Organizing service delivery on social media platforms? Loosely organized networks, co-optation, and the welfare state

ABSTRACT

Recent crisis situations have witnessed a growing number of loosely organised networks (LONs) that deliver welfare services and employ social media platforms to coordinate their actions. Focusing on the 2015 ‘refugee crisis’ in Sweden, we explore the role of LONs in Swedish resilience policy. In the absence of standardised heuristics characteristic of established organizations, the LONs refer to social media for generating a common stance on the policy problem and their relation to the state. The study indicates challenges in governing LONs, showing that although a LON may become co-opted by the state, this co-optation may lead to its demise.

KEYWORDS:

• Social media
• refugee crisis
• welfare service delivery
• resilience policy
• loosely organized networks (LONs)

Introduction

For more than two decades, researchers have described a policy system where a wide range of both public and private organizations participate in the shaping and delivery of welfare services and other public utilities (Börzel 1998; Pierre and Peters 2000; Rhodes 2000, 1996). Civil society actors are central in these networks (Brandsen and Pestoff 2006). The governance literature has contributed to our understanding of the role of civil society organizations in welfare service delivery networks, but there is much less research on the role of the loosely organized networks of individuals coming together with the help of social media to provide welfare services. The major part of the literature on such individual-based networks is focused on personalized protests rather than collective organizing of welfare services (Rheingold 2002; Nunes 2014; Shirky 2008; Bennett and Segerberg 2012, 2013). Recent years, however, have witnessed an increasing number of what we here call loosely organized networks (LONs) of individuals coalesced through social media to engage in the provision of public service, i.e. to take part in a governance network. This has especially been the case in crisis situations such as in 2015 when a large number of refugees passed through Europe (Turunen and Weinryb 2017).

There is emerging research on these types of networks in term of how they organize (Kornberger et al. 2018; Theocharis, Vitoratou, and Sajuria 2017; Simsa, Rameder, and Aghamanoukjan et al. 2018), but we know little about the role of LONs in the context of governance networks, especially in terms of their relationship to the state. Research into social media mediated protest movements indicate that the empowering capacity of social media may be limited in time to the early phases of protest, and that a more formal organizational structure is needed to ensure continuous activity (Tufekci 2014). It is thus important to learn more about how LONs, being coalesced through social media, organize and establish their relationship with the state in order to outlive the initial phase of popular engagement. Do they engage with the state like any other governance network actor? Do they vanish and disappear? Or do they develop into sites of resistance and challenge the state?

The empirical context of this paper is Swedish resilience policy and its application to what has been called the European ‘refugee crisis.’ In 2015, over 160 000 refugees sought asylum in Sweden, most of them arriving in the autumn. The number of asylum seekers in 2015 was twice as high as the year before. In addition, a large number of unregistered refugees travelled through Sweden, aiming to seek asylum in other Nordic countries. The influx of refugees was unexpected, straining Swedish authorities on the receiving end. The situation was treated as an administrative crisis. In accordance with the Swedish resilience policy, the Swedish Agency for Civic Contingencies (MSB) assumed the overall coordination function between different public agencies, primarily the Migration Agency, and a number of civil society organizations traditionally included in Swedish resilience policy, such as the Swedish Red Cross and the Church of Sweden. In addition to established civil society organizations, a number of LONs organized via social media, especially Facebook (Weinryb 2015; Turunen and Weinryb 2017). The LONs took over some central tasks the state normally considers its prerogatives, such as providing reception orientation, healthcare, temporary housing, food, and clothes to refugees. In doing so, LONs manifested the swiftness of their organizational capacity vis-à-vis the state, but some of them also carried out their service with an outspoken political message that the state had failed to do its job. The relevance of the LONs was acknowledged in the major state audits of the crisis, but the role of LONs in the Swedish resilience policy have not been scrutinized or analysed in any detail (Asp 2017; Riksrevisionen 2017; SOU 2017, 12).

