The Privatization of Protest: Demobilization and the Emergence of Political Violence in the Aftermath of Local Protest Campaigns

. The processual approach to political violence suggests a close link between nonviolent and violent tactical repertoires. Yet in doing so it excludes cases where violence appears to appear in the absence of public protest activity. This article traces how political violence emerges in the aftermath of local protest campaigns against migrant accommodation. Developing the concept of the privatization of protest, the article shows how the demobilization of protest contributes to a process where grievances are reframed into private frustrations rather than objects of political contention. Trans-formed as such, persistent patterns of intermittent political violence can sometimes grow out of private interactions, even in the absence of any consistent public protest. Applying the conceptual apparatus of frames, emotions, opportunities and to a paired process tracing of episodes of protest against Swedish migrant accommodation in 2007-2008 and 2012-2017, the article maps the causal mechanisms that create facilitative conditions for violence, sometimes long after the decline of nonviolent protest. Extending its discussion beyond the case of Sweden, it links the processual approach to adjacent discussions on the link between micro-and meso-level causes of political violence.

Processual approaches to political violence have greatly improved researchers' understanding of how individuals and collective actors overcome the strategic, normative, and emotional limitations on using violence as a means to block or promote political change (Bosi 2021).In their current usage, however, these approaches have been most closely linked to a single temporal model: the gradual escalation of large-scale social movement campaigns.As valuable as insights into this basic model of escalation have been, its predominance has meant that theoretical development has been limited to a relatively narrow set of empirical circumstances.These approaches have thereby side-lined alternative temporal patterns that are clearly present in empirical cases of political violence.The continued dominance of the gradual escalation model thereby limits the full potential of a demonstrably valuable analytical approach.
Following a discussion of the relationship between processual approaches and temporality, this article develops its argument through the detailed presentation of one pathway through which violence emerges in the absence of an unbroken escalation process: the privatization of protest.In the privatization pathway, violence does not emerge immediately from within the dynamic of highly mobilized movement environments, but from the rhythm of interactions within the private, personal networks of friends, family members, neighbours and acquaintances.However, when violence happens, its facilitative frames, emotions, networks and considerations of political opportunity remain traceable to the initial phase of protest.In the empirical section, the article demonstrates how the notion of a privatization pathway helps explain why attacks against migrant accommodation centres in Sweden, despite occurring seemingly arbitrarily, often happened in communities and against targets that had already been singled out in prior protest campaigns.As the article concludes in the final section the pathway can also be applied analogously to a range of other contexts.The article thereby makes an empirically grounded contribution to the critical discussion on temporality in processual approaches to political violence.

From structure to process
Common reviews of the study of the relationship between nonviolent collective action and political violence read as stories of enlightenment (Bosi 2021;Bosi, Della Porta and Malthaner 2019;Bosi and Malthaner 2015;Della Porta 2008;Tilly 2003).In the first half of the 20 th century, the established narrative explains, researchers treated non-state violence as a pathological reflection of social disequilibrium (Smelser 1962), or as a consequence of manipulation and collective arousal (Le Bon 2001).Starting in the 1960s, a new generation of scholars broke with their predecessors by reframing violence as "politics by other means", meaning that it should be understood as a rational, goal-oriented activity on par with other methods of political participation, claimsmaking and coercion (Oberschall 1973;Snyder and Tilly 1972).In this transition, the macro-level explanations that had once framed violence as an irrational response to structural forces were gradually replaced by meso-level concepts such as political opportunity structures, cognitive frames, and resource mobilization (Della Porta and Tarrow 1986;Haines 1984;Tilly 1978).Beginning in the mid-to late 1980s, researchers gradually adopted a more dynamic and relational understanding of the forces that drive the emergence of violence, locating events within longer series of immediate and proximate interactions between protesters, authorities and third parties (Alimi, Demetriou and Bosi 2015;Beissinger 2002;Bosi, Demetriou and Malthaner 2016;Della Porta 1995, 2013).
The historiography that has developed around the study of political violence is in part a story of how time comes to replace the static space of structure as the core of the explanatory framework (see also Melucci 1980).Against the structuralism of the pre-1960s era, the new generation of social movement scholars made way for temporal arguments through the introduction of more dynamic, fast-paced structural variables, and the linkage of structure to the model of goal-rational, strategic actors (Oberschall 1973;Tilly 1978).The trajectory accelerated in the 1990s, as a second generation of processoriented authors criticized their predecessors for failing to part with what many perceived to be a lingering structural bias, particularly in relation to the concept of political opportunities (Goodwin and Jasper 1999;Suh 1999).As a response, a vast set of publications in the early 00s and 2010s emphasized the "dynamics" of political contention, and the "emergence" of violent as well as nonviolent repertoires out of gradually developing interactions between different actors involved in the exercise of political protest (for overviews, see Bosi 2021;McAdam and Tarrow 2011).

