How should we conceptualize the structures of political systems? The standard answer in International Relations (IR), drawing heavily on Kenneth Waltz's Man, the State and War and Theory of International Politics,Footnote 1 identifies three levels of analysis (individuals, states, and international systems) and stresses the differences between anarchic top-level international systems and the hierarchic national systems that compose them.
I develop an alternative typology, based on the arrangement of central and peripheral polities in layered political spaces. I then apply this typology to Eurocentric political systems from the high middle ages to today. Rather than see no fundamental change across nearly a millennium (the system remained anarchic and for all but half a century multipolar) or a singular modern transition (with three or four centuries of fundamental structural continuity on either side) I depict a series of partial structural transformations on time scales of a century or two. I also recurrently step back to consider the nature and significance of such structural models; why and how they matter.
Although I often directly criticize the Waltzian conception of structure – which not only is widely endorsed (even by many who are not structural realists) but is the only widely endorsed conception of structure in contemporary IRFootnote 2 – my broader aim is to point toward a more accurate and truly systemic conception of political structures. My criticisms of Waltz thus arise as contrasts with the positive arguments that I present. And those positive arguments drive, structure, and make up most of the substance of what follows.
International systems, I try to show, are structured (arranged) in greatly varied and often complex ways. (They do not have just a few simple structures.) Structural change therefore is relatively common and typically the result of the interaction and accumulation of changes in intertwined elements of interconnected systems (not radical innovations or dramatic changes in core principles). Such an understanding, which is rooted in a distinctive conception of explanation (involving configurations rather than causes), grounds an illuminating comparative framework that is especially attractive today because it allows us to paint a penetrating picture of the nature and significance of globalization.
Levels, centers, and spatio-political structure
I begin by sketching a notion of what I call spatio-political structure.
Layered political spaces populated by polities
Imagine a layered political space in which lower levels represent greater spatial detail. Higher-level entities spatially encompass those on lower levels, creating a hierarchy of relative size.
Figure 1 models such an abstract political space. The entire space on the top level is (arbitrarily) divided into four pieces on the second level, which are (arbitrarily) divided into four on the third level. (The division into three levels, and the boundaries of the included space, are also arbitrary, in the sense that they could easily be otherwise.Footnote 3)
A three-level spatial grid.
The next section populates such spaces with ‘polities’ – corporate political entities capable of (at least semi-) autonomous actionFootnote 4 – to develop models of what I call spatio-political structure.
Centers and peripheries
I distinguish two interrelated types of polities: centers and peripheries.
Social and political systems are structured around specially valued ‘places;’ centers. As Edward Shils puts it ‘there is a central zone in the structure of society … Membership in the society, in more than the ecological sense of being located in a bounded territory and of adapting to an environment affected or made up by other persons located in the same territory, is constituted by relationship to this central zone.’Footnote 5
Centrality has many dimensions (e.g. political, religious, cultural, and economic). Centers may stand in varied relations to one another and to their peripheries. Whatever the details, though, how centers are related to people, places, groups, institutions, and values is essential to the structure of social and political systems.Footnote 6
Centralization and peripheralization
‘Centralization’ and ‘peripheralization,’ as I understand them here, are interlinked social processes.
Peripheries are defined not just by distance from but also by subordination to a center. ‘Peripheralization’ transforms formerly autonomous peoples, polities, or places (or peripheries of another center) into peripheries of a specific center. (I call distant but unperipheralized polities ‘marginal,’ ‘frontier,’ or ‘outlying.’)
Conversely, a powerful polity that merely sits on top of and dominates other polities is not ‘central.’ A conquering power becomes an imperial center by peripheralizing peoples, places, or polities (which are transformed in the process).
I thus avoid using ‘centralization’ in the ordinary-language sense of concentrating authority or activity in a single place. For example, I would describe 20th-century France not as ‘more centralized’ than the Roman Empire but as differently centralized. Authority was more concentrated in France and more dispersed in Rome.
Types of political systems
In IR's standard structural framing, pre-defined units (individuals and states), on (three) pre-defined levels, combine into pre-defined types of (national and international) political systems. I instead treat as empirical questions the types of polities that exist within a space, their distribution and relations, and the resulting kinds of systems.
This section develops a simple typology of spatio-political structures based principally on the number of top-level political centers and the homogeneity or heterogeneity of centers, peripheries, and their relations and, secondarily, on the relative autonomy of centers and peripheries.
In these models, the number of levels (and the space encompassed) varies with the empirical conditions of the world and the focus of the analysis. So do the kinds of polities, their distribution, and their relations. Varied centers and peripheries thus stand in diverse, complex, and often nested relations. And higher-level entities are only relatively large (not necessarily superordinate).Footnote 7
States systems: systems of single-level governance
Figure 2 depicts IR's standard (Waltzian) model of an ‘international political system’ – a states system.
A Waltzian states system (single-level governance by terminal peer polities).
The top level is unoccupied. (In Waltz's terms, the system is anarchic.) The middle level is occupied by a relatively few centers (states/units). The bottom level is occupied by a relatively large number of polities that are peripheries of a second-tier center. (The units/states are internally hierarchical.)
I call this a system of single-level governance by terminal peer polities.Footnote 8 The system is arranged around ‘terminal polities’: the most encompassing polities in the space; polities that are not parts of larger polities (but are parts of a larger political system that is not a polity).Footnote 9 These terminal polities are located on a single level. They are peersFootnote 10 in an international society. And they provide most of the governance in the system, ‘nationally’ (‘hierarchically’/vertically) and ‘inter-nationally’ (horizontally, both bilaterally and multilaterally).
This account differs from Waltz's in four important ways.
• This is only one type of international system (not a privileged or default model).
• I define a states system by its governance structure – not by the fact that it is ‘not national’ (lacks a single central government). Multiple centers on a single level provide vertical rule within their polities and horizontal governance of the broader system of which they are parts.
• Authority in the system is allocated to the terminal polities – not, as Waltz claims, absent.Footnote 11 To pretend that there is no authority in states systems (rather than that authority is distributed to states) is to fundamentally misrepresent the structure of the system, taking the authority of states as pre-given and somehow outside (rather than a defining feature) of the system.
• ‘Structure’ refers to the arrangement of the parts of a systemFootnote 12 – not something on a higher ‘system level.’ Neither the system nor its structure is on a spatial or organizational level separate from the ‘units.’Footnote 13 The top-level political system shown in Figure 2 is on the second (inter-state) level – not the (empty) top level.
This suggests re-presenting states systems, in their simplest form, as two-level systems, as shown in Figure 3.Footnote 14
A states system (single-level governance by terminal peer polities).
States and empires: systems of single-center governance
What I call systems of single-center governance have one top-tier center. I differentiate two types, based on the relations between the center and its peripheries.
Integrated polities (states)
In integrated polities, represented in Figures 4 and 5, relatively homogeneous peripheries stand in fundamentally similar relations to the center. Modern states are examples, in both their unitary (e.g. French) and federal (e.g. American and German) forms.
A unitary State (single-center governance of a unitary integrated polity).
A federal State (single-center governance of a federal integrated polity).
All German Länder stand in similar relations to the federal center. In contrast to French départements, though, they are not just subordinate administrative units. In addition to being peripheries (of the top-level center) they are (second-level) centers with their own peripheries.
