Combining transition studies and social movement theory: towards a new research agenda

This article addresses two central—yet insufficiently explored—characteristics of some social movements: i.) abrupt and rapid social mobilizations leading to ii.) the construction of novel political processes and structures. The article takes a novel approach to these issues by combining social movement literature and the notion of free social spaces with transition studies, which focuses on large-scale socio-technical transitions. This theoretical integration highlights the co-evolution between free spaces and societal transitions, and it is based upon complexity-thinking, which is essential to deal with non-linear dynamics. A key insight is that to enable bottom-up societal transitions, radical social movements need to proactively develop solid alternatives to existing societal structures, to be ready once a window of opportunity opens. This theoretical approach is empirically illustrated using the APPO-movement in Mexico in 2006.

In certain cases these sudden uprisings are followed by sustained mobilizations and the construction of novel and innovative political and social processes that clash with established political institutions. We have seen these processes in historical cases such as the French revolution in 1789 and the Russian revolution in 1917, as well as in smaller cases such as the Zapatista uprising in Mexico in 1994. A more recent example is the so-called Arab spring, in which the self-immolation of a street vender in Tunisia sparked a range of international revolts that in some cases led to the construction of new democratic regimes.
What unites these and other similar examples are two interesting characteristics that are relatively uncommon in a social movement context; they are examples of i.) abrupt, unexpected emergence of mass mobilization that did not stop at a few Bdays of rage,b ut also ii.) led to the construction of novel political structures and processes that were radically different from the established political system. Existing social movement theories often have great difficulties in providing an understanding of such innovative and often rapid transitions driven by mobilizations and involving threshold effects and scale-shifts. As I argue, there are three main and interrelated reasons for this, which are empirical, methodological, and theoretical.
Firstly, social revolutions and radical societal transitions of any kind are not very common as empirical phenomena, and they tend to occur in distinct contexts and in different time periods. This creates serious limitations for the possibilities of systematic investigations, as well as any comparison between different cases. Secondly, our established theories and methods are not well-suited to deal with complexity and emergence. For instance, it has become increasingly clear that general linear models and statistical variable-based methods are to little avail to understand non-linear dynamics (Abbott 2001;George and Bennett 2005;Stinchcombe 2005). As a consequence, scholars such as Tilly et al. (2001) and Foran (2005) have argued that new approaches are needed that move away from thinking of cause-and-effect as determinate inputs and outputs, and towards investigating mechanisms and processes in a way that also encompasses a non-linear relationship between cause-and-effect. Or, as Emirbayer (1997) has put it, we need to move from substantialist to relational thinking. Thirdly, in the social movement literature there is a persistent lack of connection between informal or clandestine small-scale forms of resistance and large-scale, organized mobilizations. These are often treated as analytically separate phenomena. In my view, these three factors together have contributed to a lack of understanding of explosive mobilizations and movement-driven radical societal transitions.
The purpose of this article is to take a novel approach towards these issues by arguing that a potentially fruitful way forward is to combine the social movement literature, particularly the notion of free social spaces (Evans 1979;Johnston 1991;Leach and Haunss 2008;Melucci 1989;Polletta 1999), with theories, insights, and concepts from transition studies (de Haan and Rotmans 2011;Geels 2002;Kemp et al. 1998) and thus approach radical societal change as a form of social innovation.
By comparison, large-scale technological transitions are relatively frequently occurring, and this rich empirical basis has contributed to the growth of well-developed theories dealing with various transition dynamics. This article argues that there are in fact strong parallels between technological transitions and societal transitions driven by social movements, which implies that there is much to gain by explicitly connecting theoretical perspectives from both these fields. Arguably, such an approach may enable one to address crucial and highly pertinent questions, such as: When do radical social innovations manage to breakthrough, change the overall social system and become the new mainstream? And when are such innovations channeled, domesticated, and stifled of their transformative potential and incorporated as institutional reforms, thereby reproducing or even reinforcing the very oppressive system they set out to challenge?
The disposition of this article is as follows; I start by broadly discussing the notion of free social spaces in the social movement literature by emphasizing the role such autonomous spaces play in fostering novel ideas that are in conflict with surrounding hegemonic ideologies. While this does provide a piece of the puzzle by explaining the emergence of innovative ideas and practices in society, we know surprisingly little about the relationship between such spaces and societal transitions. Therefore, I turn to the field of transition studies, which has incorporated the notion of complexity in its very core and focuses on how and when novel technical innovations that are fostered in niches manage to break through and radically change the overall socio-technical system. In this sense, the connection between small-scale processes within niches and large-scale transitions is well established in this field. Despite the apparent resemblances and the shared focus on similar dynamics, there have been no previous attempts to connect these two fields formally. After briefly introducing the conceptual apparatus of the transition studies framework, I use this framework as point of departure and take the first steps toward integrating and adapting it to a social movement context and terminology. To illustrate and concretize the theoretical framework, I use the APPOmovement (Popular Assembly of the Peoples of Oaxaca) in Mexico in 2006. This case is particularly interesting since it is a typical example of both a rapid and highly innovative societal transition from the bottom-up.

From free social spaces to societal transitions
Alternative ideas and deviant social practices that oppose those of mainstream society often grow in protected spaces, finding new solutions outside the competition from incumbent societal structures. There are many examples of such spaces through history, including such diverse examples as the nineteenth-century Parisian working-class cafe, the US southern black churches in the civil rights movement, and musical festivals for the white power movement. These spaces are not necessarily explicitly political, but often operate in the realm between the public and the private spheres, constituting areas of social interaction in which individuals reinforce mutual solidarity and experiment with alternative world views and social practices, partly protected from the gaze of the powerful (Leach and Haunss 2008). Students of social movements have long been aware of the importance of such spaces, labeling them, for example, free social spaces (Evans 1979;Polletta 1999), submerged networks (Melucci 1989), oppositional subcultures (Johnston 1991), social movement communities (Buechler 1990;Taylor et al. 1992), cultural havens (Fantasia and Hirsch 1995;Hirsch 1990), movement halfway houses (Morris 1986), dense subcultural networks (Diani 2013) and critical communities (Rochon 2000).
Whatever the name, scholars have emphasized the importance of these free spaces, describing them as clandestine incubators for mass mobilization and insurgencies. While the exact function ascribed to them varies in literature, some basic functions that are often mentioned are: i.) they offer protection from mainstream society and hegemonic ideologies and thus constitute a shielded space where new radical ideas, social practices, collective identities, and collective-action frames can develop, thus affecting how we perceive both new and existing problems, their causes and consequences; and ii.) they generate relations and social networks that people can draw upon to promote collective action and to disseminate their ideas to a broader audience. In some circumstances, collective action and protests may convey radical innovations that within these spaces have been growing and luring a wider audience, thus provoking a re-examination of existing values and creating pressure for change (Rochon 2000). As such, these free spaces have always been important in collective mobilizations, particularly under authoritarian regimes, but also in democratic countries where they often contribute to the strengthening of feelings of identity and facilitate spontaneous, informal mobilizations when dense organizational structures are not in place (Diani 2013).
But while the importance of such spaces has not been over-looked in the literature, surprisingly little is known about the relationship between these spaces and mass mobilization and radical social change (Leach and Haunss 2008). It remains unclear how these spaces contribute to the emergence of social movements and, as Polletta (1999) has argued, we lack a thorough understanding of how and why certain patterns of relations and ideas manage to produce full-scale mobilizations while others lead to unobtrusive resistance. Similarly, more research is needed regarding why certain movement innovations-such as certain protest tactics-manage to breakthrough and diffuse, while others utterly fail (Soule 2004, p. 303). These issues reflect a more general tendency in social movement literature concerning the lack of understanding of the link between clandestine and small-scale forms of resistance that occur offstage-what Scott (1990) famously referred to as infrapolitics-and the processes of overt, organized mass-mobilizations. These are often treated as analytically separate phenomena in the literature and we lack perspectives that focus on their interaction and co-evolution (Simi and Futrell 2009;Vinthagen and Johansson 2013).
Similarly, an adjacent issue concerns the difficulties of being able to draw a strict line between scale-shifts within social movements and when a social movement is driving a broader societal transition. For instance, when large-scale mobilizations suddenly occur and involve large parts of the population, is this a matter of a scaleshift within the movement or rather a societal transition where the movement may play a central role? This question has drawn much attention and also underlies the wellknown, and ostensibly rather arbitrary, historical/academic division between revolution studies and social movement studies (Foran 2005;Goodwin and Rojas 2015), which is arguably more of a reflection on academic convention rather than any natural boundaries among the phenomena themselves.
As I argue here, approaching societal change as a form of social innovation may prove to be a fruitful way to deal with both these issues. This implies that the shift from small-scale processes that occur within free spaces to large-scale radical societal change can be conceptualized as a type of transition. Here, I follow de Haan and Rotmans (2011, p. 92) broad definition of a transition as Ba fundamental change in the structures, cultures and practices of a societal system, profoundly altering the way it functions.3

