A promising but controversial future

Comparative politics is as old as the empirical study of politics itself. Today, even those scholars who only conduct research on a single polity find themselves ineluctably drawn into the sub-discipline. As soon as they move beyond pure description and start using a vocabulary based on generic analogies or more comprehensive systems of classification, they risk exposing themselves to comment and criticism from aggressive comparativists. For example, a student of American politics who concludes that a two-party system has been an indispensable element for this regime’s democratic stability may be challenged by those who have studied such exotic polities as Uruguay or Colombia where analogous institutions have sometimes failed to produce the same result. Indeed, in the latter case, one of the most destabilizing features may have been its oligarchic and sclerotic two-party system. Meanwhile, perhaps unbeknownst to the naïve Americanist, there are many multi-party systems in Western Europe that have been models of political stability and policy innovation.

So, even casual students of political science may not be able to escape the tentacles of comparison, no matter how hard they try. Knowing everything there is to know about some period or aspect of one’s own country’s politics could be misleading without some effort at placing it ‘in comparative perspective’. Even seeking refuge in international relations will no longer suffice. There may be only one world system to be observed (although there are several of them to be compared over time), but within that single case ambiguous ‘trans-national’ polities, such as the European Union (EU), other regional and functional ‘regimes’, and a myriad of non-governmental organizations, have emerged.Footnote 1

There have been periods of relative tranquility when the sub-discipline was dominated by a single paradigm. For example, until the 1950s, scholarship consisted mostly of comparing constitutions and other formal institutions of Europe and North America, interspersed with wise comments about more informal aspects of national character and culture. ‘Behaviorialism’ became the rage for a shorter while, during which time mass sample surveys were applied across several polities in efforts to discover the common social bases of electoral results, to distinguish between ‘bourgeois/materialist’ and ‘post-bourgeois/post-materialist’ value sets, and to search for the ‘civic culture’ that was thought to be a pre-requisite for stable democracy. ‘Aggregate data analysis’ of quantitative indicators of economic development, social structure, regime type, and public policy at the national and sub-national levels emerged at roughly the same time. ‘Structural-functionalism’ responded to the challenge of bringing non-European and American polities into the purview of comparativists, by seeking to identify universal tasks that all political systems had to fulfill, regardless of differences in formal institutions or informal behaviors.

None of these approaches has completely disappeared and all academic departments of political science are likely to have some mixture of them. But none is ‘hegemonic’ at the present moment. As one of its most distinguished practitioners described, present-day comparativists are sitting at different tables, eating from different menus, and not speaking to each other – not even to acknowledge their common inheritance from the same distinguished ancestors (Almond, 1990).

The prospective student interested in comparative politics had only to look at the dominant ‘fads and fashions’ in American political science, trace their respective trajectories and intercepts, and he or she could predict where comparative politics would be going for the next decade or more. Who could doubt that this sub-discipline of political science as practiced in the United States of America showed the rest of the world ‘the face of its future?’Footnote 2 After all, by far the largest number of professionals applying this method to describing and analyzing the widest variety of polities has always been employed in this country.

The central assumption of this essay is that the future of comparative politics should (and, hopefully, will) diverge to some degree from the trends and trajectories followed in recent years by many (if certainly not all) political scientists in the United States. As I have expressed it elsewhere, the sub-discipline is presently ‘at the crossroads’ and the direction that its ontological and epistemological choices take in the near future will determine whether it will continue to be a major source of critical innovation for the discipline as a whole, or dissolve itself into the bland and conformist ‘Americo-centric’ mainstream of that discipline.Footnote 3 In other words, this essay will not be an effort that even pretends to survey objectively and comprehensively what has been produced by comparativists – American or otherwise – in the recent past. It will be what the French call a plaidoyer, a biased plea from a particular advocate on behalf of a client who faces a critical ‘mid-career’ choice that will determine his or her status long into the future.

First, some congratulations are in order

Let me begin, however, with some self-congratulation. Thanks to the assiduous efforts of many methodologically minded colleagues (mostly Americans, it is true), many fewer students applying the comparative method neglect to include in their dissertations an explicit defense of the cases selected – their number and analogous characteristics, an awareness of the potential pitfalls involved in selecting the cases based on the latter, and to the limits to generalizing about the external validity of findings.Footnote 4 Despite many criticisms about the ‘non-cumulative’ nature of the knowledge generated by comparative politics, there have emerged some continuous lines of research in which successive generations have built (critically) upon each other’s work. At the present moment, I would cite the burgeoning field of democratization as one where this has occurred. Even in my other current specialty, regional integration, something like a ‘common tradition’ has developed – despite quite fundamental theory-based differences at the point of departure.

These important gains in methodological self-consciousness have produced (or been produced by) some diminution in the ‘class warfare’ between quantitative and qualitative political scientists. There is still some sniping and some of the former persist in asserting their intrinsic ‘scientific’ superiority over the latter, but there is more and more agreement that many of the problems of design and inference are common to both and that the choice between the two should depend more on what it is the one wishes to explain or interpret. Indeed, from my recent experience in two highly cosmopolitan institutions, the European University Institute in Florence and the Central European University in Budapest, I have encountered an increasing number of dissertations in comparative politics that make calculated and intelligent use of both methods – frequently with an initial large N comparison wielding relatively simple quantitative indicators to establish the broad parameters of association, followed by a small N analysis of carefully selected cases with sets of qualitative variables to search for specific sequences and complex interactions to demonstrate causality (as well as the impact of neglected or ‘accidental’ factors). To use the imaginative vocabulary of Charles Tilly, such research combines the advantages of ‘lumping’ and ‘splitting’ (Tilly, 1984). Hopefully, this trend will continue into the future.

The real challenge currently facing comparative politics, however, comes from a third alternative, namely, ‘formal modeling’ based on rational choice assumptions. Much of this stems from a strong desire by American political scientists to imitate what they consider to be the ‘success’ of the economics profession in acquiring greater status within academy by driving out of its ranks a wide range of dissident approaches and establishing a foundation of theoretical (neo-liberalism) and methodological (mathematical modeling) orthodoxy. This path toward the future would diverge both methodologically and substantively from the previously competing quantitative and qualitative ones. It would involve acceptance of a much stronger set of limiting initial assumptions, exclusive reliance on the rational calculations of individual actors to provide ‘micro-foundations’, deductive presumptions about the nature of their interactions, and reliance on either ‘stylized facts’ or ‘mathematical proofs’ to demonstrate the correctness of initial assumptions and hypotheses derived from them. The comparative dimension enters into these equations to prove that individual behavior is invariant across units or, where it is not, that institutions (previously chosen rationally) can make a difference.

The ‘Genealogical Tree’ of comparative politics

As a prospective or practicing comparativist, the reader will find him or herself hanging or, better, sitting somewhere in the tree depicted in Figure 1. It is a spatially schematized and temporally compressed representation of the genealogical roots, trunks, and branches that have evolved into contemporary comparative politics. Some intrepid young scholars may be agile enough to scramble horizontally from one branch to another in the canopy; most, however, will have arrived and will remain on their roost by climbing vertically up one or another of the multiple trunks rooted in past traditions of political thought.

