“If there is any great lesson we Americans need to learn, George Kennan once famously wrote, with regard to the methodology of foreign policy, it is that we must be gardeners and not mechanics in our approach to world affairs.” He was aiming at the practice and mental habits of the American diplomatic and national security establishment; however, the statement also indirectly reflected Kennan’s skepticism about the possibilities of the type of knowledge proposed from some sections of academia as a solid basis for the engineering approach. It is not that Kennan advocated for improvisation, neither that he was an epistemological dinosaur who denied the possibility of a rational and systematic exploration of the regularities of international politics. But he was well aware of the limits that both the anarchical nature of the international system and the particular fluidity and uncertainty that marked the dawn of the cold war set for the ambition of a nomothetic science of international relations with predictive potential.
Of course, many things have changed both in the real world and in the bubble of theorists of international politics since the days of the Eisenhower administration. There is also the traditional argument that sees the anarchic nature of the international system as the source of a qualitative difference between international and domestic politics. Yet, I think that Kennan’s reflections still carry a valid sobering message for political scientists in general and for those members of the corporation dealing with Latin America and other underdeveloped areas in particular. I do not intend to vindicate any sort of intuitionist epistemology, neither to resurrect any common place about “the inexhaustible diversity of social reality” as something that would preclude the development of social science tout court -although I think that political scientists need to be periodically reminded of how complex and difficult to know what social reality is. I believe in the possibility of systematic and rational explanations of social phenomena, constructed according to rigorous standards of empirical verification, and I am not willing to give up the very tangible permanent benefits of the “behavioralist revolution.”
Instead, I want to submit some ideas for debate, that are organized around two axes. First, I will point to a series of limitations of institutionalist approaches to politics -which have constituted the predominant stream among students of Latin American politics during the last 25 years- to deal with some crucial aspects of several political systems in the region. The basic argument, in this respect, is not that institutionalism has nothing to offer, but rather that it should be urgently revised and complemented -otherwise, our understanding of Latin American politics risks being seriously crippled. Second, I will contend that the current disjuncture between the theory and reality of politics reveals limitations that are deeper than the mere contents of the former. My suggestion is that the specific theoretical limitations of institutionalism provide an interesting window to reflect on some problems of the common methodological wisdom that a majority of political scientists in general, and of those educated in the United States in particular, tend to absorb these days as a central part of their professional socialization. Under this light, the problems faced by efforts to make Latin American politics intelligible become an extreme -thus especially revealing- case to show some flaws of the type of knowledge political scientists are predominantly trying to build these days.
I. The limitations of institutionalism
In 1999, Barry Ames began a review of recent work on Latin American politics by observing that “institutions are in,” noticing also that the preferred manifestation of such a theoretical vogue was the consideration of institutions as explanatory variables. This, as Ames also noticed, should hardly be surprising in itself -after all, political scientists, almost by definition, had already been aware of the fact that formal governmental structures and the rules guiding their operation shape the outcomes of political life. But it was equally well known that respect for the nuances of institutional life had not precisely been a defining attribute of the military dictatorships that plagued the area until the early 1980s. And even once the tide of the “third wave”started to sweep those regimes, politics would be for a while marked by the uncertainties and unexpected turns of multiple “games of transition.”
The hour for the institutionalist revival arrived once the attention of political scientists switched from “transitology” to “consolidology.” No matter how wide disagreements about the contents and determinants of “democratic consolidation” were, nobody questioned that the regular and predictable operation of institutions had to be one of its defining dimensions. The 1990s were thus populated by a series of debates more or less directly linked to concerns about strategies of “institutional engineering” that could favor the routinization of democratic politics. As the urgency of economic reforms climbed to the top of the agenda, concerns about “governability” and democracies’ productivity in terms of public policy change in general, and market-oriented transformations in particular, only reinforced the institutionalist bias. The continuity of elections, regular government formation, legislative activity, and reiterated interactions between presidents, legislators, and bureaucrats, simultaneously brought the accumulation of homogeneous series of comparable data, and opened the door for the importation of theories originally developed for the democracies of the OECD universe. The foundational debate triggered by Juan Linz’s concerns with “the perils of presidentialism” was thus followed by a series of discussions on the relative virtues of diverse combinations of electoral rules, party system configurations, legislative committees dynamics, federal institutions etc. The consequences, as are well known, reached beyond the academic bubble: the decade was also rich in experiments of institutional reforms, often encouraged by international organizations and based on inputs from political scientists and economists fascinated by their recent rediscovery.