The paradigmatic status of the escalation model
Despite the proliferation of processual approaches, their theories and assumptions rely heavily on a single model of contention: the continuously escalating largescale, long-term social movement campaign.The primary reference points come from a number of hugely influential studies focusing on West European labour and student protest in the 1960s and 1970s and the trajectory of the US Civil rights movement in the late 1960s (Della Porta 2008).Largely based on these studies, and subsequent studies of similar empirical cases, many existing processual approaches have converged around a theoretical model consisting in competitive escalation (between movements and security forces), the closure of political opportunities for nonviolent protest, and the dual process of radicalization and institutionalization within the original movement.While later works have added important insights on the role of emotional and cognitive mechanisms (Johnston 2014;Johnston and Alimi 2012;Snow and Byrd 2007;Viterna 2013), or further specified analytical meta concepts (Lichbach 2008;Tilly 2001), the gradual escalation model arguably remains The Privatization of Protest: Demobilization and the Emergence of Political Violence in the Aftermath of Local Protest Campaigns paradigmatic within the field (Bosi et al. 2019;Della Porta 2008).On the rare occasion that other temporal models are mentioned, it is done without analytical or theoretical distinction.
The problem with the escalation model is not that it is incorrect, but that it's paradigmatic status greatly reduces the scope of temporal patterns that have long existed on the margins of mainstream social movement approaches to violence.Going back through the literature, there are many discussions of the situational causes and consequences of riots (see McPhail 1991), of the intermittent sabotage, obstructionism and small-scale violence that occurs in workplaces and in local communities (Piven and Cloward 1991), the processes that drive the institutionalization of protest in the aftermath of violent outbursts (Park 1967).Although it is true that these early studies often lacked theoretical sophistication and empirical rigour, the ascent of social movement studies did not add to, as much as replace, our understanding of the temporal models that previously had dominated research on political violence.
It is clearly possible to imagine other temporal patterns in the relationship between nonviolent collective action and the emergence of political violence.Violence can and does emerge before, after, and simultaneously with nonviolent collective action (Lundstedt 2023).It can occur entirely in the absence of nonviolent collective action, in which case there is no relationship at all.Even among cases where there is a causal link from nonviolent collective action to the emergence of political violence, it is analytically meaningful to distinguish between those where causality runs through an unbroken, continuous sequence of interaction (as in the escalation model), and those in which violence happens only after a break, or discontinuity, in the initial sequence of nonviolent collective action.
To explore one alternative to the escalation model, this article focuses on situations where violence emerges long after the end of any visible signs of collective action.In the existing empirical literature, it is primarily visible in accounts from post-conflict settings and in research on collective memory and the generational transmission of violent repertoires (Bosi and Della Porta 2012).In these cases, and in those recounted below, violence does not emerge out of immediate confrontations between protesters and their opponents, but from the effects and transformations of collectively transferred outcomes of past protest within the dynamics of private interactions among friends, family members, neighbours and others.Hence, although the contentious interactions that occur during non-violent collective action and the dynamics of small groups are necessary for creating the general con-ditions for violence, neither is individually sufficient to explain its emergence (Collins 2008;Nassauer 2016).To date, however, no study has attempted to capture these disparate observations within a framework that is comparable to the escalation model.This article does that by developing what it calls the privatization pathway to violence.The next section will use the framework of frames, opportunities, emotions and relations to develop the concept and to highlight its differences and similarities to the escalation model.The second half of the article then applies the privatization pathway to explaining cases of anti-migrant violence in Sweden between 2012 and 2017.

THEORETICAL FRAMEWORK
Building from the paradigmatic model of gradual escalation, this section develops an analytical framework for constructing processual accounts of political violence more broadly.It thereby employs a mechanismcentred understanding of causality, in which violence is understood as the emergent outcome of mainly nonviolent interactions that cause changes in collective frames, emotions, opportunities, and relations.Going beyond the paradigmatic model, the argument stresses the equifinality of these processes, or how different combinations of mechanisms result in multiple pathways to the same outcome.In the section's second half, and in the presentation of the empirical results, the framework is employed to develop the privatization pathway to violence, or the way in which emotions, frames, opportunities and relations that have developed in times of mobilization facilitate the emergence of violence long after protesters have demobilized.