By taking a closer look at the very discussions on social media that constitute the LONs in the context of resilience policy, we aim to learn more about the role of this type of organizing, and probe into how LONs may become co-opted or not by the state (Selznick 1949; Coy and Hedeen 2005). We ask: how do LONs negotiate the policy problem they want to solve by providing welfare services and how do they negotiate this service delivery vis-a-vis the state? By the word state, we here mean national, regional, and local government. Similarly, we use the terms state and government interchangeably.

To answer our research questions, we conducted an interpretivist policy analysis of Facebook conversations of two LONs in Sweden, one in Malmö and the other in Stockholm. Our data span the period September through November 2015 and covers Refugees Welcome to Malmö and Refugees Welcome Stockholm, two large LONs created and coordinated through Facebook in early September 2015. In addition, we made an empirical revisit to the LONs’ more than two years after their inception, examining what they have become.

The paper is organized as follows. First, we place our case in the context of governance networks (Isett, Mergel, and Leroux et al. 2011), resilience, co-optation, and social media research, and discuss the organizational challenges LONs introduce to state–civil society cooperation of service production in the Swedish context. We then describe our methodological choices and empirical materials. Subsequently, we present the findings from our cases, followed by a discussion and conclusion of implications and limitations of our study.

Co-optation and resilience policy in the age of network governance

Network governance has brought a new angle to state–civil society relations by legitimising more decentralized forms of governance. Proponents of network governance argue that networks bring efficiency, new solutions and resources into hierarchical state rule (Hooghe and Marks 2003; Héritier and Lehmkuhl 2011; Kooiman 2000; Rhodes 1997). Especially in crisis situations, literature has pointed out the counter-productive nature of hierarchical steering (Koppenjan and Klijn 2004). On the other hand, the increasing power of networks may turn against the state and provide spaces of resistance to the state’s policies (Rhodes 1997; Selznick 1949), making co-optation a likely response from the state. Co-optation is a mechanism by which competing authorities reinstitute a balance of power and reclaim legitimacy. An autonomous, and possibly resistant, governance network is thus a potential target of co-optation efforts by the state. For Selznick (1949), co-optation can be classified as either formal or informal. We focus on the potential of formal co-optation, which refers to inclusion of the contesting party into the administrative structures of the mainstream authority, here the state, and thereby depriving it of its political voice and diminishing its autonomy. Co-optation ‘involves the establishment of openly avowed and formally ordered relationships’ (Selznick 1949, 13), which increases the perceived legitimacy and respectability of the state vis-à-vis the citizens and where the co-opted party may end up sharing ‘the public symbols or administrative burdens of authority, and consequently public responsibility, without the transfer of substantive power’ (1949, 261). The result of co-optation is a less pluralist and less autonomous civil society.

Recent academic and practitioner-oriented literature has increasingly turned to the concept of resilience for new ideas and practices in crisis management (Walker and Cooper 2011; Baez Ullberg and Becker 2016). Resilience here marks the transition from risk prevention to policies that aim to minimize the unwanted consequences of crises. This is done by relocating crisis management tasks to a pluralist and autonomous civil society organized in a network and mobilized during crisis situations (Sundelius 2005; MSB 2014; Dahlberg 2015). Critical research has pointed out how such neo-liberal management practices seek to transfer responsibility for social well-being by moving it from the public sector to individual citizens (O’Malley 2010; Bulley 2013; Joseph 2013). A crisis governance network requires effective communication and a shared understanding of the problem to be solved (Larsson 2017). However, at the same time, resilient organizations embrace pluralism and openly test different interpretations of the situation at hand (Termeer and van der Brink 2013) adding to the complexity of the governance network. Resilience policy thus builds upon a paradox, where civil society is both autonomous and pluralist, ascribing to a coherent interpretation of the policy problem at hand, and organized into steady, yet autonomous, governance networks ready to be mobilized by the state. Pluralism among established organizations need not be problematic as their frames of interpretation can be coordinated by the state. However, in the case of the LONs we face a new kind of contingent way of negotiating the policy problem and relationship to the state, both of which are entrusted to social media conversations rather than official crisis communication channels.