I represent this difference with ‘fills.’ The top-level center in shown both Figures 4 and 5 is dark and opaque. Merely administrative units are unfilled. The second-tier polities shown in Figure 5, however, are filled lightly and are partially transparent, reflecting both their partial autonomy (or limited subordination) and the fact that they are both centers and peripheries.
I call both types States, with a capital S. (Most readers, I suspect, will find this less jarring jargon than (the more informative) ‘integrated single-center polities.’) ‘The modern state’ is, in these terms, a type of State characterized by legal-rational bureaucratic rule. Although a world State is theoretically possible, all historical States have been parts of larger political systems.
Aggregated polities (empires)
In aggregated polities, illustrated in Figure 6, heterogeneous peripheries stand in varied relations to the governing center.
An empire (single-center governance of an aggregated polity).
Lower-level polities differ not just in size but in character and political function, as indicated by both their varied shapes and their differing fills (which, as in Figure 5, represent different degrees of autonomy). I call such a polity an empire, following the ordinary-language definition of ‘an extensive territory…often consisting of an aggregate of many separate states or territories.’Footnote 15 Most scholarly definitions also see ‘empire as a territorially expansive and incorporative kind of state.’Footnote 16
Empires that are polities in a larger system (e.g. 19th-century France) I call ‘empire-states,’ using ‘state’ in the broadest sense of polity. Empires that encompass all of a space (or a regional sub-space) I call ‘imperial international (or regional) systems.’ (The Roman Empire was an imperial international system with an empire-state at its core.)
Integrated and aggregated systems of single-center rule
Both ‘States’ and ‘empires’ – ‘integrated’ and ‘aggregated’ ‘single-center’ polities – typically arise and grow through agglomerative processes such as war, marriage, inheritance, or voluntary union. ‘States’ (with a capital S) re-integrate their peripheries into relatively similar parts of a relatively unified whole. ‘Empires’ remain closer to their origins, ‘choosing’ to rule their peripheries differentially; as parts of an aggregated (rather than an integrated) system.
For example, late-medieval Florence acquired the surrounding communities through a series of ad hoc ‘bilateral accords [with] a large number of single entities.’ ‘The plurality of asymmetrical relationships gave shape to a politico-territorial system in which a mosaic of unintegrated clusters was arranged about Florence as a central pole, each defined by separate autonomies and privileges.’Footnote 17
In Classical Athens, by contrast, the reforms of Cleisthenes in 508/7 bce created a unitary political–juridical structure. ‘The Athenians’ were spread across the whole territory of the polis. Those in physically or socially peripheral demes (districts) were no less citizens than those who lived in the city. (Classical Greek distinguished astu (‘city’ as urban space) from polis (‘city’ as polity or ‘city-state’).) And, as the system matured, a single system of law governed all citizens equally (isonomia).
Note my loose use of ‘single-center’ in reference to federal States and empires, which have peripheries that are also lower-level centers. By ‘single-center’ I mean a system with only one top-level center. Whether a system has one or more top-level centers is a crucial feature of spatio-political structure, as I understand it.
Heterarchies: systems of multi-actor multi-level governance
In Figure 7, multiple heterogeneous centers located on multiple levels stand in varied relations to one another and to their peripheries.
Heterarchy (multi-actor multi-level governance).
I describe this as a system of multi-actor multi-level governance and use the label heterarchy,Footnote 18 which combines the root arkhē (rule) or arkhon (ruler) with the prefix hetero-, indicating difference or variety. Heterarchic systems may be polities in a larger system (heterarchy-states) or not (heterarchic ‘international’ systems).
In heterarchies, governance is organized in significant measure functionally. Different authorities regulate different issues, making most places subject to multiple and varied centers that share governance. Different kinds of centers exercise different authorities in overlapping spaces.
In states systems, by contrast, political functions are concentrated in terminal polities, making governance fundamentally territorial. Centers on a single level exercise usually similar authorities in different places.
This difference is particularly clear on the top two levels of Figure 7, which I constructed with contemporary Europe in mind. In the earlier figures, largely vertical lines of authority indicate lower-level entities that are fundamentally territorial subunits of higher-level entities. The lines of authority in the top two levels of Figure 7, however, are cross-cutting, reflecting the fact that different top-tier polities (in the European case, different regional organizations) have different functions and different memberships.
Another important feature of Figure 7 is that the top-tier polities are not the predominant centers in this space. It is structured around the second-tier polities (in the European case, sovereign territorial states). Higher-level polities are only spatially more encompassing. (Larger does not mean more powerful. Top-level does not mean sovereign.)Footnote 19
The largest top-tier polity shown in Figure 7, however, is at least approximating a co-equal center – despite being on a different level. (It is only one shade lighter and just 20% less opaque than the second-tier polities.) Further flows of authority to top-tier polities would at some point require representing the partial peripheralization of the second-tier polities by reducing their shade or opacity. (One might argue that such a representation is already appropriate for the members of the EU.)
A great attraction of my framing is that it allows accurately mapping relations of complexly intertwined political systems. The Waltzian framing, by contrast, works well only in the simple case of States-in-a-states-system – a type of international system that merits neither conceptual nor historical privilege.
Types of polities and international systems
I have focused on the differences between systems with one and more than one top-level center and with homogeneous and heterogeneous centers and peripheries (rather than emphasize the presence/absence of a top-level government or anarchy/hierarchy as an ordering principle).
I have identified three types of polities.
• States (with a capital S): systems with a single top-tier center that stands in similar relations to its peripheries (Figures 4 and 5).
• Empire-states: systems with a single top-tier center that stands in varied relations to heterogeneous peripheries (Figure 6).
• Heterarchy-states: polities composed of multiple heterogeneous centers that operate on different scales and stand in varied relations to one another and their peripheries (Figure 7).
I have also identified three types of ‘international systems,’ understood as political systems composed of terminal polities; multi-center systems that are not themselves polities in a larger system.
• States systems: single-level governance by terminal peer polities (Figure 3).
• Heterarchic international systems: multi-level multi-actor governance (Figure 7).
• Imperial international systems: one actor predominates in a system that is more an ‘international’ system than a ‘national’ empire (Figure 6).
My distinction here between systems with one or more than one top-tier polities has similarities to Waltz's division of hierarchic/national and anarchic/international systems. In addition, my focus on homogeneity and heterogeneity has resonances with Waltz's conception of functional differentiation.Footnote 20
I argue, however, that a states system is not the normal or default form of international relations; that political spaces do not typically have a simple two-level national–international structure; and that political systems do not have a singular ordering principle.Footnote 21 International political systems, no less than national political systems, are structured in varied and often complex ways. And these variations are not primarily, let alone solely, differences of polarity.Footnote 22
Centralization as assemblage
Centers and peripheries are created through processes that transform the related elements and create ‘assemblages,’ as that term is used in ‘assemblage theory.’Footnote 23
Understanding systems as wholes with distinct identities and emergent (irreducible) properties,Footnote 24 assemblages are systems with parts that are ‘extrinsically’ related, in the sense that they retain a certain separateness or separability.Footnote 25 For example, an archeological assemblage – ‘an associated set of contemporary artifacts that can be considered as a single unit for record and analysis’Footnote 26 – is the product of ‘extrinsic’ ‘logics’ of deposition, preservation, excavation, and analysis.