Figure 1

The Family Tree of Comparative Politics

Its deepest root lies in something I have called ‘sociological constitutionalism’ as invented by Aristotle and subsequently nourished by such a diverse group of ‘Dead White European Males’ as Polybius, Machiavelli, Montesquieu, Benjamin Constant, Alexis de Tocqueville, Lorenz von Stein, Karl Marx, Moisei Ostrogorski, Max Weber, Emile Durkheim, Roberto Michels, Gaetano Mosca, Vilfredo Pareto, and Herbert Tingsten. Through various extensions and permutations, this has become the branch subsequently labeled as ‘historical political sociology’ with such luminaries as Stein Rokkan, T.H. Marshall, Reinhard Bendix, Otto Kirchheimer, Seymour Martin Lipset, Juan Linz, Hans Daalder, Mattei Dogan, S.N. Eisenstadt, Harry Eckstein, and Dankwart Rustow located somewhere along it during the decades immediately following the Second World War. Karl Deutsch probably should be placed here at an odd angle, since he was so single-handedly responsible for inserting a cybernetic graft into it. On the outer reaches of this cluster, one generation later, is where I can be most safely located.

The other deep root lies in ‘legal constitutionalism’ fertilized initially by distinguished Anglo-French jurists such as Léon Duguit, Georges Burdeau, James Bryce, A. Lawrence Lowell, and Woodrow Wilson, and developed during the subsequent century by scholars such as Maurice Duverger, Herman Finer, Samuel Finer, Giovanni Sartori, Carl J. Friedrich, Samuel Beer, Jean Blondel, F.A. Hermens, and Klaus von Beyme. Someone like Robert Dahl can probably be best located hanging comfortably in a hammock strung between the sociological and legal branches – which, in any case, have been converging. Samuel Huntington is another distinguished comparativist whose roost in the tree is difficult to place, although it is easier to imagine him clinging closer to this branch than to the neighboring one.

From these two taproots have been added a number of exogenous grafts during the 20th century. Political science became a voracious consumer of conceptual and methodological innovations from other, increasingly professionalized, social science disciplines – first, from social psychology with the so-called ‘behaviorist movement’ and later (and somewhat more surreptitiously) from anthropology with the ‘structural-functionalist approach’. The most distinctive product of the former was the rapid rise of comparative survey research, symbolized by the publication of the highly successful (and criticized) work, Gabriel Almond’s and Sidney Verba’s, The Civic Culture, in 1963. Today, this branch of comparative politics is routinely conducted within and often across virtually all of the world’s polities. Certainly, it is the most distinctive (and successful) contribution of American political science to the sub-discipline.

The anthropological graft has contributed much less in volume and attractiveness to the evolution of comparative politics. Its most important contribution was undoubtedly to preside over a vast extension in the range of countries brought under comparative scrutiny. When embracing ‘Non-Western’ politics and faced with the need to explain ‘elections in Albania’, ‘budgeting in Zaire’, ‘civil-military relations in Indonesia’, and ‘federalism in Argentina’, scholars such as David Apter, Leonard Binder, Lucian Pye, James Coleman, and Myron Weiner found it difficult to apply the usual legal or sociological categories and took refuge behind a variety of ‘functions’ that presumably had to be performed by analogous ‘structures’ in all political systems. After a major flurry of activities in mid-1950s to the early 1970s under the prestigious auspices of the SSRC Committee on Comparative Politics, scholars began to realize that the stipulated functions were excessively abstract and that the structures they were trying to explain often could not be assigned to a single one of them. Moreover, the entire notion of ‘systemic equilibrium’ as the central metaphor for guiding comparisons among Non-Western polities came into question when the stability of their institutions was revealed to be highly precarious. Once the key question was seen to be change, especially change in regime from democracy to autocracy or, more recently, the inverse, the approach became much less relevant (Almond et al., 1973).

Finally, comparativists have always borrowed ideas and concepts from economics, especially from such early political economists as Adam Smith, Karl Marx, John Stuart Mill, David Ricardo, Jeremy Bentham, Friedrich List, and Adolf Wagner. Albert Hirschman, although a heterodoxical figure outside the ranks of contemporary neo-liberal economists, has made several seminal contributions. But the real novelty of the past few decades has been the transfer of root assumptions, deductive thinking, and mathematical modeling techniques into the study of politics – first, in research on American politics and, increasingly, in research on ‘other people’s politics.’ The leading figures have been Anthony Downs, Thomas Schelling, Howard Raifa, Kenneth Arrow, Douglas North, Mancur Olson, Gary Becker, George Stigler, and, most centrally, James Buchanan and William Riker. As we shall see shortly, this graft from economics has opened up a radically new path to the future for comparativists.

Presently, the evolutionary tree of comparative politics resembles more a Tropical Banyan than a Florentine Cypress. It has a wide canopy of branches, certainly not a single tapered and elegant peak. Its most curious aspect, however, is the number of practitioners who roost in the canopy, and who seem content with sharing the same generic label: institutionalists. Closer inspection of the foliage reveals that it contains an extraordinary variety of flora and fauna. About all they can agree upon is that ‘institutions matter’. They differ widely on what institutions are, how they come about, why is it that they matter, and which ones matter more than others. Moreover, some of those perched up there will even admit that other things also matter: collective identities, citizen attitudes, cultural values, popular memories, external pressures, economic dependencies, even instinctive habits and informal practices when it comes to explaining and, especially, to understanding political outcomes. This urge to find shelter under the capacious tent of ‘institutionalism’ can be interpreted either as a bizarre effort to return to their legalistic origins (precisely in a world context in which such formalized constraints are manifestly inadequate for solving problems and resolving conflicts) or as a desperate attempt to make common cause with the greatest possible number of disciplinary brethren (precisely when so many of them are heading in a direction that would radically challenge their basic assumptions and methods).

At the crossroads of three paths

At the top of Figure 1, I have placed a large question mark – a decision point that will determine the future configuration and even the very viability of the whole tree. The safest thing one can say today about the future of comparative politics is that it will not be the same as in the past.

Of course, not everything is going to have to change. Comparative politics will continue for the foreseeable future to bear major responsibility for the objective description of processes and events in ‘other people’s countries’ and, hence, for providing systematic and reliable information to those politicians and to those administrators charged with making and implementing national policies concerning these countries. The end of the Cold War and collapse of the Soviet Empire have, if nothing else, led to an impressive increase in the sheer number of polities whose (allegedly autonomous) behavior has to be described. The globalization of capitalism has produced increasingly indirect and articulated systems of production, transport, and distribution that are much more sensitive to disturbances in the behavior of their most remote and marginal components. The ubiquitous penetration of the mass media has meant that happenings anywhere in the world are immediately transmitted everywhere and comparativist pundits will be expected ‘to place them in context’ for public consumption.