Analytical logic
The approach developed here traces the emergence of violence through the identification of causal mechanisms, or comparable ways in which in which events "change relations among specified sets of elements in similar ways across contexts" (McAdam Tarrow and Tilly 2001: 25).When a specific sequence of mechanisms, with similar effects, occurs in similar ways across contexts, it forms into a causal process that may then be labelled as a pathway to a given outcome (e.g., radicalization, escalation, etc.) (Lundstedt 2023).Actors and their interactions are therefore at the centre of the analysis.In contrast to micro-level analyses, however, it does not focus on the immediate circumstances that generate violence in specific situations (see Collins 2008).Instead, the aim is to develop generalizable hypotheses about how interactive processes create conditions that make violence particularly likely to occur.
To construct generalizable models, the framework builds on the logic of causal analogy.In this view, causal mechanisms and their convergences in processes and pathways are not identifiable through their substantial content but through their effects on subsequent conditions (see discussion in Alimi et al. 2015).Because an event's effects are contingent on the wider context in which it happens, and because this context is shaped by prior mechanisms, the status of a given mechanism is affected by the timing with which it occurs (Bidart, Longo and Mendez 2013).Even when two events look outwardly similar, they are only analogous as mechanisms insofar as they occur in ways and in contexts where they can have the same causal outcomes.
Most processual approaches are open to the possibility of equifinality, or the assumption that the same outcome can follow from multiple sources.As developed above, however, this possibility is rarely employed in practice.This is particularly problematic when the goal is to explain highly contingent outcomes such as political violence, which occur across a wide range of temporal and spatial contexts, happen on different scales and with different intensities, and involve very different sets of participants, victims, and relevant third parties (Lundstedt 2023).It should therefore be emphasized that the presentation of the privatization pathway rests on the assumption that this is only one among many potential combinations of mechanisms.As will be developed in the following sections, much of the analytical work that has gone into the construction of the privatization pathway rests on identifying points of contrast vis-à-vis other pathways.

The privatization pathway
The privatization pathway describes the series of mechanisms through which the frames, emotions, opportunities and relations that have developed during moments of mobilization are maintained in such a way that they facilitate the use of violence even long after demobilization.It thereby differs prominently from the gradual escalation model, in which conditions for violence develop in step with the mobilization process itself.It also differs from other potential pathways to violence, such as the emotionally driven, uncoordinated violence that emerges from shocking events, the small-group dynamics of subcultural violence, and the strategic use of violence in autonomous cells and other specialist groups (Lundstedt 2023).
The concepts of frames, opportunities, emotions and relations grows out of social movement theory, where they capture the ways that participants think about a given grievance, what opportunities they attribute to different tactical means (e.g., nonviolent or violent), what collective emotions circulate among the protesters, and how the protesters are related to one another and to their opponents (see also Alimi et al. 2015;Della Porta 2013;McAdam et al. 2001).Depending on their specific combinations and on their timing relative to each other, these basic concepts allow for the formulation of multiple and equifinal pathways.The escalation pathway, as briefly outlined above, begins with the closure of opportunities for nonviolent protest and the fragmentation of the initial movement.These processes trigger growing competition between moderates and radicals, involving the emergence and growing salience of more radical frames, the elicitation of collective emotions that support risk-taking, and the creation of groups increasingly specialized in violent methods (cf.Della Porta and Tarrow 1986).Some of these mechanisms are also visible in the privatization pathway.However, whereas the escalation pathway develops through changes in an on-going process, the mechanisms that are involved in the privatization pathway need to account both for the decline of the initial sequence of nonviolent collective action, and for its subsequent outbursts of violence from within the context of private settings.In contrast to the smooth curve of escalation, it must account for three distinct phases in the episode's development: the initial mobilization phase, the transition into demobilization, and its aftermath.
The initial mobilization phase begins with the collective framing of a common grievance, the activation of coordinated networks and the identification of opportunities for nonviolent collective action.As in the gradual escalation pathway, the absence of emotions of shock and outrage, the optimistic understanding of political opportunities, and the possibility for leading actors to coordinate the development of common strategies, greatly lower the likelihood that some participants will take to violence already at this stage (Lundstedt 2023).