In an assemblage, the whole is ‘more than’ the sum of its parts; it has a character and meanings distinct from its parts. (It is a system.) The parts, however, are ‘more than’ just pieces of a whole; they retain a separate or separable identity; they are linked (more or less tightly) into a heterogeneous entity. (It is an assemblage.)
Parts of a complex organism, by contrast, are related ‘intrinsically.’ For example, a human heart can be part of only one kind of whole.Footnote 27 Similarly, many social roles, such as parent–child and teacher–student, are intrinsically related – although families and classes, like most social groups, are assemblages.
Centers and peripheries, as I have defined them, are intrinsically related. The polities and systems of polities that they compose, however, are assemblages. An assemblage frame emphasizes that polities often are both parts of and separate from larger political systems – a fact that is well captured by a spatio-political framing. And, I want to suggest, most political systems, both national and international, are assemblages – ‘more than’ a mere environmentFootnote 28 but ‘less than’ a totalizing whole.
So what?
What can we do with this?
Similar to any good typology, it helps us to depict and explain some of fundamental characteristics and parameters of political systems and thus compare systems and explore continuity and change.
Descriptive accuracy and structural explanations
IR's standard structural typology depicts all international systems as anarchic and fundamentally the same (varying only secondarily in polarity).Footnote 29 I am in effect arguing that we should not take seriously an account that sees no structural difference between Figures 3 and 7 (both of which are anarchic and multipolar). If structure indeed means the arrangement of the parts of a wholeFootnote 30 then states systems and heterarchic international systems unquestionably have different structures.
This appeal to descriptive accuracy, I am well aware, will strike many readers as misguided. Waltz, for example, was scathing in his contempt for descriptive accuracy,Footnote 31 which is also usually downplayed, even denigrated, in both correlational and rational-actor explanations.
If the point of theory, though, is, as Waltz put it, ‘to lay bare the essential elements in play and indicate the necessary relations of cause and interdependency’Footnote 32 or ‘to single out the strongest propelling forces’Footnote 33 then descriptive accuracy is needed – not in every detail (even descriptions do not try to get all the details right) but in certain essential ways. Theoretical simplifications are, of course, required. But their aim is, as Waltz put it, ‘to find the central tendency among a confusion of tendencies, to single out the propelling principle even though other principles operate, to seek the essential factors where innumerable factors are present.’Footnote 34 This requires, in addition to abstraction, essential or fundamental descriptive accuracy.
More particularly, structural explanations claim to explain outcomes by the arrangement of the parts of systems. (That arrangement explains this phenomenon.) A fundamentally inaccurate structural theory or typology is worse than useless. And that, I am arguing, is true of IR's standard (Waltzian) anarchy plus polarity conception of international political structure.
Of most immediate relevance, states systems, pace Waltz, are characterized not just (or even primarily) by the ‘anarchic’ absence of a top-level government but also (and at least as essentially) by the ‘hierarchic’ allocation of authority to top-tier polities. States systems combine ‘national’ ‘hierarchy’ and ‘international’ ‘anarchy.’Footnote 35
One fruitful way of getting at this, I am suggesting, is to look at relations between centers and peripheries across spatio-political levels. For example, in Figure 3 the centrality of the top-tier polities is the other side of the peripheralization of the second-tier polities. Peripheralized lower-tier polities and multiple top-tier polities are equally central to the structure of the system. The structure of the system is not reducible to the relations of the top-tier polities – a fact that is even more striking if we look at the heterarchic system in Figure 7.
Typological explanations
OK. But this typology, whatever its virtues, is just that: a typology – not a theory.
But as numerous scholars have shown, in varied ways, ‘structure,’ defined as anarchy and the distribution of capabilities, (alone) explains nothing.Footnote 36 And if we want to know not only that ‘balances of power tend to form,’Footnote 37 but which balances form, Waltzian structural models prove wildly inaccurate. For example, states balance not against power but against threat, a largely perceptual (unit-level) variable.Footnote 38
Waltz ended up with ideal-type models of two kinds of political systems. His ‘theory of international politics’ ultimately amounted to an account of certain ways in which ideal-type international systems differ from ideal-type national systems.Footnote 39
That is a ‘problem,’ though, only if we expect to be able to read causal theory more or less directly off a depiction of structure. That expectation, however, is baseless. Structural models, I will argue in the section ‘The nature of structural/relational explanations’, explain in a very different way.
First, though, I develop an extended case study to illustrate how my spatio-political framework grounds illuminating comparisons and allows us to accurately chart structural continuity and change.
From medieval heterarchy to the early modern states system
In IR's standard account, Eurocentric international systems have not changed structurally from the central/high-medieval period to today. (They have remained anarchic and, except for half a century, multipolar.)
The other common narrative sees a singular modern transition (somewhere between 1500 and 1650). Such readings, however, do not identify criteria that justify depicting a decisive modern break with three, four, or more centuries of fundamental continuity on either side.
I show that structural changes cumulated every century or two to generate identifiable eras but not radical breaks. And, I will argue, the absence of structural continuity across both medieval and modern Europe has important implications for how we think about globalization.
High-medieval heterarchy
I begin at the end of the ‘high’ or ‘central’ medieval period, suppressing great spatial variation to depict political structures in western and central Europe around 1250, give or take half a century.
Political authority was divided functionally (heterarchically) into sacerdotium (supreme spiritual authority; priesthood) and imperium (supreme secular authority; empire) or regnum (secular rule; government). Two parallel political hierarchies jointly governed high-medieval (Western) Christendom.Footnote 40 (The most important political task in the medieval world was to regulate religious practice and belief, the path to eternal salvation.)
Each functional domain had four lightly layered levels.
At the top of the ecclesiastical hierarchyFootnote 41 was the Pope, the Bishop of Rome; the successor to Peter, to whom Jesus gave the keys to heaven.Footnote 42 Sacerdotal authority was exercised by archbishops, bishops, and parish priests, in nested communities of decreasing scale. As the hierarchical or physical distance from Rome increased, though, papal control tended to decline. And secular authorities everywhere regularly exerted substantial influence.
The top secular level was occupied by the Emperor, the successor to the emperor of Rome.Footnote 43 Propagandistic protestations of universal imperium aside, though, he had little power beyond the boundaries of the Empire (Germany, northern Italy, southeastern France, and the low countries) – which itself was a disparate collection of more than two hundred secular and ecclesiastical polities of diverse sizes, shapes, and powers that stood in varied de jure and de facto relations to the Emperor.
I call the second secular level regnal, using a neologism created by Susan ReynoldsFootnote 44 to identify kings as the typical actors on this level, without suggesting anything else about the character of these polities. Rulers such as the King of England acknowledged no higher secular authority. (The motto rex imperator in regno suo [the king is emperor in his realm] was being used regularly by the early 14th-century.Footnote 45) But they had limited legal, fiscal, and political rights and usually lacked the resources to exercise much control beyond their personal dynastic domains.
