Realism,Mechanism,Effect,and Comparative Contentious Politics

Abstract

Charles Tilly was instrumental in launching the comparative study of contentious politics through key contributions delineating and defining the subject matter. But his input is epistemological as well as theoretical and substantive. What characterizes his epistemology is a realist mechanistic explanation rendering mechanisms tools of both explanation and comparison. Employing the realist perspective, this article examines Tilly’s focus on mechanisms and some of the ensuing challenges for comparative contentious politics. It argues that mechanisms can explain confidently the emergence of contention in given episodes and can enable the comparison of emergence across episodes but cannot generalize easily about them. This tension between the particular and the general characterizes also the task of conceptualization because the key concepts in this field of study refer to processes. What ameliorates this tension is to conceptually decouple the effect produced in mechanism operation from the modality of such operation and to assign epistemological weight to the effect. This move is discussed with reference to distinctions among mechanisms, to distinctions among concatenations of mechanisms, and to potentiality.

Introduction

Charles Tilly tends to shun labels when describing his own approach to sociology. But one label which he actually embraces is “relational realism,” a made to measure label befitting particularly the work he produced in the last 15 years of his career. This label indicates the position that relations are to be considered real elements of social life rather than merely analytical categories for the study of social life, ontologies rather than models. As Tilly writes, “transactions, interactions, social ties, and conversations constitute the central stuff of social life” (2002, p. 72). But I suggest that Tilly’s realism is broader than this, even though not fully committed. It is not merely an ontological statement about a feature of social reality but rather a broader realist stand that aims to bridge ontological and epistemological concerns. At the same time, Tilly’s writings lack systematic engagement with the philosophy of science. This means that Tilly’s realist positions must be first searched for and assembled and, then, the blanks which they leave behind must be filled with extrapolations, even speculative ones. To describe what is realist in Tilly, as I endeavor to do in this article, is therefore an act of interpretation which is liable to misstep. Yet, I suggest that this intellectual effort, skewed though it may be, has value that goes beyond the exegesis of Tilly’s work. It has broader value because the research program on contentious politics charted by Tilly is currently alive yet not firmly on the tracks which Tilly might have hoped for it. It has leaned toward realist epistemology without being reflective about it, and this is a predicament which has arguably fed topical theoretical advancements but has not served well the overarching theoretical aims aspired by Tilly. Situating the work of Tilly in the realist tradition, therefore, will go some way to illuminate the realist pathway and related epistemological challenges conceivably facing the contentious politics program currently and into the future.

Tilly’s work aspires to develop tools for comparison and generalization, with the most fundamental tool being the notion of mechanism. To be certain, the contentious politics research program, most distinctly charted through Tilly’s collaboration with Sidney Tarrow and Doug McAdam (McAdam et al., 2001; Tilly & Tarrow, 2007), is an empirical research program more than a roadmap for theory-building, let alone a treatise in epistemology. It does recommend the notion of mechanism but not as a requirement of admission, and many of the participants of that program may well eschew the notion. Nevertheless, holding a realist yardstick, I approach this as a mechanistic comparative program. I therefore maintain that the biggest challenge pivots on its ambiguity about the mechanism and process abstraction, which is a manifestation of the tension, looming large in the social sciences for over a century, between the particular and the general.

This essay has three sections. The first offers an interpretation of Tilly’s realism, the second an amendment of the notion of mechanism which Tilly provides, and the third three considerations which may benefit future research. Thus, I start by examining the notion of mechanism and its affinity with concepts denoting change. I show that Tilly is a realist when it comes to such concepts, but his stand is in tension, first, with his occasional turn toward analytical thinking and, second, with his ambition to develop mechanism generalizations. In the second section of the article, I move on to examine comparison more particularly, suggesting that it is hindered by the expectation that mechanisms and their concatenations entail modular constituent events. I argue that the focus be placed on the effects that constitute the defining characteristics of processes and mechanisms, not the way those effects come about. This focus, as I discuss in the third section of the article, facilitates the manner in which we might conceive distinctions among mechanisms and among the concatenation of mechanisms, and has bearing also on how to approach process potentiality. Ultimately, I hold that the program on contentious politics has a purpose, but one which is epistemologically modest regarding generalizations—arguably more modest than what Tilly hoped for.

Realism and Mechanism

The social scientific perspectives which employ notions of mechanism are not necessarily realist; some are seen as analytical (Hedström & Ylikoski, 2010), postpositivist (King et al., 1994), or “unaffiliated” (Stinchcombe, 1998). Nor do all social scientific perspectives seen as realists employ some notion of mechanism. But some of the leading realist perspectives do so. Indeed, these perspectives have brought to the fore one or another notion of mechanism to uphold as superior to the notion of causality as constant conjunction. And so does Tilly’s sociology. Below I discuss what is realist in it and unpack the notion of mechanism which it formulates, drawing connections to the realist philosophy of science and contrast to analytic approaches to social science.