Transition studies
Transition has been studied in various disciplines, approaches, and perspectives, but it has only recently been the object of study in its own right (de Haan and Rotmans 2011). In recent years, we have seen the emergence of a new interdisciplinary field, congregated under the name transition studies, which focuses on understanding and influencing socio-technical transitions. This field has deep connections with various perspectives such as innovation studies and the socio-technical change literature Grübler 2003;Hughes 1993), but is also heavily inspired by complexity theory (Loorbach 2010;Rotmans 2005;Rotmans et al. 2001) through the realization that complex problems-such as transitions-require a systemic perspective.
While the underlying theoretical framework of transition studies is general and by no means confines the concept of innovation to technologies, the bulk of literature in the field has so far been on technical transitions in a market context, such as the transition from horse-drawn carriages to automobiles (Geels 2005a), from physical telegraphy to electric telephone (Elzen et al. 2004), and from sailboats to steam ships (Geels 2002). However, scholars have argued that transition studies itself Bmight be on the brink of a transition, broadening its scope from the technological to the societal,^and thus opening up a variety of research (de Haan and Rotmans 2011, p. 91). There is in fact a growing body of work that focuses explicitly on social innovation and grassroots innovation (Seyfang and Haxeltine 2012;Seyfang et al. 2014;Seyfang and Smith 2007;Smith et al. 2014Smith et al. , 2013. These studies draw upon niche-development theory ) and focus on how networks of actors and organizations in civil society generate novel bottom-up solutions that respond to the interests and values of the communities involved. The focus often lies on social innovations on the border between the technical and non-technical, such as grassroots innovations in currencies (Seyfang and Longhurst 2013), sharing economy (Martin et al. 2015), rock n' roll (Geels 2007) and transition towns (Seyfang and Haxeltine 2012).
But despite that this field is clearly advancing and the boundaries between technical and social innovation are increasingly blurred, the study of grassroots-led innovations remains under-researched and has not yet been applied to social movements (Seyfang et al. 2014). My intention here is to extend this field by stressing the social in innovations and merging it with contemporary social movement theory, in order to achieve an integrated framework that is able to account for radical innovations and societal transitions driven by social movements. In the following section, I briefly introduce the conceptual apparatus of transition studies, which is later used as the theoretical basis when integrating and adapting this framework to a social movement context.

Introducing the conceptual apparatus of transition studies
The field of transition studies consists of a number of different theoretical frameworks that focus on different aspects of the transition process. Among the most established frameworks are the multi-level perspective (Geels 2002(Geels , 2005bGeels and Schot 2007;Rip and Kemp 1998), strategic niche management (Kemp and Rip 2001;Kemp et al. 1998;Smith and Raven 2012) and transition management (Loorbach 2010;Loorbach and Rotmans 2006). What unites these frameworks is an understanding of socio-technical phenomena as complex, entangled systems, consisting of various analytically separated-but interdependent-levels and subsystems. This implies that technologies are deeply connected with each other and with social and cultural ideas and practices in a seamless web. This interdependency often serves as an obstacle for the emergence of new innovations through, for example, path dependency and lock-in processes. 1 But in some occasions, it may also lead to radical transitions; when a new innovation manages to breakthrough in the overall system, leading to cascades that often have an impact on the entire system. In this sense, these frameworks move the focus from single objects, artifacts, and innovators (e.g., the idea of Bthe great innovator^) to systems, and transitions are understood as emergent system-level phenomena with certain autonomy with respect to their causes. Thereby it is meaningless to search for singular laws or specific factors in these processes. Instead focus lies on finding patterns, dynamics, and mechanisms that drive changes in socio-technical systems, and the understanding of such patterns may provide a basis for directing and influencing transitions (Loorbach 2010, p. 165). Accordingly, the transition frameworks typically build upon within-case analysis of specific cases of socio-technical transitions (or failed transitions). Based on these case studies, focus lies on developing typologies of transition pathways in order to explain how processes and mechanisms at various levels co-evolve, for instance; how specific structural conditions (or system state) lead to certain transition patterns, which in turn tend to end up in certain transition pathways. We elaborate more on these typologies later.
The multi-level perspective (MLP) is generally regarded as the most established of these frameworks and provides an empirically tested heuristic tool that has proven useful in analyzing relationships between social innovations and transformative systemic change. Since I use this framework as the basis when introducing the integrated theoretical framework below, I only briefly introduce the conceptual apparatus at this point. In short, the MLP analyzes transitions as a complex and multi-dimensional interplay between micro-, meso-, and macro-levels (Geels 2005a;Rip and Kemp 1998).
The socio-technical regime is the meso-level and consists of rules in the form of prevailing norms, values, technologies, standards, and infrastructure that guide the involved actors and limit behavioral patterns. This level is generally stable due to the entanglement of technologies and social factors that lead to a dynamic rigidity. In other words, while incremental innovations (e.g., better-faster-cheaper innovations) can relatively easily be integrated within the existing technological regime, new and more radical innovations generally have a harder time breaking through because they may not fit with existing technologies and behavioral patterns. 2 The macro-level is conceptualized as the landscape and constitutes the wider exogenous environment that has an impact on socio-technological development. This includes both the material environment (e.g., industries, highways, and electricity infrastructure) and shared social and cultural beliefs and values. These are generally hard to deviate from and are beyond the direct influence of individual actors. Finally, the micro-level is the activities/practices occurring within niches that serve as an incubation room for new, path-breaking innovations that cannot yet compete with established technologies that are fully integrated in society (Raven 2006;Smith 2006).
Together, these analytically separated levels constitute a socio-technical system, and socio-technical transitions are understood as regime changes that occur when the interplay and linkages within and between dynamics at the different levels become connected and reinforce each other (Geels 2005c). For instance, when the existence of a strong socio-technical alternative that is fostered in niches is combined with an opening in the selection environment within the socio-technical regime, this may provide a window of opportunity where the previously hidden innovations manage to Bhit the surface.^Obviously, this also makes timing and the temporal sequence of events and processes a central factor in transitions. Furthermore, instead of explaining transitions using constant-cause explanations, these frameworks are based on evolutionary causality or Bcircular causality,^meaning that multiple dynamics within and between elements of socio-technical systems interact in feedback-loops-they co-evolve (Geels 2005b;Geels and Schot 2007).