Secular rulers on what I call the provincial level thus typically were subject only to limited direction and oversight.Footnote 46 Some, such as the Duke of Saxony, rivaled their status superiors in power and wealth, making them more centers than peripheries.Footnote 47 As John Watts nicely puts it, there was ‘a thin royal crust over a mass of independent jurisdictions.’Footnote 48
In local communities, which were overwhelmingly rural, secular authority was exercised by lords (seigneurs, Herren), with very little (and often no) higher supervision. (Most localities were outlying areas that had not been peripheralized.) Furthermore, ‘many lordships … were little more than private estates, with odd scraps of jurisdiction attached – often insecurely.’Footnote 49
Now add to this the fact that secular and ecclesiastical jurisdictions rarely corresponded. The result was a heterarchic system of overlapping and interpenetrating authorities and jurisdictions that stood in varied, complex, and often contested relations.
Although the particularism of medieval life stands out to us today, the emphasis at the time was on the unification of particularities in an all-encompassing cosmic hierarchyFootnote 50 – and the temporal hierarchical unity of Christendom, the universal polity that provided the path to salvation. Therefore, the top level, although populated by actors with limited powers, was essential to the structure of high-medieval politics.
Late-medieval political changes
In the late-medieval and early modern periods, the powers of the Pope and the Emperor deteriorated. And no new top-level actors emerged.
The decline of the Papacy was absolute. The Church never fully recovered from the Great (Western) Schism (1378–1417), when competing popes (and their backers) crassly competed for ecclesiastical controlFootnote 51 – allowing kings ‘to wield powers that had formerly been exercised by popes,’ especially taxation to support the Church.Footnote 52 Although the Council of Constance (1414–1418) restored ecclesiastical unity,Footnote 53 churches became increasingly national/regnal. The ReformationFootnote 54 then fractured the doctrinal and ecclesiastical unity of Christendom. The Italian Wars (1494–1559)Footnote 55 finally reduced the (secular) papacy to an Italian regional actor.
The decline of the Empire was (only) relative. Sigismund (r. 1410–1437) and Frederick III (r. 1452–1493) were not unsuccessful Emperors.Footnote 56 Leading regnal rulers, however, more rapidly improved their legal, administrative, fiscal, and military capabilities.Footnote 57 The formal change of name in 1512 to the Holy Roman Empire of the German Nation effectively recognized that the Emperor had become the regnal ruler of Germany. But extensive reforms between 1489 and 1559Footnote 58 kept it in the ranks of the great powers into the late-17th century.
New monarchies and the rise of a European states system
The central issue of ‘national’ political contention in late-medieval and early modern Europe was the balance between regnal and provincial centers. In France, England, and Spain, 15th-century wars rooted in succession crises – the last third of the Hundred Years' War (1337–1453); the Wars of the Roses (1455–1485); and a series of conflicts from 1412 to 1469 within and between the crowns of Castile and Aragon, followed by the War of the Castilian Succession (1475–1479) – weakened provincial princes and created opportunities that were vigorously exploited by the victorious kings. Over the following century, these kingdoms differentiated themselves from other regnal polities and significantly expanded and deepened their peripheralization of provincial centers.
The mid-1550s can be seen as a pivot point between universalistic medieval heterarchy and a particularistic system of increasingly regnal politics.
In 1519, Charles I of Spain, who also held the Hapsburg domains in the low countries, became Emperor (ruling as Charles V), dynastically aggregating an empire much larger than Charlemagne's. His aspiration to re-establish a universal Christian polity was evidenced not only by the five wars he waged against France, his only serious geo-political rival, but by his leadership of the military resistance to the Ottoman advance and his strong support for ‘Christian unity,’ especially suppressing heresy (Protestantism), by force if necessary, and encouraging the Catholic/Counter Reformation.Footnote 59
Charles's abdication in 1556, and the division of the Hapsburg holdings between his son and brother, ended any effort at universal imperium. No less importantly, in 1555 Charles accepted the Peace of Augsburg, which (under the doctrine cuius regio, eus religio [whose realm, his religion]) allowed certain princes within the Empire to establish Catholicism or Lutheranism as their polity's official religion.Footnote 60 This acceptance of particularistic religious heterodoxy and the subordination of churches to princes marked a decisive end to the functional bifurcation of high-medieval politics.
A four-level system of functionally divided governance was being replaced by a three-level system in which governance functions, both ‘nationally’ and ‘internationally,’ were increasingly concentrated in regnal polities (which initially were dynastic aggregates but in the eighteenth and especially 19th centuries became more integrated and increasingly national or territorial).
Early modern dynastic empire-states
Most polities in the emerging early modern states system were empire-states (agglomerated polities) not States (integrated polities) – let alone ‘modern’ legal-rational states, a model that ‘is hopelessly anachronistic when applied to an early modern state.’Footnote 61
‘The modern state has been constructed to create a uniformity or universality of life within its borders.’Footnote 62 ‘A single system of governance … [and] law applies to (virtually) all who find themselves within these boundaries.’Footnote 63 Early modern polities, by contrast, were variegated aggregations of ‘a myriad of smaller territorial and jurisdictional units jealously guarding their independent status,’Footnote 64 with ‘different social structures [and] different laws and institutions.’Footnote 65
Historians today often call these polities ‘composite states.’Footnote 66 And these ‘non-integrative unions’Footnote 67 were very much assemblages. ‘Their constituent parts ha[d] a meaningful and partly independent existence.’Footnote 68
For reasons of space, I focus here on France, which is usually considered the most ‘modern’ of the early modern great powers, and, secondarily, on Spain, Europe's most powerful polity in the second half of the sixteenth and the first third of the 17th century.Footnote 69
Castile/Spain
The marriage in 1469 of Isabella, future queen of Castile, and Ferdinand, future king of Aragon, laid the foundation for ‘modern Spain.’ The Crown of Aragon, however, loosely linked an Iberian core of Aragon, Catalonia, and Valencia with peripheral Mediterranean polities of Naples, Sicily, Majorca, and Sardinia, ‘each with its own laws and institutions.’Footnote 70 Castile was a geographically compact realm. Major towns, however, enjoyed substantial self-ruleFootnote 71 and noble families in the countryside were largely independent authorities.Footnote 72 And these two realms were only weakly joined – creating if not ‘a ramshackle assemblage’Footnote 73 then ‘a plural, not a unitary state’Footnote 74; a ‘polycentric monarchy.’Footnote 75
When the Hapsburg holdings were divided in 1556, Philip, king consort of England and Ireland (through his marriage to Mary I), received the crowns of Castile and Aragon (and Navarre) and the low countries. This in some ways represented a political ‘rationalization.’ ‘In place of the vast and cumbersome geographical monstrosity that passed for an empire under Charles V, Philip II would rule an empire of three logical units: England and the Netherlands, Spain and Italy, and America.’Footnote 76 Compared to a territorial or national polity, though, both ‘England and the Netherlands’ and ‘Spain and Italy’ were geographical (and socio-political) monstrosities. And even ‘the Spanish peninsula was not a single economic or administrative unit, but a complex of kingdoms and territories’Footnote 77 each of which ‘remained in its own compartment, governed by its traditional laws’ and institutions.Footnote 78
France
Early modern French kings, like their medieval predecessors, ‘assembled their kingdom piecemeal, layer on layer. They accreted different customs, legal systems, and privileges.’Footnote 79 Even Louis XIV added Flanders to his realm (in 1668 and 1678) through agglomerative compacts that produced ‘two governmental systems corresponding to very different historical traditions [that] found themselves having to work side by side.’Footnote 80
This ‘conglomeration of duchies, counties, and provinces’Footnote 81 ‘with a wide range of local institutions adapted to the many local cultures,’Footnote 82 was ‘only partially under royal control’Footnote 83 – especially as one moved away from Paris. Provincial governors were more or less independent (and often deeply rooted) powers.Footnote 84 And major towns were semi-autonomous.Footnote 85
The early Bourbon kings did significantly increase central administrative capabilities. Most notably, after 1634 intendants exercised administrative oversight in two dozen généralités (new administrative districts).Footnote 86 In typical early modern fashion, though, these were layered on top of, rather than replacements for, older jurisdictions, institutions, and offices.