Ample differences can be found under the rubric of realism. Two often-mentioned variants are critical realism and scientific realism, the latter a broad and rather elastic grouping. Some principal tenets are shared among realists of all shades. The most fundamental is the ontological position holding that social reality exists independently from the researcher and is ordered to a significant extent. In terms of epistemology, realism holds that the social reality which exists at a deeper level than that betrayed by secure data (sense data and/or data not subject to varied interpretation) is detectable indirectly and, thereby, falls within the purview of the social sciences. This epistemological tenet, taking realism away from empiricism and postpositivism, leads realists to be open to conjecture and to wide-ranging techniques of generating and analyzing data. This and the previous tenet lead many realists to engage in research-specific reflection on the relationship between ontology and epistemology. That is, they are inclined to theorize the non-operationalizable as well as the operationalizable facets of their subject matter and to juxtapose the two theorizations. Thus, they often develop narratives that feature both “analysis at the level of ontology” and “analysis at the level of epistemology.”

Tilly’s work is realistic in the above sense, except for the last part. That is, it accepts a research independent yet directly and indirectly researchable social reality but avoids systematic analysis at the level of ontology and discussion of the ontology–epistemology link. As I elaborate below, important concepts in Tilly’s work leave open questions as to whether they lean toward the ontological or the analytical end of the spectrum of concepts. Furthermore, while Tilly’s stand on causality is evidently realist, it also lacks clear reflection on the ontology–epistemology link.

The ontological discussions which Tilly does develop are limited and largely regard what he considers to be the basic ontological choices facing contemporary social science. Specifically, he delineates the basic ontological choices to be holism (whereby social structures have their own sustaining logic), methodological individualism (whereby human individuals are the basic or unique social reality), phenomenological individualism (whereby individual consciousness is the primary or exclusive site of social life), and relational realism (whereby transactions, interactions, social ties, and conversations constitute the central elements of social life) (Tilly & Goodin, 2008). Tilly presents this assembly of alternatives in different publications, typically with the aim of pushing for relational realism. It would seem, therefore, that Tilly considers ontological discourse in the social sciences to regard types of concepts and not so much individual concepts.

Nevertheless, there is reason to argue that the fundamental concepts which Tilly develops—contentious politics, democratization, social trust, durable inequality, and more—refer directly to social reality and are therefore not analytical concepts. The latter are characterized by theoretically supported coherence rather than coherence based on the claim that the phenomena they denote are coherent; they are exemplified most clearly by Weberian ideal types (Jackson, 2016, Chapter 5). The assessment that Tilly’s fundamental concepts have a realist outlook is supported by the manner in which he writes about their referentiality as well as by the very fact that they refer to relations, the privileged type of ontologies.

Here it must be said that no contemporary realist would argue for a simple correspondence theory of truth claims. Rather, rejecting the idea that concepts mirror reality, practically all realists acknowledge the gap between reality and knowledge of reality. While this acknowledgment feeds disagreement among them on the extent to which concepts can reflect reality or can capture some essence of reality, it is also a recognition of conceptualization as an epistemologically informed act. Thus, as concepts’ connotations are always limited, the social scientist must determine such connotation based on something of interest to her intellectual project, at the exclusion of whatever else might be of less interest. Accordingly, no matter how committed to realism, one conceptualizes from a point of intellectual interest. This also means that while the concept can claim to denote real occurrences, it cannot claim to denote everything that may regard those occurrences. Selective connotation implies selective denotation.

Tilly’s writings effectively accept this position. That is, they implicitly accept that a given concept can refer credibly to reality and that other concepts can refer to the same reality just as credibly from a different angle. Tilly’s move to climb up the ladder of conceptual abstraction is an indication of this. Tilly often employs the technique of climbing up the ladder of conceptual abstraction to develop substantive theories from a fresh point of view (Demetriou, 2018). This is seen, for instance, with his concept of categorical inequality, which is placed higher on the ladder of conceptual abstraction than the commonly employed terms gender inequality, race inequality, or economic inequality (Tilly, 1998). More significantly for present purposes, this is seen also with his concept of contentious politics, developed in tandem with Sid Tarrow. This concept is placed higher on the ladder of conceptual abstraction than such concepts as revolution, social movement campaign, protest cycle, labor strike wave, ethnic conflict, and insurrection. Tilly argues that the concept of contentious politics subsumes all these more commonly employed concepts and, thereby, allows not only the comparison across the phenomena that are denoted by these concepts but also the comparison of similar phenomena that elude these concepts or of phenomena that connect to more than one of these concepts. Tilly maintains this thesis as he defines contentious politics in terms that are relational and light (in the sense of Sartori (1970)): “Contentious politics involves interactions in which actors make claims bearing on someone else’s interests, in which governments appear either as targets, initiators of claims, or third parties” (2008, p. 5). Such a definition implies that, as far as Tilly is concerned, the concept connotes what matters the most about the pertinent phenomena. That the concept thus defined “brings together three familiar features of social life: contention, collective action, and politics,” is perhaps a further conceptual strength (Tilly, 2008, p. 5; see also Tarrow, 1998).