Social change as a social innovation
This main thrust of the article is the argument that the transition studies frameworks can be extended to fit also the dynamics of non-technical societal transitions. In other words, by approaching radical societal change as a form of social innovation, we may draw important insights that may help us to understand innovative and rapid societal transitions better. As I argue next, there are several important affinities and shared dynamics at play within both these fields, which motivates such a theoretical integration.
First of all, a fundamental assumption here is that, similarly to socio-technical systems, social movements are complex adaptive systems 3 that consist of multiple interacting and interdependent actors and exhibit a range of typical dynamics such as tipping points, co-evolution, emergence, non-linearity, and feedback loops, where small changes can accumulate and cascade through the system as a whole (Fuchs 2006;Törnberg 2017;Törnberg and Törnberg 2017). Accordingly, uncertainties, emergence, 0 For instance, a new and more effective petrol engine is very likely to diffuse and to become integrated in existing regimes, since it fits well with other established technologies and does not challenge their underlying logic. However, a new and radically different type of public transportation, would most likely have a harder time breaking through because it may challenge the established socio-technical regime of automobility, which includes, e.g., existing institutions, organizations, and cultural norms, but also infrastructure such as roads, gasoline stations, sunken investments, norms, etc. In short, many cities are practically built in such a way that they require a car, which makes alternative ways of transportation difficult to implement. 3 I here follow the conventional definition of complex systems as non-linear systems, characterized by emergence and far from equilibrium (Byrne and Callaghan 2014). and non-linear processes are always important features of innovation and societal change. Such non-linear dynamics are difficult to deal with using traditional variable-based approaches (Byrne 2009) that typically focus on a Bpush-type causality ( Poole et al. 2000, p. 3) and account for variations in outcomes as the result of the influence of individual causal factors. Statistical approaches have serious troubles in dealing with causal complexity 4 and emergence, i.e., mechanisms that are contingent and context-dependent, containing interactions between elements that lead to structures that are not possible to decompose linearly into separate factors or covering laws. Therefore, the transition frameworks are, based upon process-based explanations and typological theorizing, aiming to identify different causal pathways that lead to transitions. Focus here lies on analyzing multi-level interactions between agency and changing environment, time and the sequence of events (Abbott 2001). Consequently, this enables explanations to incorporate layers that range from immediate to distant explanations, i.e., macroscopic and long-term processes and structural patterns can be incorporated alongside with immediate events (Geels 2010;Grin et al. 2010). While process-based explanations and typologies are hardly revolutionary news for social movement scholars, an important difference is that-as opposed to most social movement theory-the typologies and process-based explanations used within transition studies are built upon complexity-thinking, which is central in providing conceptual tools for understanding non-linear dynamics.
Secondly, there are no strict boundaries between the innovation dynamics of technologies and those of social movements: it is not the materiality of the innovation that matters, but rather dynamics such as the interaction of the incumbent-challenger relationship and the effects of exogenous shocks. Social movements and activists that fight for social and institutional change and entrepreneurs fighting for radical technical innovations are in fact facing similar challenges-they develop counter-narratives and discourses that struggle against the predominant normative and institutional configurations in society and the structures of meaning and power that they convey.
Thirdly and finally, both technical and non-technical radical innovations often develop in protected social spaces until they are ready to face the interconnected and rigid incumbent structures of mainstream society. Such spaces not only shield against market pressure, but they also provide a space to develop values, norms, issue framing, and societal discourses that affect how people perceive new problems, their causes and consequences. In this sense, clashes between niche-innovations and a regime may be a matter of clashes of ideas, values, and practices, rather than merely technical solutions.
As a whole, stakeholders in both cases need to build-up legitimacy, gather momentum, and construct networks of contacts with influential actors and political alliances in order to spread their constantly transforming innovations. In the end, both types of actors are dependent upon structural conditions whether they succeed or not. The shared affinities between these fields are also manifested in studies that show how social innovations may play a part in reshaping society into a more participative arena. For instance, Smith et al. (2014) investigate the intersection between social movements for democratization and movements that develop technologies for social inclusion in Latin America. They illustrate how radical groups use grassroots innovations for building and mobilizing resources and opportunities to achieve broader, local transformations. In these movements, the boundaries between the social and the technical are blurred, and technologies for social inclusion are analyzed as catalysts for broader social transformations.
Following this line of argument, there are clearly parallels between the innovation dynamics occurring within technical transitions and those within a social movement context. In the following section, I take the first steps towards a theoretical integration of these fields by translating key notions from the transition studies frameworks, particularly from the multi-level perspective, to also fit a social movement context.

Combining transition studies and social movement theory: towards an integrated framework of societal transitions
The MLP provides a pragmatic starting-point for an integrated framework since it is well-established and has proven useful for studying the relationship between social innovations and systemic change. I further extend this framework by focusing on the emerging literature on grassroots innovations and integrate these perspectives with the free social space literature. Thus, the notion of social innovation is here broadened to include also non-technical solutions and processes, and the concept of niches is translated to free social spaces. The new, synthetic theoretical framework is illustrated in Fig. 1.
To concretize and illustrate the framework throughout the text, I use the APPOmovement (Popular Assembly of the Peoples of Oaxaca) in Mexico in 2006 as an empirical example. This movement is particularly interesting since it is a typical example of a grassroots-based social mobilization that (partly) led to a bottom-up societal transition, characterized by novel political structures and processes. Since the main aim here is to illustrate the applicability of the conceptual framework, the empirical case description remains relatively sketchy here.
This section is organized in two parts. First, I introduce the conceptual framework, which is then practically illustrated using the case of APPO. Secondly, I show how these concepts are used and set into motion in order to understand and ultimately to explain transition dynamics, which are finally empirically illustrated through the same case.

Social innovation
First of all, I follow the definition coined by Avelino et al. (2014, p. 16) of social innovation as Bnew social practices, comprising new ideas, models, rules, social relations and/or services.^This broad definition refers to both new social solutions and to the new processes that are used to arrive at these solutions. Here the term Bnewŝ hould be understood in terms of a contested novelty-new in relation to the established solutions and processes. In short, it can be understood as changes in Bthe way of doing things.^Of course, this means that social innovations are not by necessity progressive in any sense; they can just as well be regressive and conservative. As opposed to technical innovations, which have the primary purpose of serving the market, social innovations have a broader purpose-to serve societal needs and to fulfill societal functionalities.
Social innovations emerge when established societal structures and institutions, here referred to as the socio-political regime (see below), are considered not capable of delivering satisfactory solutions to existing social problem such as poverty, exclusion, and segregation. In this context, we may further distinguish between incremental and radical social innovations. The former includes innovations that do not provide any fundamental challenge to the existing regime and also includes, for instance, various political reforms. While such reforms may indeed lead to political changes, they rarely lead to any fundamental changes in the underlying logic of the established regime. Radical social innovations, on the other hand, bear particular potential for transformative rather than incremental changes. These innovations typically exist on the outskirts of dominant institutionalized fields (i.e., the state and market) and often provide a bottom-up challenge to the hegemonic structures of the established regime in the form of how societal needs are addressed, and the processes that are required to arrive at them, including current social relations, dominant coordination mechanisms and institutional configurations (Pel and Bauler 2014).