Furthermore, almost the entire judiciary and much of the royal administration owned their officesFootnote 87 – which after 1604 were fully transferable private property.Footnote 88 By 1664 there were more than 45,000 venal offices in the judiciary and financial administration alone.Footnote 89 And many of these offices had been created not out of an administrative rationality but simply so that they could be sold.
Militarily as well, early modern France was far from a ‘modern state.’
For example, in the Wars of Religion (1562–1598), the King raised only about a third of his forces directly.Footnote 90 Mercenaries, who are notoriously unreliable, made up another third. The remaining third was composed of temporary additions, raised and led by local and provincial nobles (and towns), that ‘were little more than the armed clients of their noble commanders.’Footnote 91
This ‘republic of semi-independent warlords’Footnote 92 ‘was but a small step forward from the feudal levies of the middle ages.’Footnote 93 The king persuaded, cajoled, and coerced more than he commanded. He allied with, rather than ruled over, the leading nobles – when he was not struggling with or fighting against them.
Louis XIV (r. 1643–1715) did largely suppress local and provincial military forces.Footnote 94 And he was able to field an army of 350,000 – more than double the largest French force during the Thirty Years' War (1618–1648), which was three times the size of the largest French army in the Italian Wars (1494–1559).Footnote 95
The Sun King's army, however, was not a ‘modern army’ but what John Lynn aptly calls a ‘state commission army.’Footnote 96 The king, rather than raise or provision ‘his’ forces, contracted with (noble) colonels, who recruited and equipped soldiers, and with what David Parrott calls ‘enterprisers,’Footnote 97 who supplied them. (The government only began to provide arms directly to soldiers in 1727Footnote 98 and bread in 1799.Footnote 99)
Furthermore, Louis XIV institutionalized venality in military office.Footnote 100 Colonels ‘were officially allowed to sell the captaincies to suitable candidates. In their turn the captains sold lieutenancies … until each commission from the lowest to the highest came to be regarded as a piece of property.’Footnote 101 The King and his ministers could only try ‘as far as possible to keep major military commands in the hands of family, clients and allies.’Footnote 102 Although this public–private partnership partly reflected the superior capabilities of local actors,Footnote 103 it was primarily a ‘compromise that the crown and war ministers needed to establish with French elites if the latter were to continue to provide large-scale financial support for the state's military activity.’Footnote 104
Periodizing early modernity
In this as in so many other ways, the early modern era was a ‘middle age’ between the medieval and modern worlds; post-medieval but pre-modern. Furthermore, as the preceding discussion hints, it falls into two rather different periods.
The earlier ‘Renaissance’ period was a time of hotly contested peripheralization, culminating in the tumultuous 1640s and 1650s: the Thirty Years' War (1618–1648), the English Civil War (1642–1651) and Protectorate (1653–1659), the Frondes (1648–1653), and the Catalan Revolt (1640–1659) and Portuguese War of Independence (1640–1668). By the last third of the 17th century, though, the French and English kings had made their principal challengers peripheralized parts of regnal realms.Footnote 105 This ushered in the later or ‘Baroque’ (or Ancien Régime) period. 1660 – the Stuart Restoration in England, followed the next year by Louis XIV's assumption of personal rule – is a conventional marker.
Baroque states, however, remained agglomerated, not integrated, polities. Although provinces (and provincial nobles) no longer had the option of armed resistance, they continued to enjoy considerable autonomy and a wide range of fiscal, legal, and political privileges. And the central government still had only the most limited local capabilities.
The interdependence of centralization and peripheralization
The concentration of political power in early modern kings, I am arguing, involved complex processes of re-centering. And this re-centering was achieved through contentious, often violent, ‘negotiations’ – and continuing renegotiations – that produced ‘more powerful and plural ‘regnal’ polities.’Footnote 106 ‘The gradual … centralisation of power in the person of the king enforced a decentralisation in geographical and institutional terms which respected the strength of provincial particularism.’Footnote 107
Early modern politics were based on what Angelo Torre nicely calls ‘empowering interactions and entwining jurisdictions’; ‘a reciprocal sequence of “crossed legitimations” between different social, juridical and political actors.’Footnote 108 Early modern kings ruled not so much over their provinces (and the privileged groups that dominated them) as in conjunction with them – a type of rule that was closer to that of their medieval predecessors than their 19th- and 20th-century successors.
Territorial and social corporations
Early-modern provinces were understood, especially in France, as territorial corporations; legally recognized entities that enjoyed privileges, in the root sense of private laws.Footnote 109 Corporate privileges were also, though, ‘assigned vertically along a hierarchy of social status.’Footnote 110 These privileged social corporations, although outside my spatio-political focus, were central to the composite character of early modern politics.
NoblesFootnote 111 enjoyed extensive social, legal, and fiscal privileges.Footnote 112 Towns – urban corporations composed of their citizen members – enjoyed more or less broad powers of self-rule and various economic and fiscal privileges.Footnote 113 Churches, both Protestant and Catholic, were also privileged.
Furthermore, in most regnal realms, select social corporations – usually some combination of clergy, nobles, and towns – were politically represented as ‘estates’ (états, Stände, staten, stati) organized in assemblies at the ‘national’ (e.g. the ‘Estates General’ in France and the Netherlands) and/or ‘provincial’ level.Footnote 114 These assemblies of estates ‘claim[ed] to represent a wider, more abstract, territorial entity – country, Land, terra, pays – which, they assert[ed], the ruler is entitled to rule only to the extent that he upholds its distinctive customs and serves its interests.’Footnote 115 ‘The territorial ruler and the Stände [estates] make up the polity jointly, but as separate and mutually acknowledged political centers. Both constitute it, through their mutual agreement.’Footnote 116
The other crucial corporate group in early modern Europe was the family/clan/dynasty. Deploying their status, children, lands, offices, and wealth, families created complexly networked alliances on and across the local, provincial, and regnal levels.Footnote 117 In the resulting systems of patronage politics, the king was not a qualitatively different actor but ‘the archpatron,’Footnote 118 ‘best visualized as sitting, spider-like, at the centre of a kingdom-wide network of patron-client relations’Footnote 119 – pursuing, like most other actors, primarily family (dynastic), not national, interests.
The logic of composition in early modern Europe
From a modern perspective, early modern regnal polities look like kludges. But rather than ‘an unsatisfactory prelude to the construction of a more effective and permanent form of political association,’Footnote 120 they were the European norm for three centuries – largely because they reflected the values of the time.