The threat of co-optation becomes actual when a LON enters a governance network, as there is little prior knowledge how they interpret the crisis situation and organize their service production in relation to the state and other actors in the field. LONs are networks of individuals converged through social media, but they perform welfare service delivery within a governance networks of organizations, as if they were an organization like any other. Yet, this is precisely what the social media network literature claims that they are not. Research on social media has emphasized the digital platforms’ unique role in orchestrating individualized engagement (Nunes 2014; Shirky 2008; Rheingold 2002), and the individualized protest possibilities that this ‘connective action’ enables (Bennett and Segerberg 2012, 2013), but we know little about how such individual engagement affects more traditional forms of policy implementation such as welfare service delivery during crisis situations. What we know so far points towards an undertheorized situation where social media gives rise to ‘complex, polymorphous, and multidimensional’ forms of organization, which nevertheless seem to require an adjacent material organization structure to survive (Theocharis, Vitoratou, and Sajuria 2017, 249). Kornberger et al. (2018) approach the same refugee crisis that we, but in Austria, through the concept of sharing economy organized via social media and focusing on the shared moral concern (rather than the commodity). Simsa, Rameder, and Aghamanoukjan et al. (2018) study similar initiatives, again in Austria, as ‘structured self-organization’ with ‘clear (although often very short time) goals’ (Simsa, Rameder, and Aghamanoukjan et al. 2018, 13). These studies point towards the centrality of understanding how such initiatives arrive at some policy interpretation, set goals for shared activities, and combine the online enthusiasm with a more durable organizational structure. However, these studies elude the broader question of the LONs’ role in governance networks and consequent relationship with the state, which is crucial for their impact on the state’s resilience policy.

Negotiating a relationship to the state

As stated above, governance networks decentralize the state’s power and may become sources of resistance to state policies. The question of co-optation in our case concerns the possible and acceptable forms of interaction between LONs and state. Drawing on social media research, Nunes (2014) states that to understand the ‘organization of the organizationless’ requires a study of the very negotiations between a multitude of individuals that constitutes such organizing. Because of the absence of formally decided organizational elements of a complete organization (Ahrne and Brunsson 2011) and the absence of standardized heuristics guiding decision-making processes (Cyert and March 1963; March and Simon 1958), the written online conversations become the place for interactive policy interpretation and central means of organizing for the LONs. Past studies have established that technical affordances affect the way such conversations are formed, and algorithms guide much of the organizational processes that follow suit (Kaun and Uldam 2017; Treem and Leonardi 2013). However, little attention has been given to qualitative analysis of the concrete meanings conveyed in these online conversations by the participants, as well as to the broader social and political context which imprints itself on the organizational negotiations that take place (Lounsbury and Ventresca 2003; Stinchcombe 1965; Pfeffer and Salancik 1978). In the case of welfare service production, digital activism requires an adjacent local infrastructure, and in this process the movement becomes increasingly organized (Kornberger et al. 2018; Theocharis, Vitoratou, and Sajuria 2017; Simsa, Rameder, and Aghamanoukjan et al. 2018), coming into contact with material realities on the ground as well as other actors, here most importantly with the state. Crucial, then, becomes how this structural accommodation of welfare service delivery is negotiated, and how this affects the relationship of the LON to the state, considering the potential for co-optation. From previous research, we know that there is a potential of co-optation as civil society becomes engaged to fulfil the organizational agenda of public administration (Selznick 1949). As Selznick put it: ‘coöptation tells us something about the process by which an institutional environment impinges itself upon an organization and effects changes in its leadership, structure, or policy’ (1949, 13).