The transition from the initial phase within the privatization pathway begins with the closure of political opportunities, the proliferation of collective disillusionment, and the fragmentation of the protesting networks.At this stage, at the end of the initial phase, there is nothing that distinguishes the privatization pathway from cases of escalation or simple demobilization.However, instead of resulting in competing patterns of institutionalization and radicalization, or the end of collective action altogether, the frames that developed in the The Privatization of Protest: Demobilization and the Emergence of Political Violence in the Aftermath of Local Protest Campaigns initial phase of contention are transferred into the intimate relational settings of private personal networks (e.g., friends, family members, neighbours).Separated from public claims-making, collective action frames are transformed in ways that present grievances less as matters of policy, and increasingly as breaches with personal security and everyday experience (Snow, Cress, Downey and Jones 1998).The emotional tone of the participants' interactions shifts from outrage to a combination of demobilizing emotions such as disillusionment and fear (Johnston 2014).In the absence of perceived political opportunities, with frames that embed grievances in the context of everyday life, and through the elicitation of demobilizing but powerful emotions, the resurgence of nonviolent collective action becomes very unlikely.However, because grievances are kept salient within these smaller settings, they can still result in momentary outbursts of violence, e.g., through chance encounters, shocking events, or from escalated interactions among the participants (see Bjørgo 1997;Piven and Cloward 1991).The result is an uneven pattern of seemingly arbitrary violence in the aftermath of demobilization, which nevertheless draws its foundations from the preceding phases.Because the maintenance of grievances is separated from the policy process and from the development of the specific issue, and because they are embedded in collective memory, violent outcomes can flare up intermittently over very long periods of time.

DATA AND METHODS
The following analyses are based on two datasets on nonviolent and violent protest campaigns targeting the creation and maintenance of refugee accommodation services in Sweden between 2012 and 2017.The article's argument is based on the comparison of 14 cases in which the emergence of violence happened after a significant break in publicly visible protest events or other attempts at claims-making (e.g., letters to the editor, petitions, correspondences with policymakers).These are taken from a larger dataset of 86 episodes where violent events happened before, during, or after any occurrence of nonviolent collective action, without significant discontinuities in the targets or scale of protest, or among the participating actors (cf.Kriesi, Hutter and Bojar 2019).For the sake of brevity, the results section illustrates the argument through the detailed presentation of two cases that, in combination, demonstrate the breadth of applications of the privatization pathway.
Each case was reconstructed using extensive triangulation of mainstream media, social media, and far right movement sources.The cases were initially identified through a keyword search in the Swedish Retriever media database, covering roughly 2500 print and broadcast media outlets at the local, regional and national level.Further data (and a small number of cases) were added by way of a manual reading of the full websites and social media of the leading far right organizations at the time (the Sweden Democrats, the Party of the Swedes, Nordic Youth, and the Nordic Resistance movement).The final part of the data collection consisted in the tracing and reading of sources that were only relevant to specific cases, such as police and court records, local blogs, and locally specific (open) social media groups.Because the latter differed on a case-by-case basis, it is not possible to efficiently summarize the sources in full.
Upon collection, the data were organized into two parallel datasets.The first is a protest event dataset containing public, non-routine attempts to obstruct, change, or revoke the decision to place migrants into the host community.This dataset is used to supply quantitative measurements of the intensity, extent, and order of protest events, as well as the methods, actors, and targets that are involved in them.The second dataset consists of more fine-grained qualitative event chronologies that organize potentially relevant events by date and (if necessary) by hour.These chronologies include data on potentially relevant policy decisions (e.g. the decision to open a specific facility), reactions in local media and among residents, protest events, responses from policymakers, anti-racist actions, and so on and so forth (see also Bjørgo 1997).While the level of detail differs from case to case (depending on the type of events involved, and the level of local media coverage), making it improper to conduct traditional process tracing, the large number of cases greatly benefits the identification of causal analogies.
The construction of the privatization pathway is part of a comparative project that also includes an additional 80 violent protest campaigns where violence against migrant accommodation emerged through other combinations of mechanisms.Hence, the analysis is deeply informed by the contrasts that emerge between the privatization pathway and five others: the moral outrage pathway, the nationalist opportunism pathway, the gradual escalation pathway (as mentioned extensively above), the subcultural pathway, and the autonomous cell pathway.Whereas the moral outrage and nationalist opportunism pathways describe the brief but rapid emergence of violence in response to shocking events -with or without the opportunistic involvement of farright activists -the subcultural and autonomous cell pathways show how violence develops in the absence of wider protest campaigns, either from ritualistic smallgroup dynamics among friends or from the strategic activities of violent specialists (Lundstedt 2023).Briefly touched upon in the previous section and in the narratives below, using the other pathways as contrast points makes it possible to construct a more robust explanation of why the emergence of violence happens so late in the privatization pathway, as compared to the others.