The results of this trend -still pretty much alive and delivering- have been far from negligible. However, it is somehow symptomatic that, during the last decade, a growing portion of the literature has been characterizing several Latin American political systems as “hybrid,” “dual,” or “fragmented;” split between “formal” and “informal” institutional logics; presenting “gray areas” between institutionalized and violent contentious behaviors; going through cycles of “serial institutional replacement;” spotted by “brown areas” beyond the effective reach of state apparatuses, where citizenship rights may become little more than legal fictions, etc. In one way or another, they all recall an old topic for anybody familiar with Latin American history -the recurrent frustration of successive generations of constitution-makers and institution-builders in general with the gap between “el pais de papel” and “el pais real.”
The conclusion is not, of course, that institutions do not matter. The basic problem has at least four dimensions (the following list has is not exhaustive, but only a quick illustration): 1) the operation of institutions is not context-proof -thus, the same institutional design may end having divergent results, depending on the socio-economic environment that surrounds it and on the more comprehensive institutional web in which it is inserted in each specific case; 2) “machiavellian moments”did not end with the end of transitions to democratic regimes. Under some critical contexts, and the formal continuity of institutional forms notwithstanding, political stalemates throughout the region demanded exercises in creative statecraft that escaped or bypassed the established institutional channels, thus reducing the predictability of outcomes; 3) the frequently frustrating results of institutional reforms poorly fitting their environments often led to what Levitsky and Murillo term “serial institutional replacement” -recurrent cycles of almost compulsive institutional change; 4) the basic assumption of institutionalist formulations -that the Hobbesian problem of political order has been satisfactorily solved, thus making it possible to enforce and guarantee effective compliance with institutional logics- does not necessarily apply in many areas of the region; 5) the unpredictability that results from some combinations of the mentioned factors has a perverse self-reinforcing logic. On the one hand, key political actors may choose strategies that either fully bypass institutions, or combine partial compliance with the development of extra-institutional logics. On the other hand, even actors willing to fully play by the formal rules of the game may be precluded by the spasmodic dynamics of recurrent institutional change from advancing along the learning curves that the consolidation of predictable institutional dynamics require.
Here I will not ever begin to attempt the exploration of possible solutions for the multiple theoretical, empirical, and methodological problems created by this picture. I just want to stress two very general aspects of its results. First, political dynamics are frequently messy, highly dependent on context, difficult to predict -closer to what Charles Lindblom once dubbed “the politics of muddling through”than to the clean, transparent, and predictable type of game that consolidated institutions are -in theory at least- supposed to guarantee. Second, the intelligibility of these dynamics is not likely to improve if we replace parsimonious institutionalist explanations by a new type of explanation parsimoniously based on some type of alternative factor. We have accumulated enough solid evidence that institutions matter; the problem is that they are not all that matters, and that they may matter to a variable extent and in variable ways, depending on their multiple interactions with other factors. What we may need, then, is not necessarily an alternative theoretical approach, but a more refined theory of institutions, capable of accounting for the complexity of those interactions.
The last point is crucial, and following Javier Auyero we could term it the problem of “gray areas.” One of Auyero’s multiple contributions to a more accurate understanding of contemporary politics -significantly coming not from a political scientist but from a sociologist working with ethnographic techniques- is the identification of “gray areas” where violent contentious strategies, rather than fully replacing institutional ones, interlock with them in original and puzzling ways. What such realities call for is neither abdication from the production of theory, nor the dismissal of institutionalist theory. Instead, we need a theory of institutions that covers the interlocking of institutional and non-institutional dynamics -good institutional theory should include theorization on the limits of institutions’ explanatory power. We need, in other words, and to make the metaphor of the title more transparent, theorization of the gray areas of political life, where actors definitely approach their options more with gardener-like than with mechanic-like dispositions.