Just as Tilly’s approach apparently follows the position that key concepts refer to reality, however perspectival the concept formation might be, it features an inclination toward analytical thinking. His penchant for formalization is a conspicuous manifestation of this inclination. Formalization refers, in Tilly’s own words, to “a variety of procedures that match descriptions of events, structures, and processes with explicit models of those events, structures, and processes” (Tilly, 2016, p. 111). Throughout his career, Tilly developed many formalizations as a means to, or in support of theorization. In a well-backed argument, Krinsky and Mische (2013) demonstrate Tilly’s efforts at formalization and explain them as his evolving attempt to grapple with the challenge of understanding and explaining the role of the actor in historical processes. The authors also acknowledge that at the base of these efforts is the logic of analytic theorization. This logic, when it does not regard speculative theorization, puts a premium on “cross-sectionality”: the extrapolation of data from the contexts to which they refer and the slicing-and-piecing-back-together of that data in ways that make them stand apart from those contexts. It is a logic followed by Tilly when he breaks down the general concepts which he creates, which is to say, when he climbs down the ladder of conceptual abstraction. Elements of his theorization of contentious politics present a case in point. Thus, the concept repertoire of contention, which is one step lower on the ladder of conceptual abstraction than the concept of contentious politics, is taken further down the ladder by way of these three conceptual pairs: parochial/cosmopolitan repertoires of contention, particular/modular repertoires of contention, and bifurcated/autonomous repertoires of contention (Tilly, 2006a, 2006b). This path down the ladder of conceptual abstraction thereby yields taxonomic theorization. It would not be far-fetched to suggest, therefore, that Tilly’s approach here is to start with realist conceptualization to collect data, and then to employ analytical reworking for the sake of theory-building via the production of taxonomies. If this is so however, it should be kept in mind that taxonomic theorization is not the same as the theorization of the generation of change. And ultimately Tilly is interested, above all else, in studying social change and its drivers. Recurring though they are in Tilly’s work, formalizations remain a side-show.

The foundation of Tilly’s interest in social change is in fact so deep that his relational realism might as well be called dynamic relational realism. For as he argues, relations emerge in time and are most typically characterized by change, even though they can become durable under certain circumstances (and be labeled “ties”); they are the products of incessant transactions, of dynamic interactions (2005, p. 7). But what the convoluted label “dynamic relational realism” helps us see is not merely that Tilly’s perspective on ontology is realist, but also that his perspective on causality is realist. Whereas logical positivists such as Hempel (1965) and postpositivists such as King et al. (1994) hold that social-scientific concepts can refer to real social entities but that causal accounts can only be formalizations of empirical regularity with necessarily inexplicit reference to extant forces, Tilly’s causal accounts unreservedly refer to extant forces. Indeed, his perspective is what realist philosophers call ontological causality. Tilly’s is effectively a perspective that postulates that social relations constitute social events in partial but crucial ways and that social events emerge out of other social events, and then it calls for the explanation of emergence precisely via the description of it. It is this perspective that allows Tilly to argue that explanatory accounts are especially useful when they explain how phenomena unfold rather than why they unfold, when they appeal to the “mechanics” of emergence rather than to covering laws or some other presumed constants of human behavior or human nature (Tilly, 2016; Tilly & Goodin, 2008). It is in light of this perspective that Tilly’s notion of mechanism becomes central, exactly because it is tasked with sorting out the forces behind emergence.

I return below to the question of how mechanisms can “sort out” the forces behind emergence. But the first two ambiguities existing in the realist philosophy of science with respect to the notion of mechanism must be addressed. First, is the notion of mechanism relevant only to the extant forces featured in a particular set of emergent events, or is it also relevant to generalization? Second, even if the question is particular and not general, is the notion of mechanism to describe extant forces in all of their complexity, or is it to reduce their complexity?

On these matters, critical realism differs from scientific realism, at least Mario Bunge’s version, and it is useful to start by quoting critical realism’s originator, Roy Bhaskar:

The real basis of causal laws is provided by the generative mechanisms of nature. Such generative mechanisms are, it is argued, nothing other than the ways of acting of things. And causal laws must be analysed as their tendencies. Tendencies may be regarded as powers or liabilities of a thing which may be exercised without being manifest in any particular outcome (2008, p. 14).

Supporting naturalism, Bhaskar means to apply these assertions to social as well as to natural reality. Thus, a “thing” might be considered a social entity as well as a natural entity, and the generative mechanisms may be of human agency (in combination with those of nature) as well of those of nature alone. Indeed, the crucial difference between natural reality and social reality is for Bhaskar not ontological but epistemological: causal laws are discoverable with a reasonable degree of accuracy only in experimental conditions, which are beyond the reach of social scientists. Critical realism therefore conceives mechanisms as ontologies that are lawful at some deep level of reality, govern potentiality as well as actuality, are not necessarily in pace with the unfolding of events, and are not generalizable through social scientists’ investigations of social reality (Bhaskar, 2008, pp. 12–14). When it comes to social reality, therefore, critical realism provides for the analysis of singularly contextual causality and cautions against generalized causality. It stresses that social change emerges or can potentially emerge out of inimitable assemblies of events in contexts that feature multiple determination, over-determination, and plural determination (Bhaskar, 2015, passim but also 162 f. 23). The role of causal accounts is thereby to approximately describe the important way in which this has happened in given contexts, respectively. Such accounts can be developed by way of conjecture as well as more direct empirical evidence, but ultimately they aim to approximately reconstruct the tendencies (powers/liabilities), actualized or not, concerning a given specific set of events. Causal accounts, in short, regard intensive knowledge rather than extensive knowledge (Sayer, 2000, pp. 19–22).