Even in the 18th century, ‘few European states had any obvious geographical, ethnic, or linguistic unity, nor was it widely felt that they should.’Footnote 121 People saw themselves (and each other) as members of multiple complexly interrelated corporate groups with different statuses. And a legitimate government was expected to respect corporate privileges. As James Collins puts it in the case of the French legal system, ‘the thicket of jurisdictions, seemingly so absurd, in fact served a very important political purpose: it protected the contracts between the king and [privileged] members of French society.’Footnote 122
‘Full incorporation of distant provinces was seldom attempted and even more rarely accomplished.’Footnote 123 The medieval world of a ‘patchwork of jurisdictions, operating under a (sometimes very light) co-ordinating authority’Footnote 124 had become more ‘centralized,’ in the sense that more capabilities and authority were concentrated in the regnal center. It remained, though, a dynastically aggregated patchwork – because that suited the political needs and aspirations of (at least the dominant groups in) the multiple overlapping communities and corporations that made up early modern regnal polities.Footnote 125
Polities and systems of polities: before (and beyond) ‘International relations’
In neither medieval nor early modern Europe did political systems sort into ‘national/internal’ and ‘international/external.’ Multiple nested systems of polities – not one international system composed of multiple national systems – has been the historic norm in the Western world since the fall of the Roman Empire (and in the non-western and pre-modern worlds more generally).
Before ‘international relations’
Consider early modern France.
Between 1560 and 1660, France experienced ‘internal’ warfare in 49 years and ‘external’ war in 47.Footnote 126 And ‘internal’ ‘French’ politics often looked more like relations between polities than within a polity.
During the Wars of Religion (1562–1598)Footnote 127 large swaths of the king's realm repeatedly fell out of his control, including Paris from 1588 to 1593, as leading families – Valois, Bourbon, Guise, Montmorency – pursued dynastic, religious, and regional interests. For example, in the fifth (1575–1576) war, the King's brother, Francis, the heir to the throne, joined the rebels (to advance his own fortunes and that of his line of the family). And the Guises, who for more than a quarter century had been the King's leading supporters, turned on him in 1587. Nobles, for family, religious, or personal reasons, regularly chose to fight against the king (or sit out a particular conflict). And the King too pursued his (dynastic, royal, and religious) interests rather than anything that could plausibly be understood as a national interest.
‘National’ boundaries were tenuous as best. Noble leaders of rebel communities as a matter of course made treaties with foreign rulers. For example, both the 1562 Treaty of Hampton Court between the Prince of Condé and Queen Elizabeth of England and the 1584 Treaty of Joinville between the Guises (on behalf of the Catholic League) and Philip II of Spain brought ‘foreign’ forces to France. Treaties were agreements between ‘princes,’ in the broadest sense of that term, not states (or even kings). Guise, Bourbon, and Valois were equally free to enter into treaties. (That the former usually were less attractive allies was an entirely different matter.)
Conversely, royal edicts ending individual wars were essentially peace treaties among the various ‘French’ parties. And they regularly returned leading rebels to their prior positions. (For example, after the first war the Protestant leader Louis of Bourbon, Prince of Condé, received the office of Lieutenant-General, which had been held by his brother, who had been killed in the fighting.) This looks very much like the treatment of foreign princes after a war: restoration of the status quo ante, with adjustments based on the ex post balance of power.Footnote 128
France, it must be emphasized, was not unusual. For example, during the Catalan Revolt of the 1640s and 1650s – more than a century and a half after ‘the creation of modern Spain’ – J. H. Elliott notes that ‘the rebels found it easier to rally support, because the oppression came from foreign [i.e. Castilian] rulers, foreign officials and foreign troops.’Footnote 129
Neither was this an early and brief ‘transitional’ phenomenon. In the mid-17th century, regime-threatening regional crises wracked all the (dynastically agglomerated) great powers – both where religion interacted with provincialism and privilege (Britain (the Civil War) and the Empire (the Thirty Years' War)) and where it did not (Spain (the Catalan and Portuguese revolts) and France (the Frondes)). And this was a continuation, on a larger geographical scale, of a pattern of provincial resistance to royal rule – contested regnal peripheralization – going back at least to Charlemagne.Footnote 130
Beyond ‘national’ and ‘international’ politics
IR's national–international binary defines political systems by whether they involve relations within or between a privileged type of polity – ‘states’ (terminal polities). This, in addition to its anachronism, makes ‘the international system’ an external, even alien, arena; a mere environment in which polities happen to interact.Footnote 131
My account instead emphasizes that ‘the international system’ is the system (whole) of which ‘states’ are parts – and that the characters of both the whole and its parts are interdependent. It also sees as normal multiple nested systems of polities. (The two-level structure of Figure 3 is but one (not especially common) arrangement of a political space.)
Even Justin Rosenberg's account of ‘the international’ as ‘that dimension of social reality which arises specifically from the coexistence within it of more than one society,’Footnote 132 reflecting the fact that ‘human existence is not unitary but multiple,’Footnote 133 largely reformulates a foundational national–international binary as unitary-multiple. It also suppresses the fact that all complex societies are societies of societies – and that federal States, empire-states, and heterarchy-states are partly defined by multiple centers/polities/societies.
Multiplicity is distinctively ‘international’ only where relatively tightly integrated territorial terminal polities create domains of relative (internal/‘national’) unity and deal with each other as top-level peers. (The more that centers are distributed across levels the more problematic the label international – which was coined in the late-18th century,Footnote 134 just as a system of single-level governance by terminal peer polities was emerging.) Only in a world of States-in-a-states-system is it illuminating to draw a fundamental distinction between national/inside/unity and international/outside/multiplicity. Mainstream IR, however, inappropriately generalizes this very particular kind of structure.
The frame of centers and peripheries, by contrast, shifts attention to the variety of relations between polities in layered (and often nested) political systems. (How, rather than whether, systems of polities are nested becomes the crucial question.) It also (properly) treats as empirical questions the locations of governance and the ways that peoples, places, and political authority are organized.
Early modern vs. modern polities
For reasons of space, I skip over the (largely 19th-century) transition to modern States (as well as transformations of the Eurocentric international system). Here I simply note four fundamental differences between early modern composite polities and ‘modern states’ understood in standard ‘Weberian’ (legal-rational bureaucratic) terms.
• Modern states were defined territorially or nationally, legitimated legally and rationally, and ruled bureaucratically. Early modern polities were defined dynastically, legitimated by dynasticism, tradition, and religion, and ruled patrimonially.
• Modern states were relatively tightly integrated polities with a single system of law and administration. Early modern polities were agglomerations of disparate territorial and social corporations governed by different laws and institutions.
• Modern states had citizens who were individually equal before the law. Early modern polities had subjects who were members of hierarchically organized corporate groups, all but the lowest of which enjoyed particularistic privileges.
• Modern states were multifunctional entities that monopolized not only law and force but identity, social policy, and economic regulation. Early modern polities provided little more than (limited) access to justice, some degree of internal order, and ruinous foreign wars (during which they offered partial protection against external attack).