Comparison between ‘real-existing polities’ will also remain the best available research method for analyzing similarities and differences in behavior and for inferring the existence of patterns of regularity with regard to the causes and consequences of politics. It will always be the second best instrument for this purpose, but as long as it remains impossible for students of politics to experiment with most of their subjects and subject matter, political scientists will have to settle for analyzing as systematically as possible variations they cannot control directly.

Figure 1 suggests that comparative politics will have to choose among three distinctive paths. It can continue along the very broad ‘institutionalist’ trajectory it has been on for the last decades, presumably adding more ‘neo-neo-neo-’ prefixes as it permutes into more specialized approaches. Otherwise, it can take a turn to either the left or the right. Whatever the choice, it is most unlikely that comparative politics will taper toward a single peak – however much some practitioners would like it to.Footnote 5 The most ‘clear and present danger’, as I see it, is that the sub-discipline’s evolution will lead to an irreversible split in the canopy with less-and-less communication or cross-fertilization between scholars perched on its different branches and more-and-more efforts to exclude dissidents from claiming the professional right to call themselves ‘scientific’ students of politics.

Those who take a sharp turn to the right towards economics will be opting for ‘simplification’. They will be led by those American colleagues who have already accepted the limited initial assumptions, exclusive reliance on individualistic ‘micro-foundations’, deductive presumptions about how these actors behave with regard to each other, and proof by ‘stylized facts’ or ‘mathematical formulae’ that characterize the path known as rational or public choice.Footnote 6

Those who choose the leftward path will opt for what I call, for lack of a better term, ‘complexification’. They will follow the lead of a less well-defined and less self-confident group of scholars who:

1. (1) Accept far fewer and less restrictive initial assumptions – indeed, who rely upon a calculated proliferation of assumptions about the identity and motives of actors and about the role of entrenched institutions and historical memories in determining seemingly ‘irrational’ behaviors.

2. (2) Are convinced that adequate micro-foundations in the present world context can not only be based on individual persons – indeed, they must also include collectivities that cannot be simply decomposed into the preferences or actions of individuals and to take more-and-more into consideration the composition effects generated by multiple levels of political power and authority.

3. (3) Choose to rely upon ‘reasonableness’ rather than rationality, i.e. on ‘improvising’ and ‘avoiding the worst’ in complex situations where optimal pursuit of marginal returns is virtually impossible given the number of actors involved, the plurality of sources of information and the unintended consequences generated by interdependent layers of political aggregation.

4. (4) Consider that the usual fallacies of composition can be converted into novel ‘laws of composition’ to explain outcomes in situations where multiple layers of different types of actors from a plurality of centers of power and authority bargain and deliberate with each other.

5. (5) Have a healthy respect for ‘real’ data – whether generated by the normal operations of the polity or invented and gathered by themselves, coupled with an abiding suspicion of simple aggregative indicators for complex phenomena, so-called ‘stylized’ facts that suppress confounding observations or simulations produced by impressive mathematical equations.

6. (6) Insist upon endogenizing as many potentially causal variables as possible, even those notoriously difficult to measure such as ‘preferences’ – rather than shoving them into the background, assuming them out of existence, presuming what values they take in a given situation or inserting new ones ex post in order to ‘prove’ the alleged rationality of observed outcomes.

The competition between the three alternative paths depicted at the top of Figure 1 is hardly going to be equal. The middle one, toward various permutations of ‘the new institutionalism’, should be the most favored choice, if only due to sheer inertia rooted in the fixed intellectual assets of most practicing comparativists. Given the profusion of qualifiers that usually precede it – historical, sociological, legal, and rational, just to name the most common – this approach is sufficiently ambiguous to appeal to a large number of them, even if in my opinion it is already subject to diminishing marginal returns, divisive specifications, and less and less capacity to deal with anomalies.

The sharp right turn toward the simplicity of formal modeling should be (and already has been) very tempting, especially in the United States, for reasons I have discussed elsewhere. Comparativists may be especially seduced by its appeal since it provides a convenient justification for eliminating what has, heretofore, been some of the most demanding requirements of the sub-discipline, namely, the need to learn a ‘foreign’ language, culture, and history and to carry out protracted field research in a ‘foreign’ setting. Dedicated rational choicers already know what the dominant preferences are supposed to be,Footnote 7 and have no need to observe directly or interview ‘exotic’ respondents.Footnote 8 Information requirements have been radically simplified and, if they are not available in an on-line data banks, they can always be smoothed out by asserting ‘stylized facts’ or just by simulating their probable distribution.

Most saliently, those comparativists who take this path may be convinced that they are likely to reap the same rewards from higher ‘scientific’ status as have the neo-liberal, mathematized economists from whom they have lifted their intellectual baggage – ‘lock, stock and barrel’.Footnote 9 They can also be assured that their work will only be understood by a small group of conoscenti.Footnote 10 Even if ‘economic science’ has been notoriously unsuccessful in predicting the rate, direction or locus of change in the macro-economy and even if it has suffered recently some notorious defections from within its ranks, it still remains very prestigious in the eyes of other social scientists. Public or rational choice is by now firmly entrenched in a dominant strategic position within several leading departments and journals in the United States – although it has recently provoked a strong reaction from non-believers within the profession gathered under the banner of ‘Perestroika’ (a substantial proportion of whom seem to have been comparativists).

The leftward turn toward complexity does not have any such disciplinary prototype to follow. At best, it can only pick up assumptions, concepts, and ideas from a scattered diversity of sources within political science. Eventually, it may receive grafts from abstruse disciplines in the physical and mathematical sciences that deal with cybernetics, advanced circuits, complex systems, chaos theory, and so forth. By far the most important political scientist who has attempted to address these issues is Robert Jervis. His System Effects: Complexity in Political and Social Life is a foundational statement that deserves to be more widely read and absorbed.Footnote 11 Such notions as ‘chains of consequence’, ‘emergent properties’, ‘indirect and delayed effects’, ‘quasi-homeostasis’, ‘domino dynamics’, ‘spiral model’, ‘trans-national advocacy networks’, ‘cheap talk’, ‘embedded liberalism’, ‘complex interdependence’, ‘multi-layered governance’, and so forth, point the ‘complexifier’ in the right direction and are applicable to all levels of political aggregation – and to their intercepts. Nota bene that these concepts differ to a significant extent from those commonly used by institutionalists, even from historical institutionalists whose approach comes closest. Capturing the nature of complexity requires not only a more dynamic conceptualization of political relations, but also one that captures emergent, informal, and even ephemeral ones. They do, however, not by any means add up to a comprehensive, coherent, or consistent theory.Footnote 12 Indeed, virtually none of these sources refers to the other and even less do they offer a cumulative perspective that can subsume previous knowledge, explain new puzzles, provide counter-intuitive answers, and stand up to repeated attempts at falsification by both its practitioners and its competitors.