Bunge’s take on the notion of mechanisms is different from the above. “I stipulate,” he writes, “that a mechanism is a process in a concrete system, such that it is capable of bringing about or preventing some change in the system as a whole or in some of its subsystems. Shorter: a mechanism is whatever process makes a complex thing work. In other words, a mechanism is the way a process proceeds” (2013, p. 21, emphasis original). In distinction to the critical realist perspective, therefore, Bunge’s notion of mechanism is about the ways extant processes functioning in concrete systems proceed, but not about the unrealized tendencies therein; it is about generative events, not generative powers. ¹ Furthermore, Bunge’s take differs from critical realism in terms of epistemology. Even though Bunge maintains that it is impossible to fully apprehend mechanisms, he argues that it is possible to conjecture about them and test the conjuncture against evidence. Mechanistic explanations are thereby theoretical reductions referring to kinds of processes, not only to given singular processes. The many examples he offers leave no doubt about this; they include some of Robert Merton’s social mechanisms, such as status ranking, distribution of power and authority, and creation of the private sphere—claims, then, about generalized patterns across contexts (2013, p. 18). In short, for Bunge mechanistic explanations regard the general as well as the particular.

Tilly mentions with approval Bunge’s approach to mechanisms but without comparing it to his own approach (2005, pp. xii, 135). After inspection, it transpires that Tilly’s notion of mechanism is similar to Bunge’s epistemologically but less so ontologically. Ontologically, Tilly stays far away from connecting mechanism to system; epistemologically, he shares Bunge’s aspiration for mechanism generalization. A mechanism for Tilly is “a delimited class of events and occurrences that alter connections among social units” (McAdam et al., 2001, p. 24; see also Tilly, 2001, pp. 25–26; Tilly & Tarrow, 2007, p. 29; Tilly & Goodin, 2008, p. 15). It is, in less technical terms, “similar events that produce essentially the same immediate effects across a wide range of circumstances” (Tilly 2003, p. 20). If the employment of the term “class of events and occurrences” in the first version of the definition indicates that at issue is the general rather than the particular, the less technical version of the definition makes it clear. Thus, Tilly seems to suggest that a mechanism regards the generalization describing analogous effects and the corresponding modalities of their emergence.

Also employed by Tilly, though not in a conspicuous or systematic way, is the description of the mechanism as a causal analogy (2002, p. 38; 2016, p. 75). This description affirms the generalized but “analogical” dimension of any given mechanism—that is, that a given mechanism’s identity from manifestation to manifestation is imperfect—but opens up a question about the “causal” dimension of it. Does “causal” refer to the force generating the immediate effects constituting the mechanism or to the capacity of those immediate effects to generate extended effects? In all likelihood, Tilly meant the former. Nevertheless, it is important to underscore that mechanisms do have the capacity to generate extended effects, no matter how analogous or open-ended these may be. Indeed, accepting that mechanisms have such causal efficacy is what allows Tilly to put at the center of his approach the proposition that concatenated mechanisms can generate a process.

What must be clarified at this point regarding the distinction between a process and a mechanism is that, for Tilly, it is an epistemological distinction, not an ontological one. Ontologically speaking, both process and mechanism refer to events generating immediate effects, such that the same extant set of events can be referred to as mechanism by a given investigation and as the process by another. But while referring to the same kind of ontology, mechanism and process designate different units of analysis. Thus, on the one hand, the analysis of a given process is to be done via the analysis of the mechanism concatenation constituting that process. Here, the process is what is to be explained, and the explanation is the description of emergence. On the other hand, the analysis of a given mechanism is to be done via the analysis of the concatenation of mechanisms of which the given mechanism is a part. Here, what is to be explained is not connected to the emergence of a named process but to other, less clearly formulated outcomes. If the emergence of a given mechanism were at issue and were the main issue, the mechanism would have been called a process instead of a mechanism.

We have seen, then, that in Tilly’s approach to sociology, one finds crucial realist positions but also analytical thinking. His key concepts are realist but formalizations creep up during some moments of his theorization. And while the mechanistic explanation he proposes for any given process is realist, explaining the way of emergence, he also aspires to mechanism abstraction. This tension between the particular and the general is explored below with respect to comparative contentious politics as well as to the notion of mechanism.

Mechanism and Comparison

The program of contentious politics studies has encompassing concerns going well beyond epistemology, a broad approach set in by the program’s founding text (McAdam et al., 2001). The overarching argument in that text is that phenomena such as revolutions, social movement campaigns, labor strike waves, and the like are generated by shared forces, which is to say, kinds of social and cultural patterns that are not particular to each of these phenomena but are rather common among them—hence the need for the term contentious politics and for a research program that compares widely across these phenomena. Nevertheless, as McAdam et al. (2001) go on to maintain that important generative social and cultural patterns be couched in terms of mechanisms and processes, the comparative program becomes one about how mechanisms concatenate to generate processes, and about how processes concatenate to generate episodes of contentious politics.