To understand the relationship of LONs to the Swedish state, we must situate the civil society sector in the context of the Swedish welfare regime. Swedish research often employs a dichotomous distinction between service and voice (protest), where the increase in the former is described as leading to a decrease of the latter (Johansson, Arvidson, and Johansson 2015; Lundberg 2017; Arvidson, Johansson, and Scaramuzzino 2017; Dahlberg 2005). Its historical legacy has meant that Swedish civil society largely sees itself – and is seen – as a partner rather than a competitor to the state (Lundström and Svedberg 2003), and increased welfare service production is generally taken to mean declining political advocacy (Herz 2016; Arvidson, Johansson, and Scaramuzzino 2017). As the public sector began to detract in the wake of the 1990s reforms, civil society shifted to assume more welfare service production functions (Amnå, 2006; Stryjan and Wijkström 1996; Johansson, Arvidson, and Johansson 2015). When the refugee crisis erupted, Swedish civil society was thus largely marked by the absence of a strong political voice, working in tandem with the state, and not challenging or attempting to improve it. Against this background, the emergence of political, self-organizing, and crowd-funded LONs, that organized service where the state declined to do so, posed a particular test to established state–civil society relations (Trägårdh 2010; Turunen and Weinryb 2017).

We will here investigate the evolving relations between the LONs and the state as an incrementally unfolding process of potential co-optation (Coy and Hedeen 2005), operationalized as changing forms of state-civil society relations (Young 2000). To do so, we turn to Young’s (2000) typology, where relations between the state and service producing civil society organizations are labelled as complementary, adversarial, or supplementary. Different types of relationships may exist both simultaneously and sequentially. In the case of complementarity, civil society organizations act as an extension of the state in their delivery of welfare services. The state gives the funds and expects civil society organizations to deliver services more efficiently than the state. Complementarity is the default case for the contemporary civil society sector in Sweden (Lundström and Wijkström 2012, 1995), and more specifically it is the default type of relationship as envisioned in Swedish resilience policy (FOI-R–4295–SE et al. 2016; prop. 2009/10:55; bet. 2009/10:KrU7; Fältguide 2016). In an adversarial relationship, civil society organizations want to bring about changes and ask, lobby, pressure, demonstrate to make the state do so. Supplementary relations have historically been least prominent in Sweden. In this case, civil society organizations provide services that are beyond or somehow different from what the state is willing to supply. There is an absence of an obvious communicative relationship here, and civil society acts independently of the state. Supplementary relations challenge the state’s legitimacy in a given policy area. Civil society actors that pursue too autonomous or radical agenda run the risk of being co-opted by the state (Coy and Hedeen 2005). The complementary and adversarial lenses are compatible with the Swedish tradition as they both function within a communicative, even dialectical, relationship with the state, upholding the distinction between voice and service (Johansson, Arvidson, and Johansson 2015; Lundberg 2017; Trägårdh 2010; Arvidson, Johansson, and Scaramuzzino 2017; Dahlberg 2005). However, the supplementary lens can be understood as challenging the status quo by acting beyond or differently than the state, and raises the potential of co-optation by the state to neutralize such initiatives.

Research design, context, and methodology

This study is an interpretive policy analysis of two LONs, analysing the participants’ discussions on the policy problem they aim to tackle, and of their relationship to the state. Given the potential for co-optation, our analysis will focus on the alternation between different forms of state–civil society relations as interpreted by the LONs in their Facebook discussions. Interpretive policy analysis approaches governance networks not as pre-conceived but as something emergent and therefore contingent (Howarth and Griggs 2006; Griggs and Howarth 2002). This approach suits our aim, where the LONs and their relation to state emerged only during the crisis and as the actual service delivery was undergoing. Interpretive policy analysis (Bevir and Rhodes 2004; Wagenaar 2011; Fischer, 2012) asks in what ways meanings brought to the policy implementation process shape actions and affect policy outcomes. Meanings in this perspective are not linguistic nuances but ‘constitutive of political actions, governing institutions, and public policies’ (Wagenaar 2011, 4). Whether a LON’s actions are predominantly negotiated as complementary – helping out the strained state, or as supplementary – describing the state as deliberately withholding from certain types of welfare provision, affects what can emerge as the ‘common interest’ informing the Swedish resilience policy. Such differences further affect the self-understanding of the LONs – for example, as partner or challenger to the state – and the possibilities of coordinated action between the state and the LON in the governance network.