Free social spaces
Grassroots movements support radical social innovaƟons (SI) within free social space.  Fig. 1 This graph illustrates the theoretical framework and the multi-level relationships among free social spaces, political regimes, and landscapes. SI is an abbreviation for social innovation. The graph is a remake inspired by Geels's (2005a)

model of innovation in socio-technical systems
The socio-political regime As indicated above, the socio-political regime is defined as the established political system that consists of the prevailing social practices, rules, norms, values, social relations, and political institutions. In this sense, the regime is the level that is replaced during a societal transition. The regime is generally characterized by stability, since the various parts are entrenched within co-dependent relations that lead to lock-in effects and path-dependency. In other words, we get locked-in to the Bcurrent ways of doing things,^and certain parts of the regime are often difficult to replace even though they may be suboptimal, because they are deeply connected and embedded within other parts of the regime.
Defining, delineating, and applying the concept of the socio-political regime in an empirical analysis is both an empirical and analytical challenge. The regime is a multilayered concept and there are no clear-cut distinctions between these layers in realitypolitical regimes are composite, nested systems and a local government can, for instance, seldom be neatly separated from the national regime since they are often based on relatively similar norms and social practices. Additionally, the concept serves two main analytical functions by i.) fixating the unit that is replaced or altered during a transition in a specific case, and ii.) pinpointing the various mechanisms and factors that may have an impact on such a transition. Thus, by necessity the term cannot be analytically exhaustive, and as I illustrate in the empirical case below, the level or delimitation one chooses for an empirical analysis largely depends on local, specific circumstances and the purpose of the analysis. This requires a pragmatic approach and awareness that certain aspects that are left out when defining the analytical concept in a certain case may still have an important impact in the transition process.
As noted above, incremental social innovations, such as reforms, tend to fit well within an existing regime since they do not radically challenge the underlying logic. Thus, political parties and leaders may often replace each other without any major friction. However, due to various stabilizing mechanism (that I elaborate on below), it is generally difficult to create and establish new, radical social innovations from within the socio-political system. So how do they emerge?

Free social spaces
In the framework suggested here, the importance of free social spaces as the locus of radical social innovations is emphasized and they correspond to the notion of niches. The radical social innovations that grow in these spaces are generally in conflict, or at least incompatible, with the existing regime. These spaces provide an incubation room for new path-breaking social innovations that cannot yet compete with the incumbent political structures and norms that are fully integrated in society. As Smith et al. (2010, p. 440) put it when referring to technical transitions, while Bchange within the regime tends to be incremental and pathdependent … 'revolutionary' change originates in 'niches.'^These protective spaces thus enable social and political alternatives to grow in performance and legitimacy and to develop new connections and ideas before they can compete in a more open, public way. 5 It should be noted that these free spaces can come in the form of both physical meeting places such as churches and cafes, but can also be structurally protected free spaces, for instance provided by linguistic codes that are opaque to those in power, which corresponds with Scott's (1990) notion of hidden transcripts. Finally, such free spaces may also transcend physical spaces, such as in the cases of social media and internet forums.
Based on insights from the transition studies literature and particularly the strategic niche management literature, I argue that free social spaces generally provide three main, interrelated functions (Smith 2006;Smith and Raven 2012). Firstly, they provide what transition scholars refer to as shielding: referring to the processes that hold at bay parts of the selection pressures from mainstream society that allows the innovation to grow. An early innovation can rarely compete on its own, not only because it needs time to grow in efficiency, but also because the mainstream selection pressures are adapted around mainstream artifacts. Within social movements, free social spaces serve as a shelter against both political repression and the hegemonic ideologies of mainstream society.
Secondly, they provide a nurturing function: referring to processes that support the development of radical innovation. In a social movement context this includes two somewhat different processes: i.) the development of collective identities, shared cultural values, and collective-action frames, which affect how we perceive both new and existing problems and their causes and consequences, and ii.) the enabling of a build-up of social networks among actors who are agitating for political alternatives and the coordination of activities based on these emerging alternative rules, norms, and perceptions. In this sense, free spaces generate social relations and connections that people can draw upon to promote collective action. This also includes mobilizing aspects such as developing strategies, sharing information, evaluating tactics, creating campaigns, and training leaders. Much of the focus in transition literature lies in Bexperiments^as the key for nurturing ), which has an interesting resonance with social movement theories. Free spaces in social movements are often referred to-in similar terms-as Bcultural laboratories^; spaces for experimenting with alternative world views, and Bto toy with unconventional ideas and experiment with new roles^ (Polletta 1999, p. 23).
Finally, free spaces provide an empowerment function. In transition studies terms, this comes in two forms: empowerment to fit and conform and empowerment to stretch and transform (Smith and Raven 2012). The first makes the innovation competitive with mainstream social and political practices in an unchanged selection environment (i.e., it transforms the innovation to fit into mainstream structures). The latter aims to undermine incumbent regimes and transmit bottom-up derived social innovations into regimes (i.e., to adapt mainstream society to the radical innovation).
In a social movement context, much of the work of revolutionary organizations seems to fulfill the role of stretching and transforming society in a way that enables a more radical transition. This is also one way to understand the role of everyday resistance (Bayat 1997a, b;Scott 1990) in large-scale mobilization-it influences by paving the way for more radical transformation, through undermining legitimacy and creating network ties. Similarly, empowerment to fit in and conform plays a role in the integration of Bnarratives of change^into mainstream terminology and adapts them for existing institutional structures and the hegemonic order. For instance, activists typically develop common scripts in response to the features of the institutions they confront: to speak to potential recruits, movements may need to use the language of Bcommon sense'^including parts that stand in opposition to their articulated ideology. This relates to the notion of frame alignment (e.g., Snow 2004;Snow and Benford 1988) and is, of course, highly central for organizations that are struggling for societal transformation.
Following the conceptualization of these three main functions, we may conclude that free social spaces can be described as clandestine incubators of revolt where radical social innovations in the form of, for example, new ideas, social practices, resistance repertoires, and conflicting values can grow under the surface, connect to each other, generate support networks, and gain momentum. The main task for the actors within free social spaces is to propagate these social innovations, and they play a central part when such innovations manage to reach a wider audience in the overall society. To understand when such revolutionary events may occur, we need to account for the structural context surrounding social innovations and the role of game-changing events.