The article takes several measures to ensure a balance between transparency and the treatment of sensitive information, particularly when referring to private social media accounts.First, while the locations are sometimes mentioned by name, there are no hyperlinks to specific conversations or comment threads.All observations of social media activity not attributable to organizations and public figures (e.g., policymakers) occurred in groups that were either open or semiopen (in the sense that they were searchable, had large numbers of members, and that anyone could become a member without qualification).Second, when translating quotes, care has been taken to alter or shorten full statements in ways that hinder identification through translation back into Swedish.If discussions were held in a format that made identification possible (e.g., in comment sections on websites where the users were already fully pseudonymized), the article uses full (translated) quotes.Overall, because of the specific ethical challenges that come with studying sensitive subjects that mainly involve private individuals (rather than collective actors), direct quotes have been kept to a minimum.The sacrifice of transparency and narrative detail, to the benefit of anonymity, should also be understood in relation to the article's intended contribution, which is primarily theoretical and analytical.
The decision to focus the empirical section on antimigrant violence in Sweden has methodological and analytical consequences.Regarding the former, the informal character of much anti-migrant protest, and the blurring of the lines between political claims and apolitical expressions of local conflict that often characterizes anti-migrant violence, makes it difficult to properly identify cases, events, and their interrelations (Blee 2009(Blee , 2017;;Klandermans and Mayer 2005).Without extensive use of triangulation, including accounts from many different types of actors, it would likely have been very difficult to develop a full image of the cases under study.On the other hand, the informal and local quality of much anti-migrant protest in the studied period (cf.Andretta and Pavan 2018; Lundstedt 2023) means that there is a larger number of sequences to observe and to analyse than would be possible on issues where people tend to mobilize at the regional and national level.
Regarding analytical consequences, it is important to note the relatively high level of conflict that characterizes migration issues.Although the likelihood that an individual sequence will result in violence is probably higher when concerning migration than many other issues, the core mechanisms should not differ.Antimigrant protest might be more likely to lead to violence precisely because it more often involves the mechanisms detailed here.For instance, the high level of politicization, and the topic's high level of salience in nationallevel media and on social media might support the maintenance of frames in the aftermath of local demobilization.It is also possible that the presence of racist and xenophobic frames increases the risk that the core mechanisms will in fact lead to violent outcomes.

THE PRIVATIZATION PATHWAY IN PROTEST AGAINST SWEDISH MIGRANT ACCOMMODATION
The privatization pathway traces how violence emerges from the context of private interactions in the aftermath of nonviolent collective action campaigns.It thereby develops through two phases: a first phase consisting of nonviolent, publicly oriented protest activity, and a secondary phase in which the same protests have turned inward, transitioning into the everyday context of conversation among friends, family members, neighbours and other private personal networks.As developed in section 3, the transition from one phase to the next is mediated by changes in political opportunities, frames, emotions, and relations.
Table 1 illustrates some of the main characteristics of the privatization pathway, based on the protest event dataset.The data show how the combination of mechanisms in the privatization pathway make for sequences that are often very extensive, and where the temporal distance from nonviolent collective action to violence can be very long.When combining the moral outrage, nationalist opportunism, escalation, subcultural, and autonomous cell pathways, the average duration from the first occurrence of nonviolent collective action and the final occurrence of violence against the same target is only 136 days.In the privatization pathway, the corresponding average is 403 days.In fact, the longest single case of the privatization pathway continues through intermittent acts of violence over the course of roughly three years (1099 days).Likewise, the average duration from a nonviolent collective action event to subsequent violence is only 23 days in the other pathways, but more than five months (168 days) in the privatization pathway.The data also reveal differences in organizational makeup and repertoires of action.Cases of privatization have limited involvement of far-right organizations, but a relatively large involvement of informal groups and individual participants who do not announce their involvement in any existing organizations (54% as opposed to 46%).Before demobilization, and before the emergence of violence, cases of privatization mostly involve small-scale, conventional, partially institutionalized methods (appeals, letters to the media, meetings and correspondence with politicians).Hence, the structure of nonviolent collective action, with its informal and largely local relational structure, and its small scale of collective action, partially reflects the structure that emerges after demobilization.
The following sections illustrate the mechanisms in the privatization pathway through two in-depth case studies that cover sequences of nonviolent collective action and violent protest in Tollarp and Ljungskile between 2015 and the summer of 2016.These are cases that illustrate the range of situations where the privatization pathway is visible.Although both campaigns ended in intermittent acts of violence, the extent and intensity of violence was higher in Tollarp than in Ljungskile.They are also cases where the quality of media reporting, the use of open social media, and the maintenance of relevant sources make it possible to develop in greater detail the day-to-day development of each sequence.Finally, they represent different combinations of contextual factors, with very different levels of support for anti-migrant parties in the 2014 national and local elections, different socioeconomic situations, and different histories of far-right organization.Tollarp is characterized by its very strong support for the Sweden Democrats, moderate median and average income, and a long history of far-right organizations operating in and out of the local community.In Ljungskile, the Sweden Democrats performed like the national average, and the community does not have a prominent history of local far right organization.On the other hand, Ljungskile has the highest average and median income in its surrounding municipality.Both localities are of roughly the same size (≈4000 inhabitants), and they are both located on the margins of municipalities with larger administrative centres (Kristianstad and Uddevalla).