The study consists of two exploratory descriptive case studies (Yin 2003) of the Facebook discussions of Refugees Welcome to Malmö (RWTM) and Refugees Welcome Stockholm (RWS), also known as We Who Welcome Refugees at [Stockholm] Central [Station] (name translated from Swedish), two LONs that emerged to provide services for refugees in 2015. To give a sense of the size of these networks, RWTM had on 2 October 2015, about a month after its inception, 7490 members. Two months later it had 9888 members. RWS had on October 2, also a month after inception, 19,059 members, and two months later 22,072 members. The number of members should be taken as an indication of the size of engagement rather than an absolute figure. Given Malmö’s location by the Öresund Bridge, it was the entry point for most refugees arriving in Sweden. In Malmö, the arriving refugees faced the decision whether to continue as transit refugees, and thus breach the law, or file an asylum application in Sweden. By the time the refugees arrived in Stockholm, the question of continuing or staying in Sweden, although still in the air, was not that new or urgently pressing as in Malmö. The immediacy of the border is one aspect that is reflected in our material. Obviously other differences exist too, but our main focus here lies on exploring the differences in the interactions between individual posts as they were presented to the individual participants, i.e. how the policy problem as well as the relationship to the state was negotiated, rather than conducting a comprehensive comparative case study.

The data set was collected through web-archiving (Lomborg 2012), which involves collecting and archiving data from social media as it occurs. As with any type of method, web-archiving has its advantages and disadvantages. One benefit is that it preserves the chronological order of the discussions, enabling researchers to look at how negotiations dialogically unfold, and how participants make sense of this evolving sequence of posts (Meredith and Stokoe 2014; Frobenius and Harper 2015). A drawback is that this type of data is not suitable for any substantial quantitative analysis. In addition, discussions on Facebook do not yield data on what happened on the ground, but they represent the ‘data’ that was available to the activists themselves as they convened online. In other words, this is qualitative data of the communications through which participants were organising themselves, coordinating service production and negotiating their stance to the state. It was on Facebook, rather than on the ground, that the LON’s (organizational) understanding of the problem and role in the crisis governance network was established and communicated. Such data does not allow us to go beyond the medium of interaction and the dynamics it reveals, but it does enable us to take seriously the ordinary utterances the activists import to the policy process (Wagenaar 2011), and it is precisely this types of online negotiations that we wish to explore.

The web-archiving was done manually by one of the authors at frequent intervals throughout the fall, clicking and opening each post, comment and reply and subsequently saving it all in the PDF format. All in all, this resulted in 833 pages from the Facebook group RWTM and 1016 pages from RWS. As some posts were re-commented on in later time periods, which bumped them up in the page feed, there is some overlap in the PDF-files, but the files cover mainly unique posts and comments for the period September 1–2 December 2015. The data thus spans the period from the creation of the LONs in September 2015, to the temporary legislation that essentially became a closure of the Swedish border to incoming refugees in November 2015. We revisited the Facebook pages in summer, 2018, and reviewed the mission statements and checked the level of activity of the LONs and the news section of their homepages.

Facebook conversations were primarily in Swedish, and all quotes included in this paper are translated by the authors from Swedish to English. The grammar and style of the original posts are kept in the translations. The translation process provides anonymity to the participants, as quotes are not directly traceably by an online search. For reasons of anonymity, we also chose to not include the date and time of any specific post.