Globalization as heterarchic re-assembly
Jumping to today, my framework depicts ‘globalization’ as an epochal change from the configuration of modern-states-in-the-modern-states-system to a post-modern configuration that is increasingly heterarchic but otherwise of uncertain shape. I have the space, though, only to offer a superficial spatio-political depiction and compare it with a few familiar alternatives.Footnote 135
Figure 7, as I noted above, was designed with contemporary Europe in mind. To model globalization we would need to add an additional layer on top (and populate the two supra-state levels less densely and much more unevenly).
The inter-state system, which previously was the highest-level political system, now is the middle-level system. Authorities and capabilities formerly held by states are being disaggregated and reallocated – ‘up,’ ‘down,’ and ‘across’ (to non-profit and business organizations). In addition, new functions (e.g. standard setting) are being created, usually at a level above states.
States, it is true, remain the most important centers. (Most governance continues to go through states, even where they are no longer its sole or direct provider.) Nonetheless, there is a clear trend toward multi-actor multi-level functional governance. And to the extent that a clear national–international distinction persists, the character of ‘the international’ has been transformed by the growing importance of the two supranational levels.
Furthermore, the downward reallocation of state authorities has been more ‘imperial’/territorial than heterarchic/functional. Even in countries where ‘devolution’ is relatively advanced, such as Spain and the UK, authority has flowed mostly to regional territorial polities (rather than being fundamentally reorganized along functional lines).Footnote 136
This spatio-political account not only easily and ‘naturally’ pulls together standard elements in depictions of globalization (such as interdependence, complexity, changes in scale, changes in state sovereignty, and the rise of nonstate actors) but shows them to be structurally interrelated. And it encompasses globalization within a comparative framework of broad applicability – rather than draw ad hoc comparisons with selected features of the modern configuration (or claim that nothing fundamental has changed because the system remains anarchic).
The familiar framing most similar to mine is ‘multi-level governance.’Footnote 137 Heterarchy, however, adds multi-actor governance. And if, as I have suggested, these two features are not only interdependent but connected with a move from territorial to functional organization then heterarchy provides a richer account of the referenced phenomena. (It also clearly makes the distinction between single-level and multi-level governance a matter of conceptual principle rather than ad hoc comparisons to a previously predominant pattern.)
The rise of nonstate actors has been an important theme in IR for half a century. Heterarchy, however, does not conceptualize political organization through the lens of (the historically contingent dominance of) states; refuses to reduce most types of political actors to undefined members of a residual class (not states); and (properly) presents the multiplicity of types of actors as noteworthy but normal.
My account also has similarities to James Rosenau's notion of ‘fragmegration,’ understood as ‘processes of integration and fragmentation [that] are unfolding simultaneously and endlessly interacting as the migration of authority [away from states] moves helter-skelter and in contradictory directions.’Footnote 138 Fragmentation (differentiation) and integration, however, are universal features of social and political systems. States systems are differentiated and integrated in one way, empires in other ways, and heterarchies in still other ways. ‘Fragmegration’ inappropriately take the assemblage States-in-a-states-system as normal (or at least an unquestioned reference point) and somehow not fragmegrated.
Similarly, the idea of an embedded state, enmeshed in increasingly complex transnational, international, and supranational practices and institutions,Footnote 139 inappropriately privileges the (unembedded) modern state and is focused on states (rather than systems) and their autonomy (rather than the structure of their relations). My framework shifts attention to the fact that polities of various types are regularly parts of varied kinds of layered systems of polities.Footnote 140
Globalization is also sometimes presented as ‘neo-medieval.’Footnote 141 The world of the 21st century, however, is not becoming ‘more medieval’ in any substantively interesting sense. It is becoming more heterarchic.
A spatio-political account of globalization is neither a complete nor a privileged structural depiction. It does, however, provide distinctive and penetrating insights that help us to see globalization as neither structurally unprecedented nor insignificant but instead involving political reconfigurations comparable in significance and similar in form to those that have occurred regularly in Eurocentric political spaces over the past millennium.
A spatio-political framing also allows us to consider the (I think very real) possibility that rather than moving in any particular direction – for example, toward a world state or back toward an ideal-type states system – varied changes in relations between varied polities on varied levels will proceed in varied ways and at varied paces in different places. As in earlier periods, we can expect to see, on a time scale of several decades, complex combinations of continuity and change in the configurations of political systems.
Multiple models of multidimensional structures
I want to conclude by (re)turning to the significance of structural analysis, looking in this section at the nature of structural models and in the next section at the character of the explanations they provide.
As we have seen, political structures are not simple things that come in only a few forms that differ in relatively minor ways. And spatio-political structure is but one dimension of the arrangement of the parts of political systems – which also, pace Waltz, are stratified and functionally differentiated, have normative and institutional structures, and have material structures that polarity barely begins to address.
Elsewhere I have touched on the substance of some of these additional dimensions.Footnote 142 Here I stress the fact of multidimensionality.
It would be difficult for any structural account to address, let alone integrate, all potentially relevant dimensions. And I doubt that that would be desirable. Depending on the facts of the case and the purpose of the analysis, only some dimensions are likely to be of concern.
Therefore, we should seek insightful models of some of important structural features a system (rather than try to depict ‘the structure’ of a political system, even if we could agree on what would need to be included, or generate a single typology of near universal applicability). By fruitfully depicting selected essential characteristics, such models provide an entry into better understanding the nature and structured operation of actual political systems.
Therefore, in considering early modern political systems I not only applied my models, to identify and focus on core features of interest, but supplemented them, to produce richer empirical accounts. For example, dynastic aggregation (an institutional–ideational feature of the structure of early modern politics) helps to explain the prevalence of provincial/regional rebellions during the long period during which previously more autonomous polities were being peripheralized. If space had allowed, I would have looked at fiscal, institutional, and administrative developments that enabled early modern royal centers to penetrate their peripheries both more extensively and more intensively. This, over a period of centuries, began to prepare the way for what ultimately became States – although that transition also required, among other things, politically subordinating social corporations and reconfiguring political relations between sovereigns and subjects toward relations between citizens and states.
Deciding which relations of which actors to include in a structural model is partly an empirical question and partly a matter of the interests of the observer. This leaves considerable room for substantive and methodological variation, even when addressing similar subject matters.
For example, Daniel Nexon and Thomas WrightFootnote 143 develop models of unipolar systems that loosely overlap with some of my models but identify fundamentally different relational configurations. (Most notably, they define empire by hub and spoke relations between a center and its peripheries.) And in a recent paper in this journal, Meghan McConaughey, Paul Musgrave, and NexonFootnote 144 develop a three-dimensional typology that in many ways overlaps with mine but focuses attention in significantly different places, in large measure as a result of adopting a contracting frame and looking at segments (rather than peripheries).
These projects are linked with mine, it seems to me, by a focus on thoughtfully selected sets of ‘middle-range’ configurational models. (‘Structure’ is used not with the definite article but with adjectives – e.g. spatio-political.) These sets of models identify relations that enable ‘big picture’ comparative explanations/understandings. (The types in a typology mutually illuminate one another, further highlighting distinctive forms of organization through contrast.) And such typologies, by specifying a set of expected alternatives, facilitate exploring possibilities for and patterns of continuity and change – and, by what they leave out, facilitate interrogating those types and expectations.
The nature of structural/relational explanations
Structural models both explain different things (that are otherwise hard to explain) and explain differently – through configurations (not causes).
