation of the injunctions of international law’ (Bull, 2002 [1977]: 104). He effectively sanctions the sacrifice of an individual state as a means for maintaining the freedom of the society of states, as a collectivity (Bull, 2002 [1977]: 17). The priority he assigns to the good of the majority over the individual reveals the utilitarian underpinnings of his position.

The ‘anarchical’ freedom endorsed by Bull, then, is not the individual freedom of each member state, as a sovereign unit, but the collective freedom of the society of states. This notion of anarchy as collective freedom clashes with anarchy as formal, sovereign equality, a clear inconsistency in Bull’s argument. The inconsistency arises because Bull much like Waltz is torn between two conflicting intuitions – one points to moral and legal arrangements, the other to considerations of power, balances of power and great powers.

Great power politics has a prominent place in Bull’s and especially in Waltz’s framework. As Waltz remarks, ‘the theory, like the story, of international politics, is written in terms of the great powers of an era’ (Waltz, 1979: 72). ⁶ Great powers interaction is the factor that produces the distribution of system-wide power. Once such a distribution emerges, the less powerful states are advised to adopt a policy of balancing as opposed to bandwagoning (Waltz, 1979: 126). This argument prioritises the de facto agency of great powers even though – formally – all states count as equal. Bull echoes this view: great powers enjoy a special but strictly informal status inside international society. He quotes with approval Vattel’s definition of a balance of power, as a condition where ‘no one power is in a position where it is preponderant and can lay down the law to others’ (Vattel quoted in Bull, 2002 [1977]: 97). The idea of ‘laying down the law’ refers to legislative (sovereign) power. Formally, great powers do not possess extra legislative powers: they cannot dictate the rules of the game in international society. Still, informally, they have executive prerogatives – in waging anti-hegemonic wars to restore the balance of power (Bull, 2002 [1977]: 201). Bull assumes that anti-hegemonic wars would be rare and that the anarchical nature of the international society can be preserved if states, and the large players, learn to curb their ambitions for world domination. Unless learning occurs, there is no reason to expect like Waltz that the international system would automatically reproduce itself as a balance of power. ⁷

A system where some actors hold a superordinate status is a hierarchical system. Bull and Waltz incorporate elements of hierarchy within their theories of anarchy by granting a higher, informal standing to the great powers. But recent critics of anarchy in IR have urged a more radical, complete, turn towards hierarchy. This criticism can involve attacks on anarchy (Donnelly, 2009, 2015) as well as a defence of hierarchy (Donnelly, 2006, 2009; Hobson, 2014; Hobson and Sharman, 2005; Lake, 2001, 2009; Mattern and Zarakol, 2016). Assessing the former, anti-anarchy argument is relevant for the present discussion.

Moving beyond anarchy in IR? The challenge of conceptual atomism

In an article in International Theory, Donnelly (2015) contends that anarchy, far from being fundamental to IR, has had an impact confined to the perimeter of Waltz’s structural realism. As the dominance of structural realism is now fading, IR scholars should embrace the alternative, more productive category of hierarchy. The contribution of this anarchy critique is that it provokes us to rethink the foundational category of IR discourse. But because it is encumbered by conceptual atomism, it produces a distorted picture of Waltz’s holist account of anarchy and of theories of anarchy in IR more generally.

This atomist attitude is disclosed in Donnelly’s strategy of studying definitions of isolated concepts and even isolated words. His 2015 article tracks the frequency of references to the world anarchy in the IR literature since 1895 and concludes that this word appears seldom in IR texts predating Waltz’s (1979) magnum opus: the median number of post-1979 references increases 10-fold, though the trend is subsiding (Donnelly, 2015: 393–395, 402). Notice that this observation neglects the fact that one and the same word can designate different concepts. Some authors from the sample like Edward Gulick (1943) conceptualise anarchy as a weak order tied to a balance-of-power system of states at odds with a system tightly regulated by international law (1934: 34). Others like Friedrich Kratochwil (1989) associate the concept with the absence of a common legal order, government or state (1989: 45, cf. 66) – a meaning traceable to Thomas Hobbes’ (1968 [1651]) notion of a ‘state of nature’ (Hobbes, 1968 [1651]).