We opted for qualitative coding of the data, using NVivo software that enabled us to identify and contextualize intricate negotiations in the two Facebook groups. The analytical process began by us reading through all posts chronologically. Our attention was drawn to the striking difference in the levels of antagonism towards the state in RWTM contrasted by RWS’s more pragmatic orientation. However, even though the level of antagonism in Malmö was high in the beginning it abated as the crisis prolonged. This led us to think of the different ways the LONs polyphonically negotiated their perception of the policy problem. Consequently, our coding scheme builds on the distinctions the LONs drew in identifying the type of problem they emerged to alleviate. This has been coded by identifying different types of political demands put forward by the LONs. More specifically, three codes emerged as dominant: 1) pragmatic talk on logistical issues, 2) demands to bring back the welfare state, and 3) calls for political change. In the second stage of the analysis we focused on the LONs’ interpretation of the state-civil society relations. Here our main focus concerned the organizing of the LONs’ activities and how the state is envisioned in this respect. Central codes that emerged in this case were: 1) action in conflict with the state’s policy or lack of it, 2) action in cooperation with the state, and 3) action explicitly disregarding the relevance of the state. These codes functioned as broad themes within which we focused on the interactive dynamics of negotiating the LON’s understanding of the policy problem and its relationship to the state.

The challenge and opportunity with our Facebook data is that it consists primarily of conversations that are composed of different posts being addressed to each other. Most studies on social media focus on the users’ experiences, with the content of social media posts, or social bonds established between individuals (Snelson 2016; Papacharissi 2010). Interpretivist interest in the interactive mode of social media seems scant among social sciences; however, there are some studies in the field of conversation analysis, which have inspired also our inquiry, especially our focus on the sequence of postings (Paulus, Warren, and Lester 2016). Such interactions acquire coherence and meaning only if read as a sequential and polyphonic flow of conversation (Meredith and Stokoe 2014; Frobenius and Harper 2015). This requires extensive contextualization of individual posts in the sequential evolution of the conversation, something that standard methods of coding focusing on the content of individual statements or more quantitative methods building on keywords have difficulty to capture. The openness of social media contradicts clear hierarchical dictates and leans towards negotiated organizing, where communication of decisions become an ever evolving chain of new posts that each ‘hook onto’ previous posts at least at two different chronological levels: the immediate post–reaction sequence, and the sequence of the whole discussion thread as it appears to individual members reading it. Such a chain is open to multiple interpretations of the problem at hand, the services to be organized, and the political context of these activities. We here trace such conversations where the meaning-bearing unit is a sequence of posts, not the individual post as such. Due to space limits, we cannot recount all the contextualizations that were necessary for the analysis. Therefore, we will show exemplary extracts from the conversations and the broader context of the discussion will be explained. As our aim is descriptive and exploratory, rather than searching for causalities, we believe that this method can contribute to our understanding of how the LONs evolved and negotiated the policy stance, even though each quote can only be claimed to be an example rather than a full representation of the entire material.

Empirical analysis of the two LONs

We will begin by describing different ways political demands were articulated in the LONs, pointing to the perceived policy problems. By looking at the ways policy problems were interpreted, we then trace the stances towards the state that were adopted in the Facebook conversations of the LONs.

Interpreting the policy problem

In Malmö, the border town, the international dimension was one of the cornerstones of the problem formulation. Many posts referred to the Dublin Convention and the EU’s carrier liability rules, which sanction transport companies to assume responsibility for passengers’ valid travel documents or face costs of repatriation. In practice, this meant that transport companies denied entry for those passengers who could not produce valid travel documents. The concrete problems brought about by the Dublin Convention and the aviation rules were presented primarily as political, and not administrative. The following post was a requested reposting of a political speech delivered in Malmö at a local demonstration summarizing this problem formulation:

Some talk about a refugee crisis, but this is in fact a politicians’ crisis with devastating consequences for people on the run. A trip from Istanbul [to Sweden] costs 80 euros, but because of political decisions it costs the lives of refugees. [It is unconceivable] THAT YOU ARE NOT ASHAMED!