The first phase
The first phase begins with the identification and framing of grievances, the attribution of opportunities for nonviolent protest, and the activation of relatively coordinated networks of protesters.The framing of grievances is necessary if the protesters are to make any claims at all, and if they are to successfully mobilize bystanders.The attribution of opportunities and the activation of coordinated networks of protesters are necessary for protest to remain nonviolent throughout the initial phase.Because the participants believe that nonviolent means are sufficient to achieve political change, and because the actors making this interpretation have some degree of control over the direction of the campaign, it is considerably less likely for violent strategies to emerge on the margins.If the latter does happen, however, the campaign transitions into a different pathway altogether, falling outside of the scope of this article.
In Tollarp, the first phase began in the fall of 2015.In a context of rapidly growing refugee migration into Sweden, the national-level Migration agency was pushing the municipal government was to increase its capacity for accommodating unaccompanied migrant youths.Because of a recent wave of arson attacks and other attacks against migrant accommodation facilities throughout Sweden and internationally, the municipal government decided not to announce its commission of an empty lot for the construction of a mid-sized residential care unit on the margins of central Tollarp.
In the beginning of December, rumours about what was happening at the lot were circulating intensively in local social media and among neighbours (see Tollarp's Facebook page).Responding to these rumours, a neighbour to the facility walked onto the premises and into the foreman's office, taking with him a copy of the building instructions.Immediately afterward, a small group of neighbours went to the local newspaper, entered discussions on social media, and wrote an open letter to the municipality in which they denounced the plans for the care unit, and the way that the planned facility had been kept secret from those living in the area.In the letter and in the ensuing discussions, the participants opted for a meeting with local political representatives and for an attempt to challenge the plans through a formal appeal (Maunula 2015).
The framing and interpretation of opportunities that emerged after the Tollarp plans had been exposed remained throughout the initial phase of the campaign.Focusing primarily on the municipality's failure to anchor the plans within the neighbourhood and drawing on themes of centre-periphery relations and the defence of property values, the participants remained optimistic about their chances to block or alter construction through formal channels.Either the plans could be stopped through conversation with local politicians, or through the exercise of the neighbours' right to appeal.Expressing emotions of outrage and disappointment, the participants in the social media discussions agreed that the municipal leadership in central Kristianstad failed to take the interests of peripheral areas seriously in the policy process (see Tollarp's Facebook page).Hence, the initial phase saw the development of a framing that centred on centre-periphery relations, highlighting the relationship between the proposed facility and the wider policy process.This coincided with the perception that political opportunities were open, and the elicitation of collective emotions conducive for collective action.Only to a limited extent did these initial interactions draw on xenophobic depictions of the youths that were meant to live in the coming facility, and it was rare to find expressions of fear in the various written materials that emerged during this period.
In Ljungskile, the object of grievance was the repurposing of a disused hotel into a home for unaccompanied minors that had originally been allocated to the municipality of Gothenburg.While most participants on local social media, as well as many writing in the local newspaper, nostalgically lamented the loss of the hotel, a small group of parents feared that the opening of a home for unaccompanied minors would threaten the safety of the children at a nearby preschool.Fearing the presence of what they referred to as "traumatized, psychotic migrants" as well as the risk of spillover from far-right violence, the parents doubted the legality of the project, acting through repeated correspondences with local policymakers and inquiries to responsible inspection authorities.Like in Tollarp, they also demanded the organization of a municipal information meeting where the neighbours could express their claims directly to policymakers, and they promoted a petition for other locals to sign (see Ljungskile's Facebook page).
The beginning of the first phase in Ljungskile saw a similar development of a policy framing, an optimistic interpretation of political opportunities, and the elicitation of mobilizing emotions within a limited network of protesters.The participants placed blame on the responsible authorities, they identified formal means through which they could potentially stop the project, and they felt disappointed and angry that their specific neighbourhood had been singled out for the project.Framing the opening of homes for unaccompanied minors either as an expression of unequal centre-periphery relations, or as a threat against children's' safety, the actors that tried to mobilize support for their cause presented the two facilities as concrete expressions of more general issues, although located in the middle of local, everyday experience.Publicly announcing the possibility to challenge the projects on legal grounds, and through direct deliberation with policymakers, both campaigns placed considerable stock in nonviolent and even institutionalized means, rather than more disruptive or nonconventional forms of protest.By mobilizing primarily within very limited and highly coordinated groups (the immediate neighbours and parents with stakes in a specific preschool), both campaigns contained very limited internal variation.These elements made possible the transition to the privatization of protest in the second phase.