However, this program and the meta-theory that justifies it ushered in a certain controversy which must be cleared before the program’s epistemological possibilities and limits can be contemplated. The controversy surrounds the expectation that neat substantive findings about the mechanism–process relation are possible. Thus, reviewers of McAdam et al. (2001) complain that the analyses developed in the book fall short of the rigorous mechanistic analysis seemingly promised. Their arguments point to analyses in which the links between identified mechanisms and processes appear weak, and relatedly to ones in which the same processes appear to be connected to different sets of constituent mechanisms (Earl, 2008; Koopmans, 2003; Oliver, 2003; Rucht, 2003). The assumption on which these reviews operate seems to be that extant mechanisms and processes are to fit each other neatly, as in the way Russian dolls fit each other. However, while the manner McAdam et al. (2001) frame meta-theory might have fed this assumption, in practice the authors themselves do not seem to consider it credible. Rather, the tacit assumption under which they analyze episodes of contention is that of multiple realizability, holding that the same kind of effects can be generated through different pathways. This means that the same process can be generated by different sets of mechanisms across episodes of contention. The further assumption that similar pathways may generate divergent effects might have been tacitly present in their thinking as well.

What can be added to this is that, at the level of epistemology, the generative relation between mechanisms and processes must always remain incomplete. Thus, while Tilly is clear to suggest that the emergence of a given process must be explained by reconstructing the concatenation of mechanisms generating the process, he is not clear on the efficacy of such explanations. What needs to be submitted clearly is that no account of mechanism concatenations can exhaust all the elements generating the process. The same way that a description of a mechanism is a reduction for Bunge, the mechanism concatenation generating a process must also be considered a reduction, an emphasis on some key elements of “the way of the process” at the exclusion of secondary generative elements. Clearly, this is what Tilly does in practice; for in any analysis of a process he develops, it transpires that the generation of the process is the work not only of the mechanism concatenation which he specifies but also of ancillary elements which he also presents—contingent factors, background conditions, singular events, and so on.

So, let us spell out what is in between the lines of McAdam et al. (2001): to expect that the emergence of a mechanism or a process remains modular across circumstances is too much to expect when in question is the social reality of contentious politics. To be sure, different realms of reality may lend themselves differently to the study and theorization of modular emergence. Where systems set strict parameters for mechanisms, modular operations may exist (Demetriou, 2012). Biologists, for one, seem to expect molecular mechanisms to operate in a similar fashion from “start to finish” across manifestations (Craver, 2006). The social scientists who focus on stable and well-bounded social entities, such as formal organizations, may as well consider operations in similarly conceived mechanistic terms. For example, a tradition in program evaluation research understands mechanisms as the ways effects are produced through programs and interventions, which then allows it to search for configurations within bounded sets of interaction (Pawson & Tilley, 1997). But the ramified realms characterizing contentious politics resist attempts to standardize the emergence of either mechanisms or processes, as the critics of McAdam et al. (2001) demonstrate.

Where does this leave comparative contentious politics? To start answering this question, let us return to Tilly’s least technical version of the definition of mechanism. To recall, in this version, a mechanism refers to “similar events that produce essentially the same immediate effects across a wide range of circumstances.” Note that while the given effects are to be “the same” across circumstances, the given events are to be “similar.” Thus, Tilly emphasizes the regularity of the effects (emphasis with exaggeration, since the effects can never have the same form) and downplays the regularity of the events. Instinctively, it seems, he assigns gravity to the effects, and this instinct may well be at play when he names and defines the various mechanisms he employs. For the names of these mechanisms are the names of the effects alone, while the definitions essentially state that this or that kind of effect are produced, indicating nothing about how production happens. For example, the mechanism actor constitution, which is named by virtue of the effect “actor constitution,” is defined as “the emergence of a new or transformed political actor”; and the mechanism coalition formation, which is named by virtue of the effect “coalition formation,” is defined as “the creation of new, visible, and direct coordination of claims between two or more previously distinct actors” (Tilly & Tarrow, 2007, p. 216).

What I suggest, therefore, is to take the guesstimated instinct of Tilly to its logical conclusion. That is, drop completely the expectation that what constitutes any given mechanism are modular events, that is events that are to retain a standard outlook across the mechanism manifestations. Rather, hold that the effect defining the mechanism recurs in an analogous form across mechanism manifestations and that it has an emergent outlook that cannot be specified in any general way other than the recognition of the quality of emergence. The last part of the assertion is not trivial, for it is about the recognition that effects are typically generated in the overdetermined and temporal ways indicated by the realist philosophy of science, and not in some “billiard-ball fashion” where the effect is matched by a specified cause. (Neologisms seldom stick, but I have dared to give a name to such a notion of mechanism: m-outcome [Demetriou, 2012].)

To be sure, a notion of mechanism so conceived seems far less ambitious than one packaging effects and modular constituent events; what it amounts to may seem little more than just the naming of effects. Nevertheless, I maintain that the comparative program in contentious politics does give it purpose.

For one, the naming of effects is no small feat when a research program manages to delineate many of them, develop consensus on their definitions, and call them in empirical analyses. And this has been to a considerable extent the case within the contentious politics scholarship, even though it does not think of mechanisms as effects. The list of mechanisms and processes compiled in the appendix of Tilly and Tarrow (2007), which at the end of the day is a list of effects, is one with the antecedent and subsequent history of employment in the field, indicating a remarkable achievement notwithstanding occasional disagreement on definitions, labeling, and the like. Mechanisms and processes from this and other sources have provided useful tools to the researchers of contentious politics, supporting their ability to apprehend, analyze, and narrate. Tilly’s often-used term “robust mechanism” may be taken to refer to effects that research finds recurrent as well as salient (Tilly, 2005, p. 138).