Socio-political landscape and game changers
The socio-political landscape is the third heuristic level in the framework and constitutes the broader environment that has an impact on socio-political development. This includes various material/technical, institutional, and social-cultural factors that form a wider and relatively stable structural context for both the regime and free social spaces. Or put differently: landscape refers to the Brules of the game,^and constitutes the patchwork of societal systems in which the sociopolitical regime is embedded.
There are three important differences in how this concept is defined and used here, compared to the transition studies frameworks.
First, while MLP typically refers to the socio-technical landscape as various inherently exogenous and objective contextual factors (see, e.g., Geels 2005c; Geels and Schot 2007), I believe it is important to acknowledge that these structures and processes do not necessarily exist Bout there^as external entities, but are often, consciously or not, co-constructed by actors of change. As social movement scholars have argued for some time in relation to, for example, political opportunity structures (Ferree 2002;Gamson and Meyer 1996;Kurzman 1996), such systemic weaknesses are often difficult to assess objectively through scholarly methods. Rather, the analysis is a matter of interpretation and exploitation by various players in contentious politics (cf. subjective and objective opportunity structures). This means that the level of exogenity and temporal scale may differ across different types of landscape changes: some may be more exogenous than others, and different actors may interpret landscape changes differently. Consequently, the notion of landscape may include a variety of phenomena that are fundamentally different in kind, ranging from relatively objective and undisputable contextual developments like declining oil reserves and climate change, to more discursively constructed factors like ideology and cultural beliefs. However, they are united in that they refer to trends at the macro-level that are beyond the immediate reach of individual practices.
This brings us to a second important difference. While material, political, and institutional aspects of landscapes are well-developed in the MLP, the socio-cultural aspects have remained largely under-theorized. This may not be very surprising considering that the focus of transition studies has been on technological innovations and most attention has consequently been on various Bhard^factors such as material infrastructure and the entrenchment with other artifacts. Thus, while more Bsoft^factors such as culture and ideology undoubtedly also matter in the context of technological innovation, these factors have arguably an even more decisive role when it comes to social innovations.
Broadly speaking, cultural power describes the collection of established world views, beliefs, assumptions, and values in society that help defining the boundaries of common-sense Breality,^either by ignoring views outside those boundaries or by labeling deviant opinions Birrational,^Bunrealistic,^Btasteless,^or Birresponsible ( Lears 1985, p. 572). As such, cultural aspects are central in preserving and reproducing dominant institutions in society by justifying, legitimizing, and normalizing the existing institutions.
These factors are hard to place comfortably within any of the analytical levels in the framework. They are often simultaneously part of the broader selection environment that socio-political regimes and free social space actors must relate to, but at the same time they are also a central aspect of established regimes, constituting entrenched sociocultural arrangements that, similar to, e.g., material infrastructure in the case of technological systems, make novelties hard to introduce. After all, established worldviews, beliefs, and values are often what free space actors are up against and what they ultimately aim to replace with something different.
But while the relation between regimes and landscape is undeniably complex, it is nonetheless necessary to keep them analytically apart in order to study the relation between them. Existing regimes necessarily relate to and adapt to the set of (typically) slowly changing, dominant discourses and cultural beliefs in society, and in turn carry out institutional practices that reinforce and reproduce these discourses. For instance, liberal democracy and the notion of property rights are ideological beliefs that are wellestablished in most western countries and thus constitute a slowly changing and relatively exogenous context that is both upheld by and serves to legitimize political regimes. In this sense, it is not a linear but circular interaction between regime and landscape, as culture and hegemonic ideologies are typically embedded in institutional practices.
Accordingly, powerful and established groups have a lasting influence on the shape and meaning of culture in society at large, and it is generally in their interest to retain status quo and reproduce hegemonic cultural norms and ideologies since these are typically aligned with the interests of the established regime. However, as Weber (1978) noted, while these ideas was originally created to serve the powerful, they may come to have a life on their own, constraining rulers as well as those they rule and forcing elites to preserve their legitimacy by relating to these ideas. Thus, the dominant culture has elements that serve existing power relations and those that subvert them.
This implies that there is a certain ambiguity at play here. On the one hand, dominant culture and established political regimes tend to be relatively stable due to the entrenchment of beliefs, ideologies, worldviews and practices, and the fact that institutions and various material factors provide a relatively solid foundation for social patterns to lean on (Elder-Vass 2017). Cultural hegemony thus sets the boundaries of permissible discourse and discourages the clarification of social alternatives. But on the other hand, culture is not a static, unified system, but complex and partly contradicting processes, often riddled with gaps and inconsistencies that may create friction between the regime and the landscape. For instance, various societal processes such as growing class divisions and ethnic segregation may challenge and undermine the idea of liberal property rights. This means that there are always, to some extent, tensions and stress in the dominant order. Cultural hegemony can thus be understood as Ba process of continuous creation which, given its massive scale, is bound to be uneven in the degree of legitimacy it commands and to leave some room for antagonistic cultural expressions to develop^ (Adamson 1983, p. 174).
Returning to the notion of free social spaces, this conceptualization of the landscape further accentuates that free spaces do not only constitute a shelter against physical repression, but also a space for Bdreams of possibilities that lie outside political discourse^ (Mukerji 2014, p. 349). Activists in these spaces thus Bstruggle against pre-existing cultural and institutional narratives and the structures of meaning power they convey^ (Davis 2012, p. 25) partly through the construction of embryonic counter-cultures that violate the hegemonic cultural order and the prevailing common sense. In this sense, the line between dominant and subordinate cultures is a permeable membrane, not an impenetrable barrier, and the relative openness of the cultural context differs from case to case.
Third and more specifically, I find the complementing notion of Bgame changers ( Avelino et al. 2014) as useful in empirical analyses of social mobilizations. This notion refers to specific events that are perceived to change drastically the selection environment (i.e., alter the Brules of the game^). Often, but not necessarily, these game changers embody landscape developments, in the sense that pressure from the landscape finds a concrete expression through a specific event that is perceived, interpreted, and co-constructed by actors and stakeholders who draw upon these events in order to pursue their own agenda. These actors can be both regime actors defending the establishment and radical actors within free spaces fighting for new societal configurations. Often, these groups have different and directly opposing narratives of these events.
A typical example of a game changer is police repression: when a longstanding culture of systematic political repression culminates in a specific event of violent repression (which is narrated and framed by actors who are critical of the regime as a symbol of the broader repressive culture). In this sense, while the specific event may be performed by regime actors (e.g., the police), it may connect with and represent broader landscape developments. Other examples include the environmental crisis, the economic crisis, but also rumors and symbolic events that may trigger large-scale mobilizations. In this sense, this concept is related to della Porta's (2018) notion of Bsignals,ŵ hich indicates perceived cracks and vulnerabilities in the regime, thus providing windows of opportunities that movements may exploit. The close interconnection between such game changing events and how they are interpreted and exploited by free space actors is illustrated by the arrow going in both directions in Fig. 1. This relationship is of course reciprocal: major external events and certain landscape changes may also have a strong impact on existing movement discourses and narratives. An example would be when discourses that have existed for decades are triggered by events such as the financial crisis, thus provoking revitalized interests in these narratives.