Moving from words to concepts, Donnelly argues that in IR the concept of anarchy has one core meaning – a domain without an overarching superior. This meaning has three variations: an absence of government, authority or rules (Donnelly, 2009: 51, 2015: 410, 412). On this basis, he criticises Waltz for confusing the demarcational with the structural aspects of anarchy (Donnelly, 2015: 408, 411, 414) and of inability to explain state conduct:

If anarchy means absence of a government – the only sense in which it applies across the whole of international relations – it simply is not true [as Waltz claims] that ‘self-help is necessarily the principle of action in an anarchic order’. (Donnelly, 2015: 412)

Thus, ‘[t]he Waltzian project of employing anarchy as a master explanatory variable has failed. Anarchy provides no significant analytical pay off for the discipline as a whole’ (Donnelly, 2015: 413; emphasis added). The contention is that anarchy cannot explain the self-help behaviour of states. This criticism however mixes up demarcation with explanation. It might be remembered that self-help is an offshoot of the demarcational (not the structural) concept of Waltzian anarchy. Demarcation is a descriptor: it describes the boundary conditions of the system of international politics. Since for Waltz, the international system is non-hierarchical, each state is left on its own devices – the system is one of ‘self-help’. To supply an explanation of what happens inside this type of system, it is necessary to fall back on the second, structural concept of Waltzian anarchy. The structural concept shows that states with different capabilities tend to reproduce the same (balancing) pattern.

Waltz’s theses have been misrepresented and it is important to understand why. The main reason is that Donnelly attributes to Waltz a single concept of anarchy, absence of world government (Donnelly, 2009: 51, 2015: 405, 414, 418), whereas Waltz relies on two distinct ones. This concept is called upon to perform two tasks: demarcational (descriptive) and explanatory. But if one and the same concept is used to describe and to explain something at the same time, the resultant explanation would be true by default. The culprit here is conceptual atomism. It prompts the researcher to reduce analytical complexity to a bedrock defined by one master concept – the concept of anarchy in this case. Such reductionism is misguided: theorists often use multiple concepts of the same thing in crafting arguments. Theory so construed is a set of coherent and mutually supporting arguments about something.

The reductionist method of conceptual atomism also collapses the distinction between a whole and its constituent elements. With respect to the anarchy–hierarchy controversy, it prevents us from distinguishing the idea of a system whose overall structure is anarchical and which incorporates limited elements of hierarchy from the altogether distinct idea of a system whose overall structure is hierarchical and which contains certain elements of anarchy. Thus, when Donnelly (2006) explores the phenomenon of ‘hierarchy within anarchy’, he assumes away the basic distinction between system and elements, focussing instead on a derivative distinction between formal and informal elements. But even if some of the elements of an anarchical international system are hierarchical (such as relations of inequality) – either formally or informally – this would not change its overall character of anarchy as long as these hierarchical relations are not defining of it as a whole.

Last but not least, the anti-theory ethos in the guise of conceptual atomism promotes explanatory arbitrariness. By severing the link between a concept and its theoretical context, atomism removes the frame of reference that allows us to establish whether the concept under study is analytically productive or empirically veridical. Take Donnelly’s claim that Waltz has an unduly narrow concept of hierarchy. According to Waltz ‘hierarchy equals common government’ but, as Donnelly (2009: 51–52, 2015: 409) observes, hierarchy represents a broader notion than government: government is an institution of rule; hierarchy does not have to be institutionalised. And in contradistinction to the concept of anarchy with its singular meaning (no common government), hierarchy comes in various forms and shapes. It can be ‘heterarchy’ (Donnelly, 2009: 50, 64–65, 2015: 416n), a variant of polyhierarchy containing multiply ranked hierarchical orders. Waltz, in short, has neglected the variety of hierarchical forms found in state systems. These remarks are illuminating but they are arbitrary as a critique of Waltzian anarchy because they do not respect its theoretical context.

The correct procedure for probing the coherence of Waltz’s concept(s) of anarchy is holist – it requires examining the containing (structural realist) theory and its basic premises. One such premise holds that a system constitutes a common interaction domain, governed by a unitary organising principle or structure (Waltz, 1979: 115–116). Provided that the system is hierarchical, and given that its structure must be unitary, it follows that this structure is necessarily one of monohierarchy. The structure of polyhierarchy recommended by Donnelly presupposes the opposite – that the system is segmented into discontinuous subsystems, governed by different (non-unitary) kinds of structuring principles – hence it cannot serve as a valid counterargument to Waltz’s position. An illustration of polyhierarchy is the Mediaeval European system. It was segmented into two subsystems: its units (principalities, free cities) were subject to the authority of the Papacy as a religious subsystem and, parallel to this, to that of the Holy Roman Empire, as a political subsystem (see Haldén, 2017). This co-presence of segment-specific, overarching authorities none of which has superior authority inside the system as a whole is excluded ab initio from Waltz’s structural realism due to his assumption that the system comprises one common domain with a unitary structure (either anarchical or hierarchical).