Furthermore, the notion of the mechanism proposed here does not change the original epistemological intension of comparative contentious politics. Thus, Tilly’s epistemological distinction between mechanism and process holds under this notion of mechanism, and the task remains to explore how mechanisms (as effects) concatenate to generate processes. What is now different is the dissolution of any expectation that modular events would generate the same mechanism across episodes, and that the same concatenation of mechanisms would generate the same process across episodes. The critics of the research program on contentious politics may now complain about a program with modest epistemological promise, but not about one which cannot deliver on its promise. Rather, the overarching epistemological question remaining at this stage is how to tweak the comparative study of contentious politics so it does not become theoretically aimless. It is with this concern in mind that three sets of considerations about comparison are briefly discussed below.

Mechanism Types, Mechanism Concatenation, and Potentiality

Of the three sets of considerations, the first regards potentially useful delineations of mechanism types. A taxonomy proposed by McAdam et al. (2001) will be discussed in particular, revisiting its logic and questioning its utility. The second set of considerations regards mechanism concatenation. Above all, the topic discussed is the distinction between mechanisms precipitating a process and mechanisms constituting a process. Tilly only hints at this important distinction, leaving under-discussed pertinent dimensions of it, such as the element of temporality in concatenations. The third set of considerations regards potentiality, which at the level of theory is a topic eschewed by Tilly altogether, though it may appear in his concrete historical explanations. This is a topic worth considering as it is central to realist ontology and epistemology.

McAdam et al. (2001) distinguish among relational, cognitive, and environmental mechanisms. They note that these distinctions only amount to a tentative typology, and in fact they hardly discuss it, but it can be cut through succinctly here. An important clarification regarding relational and cognitive mechanisms can be readily made drawing from the preceding discussion. The authors do not specify whether what distinguishes relational and cognitive mechanisms regards effects alone or the package of effects and constituent events. For example, are we to expect relational mechanisms to be defined only by effects showing relational change or by a combination of such effects and constituent relational operations (transactions, interactions, etc.)? The preceding discussion suggests that it is untenable to expect the defining characteristics in this typology of mechanisms to regard effects and constituent events together. For that reason, and as per Alimi et al. (2015), the mechanism better be simply considered relational when the effects produced are of relational change (e.g., the mechanisms of social appropriation, uprooting, and diffusion), and cognitive when the effects produced are of cognitive change (e.g., the mechanisms threat attribution, legitimization, and polarization) (pp. 29–30, 175). Accordingly, it does not become conceptually problematic if, say, a cognitive mechanism is generated by unfolding relations. Where a conceptual challenge does exist, however, is with those mechanisms which are characterized by effects combining cognitive and relational change. The mechanism boundary formation, about which more is said below, is a case in point. ²

The same clarification applies to environmental mechanisms, but here there is an additional difficulty. While Tilly and his coauthors might have conceived this type of mechanism as a better way to conceptualize factors previously placed under the terms “opportunity structure” and “resource mobilization,” it is unclear to where exactly they direct their attention. That is, from McAdam et al. (2001) and from what Tilly says later about the book (http://understandingsociety-media.s3.amazonaws.com/tilly%20interview%20-%20Medium.m4v), it is unclear whether environmental mechanisms are defined by changes that are material or by ones operating in an environment conceived as one somehow enveloping the episode of contention. The example of resource depletion given by McAdam et al. (2001) does not fully clarify the matter, though it may indicate materiality more than the environment. What can be said here is that, if the mechanism is about changes in interaction with material elements, then we deal with a definition operating at the level of ontology (though not deep ontology). And if the mechanism is about an environment conceived as enveloping the episode of contention, then the definition operates at the level of epistemology and not ontology; for an environmental mechanism so understood is ontologically reducible to (social) relational mechanisms, with the only difference being the point of view of the research. In this case, the typology of mechanisms at hand does not operate on consistent differentia, since ontology differentiates relational from cognitive mechanisms and epistemology differentiates relational from environmental mechanisms.

This typology, therefore, may not be the best guide for the development of theory. More fruitful may be other ways of drawing distinctions among mechanisms. For example, distinctions can be made based on different realms of interaction, epistemologically determined. This is exemplified by the notion of environmental mechanisms understood as a family of mechanisms indicating changes in the environment of contentious actors. As Alimi et al. (2015) suggest, other similarly conceived families of mechanisms may, respectively, regard interactions among actors in a given movement, interactions between such actors and opponents, and interactions between such actors and the public. The theoretical shortcoming of this sort of distinction, however, is that the mechanisms featured in each family are not mutually exclusive—hence no typology of mechanisms, strictly speaking. ³

Moving to the second set of considerations, regarding the concatenation of mechanisms, one moves to the heart of what is comparative about contentious politics. The issue is wide ranging, but if the aim is to be as realist as possible about concatenations, the challenge of understanding and describing generative causality is key. In general, the challenge has two parts. First, in the context of the study of a given extant process, the challenge is to describe the concatenations of mechanisms that are generative of the process, as opposed to concatenations of epiphenomenal or incidental sequences of events (what Bunge calls kinematic descriptions of processes). No ready guidelines about how to tackle this challenge are available, and the wisdom that gingerly builds through empirical expertise may ultimately be the best guide. Second, in the context of theory-building, the challenge is to generalize about generative concatenations concerning a given process when the evidence suggests that they vary considerably across the manifestations of that process. As mechanism concatenations typically entail developmental patterns, interactive patterns, constitutive relations, and layered relations, the positivist logic available for comparative study would not do. And as inferential logic cannot be based on statistical probability, generalizations need to be rather modest. Knowledge accumulation would therefore have to stem from an incessant dialog between empirical investigation and conceptual refinement. Thus, Tilly’s cautious wager that regularity is to be found at the meso level of interaction, where mechanisms typically concatenate, cannot be read simply as a wager to be proven right or wrong by accumulated evidence; it is not merely an empirical issue, but also a conceptual one. And Tilly’s writings are suggestive of further conceptual refinements on matters concerning mechanism concatenation.