Empirical illustration of the conceptual framework: the APPO case
Let us now exemplify and concretize these concepts using the APPO-movement as an empirical case. It should be emphasized that the purpose of the case is not to represent the entire framework, but rather to illustrate and concretize central aspects of it. The empirical material was collected during two separate fieldtrips to Oaxaca: one in the beginning of 2006 a few months prior to the uprising, and the second toward the end of 2015. The material consists primarily of interviews with various activists and leaders who were involved in the movement.
The socio-political landscape in the state of Oaxaca in Mexico has for long been characterized by extensive poverty and systematic discrimination that particularly targets the large indigenous population (Morris and Klesner 2010). Oaxaca and the neighboring states Guerrero and Chiapas belong to the three most impoverished states in Mexico. Eighty percent of the state's municipalities do not meet federal minimums for housing and education, and roughly half of the municipalities are considered to live in conditions of high or very high marginalization. The implementation of neoliberal economic reforms has further exacerbated the social and economic despair in the state, benefitting primarily the regional elite with connections to the tourism industry, while the large portion of the population that depends upon small-scale farming and agriculture is worse off now than ten years ago, resulting in the fourth highest out-migration rate in Mexico (Cohen 2004;Magaña 2010;Stephen 2013).
Regarding the socio-political regime, the political corruption in the state is palpable and has a history of seventy years of corrupt rule by the Institutional Revolutionary Party (PRI), as illustrated in Bishop Don Samuel Ruiz statement that: Bwhile Mexico's one-party system belongs to the history, Oaxaca still seems to be trapped in it.^The long period of rule has made the PRI practically entrenched with the governmental body, and the local government is infamous for its highly centralized structures, highlevels of corruption and electoral fraud, and for systematically excluding the indigenous population from political participation (Rénique 2007). Hence, there have historically been tense relations between the local government and civil society, in particular the large indigenous population, and these relations were further impaired under the corrupt and authoritarian regime of state governor Jose Murat (1996Murat ( -2000, and later under Governor Ulises Ruiz (2004Ruiz ( -2010. During this period, human rights organizations witnessed increasing political violence and repression in the state where civil society is particularly targeted (LASA 2008;Martínez Vásquez 2007;Stephen 2013). As a whole, this has led to increased polarization between indigenous communities and the local government, resulting in a relatively limited institutional and discursive space for developing concrete political alternatives to the established regime.
Parallel to the established political, governmental system, there are politically independent indigenous communities that are organized in municipalities (municipios). These communities have a long tradition of autonomy in relation to the state and constitute a form of free social spaces; providing the above mentioned functions such as shielding, nurturing, and empowerment and thus affording a protected and autonomous space where radical social innovations in the form of novel social and political structures and processes have been protected, developed, and implemented at a smallscale. A majority of the municipalities are governed on the basis of usos y costrumbres (roughly translated to Bcustoms and habits^): a euphemism to emphasize that the people as a whole exercise authority without electoral processes, but arrive at decisions in communal assemblies. In the communities, this includes a well-developed system of self-governance, which is protected by law in the Mexican constitution since negotiations after the Zapatista rebellion in the neighboring state of Chiapas in 1994. These alternative political and cultural systems exist parallel to, and often in direct conflict with, the governmental political structures and practices, and they provide different infrastructures, political institutions, value sets, and social practices compared with the incumbent governmental institutions. Throughout history, grassroots movements within these indigenous communities have advocated for these social innovations, which also have, to various extents, constituted a challenge to the established political regime in Oaxaca. While this has occasionally led to a series of political reforms, the lack of opportunities, poverty, physical distance, cultural norms, language barriers, and high political repression and militarization in the area have obstructed any broader coalitions between the communities, thus preventing any major breakthrough.
Consequently, while alternative values and ways of governing have grown protected within the communities, these spaces and groups have historically been sparsely connected through rather precarious social networks that are marked by political factions and relatively low-levels of cooperation (this corresponds to the scattered arrows at the bottom in Fig. 1). The relative isolation of these spaces has thus been both a necessary condition for the development of social innovations, as well as a barrier for their potential breakthrough. Noteworthy here is also the rural school teachers in the state, who have historically played an important role for the communities by serving as brokerages, both bridging the indigenous communities and providing linkages to the surrounding society (Hernández Navarro 2011). This group is a central factor behind the imminent game changer that came to initiate a radical societal transition.
On 1 May 2006, members of Section 22, the most radical teacher union, organized an annual protest encampment in the main plaza of Oaxaca and demanded raises and increased educational resources for poor pupils, but also urged bringing to a halt the increasing political repression in the state. The camp consisted of a few thousands of teachers, but not many other organizations were present. In the early morning of 14 June 2006, thousands of municipal police and troops from the Federal Preventative Police attacked the sleeping teachers and violently ousted them from the camp. While the protest itself was standard practice, the violent attack by the police became a game changer that contributed to the unification of the previously scattered organizations and communities and led to massive popular mobilizations with several hundred thousand protesters throughout the entire state.
Within just a few days, a large number of community-based associations, labor unions, women's groups, indigenous federations, left-wing political formations, student groups, alternative media, and artist collectives came together to form The Popular Assembly of the Peoples of Oaxaca (APPO). While APPO was first based on a few political demands such as an immediate halt of political repression and the steppingdown of Governor Ulises Ruiz, these basic, reformatory demands quickly grew into a full-blown political platform, where radical and subversive social and political changes were demanded. Thus what began as a rather specific trade union struggle, quickly emerged into something completely different. In this way, this event became a forceful symbol of injustice; an embodiment of the extensive corruption and longstanding culture of political repression against the civil society in the state. Of course, the grassroots actors in the communities were not merely passive observers in these processes, but they actively and consciously framed this event in order to illustrate the weaknesses and deficiency of the local government.
APPO called for the replacement of the hierarchical government with a system of multiple popular assemblies that are based on direct democracy and participatory processes on a local level, with no political parties and with a maximum of local autonomy. As the popular mobilizations spread, many districts in the state declared themselves autonomous and, during a number of months, the police, politicians, and military were banned from large parts of the city. Women's groups took control over the majority of the media stations and both governmental buildings and university buildings were occupied. The basic idea was to force Governor Ruiz to resign by way of preventing the institutional government from carrying out its functions through aggressive civil obedience.