Far from exposing Waltz’s theory of international anarchy, and presumably the IR discourse of anarchy, as incoherent, Donnelly’s criticism is itself burdened. Its shortcomings stem from its conceptual atomism. Theoretical holism corrects for this by reminding us that a concept does not mean something by itself but solely in connection to other concepts inside a theory. In Quine’s words, ‘Our statements about the external world face the tribunal of sense experience not individually but only as a corporate body’ (Quine, 1951: 38). As we have seen, Bull and Waltz worked with the same concepts of anarchy, yet diverged in their conclusions about the ramifications of interstate conduct because these concepts were nested inside different theories (scientific versus normative). It is counterproductive to try to ascertain what ‘anarchy’ – or any other free-standing concept – means outside a theoretical context.

In closing, a trio of holist lessons may be drawn. The first is that scaling up the standpoint of analysis in IR from concepts to theories opens the door for a proper, contextualised understanding of international politics. Second, we must not forget that anarchy is responsible for the peculiarity of the international system. Any attempt to present it as just another variant of a hierarchical social system where the relations of agents are ordered by a superior body is bound to distort its sui generis character. In international relations, the participant states are engaged in mechanisms of self-regulation and self-policing without oversight by a global sovereign. Third, the issue is not whether the anarchical international system contains some elements of hierarchy (Deudney, 2007; Donnelly, 2006, 2009) but whether the system – as a whole – is structured as hierarchy or, alternatively, as anarchy. This leads us back to the problem of structure. Because the structures of (social) systems, both in the international sphere and outside it, are unobservable, they cannot be studied by purely empirical means. For those interested in structuralism, be it scientific (a la Waltz) or normative (a la Bull), theory is indispensible. One final question to address is whether theories of international anarchy can illuminate an increasingly globalised realm which departs from the state-centred anarchical structure depicted by Waltz and Bull.

Conclusion: International anarchy and the promise of normative structuralism

The future promise of structuralist theories of international anarchy, I wish to suggest, lies in their normative appeal. To construe the international anarchy in normative terms is to envisage a normative structure that has a value dimension – a society of free states committed to the value of anarchical (equal) freedom. Although this theme was discussed in relation to IR theorising in the section on Bull above in these concluding pages it is pertinent to link it to general political philosophy as this affords a comprehensive overview of the recent anarchy–hierarchy controversy in IR. It is noteworthy that any society of states, not just Bull’s version prominent in IR, qualifies as anarchical on two conditions: that its form is anarchical, implying that states are its paradigmatic members, and that the participating states recognise each other’s anarchical freedom.

Philosophers writing on international relations have articulated diverse theories of an anarchical international society along these lines. Some have borrowed inspiration from Hegel’s idea of political communities as loci of value. For Michael Walzer (1992 [1977]), international society is a legal association for the protection of actual political communities. Each state member of international society has legal rights to independence and territorial integrity because it protects a self-determining political community which realises the rights to life, liberty and community of its members, men and women (Walzer, 1992 [1977]: 54, 57, 61). Individual rights, and specifically the rights of individuals who share membership in a political community, are the principal object of moral concern. That a state is free means that its citizens, as a community, are free to choose their preferred way of life. Mervyn Frost (2009) has proposed a different neo-Hegelian argument which integrates the rights of states and the rights of individuals into a two-tiered realm of ‘double anarchy’. Anarchy here epitomises freedoms (rights), and it has two dimensions: international and global. The rights of individuals cannot be adequately protected in the global domain of rights, Frost (2009: 78–115) contends, without the concerted effort of states in the international domain: to this end, states must act together, as an international society.

An alternative, Kantian image of international society portrays it as a juridical order between free states. A free state with separation between executive and legislative powers – a ‘republic’ in Kant’s terms – recognises the rights of its citizens. But, as Terry Nardin (2011) has noted, international society is a structure with various modalities and it may bring together states not all of which are genuine republics. Each individual state therefore must retain its autonomy (right to independence) inside international society: only if it is free to determine its own constitutional structure can it ensure respect for rights (Nardin, 2011: 2065). On Kantian premises international society is grounded on the principle of individual freedom – for both states and human persons – and cannot be an enterprise of collective freedom, pace Bull’s utilitarian argument.