Sending present and future students of ‘other people’s politics’ down such an un-explored (and un-fashionable) path may well seem like the height of folly. Quite understandably, apprentice political scientists need to identify with a research program that is already flourishing and promising – and, not coincidentally, likely to provide them eventually with better career opportunities. Comparativists may be even more susceptible to such a bandwagon effect since they are unusually dependent upon theory to identify the basic analogies between cases and to sustain the external validity of their findings. While I would argue that embracing complexity would be more ‘progressive’ in the Lakatosian sense than the other two (both of which I regard as already ‘degenerative’), I would be the first to admit how difficult it is going to be to convince younger scholars at an especially vulnerable moment in their career that this is the case.

By now, it will come as no surprise to the reader that I am strongly in favor of ‘tilting’ the future evolution of comparative politics toward embracing rather than rejecting complexity – even if I am manifestly incompetent to lead the way myself.Footnote 13 Perhaps, it is psychologically understandable that just when the surrounding political world has rapidly become more interdependent across units and more assertive within them that scholars of politics would seek refuge in simplicity, parsimony, and consistency. To many of those contemplating the sheer ‘messiness’ and ‘noisiness’ of contemporary politics at the national, sub-national, and supra-national levels, it must be profoundly comforting to imagine that, under all of this, there lies some easily specified and widely applicable ‘model’ for explaining what is going on. How reassuring it must seem to assume that ‘clear models about actors and preferences, strategical interaction (i.e. ‘game theory PCS), endogenization of variables one-at-a-time’Footnote 14 constitute an adequate response. The fact that this model may not be very good at pre-dicting future outcomes becomes less important than its utility in retro-dicting past ones. If, however, it is still true that ‘a theory can be judged by the range and apparent verisimilitude of the predictions it makes about the world’ – and not just by its formal elegance or logical consistency – then, I can see no viable alternative for us comparativists than to confront the messy and noisy world in which we live and design our theories accordingly.Footnote 15 And the place to start is by changing the basic concepts and classification systems that one needs to control for similarities and to identify differences. And the best tool for producing these building blocks is the ‘ideal type’. It combines a multitude of discrete variables into a recurrent ‘qualitative’ pattern of interrelations that invites attention to differences in type, not differences in magnitude.Footnote 16

Coping with a messy and noisy world

The core of my argument has been that comparative political analysis, if it is to remain significant, productive, and innovative in the future, has to reflect the ‘real-existing’ environment from which it should draw its observations and to which it should refer its findings. Take, for example, the admonition made above by a comparativist advocate of rational choice, Charles Boix. His assumption, I repeat, is that ‘clear models about actors and preferences, strategical interaction (i.e. ‘game theory PCS), endogenization of variables one-at-a-time’ constitute a threesome that is capable of generating non-trivial findings about politics in the contemporary environment. But what if what is needed are ‘fuzzy and under-specified models about a plurality of types of actors with preferences that are contingent upon differences in political setting’, ‘strategic interaction between a large number of players at different levels of aggregation with inconsistent payoffs, constant communication and multiple interdependencies’, and ‘endogenization not of single discrete variables, but of patterns of multiple variables within the same time frame’? Would not such a transposition from the simplified world of conceptual clarity, stylized two-person games, and ‘stepwise’ causality risk producing findings that bear no relation to the complexity of the ‘real-existing’ world of politics? My contention is that if their concepts, assumptions, and hypotheses fail to capture at least some of the core characteristics of their subject matter, comparativists will at best report only trivial or irrelevant findings. They will address problems and provide answers to issues that are primarily internal to their own scholastic paradigm. These are not likely to be the problems that citizens and rulers have to cope with or the answers they expect comparative political research to provide.

One thing that differentiates comparativists from their colleagues who study one polity or one international system is supposed to be greater sensitivity to contextual factors that are so deeply embedded that they are often taken for granted or treated as ‘exceptional’ or ‘unique’. Inversely, they should be especially well equipped to identify and incorporate the trends that affect virtually all of the world’s polities. Two of these trends, in my opinion, are sufficiently pervasive as to affect the basic design and conduct of comparative research. They are: (1) increased complexity and (2) increased interdependence. However independent these may be – for example, logically speaking, a polity may become more complex without increasing its interdependence upon other polities and a polity may enter into increasingly interdependent relations with others while reducing its internal complexity through specialization – these two trends tend to be related and, together, they produce something that Joseph Nye and Robert Keohane have called ‘complex interdependence’.

One major implication that I draw from this is that complex interdependence has an increasing influence not only on the substance of politics but also upon its form. It is changing, in other words, the units that we should be using for specifying our theories and collecting our data and the levels at which we should be analyzing these data.

Complexity: This undermines one of the key assumptions of most of traditional comparative political research, namely, that the variable selected and observed with equivalent measures will tend to produce the same or similar effect(s) across the units being compared.

Interdependence: This undermines the most important epistemological assumption in virtually all comparative research, namely, that the units selected for comparison are sufficiently independent of each other with regard to the cause–effect relationship being examined.Footnote 17

Complex Interdependence: The ‘compound’ condition makes it difficult, if not impossible, to determine what constitutes an independent cause (and, hence, an independent effect) and whether the units involved have an independent political capacity to choose and implement (and, therefore, to act as agents connecting cause and effect).

When Aristotle gathered data on the ‘social constitutions’ of 158 Greek city-states, he set an important and enduring precedent. The apposite units for comparison should be of the same generic type of polity and at the same level of aggregation. And they should be more-or-less self-sufficient and possess a distinctive identity. Since then, almost all theorizing and empirical analyses have followed this model. Most of all, the vast proportion of effort has gone into studying supposedly ‘sovereign’ states whose populations shared a supposedly unique ‘nationality.’ It was taken for granted that only these ‘sovereign-national’ polities possessed the requisite capacity for ‘agency’ and, therefore, could be treated as equivalent for comparative purposes. Needless to say, large N comparisons of all United Nations member-states rested on this fiction. Even area specialists working with geographically or culturally denominated subsets of countries in Latin America, Sub-Saharan Africa, the Middle East and North Africa, and South, South-East or North-East Asia occasionally have had to face this issue of inference and external validity. Was Honduras in the 1950s when a substantial portion of its territory was owned by foreign banana producers ‘really’ comparable with the much larger Brazil whose (then) major export resource, coffee, was in native hands? What is the utility of comparing the fiscal system of Kuwait that rests virtually exclusively on petroleum-derived revenues with that of Jordan that depends largely on foreign aid and its own citizens? Comparativists who pioneered the application of their method to sub-units of the same federal or decentralized polity had to limit themselves to circumscribed issues for which these sub-units possessed some equivalent autonomy of action. For example, it hardly made much sense to study differences in the application of military conscription across US States – with the possible exception of the ‘Free Republic of Berkeley’ during the 1960s and 1970s.

In the contemporary setting, due to differing forms of complexity and degrees of interdependence, as well as the compound product of these two, it has become less and less possible to rely on the properties of sovereignty and nationality to identify equivalent units. No polity can realistically connect cause and effect and produced intended results without regard to the actions of others. Virtually all polities have persons and organizations within their borders that have identities, loyalties, and interests that overlap with persons and organizations in other polities.