Waltz's theories of theory
I begin developing this argument by looking at Waltz's account of theoretical explanation. Although this may seem a digression through material considered simply because it is well known, it in fact nicely introduces the character of structural explanations – and some differences from the causal explanations that are privileged in some parts of mainstream social science.
Waltz had two accounts of theoretical explanation, each associated with a different strategy for studying systems and their structures.
The first chapter of Theory of International Politics is titled ‘Laws and Theories.’ There Waltz argued that ‘laws establish relations between variables’Footnote 145 but do ‘not say why a particular association holds.’Footnote 146 ‘Theories show why those associations obtain.’Footnote 147 ‘Theories explain laws’Footnote 148 by identifying causes,Footnote 149 understanding ‘a real causal connection’ to involve establishing ‘the relation between an independent and a dependent variable.’Footnote 150
This is the mainstream ‘neo-positivist’ correlational conception.Footnote 151 Such explanations, though, whatever their virtues, are incompatible with systemic explanations, in which variables are neither independent nor (merely) dependent but interdependent. In systems ‘the impact of one variable … depends on others.’Footnote 152
In practice, despite the language of systems, Waltzian structural theory followed the analytic independent-variable causal model. For example, a central substantive claim of Theory of International Politics is ‘balance-of-power politics prevail wherever two, and only two, requirements are met: that the order be anarchic and that it be populated by units wishing to survive.’Footnote 153 There are no emergent system effects here. Anarchy (alone) is the (analytic, independent-variable) cause of balancing, a law-like regularity.
Such explanations, which bracket or break the interconnections that make systems systems (and significant),Footnote 154 are the norm in contemporary ‘structural’ IR. Consider, for example, the literature on ‘the effects of anarchy.’Footnote 155 Similarly, polarity, Waltz's other dimension of international structure, typically is treated as an independent variable (with bipolarity, e.g. treated as having the same effects, ceteris paribus, in all international systems).
Truly systemic structural/relational models explain in a very different way – that corresponds to Waltz's second theory of theory.
‘A theory is a picture, mentally formed, of a bounded realm or domain of activity … a depiction of the organization of a domain and of the connections among its parts.’Footnote 156 ‘Once the system is understood, once its principle of organization is grasped,Footnote 157 the phenomena are explained. … “Understanding” probably means nothing more than having whatever ideas and concepts are needed to recognize that a great many different phenomena are part of a coherent whole.’”Footnote 158 Explanation thus understood, rather than identify causes, ‘ma[kes] manifest’ ‘the significance of the observed.’Footnote 159
This vision of theory and explanation can be seen when Waltz wrote of ‘relations of cause and interdependency’Footnote 160 and of ‘connections and causes.’Footnote 161 Similarly, theories, Waltz claimed, ask both ‘What causes what?’ and ‘How does it all hang together?’Footnote 162
In practice, however, Waltz only seriously pursued explaining law-like regularities.Footnote 163 Nonetheless, I draw attention to his picture theory because it is genuinely systemic and fits well with my account of spatio-political structure – and thus can ground an appreciation of the nature of structural/systemic explanations and their possible epistemic achievements (and limits).
Systemic/structural/relational explanations
If systems are parts of particular kinds arranged and operating in particular ways to produce wholes with emergent properties then systemic/structural explanations must seek to identify interdependencies and connections that give a system a particular character.
‘Structurally we can describe and understand the pressures states are subject to.’Footnote 164 This formulation, whatever its insight, is inappropriately individualistic (seeing states as externally constrained autonomous actors rather than parts of assembled wholes). Waltz got a bit closer to what I have in mind when he talked about structures ‘shaping and shoving’ statesFootnote 165 – especially if we emphasize shaping. Or, as I would put it, assembled systems position actors in ways that recursively (re)shape their character and interests, (re)set the parameters of the possible (both for them and for the system), and, through these structured relations, regulate (but do not cause or determine) actions and interactions.
Waltz claimed that systemic/structural theories provide ‘a purely positional picture of society.’Footnote 166 There are no social positions, though, in his anarchic void. And, going back to the underlying idea of depiction, there is not much to depict.
In my account, by contrast, polities have very particular positions – and associated rights, responsibilities, and roles. The resulting structure (of positions) creates a particular type of system that regulates political actions and interactions in particular ways. Change involves re-positioning polities. And my models try to depict the organization of a complex whole; to show how ‘a great many different phenomena are part of a coherent whole.’Footnote 167
Structural models neither identify causes nor explain law-like regularities. Rather, they suggest that certain ‘things’ tend to result when (other) ‘things’ are arranged and operate this way – and that these organizing relations help us understand some of what typically happens (and does not happen).Footnote 168 (That is what we should expect if the world is more or less like this.)
In IR, Ole Waever and Patrick Jackson have presented similar accounts of structural, and more broadly ideal-type, explanations, drawing on (or working off of) Waltz's picture theory of theory.Footnote 169
‘Scientific researchers trace and map how particular configurations of ideal-typified factors come together to generate historically specific outcomes in particular cases.’Footnote 170 This, it seems to me, nicely explains what I tried to do in the case study above.Footnote 171
‘Application of theory then takes the form of assessing the fit between the model and things in the world.’Footnote 172 ‘Actual events almost never look like that ideal, but keeping the ideal firmly in mind helps us make sense of what actually did happen.’Footnote 173 My models, I have argued, do fit the world rather well and help us make sense of what has happened. They render certain actions, interactions, and outcomes understandable; explicable (by analogy to the model).
Toward explanatory pluralism in IR
Nothing I have said is meant to criticize causal theories or explanations (properly applied, in their place). Rather, I reject the view that scientific explanations have a singular character or structure. As King, Keohane, and Verba famously put it, although there are multiple ‘styles’ (methods) of social scientific research there is only ‘one logic of inference.’Footnote 174
This is just plain false.
For example, Physics makes important use of laws, which explain by showing that something is an instance of a universal empirical regularity. But Waltz, as we saw, denied that laws explain. And, in both of his senses of explain, they don't. (They neither identify a cause nor specify a form of organization.)
Mechanismic explanations, which are common in Biology, explain still differently – by showing how a result is produced (not which independent variables caused it – and usually without appeals to laws).Footnote 175 Agent-based models explain in a similar but still slightly different way. And even if we are reluctant to call ‘explanations’ the results of purely predictive models (e.g. data-mining algorithms; or rational-actor models in some understandings) they represent still another standard type of scientific inference and knowledge.
These (and other) kinds of theories, models, and explanations look at different things in different ways to help us understand different parts of the world, differently. I have argued that, among the many things worth knowing about political systems, knowledge of their structure – and of their spatio-political structure in particular – often is valuable. And, I would add, for the kinds of knowledge that such structural models provide, no better explanations are available.
If this sounds a lot like where Waltz started then my work here can be read as an attempt to save Waltz from Waltz; to vindicate the systems theoretic (and more broadly relational) project that he claimed was his inspiration by rescuing structural theory from what happened to it in and through the influence of Theory of International Politics – which (for better or, as I have suggested, worseFootnote 176) has for forty years largely defined ‘structural theory’ in IR.
Acknowledgements
I thank the editors and reviewers at International Theory, Lucas de Oliviera Paes, and the participants at seminars at University of British Columbia and American University where portions of this paper were presented.