Consider Tilly’s discussion of social boundaries, in which he distinguishes between mechanisms that constitute boundary change and mechanisms precipitating boundary change (2005, Chapter 9). A social boundary for Tilly involves both actors’ understandings of the boundary and social relations relevant to the boundary, which is why social boundary formation has to be considered a socio-cognitive mechanism rather than an exclusively cognitive or an exclusively relational mechanism. On the one hand, accordingly, mechanisms constituting boundary change regard change of the pertinent understandings, the pertinent social relations, or both. They include such mechanisms as boundary erasure, boundary activation, and boundary site transfer. On the other hand, the mechanisms precipitating boundary change are the ones that may potentially lead to such change as the aforementioned. They include such mechanisms as encounter, conversation, and imposition. Now, the distinction between constitutive and precipitating mechanisms is epistemological, not ontological. It makes sense given research revolving around social boundaries; if at the center of the research were, say, the exercise of power, then imposition might have been considered a mechanism constituting power exercise. And it needs to be said that, in his discussion, Tilly does not mean to suggest that the mechanisms constituting social boundary change necessarily constitute the mechanism of social boundary formation. Nevertheless, if social boundary formation became the topic of investigation—which would then lead us to refer to it as a process, not a mechanism—we could make a distinction between mechanisms that constitute the formation of the boundary and mechanisms that precipitate its formation. More abstractly therefore, what we have here is the distinction between mechanisms that directly constitute a process and mechanisms that otherwise and potentially drive the process. This is a distinction about the mechanism–process relationship, not simply about mechanisms.

Temporality is pertinent to the foregoing. In his discussion of social boundaries, Tilly suggests that mechanisms precipitating boundary change are temporary prior to mechanisms constituting boundary change. But I would argue that such temporal ordering is not necessarily sound and that it is in fact counterproductive to expect it to exist. One needs to remember that, exactly like the situation with mechanisms, it is a given set of effects that defines a process and not the way the process unfolds. For example, the process of institutionalization in contentious politics is defined as the “incorporation of performances and political actors into the routines of organized politics,” a definition that does not relate to the way such incorporation develops (Tilly & Tarrow, 2007, p. 216). The defining set of effects, however, is not necessarily something that emerges only at the last stage of the processual unfold. (Let us leave aside the problem of the process’s beginning and endless antecedent determinants.) Rather, it is possible, even likely, that in the course of a given process the effects constituting and thus defining the process are produced incrementally at different points in time through recurrent “pushes,” such as loop effects and the like. This means that mechanisms constitutive of the process and mechanisms precipitative of the process may coincide in time or be variously ordered in time. With respect to the aforementioned institutionalization process, for example, the mechanisms of defection (“exit of a political actor from a previously effective coalition and/or coordinated action” [Tilly & Tarrow, 2007, p. 215]) and demobilization (“decrease in the resources available to a political actor for collective making claims” [Tilly & Tarrow, 2007, p. 217]) may precipitate the process during any or all stages of the processual unfold; by the same token, mechanisms constitutive of this process, such as elections, can likewise have a varied temporal place in the processual unfold.

In short, temporal sequences concerning the distinction between constitutive and precipitative mechanisms is a matter of research examining effects. Indeed, more generally, what would clarify research and thereby help theory-building is precisely the clarification of the various sorts of effects and the ways they influence each other in any given extant process. How, if at all, do the many effects which together constitute a given process vary qualitatively? How, at any given point in a given process, are these effects constituted piecemeal from various directions, and how are they precipitated by other effects? The first part of the latter question connects constitutive mechanisms with the process; the second part connects precipitating mechanisms with the process.

The distinction between constitutive and precipitative mechanisms speaks to realist sensitivities, and the field is open for theoretical and empirical elaboration (see e.g., Dasgupta, 2017; Ylikoski, 2013). It may be the case, for example, that in the face of a given concrete process the distinction proves to be a continuum, with some mechanisms being directly constitutive, some less directly so, some only remotely so, etc. Of course, this distinction is not the only relevant one to draw regarding mechanism concatenation. In fact, even the very term concatenation could be broken down. Thus, when we say concatenation, are we evoking connections that, for example, reinforce, weaken, compound, simplify, accelerate, slow down, spread, converge, and/or shift effects? But it should be kept in mind that distinctions are easy to draw when way to analytic thinking is given. But if they are to be something more than logical exercises, they must benefit the understanding of processual emergence.