Nor can one be assured that polities at the same formal level of political status or aggregation will have the same capacity for agency. Depending on their insertion into multi-layered systems of production, distribution, and governance, their capacity to act or react independently to any specific opportunity or challenge can vary enormously. This is obviously true for those national states that have entered into supra-national arrangements such as the EU or signed binding international treaties such as those of the IMF or the WTO. Not only do they occasionally find themselves publicly shamed or found guilty by such organizations, but they also may regularly anticipate such constraints and alter their behavior accordingly. Moreover, many contemporary national polities have granted or been forced to concede extensive powers to sub-national units and, in some cases, these units have entered into cooperative arrangements with equivalent units in adjacent national states.

From these observations, I conclude not only that it is literally absurd to compare only at the level of individuals but also that comparativists need to dedicate much more thought to the collectivities they do choose and the properties these units supposedly share with regard to the specific institution, policy, or norm being examined. Try to imagine someone studying the commitment to environmental policies across European polities without reference to the EU. Or another scholar comparing the human rights record of African states without taking into consideration the conditionalities posed by bilateral and multi-lateral foreign aid programs. I would admit that in neither of these examples should one presume that all variations in behavior or outcome can be explained by supra-national linkages. There still remains a great deal of difference that can only be explained by conditions within national polities, but exorcising or ignoring the complex external context in which these units are embedded would be equally foolish.

But what is the method one should apply when comparing units in such complex settings? The traditional answer is ‘to tell a story’. After all, what does a political historian – comparative or not – do but construct a narrative that attempts to pull together all the factors within a specified time period that contributed to producing a specific outcome. Unfortunately, such narratives – however insightful – are usually written in ‘ideographic’ terms, i.e. those used by the actors or the authors themselves. Systematic and cumulative comparison across units (or even within the same unit over time) requires a ‘nomothetic’ language, i.e. one that is based on terms that are specific to a particular approach or theory, not to a case. A first step for prospective ‘complexifiers’ would be to invent or re-invent concepts so that they were more capable of grasping ‘fuzzy’, ‘contaminated’, and ‘layered’ interrelationships among individuals and, especially, organizations (since the latter are much more salient components of contemporary political life).

For example, I experienced such a need when I began to think more seriously about the range of likely outcomes that the EU might be heading for.Footnote 18 To fill this prospective space, I had to resort to pseudo-Latin and to define four ‘ideal type’ configurations that I called: federatio, confederatio, consortio, and condomino. I also delved into the language that European officials and politicians were inventing in order to make sense of what they were doing and discovered such odd things as subsidiarity, co-decision, proportionality, additivity, complementarity, transparence, géométrie variable, co-responsibility, juste retour, transposition, mutual recognition, pooled sovereignty, home country control, economic, and social cohesion, sustainable convergence, euro-compatibility, opting-in and opting-out, comitologie, concentric circles, le spill-over, l’engrenage, la méthode communautaire, la supranationalité, and, of course, l’acquis communautaire. And, believe it or not, this is just the tip of the Euro-speak iceberg!

Polities not (just) states

The practice of comparative political research does follow and should imitate changes in ‘real-existing politics’ – but always with a considerable delay.Footnote 19 As I mentioned above, the most important generic changes that have occurred in recent decades involve the spread of ‘complex interdependence’. There is absolutely nothing new about the fact that formally independent polities have extensive relations with each other. What is novel is not only the sheer magnitude and diversity of these exchanges, but also the extent to which they penetrate virtually all social, economic, and cultural groups and almost all geographic areas within these polities. Now, it takes an extraordinary political effort to prevent the population anywhere within national borders from becoming ‘contaminated’ by the flow of foreign ideas and enticements. ‘Globalization’ has become the catch-all term for these developments, even if it tends to exaggerate the evenness of their spread and scope across the planet.

Globalization has certainly become the independent variable – the ‘first mover’ – of contemporary political science. It can be defined as an array of transformations at the macro-level that tend to cluster together, reinforce each other, and produce an ever-accelerating cumulative impact. All these changes have something to do with encouraging the number and variety of exchanges between individuals and social groups across national borders by compressing their interactions in time and space, lowering their costs, and more easily overcoming previous barriers. By most accounts, the driving forces behind globalization have been economic. However, behind the formidable power of increased market competition and technological innovation in goods and services lies a myriad of decisions by national political authorities to tolerate, encourage, and, sometimes, subsidize these exchanges. The day-to-day manifestations of globalization appear so natural and inevitable that we often forget they are the product of deliberate decisions by governments that presumably understood the consequences of what they have decided to laisser passer and laisser faire.

Its impact upon specific national institutions and practices is highly contentious, but two (admittedly hypothetical) trends would seem to have special relevance for the conduct of comparative political inquiry:

1. (1) Globalization narrows the potential range of policy responses, undermines the capacity of (no-longer) sovereign national states to respond autonomously to the demands of their citizenry, and, thereby, weakens the legitimacy of traditional political intermediaries and state authorities.

2. (2) Globalization widens the resources available to non-state actors acting across national borders and shifts policy responsibility upward to trans-national quasi-state actors – both of which undermine formal institutions and informal arrangements at the national level, and promote the development of trans-national interests and the diffusion of trans-national norms.

If either of these is true (and especially if both are), then, a major ‘paradigm shift’ is going to have to occur – whichever of the paths from the canopy in Figure 1 you choose.

Comparativists have occasionally given some thought to the implications of these developments for their units of observation and analysis, but have usually rejected the need to change their most deeply entrenched strategy, namely, to rely almost exclusively upon the ‘sovereign national state’ as the basis for controlling variation and inferring similarities and differences in response to the remaining variation in (allegedly) independent conditions. They (correctly) observe that most individuals still identify primarily (and some exclusively) with this unit and that national variables, when entered into statistical regressions or cross-tabulations, continue to predict a significant amount of variation in attitudes and behavior. Hence, if one is researching, say, the relation between gender and voting preferences, the relationships will differ from national state to national state – and this will usually be greater than the variation between sub-units within respective national states.

While I would concede this assumption for comparative analyses when based exclusively on behavior and attitudes at the individual level, I am convinced that the same does not hold for the behavior of meso- and, especially, macro-units of binding collective choice. Due to differing forms of complexity, differing degrees of interdependence, and differing compounds of these conditions, no polity can realistically connect cause and effect through its own institutions and policies without regard for the actions of others. Virtually all have persons and organizations within their borders that have identities and loyalties that overlap those of other polities; virtually all decisional units within national states are affected by ‘extra-national’ events over which they have limited control. The days when such exchanges only passed through Foreign Offices and were governed by international treaties or formal bilateral arrangements are over. Regions, provinces, and even municipalities engage in external relations; trade unions and professional associations become part of overarching regional and international peak organizations; economic sectors and industries include firms from many different countries; social movements regularly exchange their programs and adopt each others’ tactics. Admittedly, political parties remain among the most national in their organization and ideology, but they do meet with each other often, join ‘internationals’, coordinate their appeals, and sometimes even support each others’ candidates.