The dynamics of radical societal transitions
The next step in the analysis is to use the analytical concepts introduced above in order to explain societal transition dynamics. Just like socio-technical regimes, socio-political regimes are generally characterized by general stability and incremental change. Reformative or incremental social innovations (e.g., political reforms) that fit within the logic of existing political structures and processes are often integrated within the current regime without causing any radical change. Transitions, however, occur when the linkages within and between causal mechanisms at the different levels introduced above hook up and reinforce each other, leading to bootstrapping processes and positive feedback dynamics (see Fig. 1), i.e., self-reinforcing processes.
The transition studies frameworks aim to develop theoretical typologies in order to explain how these mechanisms co-evolve and lead to certain general transition pathways (Geels and Schot 2007;Grin et al. 2010). These typologies then provide the theoretical basis for an analysis of transition processes and help us to know what to look for in these complex processes. By separating the system state that conditions the active social mechanism (referred to as patterns) that lead to certain manifestations in societal systems (the transition paths), this allows for a greater explanatory power (see Fig. 2). This also allows us to move from individual cases by making the analysis of First of all, a necessary and fundamental condition for any transition to occur is the combination of radical social innovations that are fostered in free social spaces, together with an opening in the selection environment within the socio-political regime. In other words, the regime gets destabilized, which opens up Bcracks^or windows of opportunities that grassroots movements may exploit to advance their social innovation. Such structural destabilization can be rooted in the relationship between the system and its environment, as well as the inconsistency or inadequacy from within the system itself. 6 In the case of the former, this is referred to in transition terminology as tension: when external events and circumstances create increasing pressure on the regime and thus there is a need to re-organize to fit the new structural conditions better (de Haan and Rotmans 2011) (see Fig. 2). This can be structural (e.g., physical, infrastructure, economical, formal) or cultural (e.g., cognitive, normative, ideological). Another factor for regime destabilization is stress within the regime; that is, that the regime is internally inconsistent and can no longer provide solutions to basic problems and social needs. A typical sign of stress is an inconsistency in what is done and what is preached (e.g., when the system is designed for one thing, but its practice is different; a typical example may be political corruption).
Tension and stress are in practice often inter-connected since external landscape changes and game changers can often highlight and reinforce existing problems and inadequacies within the regime. Here, free social space actors may play a central role in seeking to represent regime weaknesses to their own advantages, thus framing and exploiting such events in order to bring forth their social innovation and thereby increase the stress and tension within the system. These actors advocate upcoming, alternative social innovations that challenge the regime, thus also exerting pressure from below by providing alternatives to the current functioning of the regime.
These conditions of tension, stress, and pressure are necessary but not sufficient for transitional change. As illustrated in Fig. 2, they form conditions for the patterns that in turn produce societal transitions. But as studies of socio-technical transitions suggest, there is a strong internal tendency toward stabilization within the regime, and social innovations that perturb their deep structures often have their transformative potentials stifled and domesticated into incremental changes, thereby being integrated into the incumbent regime, unless there are intervening developments. There are of course many examples of this tendency in a social movement context, including the institutionalization of the radical green movements as they were transformed into political parties; when radical ideas and narratives are commodified and sold as products; and when citizen empowerment initiatives act as a Trojan horse for pursuing a neoliberal 6 Overall, these types of system perspectives may come with a certain structural-functionalist taste, which has been often and rightly criticized for overemphasizing maintenance, equilibrium, and homeostasis and thus being inherently conformist in nature and incapable of dealing with changes in social systems. However, the transition studies frameworks build upon a complex system approach that understands society as inherently innovative and capable of revolutionary change; societies are seen as peripatetic boundary-testing entities with internally generated innovative power. Hence, in the complex system approach, focus is moved from the notion of negative feedback (which was central in the early system theories) to instead accentuating positive feedback, which facilitates studying how systems can endogenously go through abrupt, radical, qualitative changes (Harvey and Reed 1996;Törnberg 2017). ideology (Swyngedouw 2005). As a matter of fact, the field of transition studies suggests that the default evolutionary course for social innovation is indeed incremental change and thus system reproduction (Geels and Schot 2007;Pel and Bauler 2014).
Accordingly, most transition patterns that are illustrated in Fig. 2 are indeed incremental and generate various forms of institutionalized innovations. For instance, reconstellation represents top-down change where an existing (or a newly emerging) societal constellation gets power by influence from outside the societal system. An example is when a foreign country invades another country and replaces the existing regime with another, or when an authoritarian regime is pressured from the outside in order to implement democratic reforms. This typically leads to a top-down transition path. Adaptation represents internally induced constellation change where an existing regime adapts in response to conditions for transitional change, which often paves the way to a transformation path. An example is when a military regime responds to a massive popular mobilization by implementing reforms or undergoing a transition to a hybrid state. A potentially more revolutionary process is the empowerment pattern, which represents bottom-up constellation change; that is, when a new social innovation emerges from below. As I empirically illustrate in the case of APPO below, a potential for innovative and radical societal transition resides within these grassroots-based social innovations and this pattern tends to lead to a bottom-up transition path.
Of course, in reality these transition patterns are often intertwined and may occur simultaneously. The classification of transition paths in Fig. 2 are thus based on the patterns that dominate in the course of a transition. Each of the transition paths have in turn a number of sub-paths and, based on meta-studies of a large number of case studies, de Haan and Rotmans (2011) have identified a total of eleven ideal types of transition paths. These paths can be both gradual and radical, depending on conditions and the timing and nature of multi-level interactions. But since the focus in this article lies on grassroots movement-driven social innovations and radical societal transitions, the most interesting path here is clearly the bottom-up transition path. Therefore, for reasons of clarity, I chose to include only the bottom-up transition sub-paths in Fig. 2, namely: i.) reconfiguration, when radical innovations in free social spaces are empowered by the regime and take its place, ii.) substitution, when innovations in free spaces are empowered, scaled-up, and replace a prior regime, without the help of the established regime, iii.) backlash, when innovations in free spaces initially gain power, but for some reason fail to become the new mainstream.
Finally, following the work of Avelino et al. (2014), I believe that it is fruitful in a social movement context to distinguish further between two types or levels of societal transitions: system innovation and societal transformation. The former refers to a Bprocess of structural change at the level of societal sub-systems with functional and/ or geographic delineations^ (Avelino et al. 2014, p. 15), while the latter refers to more fundamental and persistent changes across society that exceed specific sub-systems. 7 Examples of the latter may be the industrial revolution, democratization, female emancipation, and the abolishment of slavery. As Avelino et al. (2014) note, such rare, large-scale transformations are the emergent result of co-evolutionary interactions between local system innovations, new political institutions, and changing societal narratives. Of course, distinguishing between these two levels of societal transitions relates to the multi-layered nature of regimes mentioned above, and the fact that a local system transition may occur in one level of the regime (e.g., in a single province), but then another level of the regime (e.g., the nation state) might intervene and thereby interrupt a full-scale societal transformation. This implies that a full-scale social revolution within one country is indeed a difficult task due to the mutual interdependencies within the world system.

Empirical illustration: a bottom-up transition
Returning to the empirical case, the violent attack against the teachers in a transient way changed the power relations in the state. The tensions and stresses that had been relatively hidden were suddenly made visible and the explicit violation of basic human rights became an opening for introducing new ideas of public life-a new logic of power. Hence, it was the conditions of regime destabilizations in the form of tension and stress in the system (e.g., disrupting of the structural power of dominant institutions), combined with a pressure from the communities below that enabled the social mobilizations to grow rapidly and provided a window of opportunity for new, emerging social innovations from below. As argued above, these types of windows of opportunity do not generally open up by themselves as the result of some external, objectively given processes beyond the agent-dimension. Rather, it is often the actions from below that are the key in forcing such windows ajar: it is the conscious, enduring work of the players in contentious politics that opens and exploits cracks in dominant power structures. The conditions of stress, tension, and pressure within the system further increased as the groups and communities started to scale-up, get connected, and find common platforms. These processes linked up and reinforced each other-the more momentum within the movement meant that more meeting places were formed. 8 As a consequence, the social innovations (usos y construmbres) growing within the communities started to align. In this step, intermediate meso-level organizations often play a crucial role in providing coordination and spreading information. 9 APPO and the teacher union Section 22 provided these functions here. By serving as brokerages in the networks of indigenous communities and by increasing the interactions among these groups, they created a more dense and coordinated network of actors and laid the ground for aligning (action) frames, common values, expectations, and joint visions. In this way, they provided an important function by merging the various social innovations into a dominant design-a process that was accurately captured in APPO's slogan Bfrom many no's to one yes.^Consequently, the imminent social innovations advocated by these actors were transformed from a fragmented and small-scale existence to an increasingly concrete alternative and thus a growing challenge to the incumbent regime. This is represented in Fig. 1, where the many arrows are merged into one. By successively providing competitive functions compared with the established political regime, APPO was thus growing into an alternative, minor regime, thus initiating a societal transition. Thus, what first looked like a mere revolt, an explosion of popular mobilization against a hated governor, initiated a broader societal change-the rejection of a governor became the rejection of a political regime.