The idea is not in itself particularly novel. Already in the late 1990s, for example, James Malloy and Kurt Von Mettenheim, summarizing the results of a series of studies dealing with the most recent evolution of democratic politics in the region, noticed that “[c]ontemporary Latin America illustrates that leadership or statecraft, often ignored by theorists bent on developing scientific, nomothetical, or predictive models, becomes particularly important in moments when more routinized ways of government and politics give way to more open and volatile moments of transit to or founding of new regime forms. [...] On the other hand, while statecraft or leadership has an idiosyncratic dimension, there is no reason why one cannot examine strategies of leadership after the fact and draw from them, not nomothetic generalizations, but generalizations in the form of prudential rules of action that could serve leaders and analysts confronting analogous situations.” The allusion to “scientific, nomothetical, or predictive models” provides a connection with the other main aspect of this essay -namely, that political scientists may these days be poorly prepared for the task of developing the required type of theory. In other words, we are trained to deal with politics in mechanical terms, and a mechanic’s tool-kit only can accomplish so much when dealing with gardeners.
II. Beyond substantive theoretical contents: the epistemological blind spots of the profession
Now, this is not the place to go as far back as discussing the qualitative differences between the ontology of social and natural phenomena -although doing so more often than we usually do would not hurt. I will admittedly cut thick, and point to some implications of the supreme goal under which our professional socialization usually takes place. In the end, we are usually told, we want to build, as parsimoniously and comprehensively as possible, the type of theory that, through the systematic testing of deductively obtained hypotheses, would make the development of cumulative knowledge possible, based on the verification of law-like statements. The path recommended for this ultimate goal is to a great extent signaled by a series of dichotomous oppositions that carry a high-risk of intellectual hemiplegia -nomothetic vs ideographic (or social science vs history); deductive vs. inductive; descriptive vs explanatory; theory-testing vs. theory-development, etc. (in all cases, the reader may have noticed, the “positive” pole is listed in first place). In other words, get your KKV, read it carefully (this is, of course, good advice), and stick to its recipes (this is not).
My points are the following. First, that we seem to have a really hard time accepting that sometimes the best answer we can provide is a very tentative “well, it depends”. Second, starting our work with the assumption that parsimony is an absolute value constitutes a considerable handicap. Actually, parsimony is always at one extreme of a trade-off with explanatory accuracy, and I honestly have a hard time seeing why we should have a preference for widely encompassing parsimonious theories that can only provide very limited insights into phenomena that are by nature extremely complex. Third, that the obsession with the falsification of deductive theory does not help; theory has no value in itself, unless it contributes to a better understanding of the different concrete situations of each case. This last aspect has a paradoxical consequence. On the one hand, we are reluctant to get our hands dirty with cases and topics that are too fluid or “muddy” (the “gray area”) to allow any sort of clear theoretical generalization; on the other hand, we often dare test models to explain processes that could only be properly explained with a more long-term, historical perspective -for example, when we identify on-going “critical junctures.”
Do I need to repeat that I am not recommending abstention from theorization? I am rather arguing two things. One, that we can only theorize so far with the lack of data-mining and thick description that we have been able to accumulate regarding some of the most challenging problems. We desperately need in-depth case studies, qualitative descriptions, historical narratives -many of which could be constructed to a significant extent based on evidence that is already available. Second, that there are intermediate forms of valid, rigorously-constructed conceptual generalization, that are still necessary short of the goal of cumulative research outcomes based on deductively-derived predictions. In other words, law-like formulations do not constitute the only legitimate forms of “theory” -actually, their development is contingent upon the previous availability of some building-blocks that we only too often despise as second-rank theoretical products (conceptual refinement, ideal-types, classifications). These products are not only necessary; for some problems, they may also be the best that we can do, at least for a while. But even assuming that they will constitute intermediate steps in the development of more ambitious theoretical buildings, they constitute valuable tools themselves -particularly if we agree that by putting together a good description that clearly shows how things happen may in itself already entail considerable progress towards explaining why they happen.