Finally, moving to the last set of considerations, regarding the issue of potentiality, one moves to critical realism’s turf. As seen, critical realism emphasizes that given situations feature potentialities as well as actualities, suggesting that social science pays attention to both. It is fair to say that Tilly’s work has not done so, and one can only speculate as to why the omission. That Tilly seems to have studied Bunge but not Bhaskar may have something to do with it, though his turn away from the structuralism which he had earlier embraced may be more relevant. Structural theories have long been the perspectives from which potentiality could be theorized most confidently, namely, as tendencies of structure. But of course, critical realism’s approach is not structural. Not only are the tendencies which it suggests topical and not general, they are also processual. They are elements coming together in time to form potentialities, no matter whether the length of time is short or long.

This simple insight that potentialities are part of processes can enrich the analysis of processes. While critical realism’s linkage of tendencies with powers/liabilities renders their empirical investigation very demanding, the typical processual analyses of episodes of contention can approach the question of tendency with less difficulty—if also with less promise. Decoupling tendencies and powers/liabilities, this approach can understand the tendency of the process simply as the momentum of the process. This implies that the operations generating processual effects at any given moment have such a footing as to make the future generation of these effects likely. Accordingly, if an analysis is confident enough to reconstruct “the way” of the process, it can offer a description of the processual momentum as a counterfactual, which is to say, a description of effects that were likely to be realized but ended up not realized. In addition, such analysis of potentiality can include not only the unrealized potential effects that point to the same direction as the realized processual effects but also those pointing to the opposite direction—counterfactuals, then, covering both “the more” and “the less” of what is produced. Of course, analyses in historical sociology and social movement studies have long discussed counterfactuals, thereby engaging the question of potentiality. What is uncommon, however, is for such analyses to draw explicitly from the mechanistic analysis of processes. ⁴

But while the attention to potentiality can provide comparative contentious politics with a new object of analysis, it is an object which does not lend itself easily to generalization. It is even more difficult to generalize about the potentiality of processual unfolds than it is to generalize about their actuality. For while the latter are singular in their path, the former regard many (imaginary) paths. No wonder that the structuralist efforts to tackle the mess of potentiality did not go far. Nevertheless, while comparative contentious politics left structuralism behind, it did not have to turn away from the question of potentiality. A modest take on it seems possible and worthwhile.

Conclusion

Charles Tilly’s knowledge of philosophy is not fully shown in his writings. His decision not to delve into the philosophical debates surrounding realism, the tradition in which he placed himself, was likely measured. In an interview he gave Daniel Little in 2007, he had this to say: “Too much of the debate in sociology today is unselfconscious philosophical disagreement. It is disagreement about epistemology and ontology rather than about explanation, and I don’t think that is such a great position to be in” (http://understandingsociety-media.s3.amazonaws.com/tilly%20interview%20-%20Medium.m4v). This article has taken a somewhat different view on the role ontological and epistemological discussions might have on explanation: while such discussions may indeed create a diversion from explanation, as Tilly suggests they have done, they may also clarify meta-explanation. This article has therefore endeavored to show how reflection on the realist philosophy of science may clarify the ways scholarship can explain and compare episodes of contentious politics.

To the extent that it stays close to Tilly’s approach, this scholarship uses mechanisms as tools for both explanations and comparison. However, while confident explanations via mechanism concatenations abound, comparisons of mechanism concatenations are fewer and less confident. This is especially so with comparisons aiming to offer generalizations about mechanism concatenations. It is not too far-fetched to suggest, in fact, that an interest in generalization of this kind hardly exists in this scholarship. Rather, this scholarship seems to operate on the assumption that knowledge grows less by extension and more by intension—entailing, as Abraham Kaplan puts it, that “relatively full explanation of a small region (of behavior, or of whatever the subject matter may be) is then carried over to an explanation of adjoining regions” (1963, p. 305). But the choice between extension and intension in knowledge growth, and indeed the choice between knowledge of the particular and knowledge of the general, is not simple. For even the more stanch devotees of particular knowledge and knowledge growth by intension would admit, concepts—those indispensable vehicles of knowledge—are transposable.

In concluding this article, therefore, a point about conceptualization that was left implicit in the article can be made explicit. That is, from a processual-realist ontological point of view there is no distinction to be drawn between social entity and social process. Tilly would likely concur, even though nowhere does he put it in these terms. Thus, even though social entities acquire a form that makes them appear stable (which is why we can name them), in reality processes recreate that form. This means that concepts refer to processually constituted social entities, and this is a position with implications for theory-building. Epistemologies leaning on analyticism (including constructivism) hold that conceptual refinement can happen without considering generalized causality; postpositivist epistemologies hold that conceptual refinement and generalized causality attainment are distinct endeavors, the first operating on the empirical realm and the latter on the theoretical realm. But for realist epistemology, these solutions are not good enough. Thus, both concept formation and general causality theorization are for Tilly and other realists’ exercises that answer to social reality, not to other theories. And these two exercises are for them fully interconnected, for to elaborate on a concept is to elaborate on the modalities constituting the happenings which the concept denotes. As concepts are not ontologies of statics, in short, conceptual development cannot but be connected to causal elaboration.

Therein is the key tension of comparative contentious politics: to generalize the ways of processes. Bracketing the events constituting mechanism, as this article proposes, is a strategic move forward. It enables a less problematic definition of a mechanism, reduces theoretical expectations, and enables research. But in the long run of this way forward, Tilly may be correct in evoking that the quest, if it is not to remain aimless, needs to be for abstractions of mechanism generation and of mechanism concatenations.