Demobilization and the transition from the first phase
The transition to the second phase begins with the reversal of the attribution of political opportunities, the elicitation of collective disillusionment, and the subsequent fragmentation of the protesters' networks.The reversal of political opportunities and the elicitation of demobilizing emotions are tightly bound to each other, as leading participants experience and narrate disappointing encounters with decision-makers, as they note the scant effects of nonviolent collective action, or as they realize that they have overestimated their formal opportunities.
In both Tollarp and Ljungskile, the participants were soon disappointed with the lack of outcomes for their respective strategies.In Tollarp, this was visible in three ways.First, discussions on the local Facebook group soon concluded that no one among the participants had the legal right to appeal the plans (see Tollarp's Facebook page).Second, the requested information meeting turned into a disappointment, as the municipality decided to send professional civil servants rather than elected political leaders.As one person in attendance put it, it felt like "being run over for a second time".During and after the meeting, the local village association came out in defence of the plans, revealing the loss of a potential ally (Haraldsson 2015).In its aftermath, the information meeting came to be used as more broadly a symbol of the municipal government's disregard for the peripheral areas, even among people who wanted to advance migrants' right to stay in the municipality (Johansson 2016).Third, a petition that the participants had launched earlier in the fall largely turned out to be a failure, adding up to no more than a few hundred signatures, many of whom did not reside in the locality or even in its surrounding municipality (Skrivunder.com 2016).In Ljungskile, the municipal government did not agree to organize an information meeting (Hvittfelt 2016), and the initial petition only drew 200 signatures (see Ljungskile's Facebook page).In the social media groups, in letters to the editor, and in local media quotes, these events combined to reverse the participants' previously positive perception of political opportunities, making way for the collective expression of disillusionment.
The negative attribution of opportunities that happened in this period had remarkably similar effects on relational structures and outward activities in the two localities.Reports of publicly oriented activities ceased.Except for one petition that opposed the Tollarp plans as part of a wider range of construction projects for migrant accommodation in the Kristianstad countryside, there were no more examples of nonviolent protest in either town.Meanwhile, the number of individuals who participated in the discussions diminished rapidly, and participants in both locations gradually moved into closed Facebook groups.Examining those who remained in the public groups, it is possible to see a concentration to tight groups of family members and friends, something that is also reflected in the participants' references to real-life discussions (see Tollarp's Facebook page).Mirroring the fragmentation that went on in this period, public discussions also occasionally came to focus on what the participants saw as "yet other" closed groups, none of which appeared to result in concrete action (see Tollarp's Facebook page).Changes in the attribution of opportunity and in the elicitation of emotions thereby coincided with a turn toward smaller and even more informal relational settings.

The second phase
The fragmentation of protest networks and the demobilization of public protest campaigns do not necessarily make the salience of the grievances go away.
However, with the decline of public protest and the transformation of the participating networks, the way that grievances are framed changes shape.As most of the relevant interactions transition into the informal settings of friend groups, families, workplaces, and as the purpose of these interactions change from strategizing to informal exchange, they become increasingly tied to experiences of everyday life.Hence, the framing of grievances increasingly closes in on their effects on the participants' everyday existence.Rather than being framed as a threat against collective interests and values, the identified grievance comes to be seen as a threat to, or breach of, individual safety and everyday experience.
With the demobilization of protest in Tollarp and Ljungskile, visible interactions shifted in terms of framing as well as in their emotional content.Especially visible in Tollarp, comments on the second petition and in the public Facebook group suggested that the grievance, while ostensibly the same, had now been reframed as a threat to everyday routines and individual safety, and that this altered framing elicited a sense of fear among the participating locals.This transformation of the overall framing also involved the growing involvement of explicit xenophobic and Islamophobic themes, much as had been the case in the initial phase of protest in Ljungskile.As a commenter on the petition put it, I DON'T want my 12-year-old DAUGHTER to go to school with 18-22 year-old MEN who claim to be 16-17 years old.Because these MEN have been raised since they were breast-fed that women have no value and that their ONLY purpose is to birth MEN and make sure that the MAN has everything he needs.They also believe that the day a girl gets her PERIOD is the same day they become WOMEN and it's the prophet Allah's way of saying that this woman is READY for MARRIAGE and giving birth to CHILDREN.I had my period in fifth grade, 11 YEARS OLD!!!!! (Skrivunder.com, 2016, capitalization in the original).