My conclusion is that it has become less and less appropriate to rely on the properties of sovereignty, nationality, and stateness for identifying the relevant units for theory, observation, and inference. No doubt, comparative politics at the descriptive level will continue to dedicate most of its effort to formally sovereign national states. That is the level at which such information is normally consumed by policy-makers, the media and the public at large. But at the analytical level, it will have to break through that boundary and recognize that units with the same formal status, e.g. all members of the United Nations or of some regional organization, may have radically different capabilities for taking and implementing collective decisions. In other words, comparativists must give more thought to what constitutes a relevant and equivalent case once they have chosen a problem or puzzle to analyze and to do so before they select the number and identity of the units they will compare.

One innovation in research design would be to compare units at different levels of spatial or legal aggregation, provided they had similar properties and capacities with regard to the problem being studied. Another would be to ignore the boundaries imposed by area studies and try to identify units of analysis that share similar patterns of complex interdependence, regardless of their cultural or geographic propinquities. Yet another would be to shift to functional criteria and compare economic sectors with similarly layered production and marketing arrangements or ethnic groups with similarly proportioned and conflictual relations with titular majorities. The most audacious would be to search for similarly configured, territorial or functional, patterns of power and authority and the effects they produce – wherever they were located spatially, culturally, or even temporally.Footnote 20^, Footnote 21

The most difficult challenge, however, will come from abandoning the presumption of ‘stateness’. Sovereignty has long been an abstract concept that ‘everyone knew’ was only a convenient fiction, just as they also ‘knew’ that almost all states had social groups within them that did not share a common political identity. One could pretend that the units were independent of each other in choosing their organizations and policies and one could get away with assuming that something called ‘the national interest’ existed and, when invoked, did have an impact upon collective choices. But the notion of stateness impregnates the furthest corners of the vocabulary we use to discuss politics – especially stable, iterative, ‘normal’ politics. Whenever we refer to the number, location, authority, status, membership, capacity, identity, type, or significance of political units, we employ concepts that implicitly or explicitly refer to a universe composed of states and ‘their’ surrounding national societies. It seems self-evident that this particular form of organizing political life will continue to dominate all others, authoritatively allocate most resources, enjoy a unique source of legitimacy, and furnish most people with a distinctive identity. However, we may recognize that the sovereign national state is under assault from a variety of directions – beneath and beyond its borders – its ‘considerable resilience’ has been repeatedly asserted.Footnote 22 To expunge it (or even to qualify it significantly) would mean, literally, starting all over and creating a whole new language for talking about and analyzing politics.

The assiduous reader will have noted that I have already tried to do this by frequently referring to ‘polity’ when the normal term should have been ‘state’. I confess that I first became aware of this lexical problem when working on what is, admittedly, an extreme case – the EU. I then asked my reader to try to imagine a polity that did not have the following: (1) a single locus of clearly defined supreme authority; (2) an established and relatively centralized hierarchy of public offices; (3) a pre-defined and distinctive ‘public’ sphere of compétences within which it can make decisions binding on all; (4) a fixed and (more-or-less) contiguous territory over which it exercises authority; (5) a unique recognition by other polities and exclusive capacity to conclude international treaties; (6) an overarching identity and symbolic presence for its subjects/citizens; (7) an established and effective monopoly over the legitimate means of coercion; (8) a unique capacity for the direct implementation of its decisions upon intended individuals and groups; and (9) an uncontested potential ability to control the movement of all goods, services, capital, and persons within its borders (Schmitter, 1996).

Now, the EU has yet to acquire these properties, which makes the EU quite different from well-established national states that have been losing some of them. I will also admit that there is a lot of variation in stateness among such units and that the actual and aspiring members of the EU have moved much further in this direction than others. Europe is unique as a region in which supra-national policy cooperation has been generating norms – some 80,000 pages of them it is alleged – that condition the choices and organizations of its member polities, literally on a day-to-day basis. Nothing remotely like this has come from NAFTA, MERCOSUR, CACOM, or ASEAN. Nevertheless, the polities of North America, South America, Central America, and South-East Asia are all ensnared in a growing network of supra-national norms and even adjudication mechanisms that call into question many of the nine ‘imaginary’ dimensions set out above. ‘Conditionality’ may be a vague term and its efficacy is often doubtful, but no one can question that organizations such as the United Nations, the World Bank, the International Monetary Fund, the World Trade Organization, and a myriad of other global and regional ‘inter-governmental organizations’ are having some effect ‘beyond the nation state’. And this is not to mention the much more numerous non-governmental organizations with direct links to political groups ‘beneath the nation state’.

Before comparative politics can embrace complexity, it will have to admit to a much wider variety of types of decision-making units and question whether those with the same formal status are necessarily equivalent and, hence, capable of behaving in a similar fashion. As we shall now see, this is very likely to involve paying much more attention to ‘patterns’ than to ‘variables’.

Patterns not (just) variables

Contemporary comparative politics has tended to focus on variables. The antiquated version tried to explain the behavior of whole cases – often one of them at a time. The usual approach has been to choose a problem, select some variable(s) from an apposite theory to explain it, decide upon a universe of relevant cases, fasten upon some subset of them to control for other potentially relevant variables, and go searching for ‘significant’ associations. Not only were the units chosen presumed to reproduce the underlying causal relations independent of each other but also each variable was supposed to make an independent and equivalent contribution to explaining the outcome. We have already called into question the first assumption, now let us do the same with the second.

Complexity requires that one attempt to understand the effect(s) of a set of variables (a ‘context’ or ‘ideal-type’) rather than those of a single variable. And, normally, the problem or puzzle one is working on has a multi-dimensional configuration as well. In neither case is it sufficient simply to standard score and add up several variables (as one does, for example, with such variables as economic or human development, working class militancy, ethnic hostility, quality of democracy, rule of law, etc.). The idea is to capture the prior interactions and dependencies that form such a context and produce such an outcome. In other words, the strength of any one independent variable depends on its relation with others, just as the importance of any chosen dependent variable depends on how and where it fits within the system as a whole.

There is another way of expressing this point. In the classical ‘analytical’ tradition, you begin by decomposing a complicated problem, institution, or process and examining its component parts individually. Once you have accomplished this satisfactorily, you then synthesize by putting them back together and announce your findings about the behavior of the whole. But what if the parts once decomposed change their function or identity and, even more seriously, what if the individual parts cannot be re-composed to form a convincing replica of the whole? In complex political arrangements, the contribution of the parts is contingent upon their role in an interdependent whole. We comparativists have long been aware of the so-called ‘ecological fallacy’, namely, the potential for error when one infers from the behavior of the whole, the behavior of individuals within it. For example, just because electoral districts in the Weimar Republic with a larger proportion of Protestants and farmers tended to vote more for the Nazi Party (NSDAP) is no proof that individual Protestants and farmers were more likely to have voted for that party. This can only be demonstrated by data at the apposite level. But what is more important in today’s complex world is the inverse, i.e. ‘the individualistic fallacy’. This consists in simply adding up the observations about individuals and proclaiming an explanation for what they do together. Hence, the more ‘democratic’ the values of sampled persons, the more ‘democratic’ their polity.Footnote 23 Rational choice analysts do this routinely (with the ever-present caveat that their subjects always act ‘under constraints’). While I would admit that this may work reasonably well where the political process being studied is itself additive, i.e. voting, it can lead to serious fallacies of inference when ‘rational’ individuals interact unequally within pre-existing institutions and networks. Just try to imagine the re-composition of individual preferences and rational choices into a model that would try to predict, say, the level of public spending or the extent of redistribution across social classes!