What unites such diverse writers is the conviction that international society is animated by ‘anarchical’ freedom (see also Cerny and Prichard, 2017). This is a kind of equal freedom which translates into value pluralism. Value pluralism expresses a distinctively modern way of connecting value to political authority. In earlier historical times, the justification of political authority was based on a hierarchy of value – Platonic forms, God’s will or natural law prescribing universal principles of thought and action for all rational beings. Such hierarchy of value identified the best way of life, but consensus about it has become elusive in our age of post-Enlightenment (Berlin, 1979). Instead of agreeing on what is best, as moderns we can at most agree on what is good – creating the prospect that different visions of the good life can coexist inside an anarchical international society.

It is instructive to consider the current anarchy–hierarchy debate in IR at the background of this more general philosophical problematic. From this general perspective, one important question is, what is the appropriate institutional form for organising political authority in the international sphere? This is a question about the basic form or structure of the international system as a whole and not merely about some of its aspects. Two principal groups of interlocutors can be distinguished depending on whether they prefer an anarchical, or a hierarchical structure. Proponents of hierarchy (in IR, see Donnelly, 2006, 2009; Hobson, 2014; Hobson and Sharman, 2005; Lake, 2001, 2009; Mattern and Zarakol, 2016) think of world politics in terms of a hierarchical authority covering the globe. The form of such authority is not international but global – it can range from a unitary world government (Craig, 2008; Nielsen, 1988; Wendt, 2003), to a more loosely structured global federation of states (Nardin, 2011: 2065–2067), to an ensemble of issue-specific global institutions. When normative theorists appraise the formal structure of this global authority, they point to the danger of empire (Cohen, 2004). Exponents of international anarchy (Rawls, 1999: 36) push this normative argument further. Appealing to Kant’s warning (Kant, 1991 [1795]: 113), they argue that a globally extended political authority would most likely be tyrannical, providing a singular conception of the good for all of its members. Because this debate is about values, it cannot be settled analytically: we shall have to choose between anarchical freedom and competing values (see also Cerny and Prichard, 2017; Karatzogianni and Robinson, 2017).

In their turn, advocates of globalisation criticise international anarchy by redirecting our attention from formal structure to process. They question the primacy of the sovereign state, and thereby the idea of international anarchy as a horizontal order between sovereign states. Theorists of globalisation (an overview is Held and McGrew, 2007) claim that the state has lost its capacity to control the distribution of political and economic value, as this domain is now global (Strange, 1996). In these changed circumstances, it is imperative to take account of global processes such as transborder market transactions (Strange, 1996: 7–13), ‘governmental networks’ of technocrats (Slaughter, 2004: 12–14) and ‘governmentality’ (Foucault, 2002) where local populations are subject to managerial control exerted from a distance. The logic of such arguments reflects the values of administrative control and economic efficiency and not that of anarchical freedom.

In essence, then, what critics allege is that the discourse of international anarchy is an anachronism. The anarchical society projected by this discourse has the architecture of a system with a relatively clear-cut set of rules and clear-cut set of actors, states. But all this seems to be a relic of the past. At present, we live in an era of global governance (Rosenau and Czempiel, 1992) manifest in the fragmentation and diffusion of authority among various actors, above and below the level of the state. The result is a patchwork of global hierarchies (see Cerny and Prichard, 2017). If so, what degree of agency is left to states in a world permeated by global processes?

The principal worry is that global processes and global actors may not operate in ways that are just or democratic. The agencies in a position to regulate actors with enormous resources but no public accountability – such as private security companies or private banks with a global reach – are states (Brown, 2000: 16–17; Nardin, 2011: 2070). For the time being, states remain the key agencies capable of enforcing standards of global human rights. And if, to return to Frost’s diagnosis, the protection of global human rights and freedoms depends on the concerted effort of states acting as an international society, it is critical to ensure that the freedom so realised would not be unequal freedom achieved at the expense of some actors but ‘anarchical’ or equal freedom. The task that lies ahead for theorists of international anarchy is to spell out the rules of a normative structure that would protect the anarchical value of equal freedom in a world no longer dominated by states.