System innovation: towards a societal transformation
In early September 2006, APPO declared itself the governing body of Oaxaca, stating that it was the only legitimate government in the state and gradually came to control the three main governmental pillars of the executive, judicial, and legislative powers (Davies 2007). Parallel to this development, a large number of communities in the countryside declared themselves autonomous and ejected the previous local government. This period, which lasted for about five months, has been called BThe Oaxaca Commune,^in reference to the Paris commune of 1870 (Esteva 2010).
Clearly, the local government had a hard time dealing with these radical demands since they could not be properly implemented through incremental changes within existing regime constellations. The demands simply cut against the cultural and institutional logic of the current regime. Likewise, APPO refused to adapt to existing political and social practices, although there were in fact intense discussions within the movement as to whether to take a more reformative path by transforming the movement into a political party.
This prevented a transformation path from occurring, which would have domesticated the radical demands into institutional reforms. Instead, an empowerment pattern emerged, which lead to a bottom-up transition path in the form of a substitution path: when a small-scale social innovation is empowered by free social space actors and grows in order to fill the gap created by the retreating government. It is typical of such bottom-up transition paths that they do not involve the gradual gaining of societal support, but rather consist of a more abrupt form of scaling up. As the APPO was providing societal functions previously administered by the formal regime (e.g., garbage collection, policing, protection) and establishing alternative organizational and political structures based on ethics, values, and an understanding of democracy radically different from the established, they were initiating a system innovation (i.e., a local societal transition). An important factor here is that the national government of Mexico chose not to intervene at this stage, thus providing a space for the movement to advance. 10 Finally, on 28 October, the federal government gave APPO an ultimatum: Bhand over Oaxaca or we will take it^ (Denham 2008). Following this, the state and federal government moved in with massive lethal force and deployed thousands of troops from the Army, Navy, and the Federal Preventive Police, and took back control over the city after days of intensive fights on the streets. These events marked the beginning of the end for the movement.
While analyzing the processes behind the collapse of the movement goes beyond the scope of this article, we may conclude that although APPO was successful in performing a local transition (a system innovation), another part of the composed, multi-layered regime (the state) moved in and intervened, thus hindering a full societal transformation. This transition path is in-line with de Haan and Rotmans (2011), who argue that bottom-up transition paths such as substitution often risk a backlash-path. This is quite common when an emerging power gap is filled too quickly by an underdeveloped innovation, which is often unable to consolidate the fast growing demands.
In conclusion, the theoretical framework illustrated here addresses the two main features of social movements that we set out to explore-rapid and innovative transitions driven by social mobilizations. As we have seen, the radical social innovation that pervaded through, and was brought forth by, the APPO-movement did not emerge from the thin air-it had been fostered, developed, and realized in miniature within the indigenous communities. The breakthrough was enabled by a combination of structural conditions that were concretized and enacted by a game changing event that opened a window of opportunity for alternative ideas and practices. This is also the reason why the mobilizations in this case did not stop at a few spontaneous days of rage, but managed to build new, innovative political structures and processes. These processes were primarily built on a traditional indigenous assembly style with autonomous selfgovernance, both legally and in practice-they were born at the grassroots level from the deepest entrails of Oaxacan society.

Conclusion
The main thrust of this article is the argument that there are strong parallels between dynamics occurring within socio-technical transitions, and those of broader societal transitions that are driven by social movements. Consequently, the article paper is taking the first steps towards a theoretical integration of ideas, concepts, and theories from transition studies and the social movement literature. Returning to the three main theoretical, methodological, and empirical issues that are raised in the introduction of this article, such an integrated approach contributes by addressing all three of these.
First, this integrated approach provides an encompassing framework that explicitly emphasizes the role of free social spaces in large-scale transitions. This helps bridge the preexisting theoretical schism in the social movement literature between clandestine and small-scale forms of resistance and overt, organized mass-mobilizations. Instead of analytically separating these processes, this integrated framework focuses on how they interact and co-evolve, and under what circumstances radical social innovations that have been fostered within free spaces have the potential to initiate societal transitions. It is important to emphasize that the purpose here is not to replace existing theories in the field, but rather to provide a conceptual bridge that combines existing theoretical approaches. In fact, MLP was originally designed as a way of overcoming similar problems of separation within the technological innovation literature and to combine seemingly contrasting perspectives such as Science and Technology Studies (STS) and evolutionary economics ).
Second, the framework approaches transitions as emergent, systemic phenomena and uses multi-level theorizing, focusing on the co-evolution of mechanisms at different levels. This methodological focus fits well with recent trends within studies of revolutions and social change. An emerging scholarship in this field advocates looking beyond Befficient causation,^toward causal chains and sequences and to approach radical societal change as Bconjunctural amalgams of systemic crisis, structural opening, and collective action, which arise from the intersection of international, economic, political and symbolic factors^ (Lawson 2016, p. 109). Accordingly, there has been an upswing of interest in using analytical narratives and processual ontology to study idealized causal pathways in revolutionary processes, which bear striking similarities to the type of explanations advocated here (see, e.g., Fligstein and McAdam 2012;Foran 2005;Jackson 2006;Sewell Jr 2005;Tarrow 2012). Additionally, since the framework builds upon a complexity-sensitive terminology, this facilitates the integration of computer simulations that may enable us to unpack emergent micro-level mechanisms such as diffusion processes and feedback dynamics, and integrate these insights into the broader theoretical framework (for an example of this, see, e.g., Törnberg and Törnberg 2017).
Third, as noted in the introduction, radical societal transitions are relatively uncommon. By broadening the notion of social innovation and providing a common conceptual apparatus, this opens up the possibility of systematically comparing and generalizing certain innovation dynamics from one field to the other and constructing and employing causal typologies that are relevant within both societal transitions and sociotechnical transitions. The illustrative case study in this article has provided an example of how this can be done in practice, but a similar approach is arguably applicable in a broad range of cases, including, e.g., Occupy Wall Street, the Arab Spring in Egypt, and the rise of Podemos out of the Spanish Indignados Movement.
As a whole, it has been argued that the approach suggested here offers a promising way to understand when and why social movement-driven societal transitions occur and what pathways they may take. This pertains to important questions such as: Bin what circumstances do social innovations manage to breakthrough and lead to radical societal changes?^and Bwhen are they domesticated and channeled into institutional reforms?^While reality may not be as dichotomized as standing between revolutions or reforms, an important insight that has emerged from this study is that radical change does not happen by fighting the old, but through building the new. Thus, to spur and possibly to guide societal transitions in desirable directions, experiences from the field of transition studies assert that radical social movements need to be Bahead of the game^-they need to develop concrete alternatives proactively to the existing societal structures and be ready to exploit a window of opportunity. Free social spaces are indeed vital in these processes as they provide shielding, nurturing, and empowering functions and are thus a space where new social innovations may grow. This also explains why small, peripheral groups may acquire an unexpected influence in the course of a societal transition.
But finding such free spaces does not necessarily mean having to move to an anarchistic eco-village. These spaces are not by necessity isolated in a physical sense, but might just as well be a social haven where critical social movements are free to develop new stories and counter-discourses that go beyond the constraining narratives of the hegemonic regimes of today-stories focused on building the new, able to conjure new specters, to haunt the dream of the old.
While the result from this study should not be seen as a complete and delivered framework, it does serve as a demonstration of a possibility and a starting point that is sufficiently developed to provide possible avenues for further exploration. Future studies should use in-depth case studies in order to test in a robust way the theoretical framework suggested here and to develop it further.