“Scholars, indeed all men, Albert Einstein wrote, must move forward in the faith of that philosopher who held that there is no problem the human reason can propound which the human reason cannot reason out.” I agree. So the argument is not that we should stop trying to do better: prediction is a worth-keeping goal, and we should definitely not water-down our standards for falsification. However, whether it constitutes an actually attainable goal is an empirical matter; a priori, all we have is our faith that it may be possible and our conviction that the enterprise is worth trying, but nothing justifies the dogmatic assumption that a social science that fits the canons of physics is actually viable. Now, there is a somewhat paradoxical corollary to this: while on the one hand we should perhaps be more modest about the type of theory we may be able to develop -at least in the foreseeable future-, on the other hand we should have a higher estimation for what we can actually achieve. After all, the reluctance to admit that “well, it depends” is the best we can do for now may to some extent result from lack of awareness of how much progress being able to tell what “it depends” on already represents.
This in turn requires a higher sensitivity to the diversity, complexity, and temporality of social phenomena than we are frequently willing to acquire, which in turn leads to more prosaic motivations than we are willing to admit. Hyper-specialization does not help, neither does parochialism -both geographical and disciplinary. Generally speaking, and particularly excepting a few sub-tribes, political scientists are not good neighbors: economists aside, we tend to pay little attention to what is going on beyond the boundaries of our discipline -and I find revealing that neighbors willing to borrow do not find too much worth-borrowing among what we have to offer. The exception of economics is revealing too. Crossing some disciplinary boundaries may entail acknowledging that Political Science is actually less “scientific” than we would like it to be -or that some traits we like to praise as credentials of science are not. Such acknowledgments are seldom painless (humility rarely is) -they may have severe consequences in terms of reputation, self-esteem… even funding.
Such issues are, of course, by no means exclusive of political scientists; as Pierre Bourdieu has masterfully shown, the preferred self-image of Homo academicus tends to be a more gratifying one -we are uninterested souls in search of truth out of love for humanity, and our disagreements are a matter of empirical evidence and rational discussion. But political scientists seem to be particularly ill-prepared to turn their own professional selves into objects of inquiry. We may occasionally be willing to go far enough to accept that ideological or political preferences orient the problems we pick and somehow shape some of the answers we are inclined to explore; however, considering that our most ethereal choices may be actually determined by the positions we hold in a professional field shaped by power relationships and struggles around reputational resources is a totally different thing. We are human, after all, and we do this, among other goals, to make a living; we depend on our symbolic capital for important things. The assimilation of theories, methodology, techniques, take years of painstaking efforts, upon which we build professional strategies; the possibility that the intellectual capital so painfully accumulated may not be as valuable or sufficient as we once thought can only cause anxiety. What Adam Przeworski calls “retooling processes” should be part of our professional routines; the truth is, however,that most of us find such processes too costly in more than one sense.
As a result, should the discovery that our positions in scientific controversies may be at least partially shaped by our professional stakes be that shocking? As I said, this is a risk that affects social science in general -and not only social science: in Bourdieu’s words, the inclination to “mistake the things of logic for the logic of things” constitutes the quintessential malady of the scholastic condition. And, in terms of institutional determination, we should begin by reminding ourselves that in few domains of social life do institutions weigh as heavily as they do in academia. But while sociologists, anthropologists, historians can all show, if certainly not mainstream, at least consolidated traditions of enquiry around the socio-historical determinants of the ways they do what they do, political scientists do not -actually, the collective memory of the profession has become surprisingly short. But perhaps that too should be hardly surprising: if we are willing to believe that our objects to be by nature ahistorical atoms making rational decisions, why would we regard ourselves otherwise?
There are many reasons that make the construction of a social science infinitely complex, and analyzing the (under) development of the field in terms of a sort of “modernization theory” is a good recipe to blind ourselves to some of those reasons. When we build images of the social world, we are simultaneously building self-images -something that already in itself a priori suggests a high probability that we experience the not completely conscious temptation to indulge ourselves with a more gratifying image. Being aware of the risk, and willing to turn ourselves into an object of reflection and self-analysis, is a necessary requisite for a more methodologically self-conscious practice. We may not completely like the image such an exercise provides in return, but it would certainly help to do a better job.
 That would place us, we seem to fear, one step below in the “cascade of despise,” closer to historians and anthropologists and farther from economists. Would it help make it less painful if we recalled that old joke according to which an economist can be defined as a person who spends half of her time making predictions, and the other half trying to explain why reality ended not behaving accordingly?