Even among less explicit commenters, the sense of fear proliferated.Unlike what had been the case in December, there were no attempts to prescribe potential ways of blocking the facility.Instead, a lengthy exchange between two locals, both of whom had long been involved in opposing the plans, ended with a suggestion to move out of the neighbourhood for fear of personal security, and hopes that the winter would be long enough to stall construction beyond the planned opening date (see Tollarp's Facebook page).
The negative interpretation of political opportunities, the fragmentation and submersion of networks, the elicitation of demobilizing emotions, and the reframing of grievances as a threat to individual and everyday security, combines to form conditions where violent outcomes are especially likely to be seen as viable, accessible, and potentially desirable.As small and fragmented groups see no opportunities for nonviolent collective action, as they take part in the elicitation of demobilizing collective emotions, and as their frames move further away from the policy process, it is unlikely that they will make another attempt at conventional or institutionalized methods.On the other hand, the continued salience of grievances and the elicitation of fear maintain the desire for something to be done.These mechanisms can then be channelled into chance encounters with migrants and policymakers, in escalating interactions among peers, and in other settings where fear and disillusionment can provisionally be transformed into mobilizing feelings of anger and elation.
The months that followed the demobilization of protest included a range of intermittent acts of violence.In Ljungskile, these were limited to repeated acts of property damage and vandalism, including extensive use of racist graffiti (NyheterSTO 2016).Among the perpetrators were a group of teenagers without known ties to the far right (Norlin Persson 2016).In Tollarp, the attacks involved the use of a Molotov cocktail a few days before the facility's opening date, as well as a series of attacks against the other facilities that had been mentioned in the petition a few months earlier (Sörenson 2016).Although it is not possible, based on available data, to draw direct links between discussions on social media and the attacks, it is significant that these attacks occurred in a context where the framing of migrant accommodation and the negative attribution of opportunity had moved the former outside of the capacity of political claims-making and into a concrete threat to individual and everyday experience.

CONCLUDING DISCUSSION
While differing considerably from the paradigmatic model of escalation, the privatization pathway similarly provides conditions necessary for the emergence of violence.The closure of opportunities presents nonviolent tactics as ineffective, while the subsequent fragmentation of networks cuts off the practical conditions necessary for collective mobilization.The reframing of the grievance as an urgent threat to individual safety and livelihood, and the integration of these frames into everyday interactions, allow some participants to expect otherwise drastic options to be considered socially legitimate.As demobilizing emotions of fear and disillusionment are provisionally transformed into anger and elation, some will be moved to act on what those in their surroundings only speak of.
Throughout its two phases and their transitions, the development of the privatization pathway is dependent on the absence of contingent as well as structural factors in the surrounding context.First, it is necessary that the initial phase of protest does not become an object of the type of gradually mounting repression that authors observe in the escalation process.Instead, the negative attribution of opportunities should be a reaction to events and experiences that are mundane enough to warrant disappointment and disillusionment rather than highly mobilizing feelings of outrage and shock.Second, to persist throughout the demobilized phase, it appears necessary that the grievance can plausibly be experienced in the participants' everyday life.The privatization process is unlikely to be seen in the aftermath of campaigns purely concerning abstract policies (e.g., industrial policy, general environmental policy), if the participants do not manage to frame the issue in ways that link it to everyday experience.
Above and beyond the present examples, the privatization pathway is useful in a wide variety of analogous settings, i.e., where the occurrence of violence happens despite an extensive break in the preceding sequence of nonviolent collective action.Its mechanisms can also be seen in more complex cases, where frames and relations that were once developed in the context of prior sequences of collective action are reignited as collective violence in response to new grievances.Such cases include urban riots, extensive series of intermittent ethnic violence, and the long tails of violence that sometimes come to characterize post-conflict settings.In these and other settings, the privatization pathway helps provide a framework for capturing the political substance of violence that often appears apolitical and random in a short-term perspective.
On a broader scale, the critical discussion of temporality in relation to processual approaches to political violence should inform better and more imaginative theoretical and empirical work.Once we note the arbitrary paradigmatic quality of the escalation pathway, it is possible to extend the framework without abandoning the enormously useful insights that past works have provided.The difference, however, is that these insights should be understood in relation to a single temporal model rather than to the emergence of political violence as a general phenomenon.Future research should therefore continue to explore the temporal variance that exists within the empirical field, and how these differences relate to the mechanisms and processes that lead people to use violence for political ends.
The Privatization of Protest: Demobilization and the Emergence of Political Violence in the Aftermath of Local Protest Campaigns

Table 1 .
Differences between privatization and other pathways.