These, admittedly primitive, thoughts about complexity imply a very significant change in conceptualization – how one defines and circumscribes what it is one proposes to use as independent, intervening, or dependent variables. The classical advice is to make these concepts as precise as possible, so that they can be recognized inter-subjectively, and measured unequivocally. Applications beyond these limits are said to be ‘stretched’ out of shape for historical or cultural reasons and, therefore, invalid for comparative purposes (Sartori, 1970). Of course, the ultimate simplification is to reduce the variable to a single dimension that can be accurately represented by numbers, by symbolic logic, or by mathematical equation. Most rational choice and a good many institutional comparativists are quite proud of their ability to do this – and, thereby, to insure both clarity and parsimony in their work.

A ‘complexifier’ is more likely to be interested in a set of variables whose boundaries and interrelationship are intrinsically unclear. What does the explaining or is the object of explanation something ‘fuzzy’ that may not take on the exact same observable properties in every case. His or her concepts reference something that is similar but not identical. They may share certain properties but not all of them and when they are broken down into sub-types, the categories may be nominal and not ordinal, i.e. they may cluster according to different and not the same criteria. Just think of some of the concepts that political scientists use almost everyday: power, state, democracy, legitimacy, pluralism, capitalism, competition, hegemony, accountability, responsiveness, and bureaucracy. These and many more are ideal types and they are blurred, radial, and fuzzy in nature. A qualified observer knows what they are and grasps what they may have in common, but hesitates before specifying them in identical ‘non-stretchable’ terms. Take, for example, the label democracy when applied to the United States. We find it relatively easy to dismiss the fact that this polity did not accord full citizenship to women until the 1920s and yet still deserves to be classified as a democracy before that time, but what about the ‘Single-Party South’ and the systematic (and widely tolerated) suppression of the right to vote for Afro-American ‘citizens’. By any rigorous standard, the prevalence of this condition and practice should disqualify American democracy until after the Civil Rights Movement of the 1960s and, yet, I suspect that most American political scientists would strongly contest such a ‘de-classification’ of their democracy. I have yet to find a comparative study of ‘stable Western liberal democracies’ that does not include the United States – whatever the time frame. The same generic argument could be made about the political manipulation of the judiciary by the executive power in France until quite recently or the ‘bizarre’ survival of the un-elected House of Lords in Great Britain or the non-rotation in power of Swiss political parties. There have been quantifying simplifiers who scored Guatemala and Colombia as democracies when the military in the former had veto power over elected civilians (as well as being an egregious violator of civil rights), and the latter was ruled by martial law over 1/3 of its territory – just because regular competitive elections continued to be held.

My contention is that fuzzy ‘ideal-typical’ concepts are virtually indispensable in political science, even if attempts (and there have been many) to pin them down to identical, least of all quantifiable, measures have failed. Moreover, in a world of steadily increasing ‘complex interdependence’, comparativists will have to rely more and more on such concepts, both to do the explaining and to specify what has to be explained. The quest of simplifiers for precise definition and uniform measurement (not to mention parsimonious explanations) will mean that some of the key aspects of politics in our times will be excluded conceptually and dismissed as potential subjects of research. Just think of all those elements of contemporary politics that involve lengthy chains of causality, the intervention of indirect or delayed agents, the impact of unintended consequences, the possibility of multiple equilibria, the cooperation of several layers of authority, the emergence of new (and, often, contradictory) properties, the ‘chaotic’ effect of minor variations, the concurrent presence of discrete causes and their compound impact, the unexpected resistance of entrenched habits and standard operating procedures, the effect of random or unique contingencies, the role of anticipated reactions, the ‘invisible constraints’ imposed by established powers, not to mention, the inability of any actor to understand how the whole arrangement functions.

Let me illustrate this point by invoking two concepts that I have used to describe and analyze the EU – my assumption again being that this regional polity represents an extreme case, even a reductio ad absurdum, of the direction in which all polities are heading (admittedly at very different paces):

Multi-level governance: An arrangement for making binding decisions that engages a multiplicity of politically independent but otherwise interdependent actors – private and public – at different levels of territorial aggregation in more-or-less continuous negotiation, and that does not assign exclusive policy compétence or assert a stable hierarchy of political authority to any of these levels.

Poly-centric Governance: An arrangement for making binding decisions for a multiplicity of actors that delegates authority over functional tasks to a set of dispersed and relatively autonomous agencies that are not controlled – de jure or de facto – by a single collective institution.

To the extent that the polities being compared have something like these characteristics, they are both spatially and functionally complex. Nota bene, these are not just ‘supra-national’ properties, even if they are most prominently on display in the EU. A polity may be relatively free of constraints imposed by global or regional organizations (or by hegemonic neighbors) and still find itself in a situation of multi-level and polycentric governance due to the autonomous behavior of sub-national units or functionally specific agencies. Consider, for example, the United States, which under its present government, is most emphatic about its lack of accountability to international law and organizational constraints (that it does not control) and yet on stem-cell research it has been unable to exert sovereignty over sub-units and agencies that oppose the policy of its central state. Infra-national units can contribute just as much to the complexity of politics as supra-national ones.

In addition to major implications for the operationalization of indicators – i.e. simple aggregate measures of individual or collective behavior will not do – complex interdependence within and between polities raises the prospect of major fallacies of inference between different levels of analysis. Properties reliably observed at the micro-, meso- or macro-levels may be poor predictors of what happens at other levels, lower and higher. Just as one cannot simply add up a large numbers of ‘civically cultured’ and ‘democratically minded’ citizens in order to produce a democracy, one cannot guarantee that conditionalities imposed by regional organizations or hegemonic democratic neighbors will suffice to ensure a successful transition from autocracy to democracy – or even to have an important impact. The correct inference may depend on context, i.e. upon the multiple layers involved and the possible existence of a plurality of competing centers of authority.

It is no coincidence that both of these descriptive qualifiers are attached to the same substantive concept: governance. No student of comparative politics over the past two and a half decades can have ignored the amazingly rapid and widespread diffusion of this concept and yet few concepts have ever been as fuzzy and polysemic. Whatever its opportunistic origins or the many (all ambiguous) meanings attached to it, I am convinced that behind the notion of governance hides an important message about changes in the practice of ‘authoritatively allocating values’ (as David Easton so eloquently put it). Government, i.e. doing so through a hierarchically disposed and legitimately recognized set of public institutions, is less and less capable of making such allocations – especially when confined to a single level of spatial aggregation. What it takes is something much more complex, which I have elsewhere described as

The core of such a complex arrangement rests on horizontal forms of interaction between actors who have conflicting objectives, but who are sufficiently independent of each other so that neither can impose a solution on the other and yet sufficiently be interdependent upon each other so that both would lose if no solution were found.Footnote 25 In both modern and modernizing societies, some of the actors involved in governance are non-profit, semi-public, and, at least, semi-voluntary organizations with leaders and members; moreover, it is the embeddedness of these organizations into something approximating a civil society that is crucial for its success. These organizations do not have to be equal in their size, wealth, or capability, but they have to be able to hurt or to help each other mutually.

Also intrinsic to governance is the notion of regularity. The participating organizations interact not just once to solve a single common problem, but repeatedly and predictably over a period of time so that they learn more about each other’s preferences, exchange favors, experience successive compromises, widen the range of their mutual concerns and develop a commitment to the process of governance itself. Here, the code words also tend to be fuzzy: trust and mutual accommodation – more specifically, trust and mutual accommodation between organizations that effectively represent more-or-less permanent social, cultural, economic, or ideological divisions within a society or across several of them.

Note also that governance is not only about making decisions via deliberation and negotiation but also about implementing policies. Indeed, the longer and more extensively it is practiced, the more the participating organizations develop an on-going interest in this implementation process since they come to derive a good deal of their legitimacy (and material rewards) from the administration of mutually rewarding programs.

Governance is not a goal in itself, but a means for achieving a variety of goals that are chosen independently by the multiple layers of actors involved and affected. Pace the frequent expression, ‘good governance’, resort to it is no guarantee that these goals will be successfully achieved. It can produce ‘bads’ as well as ‘goods’. Nevertheless, it may be a more appropriate method than the more traditional ones of resorting to public coercion or relying upon private competition.

Moreover, it is never applied alone, but always in conjunction with state and market mechanisms. For ‘governance’ is not the same thing as ‘government’, i.e. the utilization of public authority by some subset of elected or (self-) appointed actors, backed by the coercive power of the state and (sometimes) the legitimate support of the citizenry to accomplish collective goals. Nor is it just another euphemism for the ‘market’ i.e. for turning over the distribution of scarce public goods to competition between independent capitalist producers or suppliers. It goes without saying that, if this is the case, the legitimacy of applying governance to resolving conflicts and solving problems will depend upon different principles and operative norms than are used to justify the actions of either governments or markets. This concept of ‘governance’ will not suffice to bear all of the weight imposed upon future comparativists who turn toward complexity. But it is definitely one starting point for those setting off in that direction.

Concluding thoughts

I conclude with three disciplinary suggestions:

1. 1. Political scientists should abolish the distinction between comparative politics and international relations and re-insert an ontological one between political situations that are subject to rules, embedded in competing institutions, and not likely to be resolved by violence, and those in which no reliable set of common norms exists, where monopolistic institutions (including but not limited to states) are in more or less continuous conflict and likely only to resolve these conflicts by force or the threat of force. It used to be believed that this line ran between politics within states and politics between states. This being no longer the case – the probability of war has been greater within the former than between the latter for some time – there is no generic reason why these two ‘historical’ sub-disciplines should be kept apart. How about separating the students of politics into those working on ‘ruly’ and on ‘unruly’ polities, whether they are national, sub-national, supra-national, or inter-national?

2. 2. Comparativists should attempt to include the United States in their research designs when it seems apposite, but they should not expect their Americanist colleagues to join them – at least, not for some time. The present direction of politics in the US is virtually diametrically opposed to the trends I have noted above. Americans (or, better, their present leaders) have reacted with hostility to the prospect of ‘complex interdependence’ and made all possible effort to assert both their internal and external sovereignty. They have repeatedly denied the supremacy of supra-national norms and the utility of international organizations by refusing to regard those legal or organizational constraints that do exist as binding when they contradict or limit the pursuit of so-called national interests, and by withdrawing from them when it seems expedient to do so.

3. 3. Comparativists – whether of ruly or unruly politics – should be equipping themselves to conceptualize, measure, and understand the great increase in the complexity of relations of power, influence, and authority in the world that surrounds them. Admittedly, ‘complexity’ is still only a specter haunting the future of their sub-discipline and the answer to meeting this need probably cannot come only from within their own ranks. Hopefully, comparative politics will attract successful ‘grafts’ of theory and method from disciplines in the physical and mathematical sciences that deal with analogous situations, but in the meantime the challenge should be met and the opportunity seized by us. Just picking up a few scattered concepts from within political science, such as multi-layeredness, polycentricity, and governance – as I have done – will not carry comparativists far enough. Although, if my experience in studying what must be the most complex polity in the world, the EU, ‘real-existing’ politicians and administrators who have to cope with all of this contingency and complexity are inventing expressive new terms everyday. We should be listening to them, as well as to scholars in other disciplines, to pick up on these emerging arrangements, specify them more clearly where this is possible, and search for points in our theoretical frameworks where they can be inserted. Contrary to the ‘simplifiers’ who are genetically opposed to dialoging with their subjects, we ‘complexifiers’ have a need and obligation to take seriously what they tell us they want and what they claim they are doing.

I cannot escape the conviction that this is the most promising path forward for the sub-discipline. The emergence of a new instrument, qualitative comparative analysis (QCA), has given it some initial methodological momentum. Its originator, Charles Ragin, has been quite articulate about the implications it has for theory, in general, and the comparative method, in particular (Ragin, 1987, 1994). Especially in its ‘fuzzy set’ and ‘two-step’ versions, QCA offers a far more appropriate estimation technique for analyzing complex causality than the usual ones from social statistics.Footnote 26

And it also seems uniquely capable of explaining something that I think will become more and more salient in the future, namely, equifinality. Since its Aristolelian origins, the comparative method has been applied mainly to explaining differences. Why is it that polities sharing some characteristics, nevertheless, behave so differently? This has allowed the sub-discipline largely to ignore what John Stuart Mill long ago identified as one of the major barriers to developing cumulative social science: the simple fact that, in the ‘real-existing’ world of politics, identical or similar outcomes can have different causes. Perhaps, it is only because my recent research has focused on two areas where this phenomenon has been markedly present: European integration and democratization that I am so sensitive to this ontological problem. In both of these sub-fields, the units involved had quite different points of departure, followed different transition paths, chosen different institutional mixes, generated quite different distributions of public opinion and, yet, ended up in roughly the same place. Granted there remain significant quantitative and qualitative divergences to be explained – presumably, by relying on the usual national suspects – but the major message they suggest is that of equifinality, i.e. convergence toward similar outcomes.

Of course, not all of the world’s polities are converging toward each other either in institutions, policies, or behaviors. Neo-neo-neo-institutionalists will have plenty of differences to explain into the distant future. Simplifiers will no doubt come up (ex post) with plenty of arguments why actors have rationally chosen different rules and policies. In other words, there will still be lots of room in the broad canopy of comparative politics. All I have been trying to do in this essay is to ensure that it will have a secure place and adequate rewards for those who choose to embrace complexity.