The requested reposting indicates identification with the speech, and its view on the political nature of the crisis is applauded in the Facebook posts. The quote shows that political decisions, politicians in a broad sense, not specifically Swedish, have created the crisis, and that it is up to the political decision-makers to solve it. Anger is articulated towards these politicians, as they are bashed and shamed.

At the same time, a strong image of Sweden as a welfare state, or rather the need to retrieve the welfare state whose absence the crisis has made all too tangible for many, is articulated:

Regardless if it is about waiting for a place [in the facilities of the Migration Agency] or not it should not look like that [–] Sweden has chosen to receive this sum of refugees which means that they have to use the resources available so that everyone will get to sleep in a proper way and not on the floor.

This post is responded to by many participants who point out that there is a difference between waiting a couple of hours and the whole night, which then leads to the articulation of clear differences in the work ethics of the Migration Agency that holds to its working hours and the participants who work around the clock. The statement that Sweden ‘has chosen’ to receive a specific number of refugees also implies both power and agency on the part of the state, which in turn can be held responsible, and criticized for the current situation. However, this responsibility is disputed and debated in the Facebook conversations, leading to a more or less consensual position that the indignity the refugees are subjected to is at the core of the problem formulation. In the example below, one participant recounts a newspaper report on refugees’ ‘miserable situation at the Migration Agency’ and concludes:

It is true that the pressure on Sweden and the Migration Agency is extremely heavy but this is not dignified! When I think of the warmth and caring that met the refugees at the central [train station of Malmö] I feel even more grateful for the efforts that were made there. But how can the newcomers’ situation be improved???

These three examples of the interactively shifting problem formulations show the volatility of policy interpretations that surface in the Facebook discussions. Even if the Dublin Convention and the political situation are often recognized as the ultimate sources of the crisis, the following articulations together produce the inefficiency of the Swedish (welfare) state to give the refugees a dignified reception as the main policy problem for the participants to take care of. This problem formulation rests on the posited expectations put on the Swedish welfare state and the perception of the state’s potential intentionality in not living up to its own standards with regards to the incoming refugees.

In Stockholm, rather different problem formulations emerged. The political and administrative deficiencies of the politicians and the welfare state were not the main focus of the Facebook discussions. Instead, the problems that the participants articulated concerned almost exclusively how to navigate the logistical aspects of the refugee reception work. Some participants related this to the Dublin Convention, as in Malmö, but it was contextualized differently than in Malmö. This is how RWS explain their work:

In Sweden, the Migration Agency receive the people who choose to seek asylum here, and any other category of refugees should under the Dublin Convention not exist. The so-called transit refugees who still came to Stockholm without seeking asylum therefore had no place to go while waiting for the next train north. Meanwhile lots of Stockholmers went to the Central [train] station to give all arriving [refugees] a dignified reception and see if they could help them in any way.

The problem for RWS emerged from the new category of ‘transit refugees’ that did not fall under the state’s jurisdiction, but who nevertheless should be welcomed somehow. Hence, civil society took up the task to organize, for example, food supplies, clothes, temporary housing, legal counselling and advice for the refugees for ‘a dignified reception’. The shortcomings in the official refugee reception were recognized, but not politicized further:

NEEDED AT THE MIGRATION AGENCY – The Migration Agency really has the responsibility to care for those who are seeking asylum in Sweden but in their offices around Stockholm people are forced to wait for several hours without food and drink. WHAT YOU CAN DO: 1. Call your nearest Migration Agency office and ask if there is a need for sandwiches or drinks. 2. Then write in this thread WHAT is needed at the Migration Agency, WHEN it is needed, and the ADDRESS to the office. 3. Go there with what you can [bring]!

At the very outset, the position of RWS is negotiated as helping the state out. The post urges cooperation with the Migration Agency and directs activism to logistical issues. This post was replied to with pieces of information about what individual participants did and what remained to be done. The same strategy is used to meet criticism directed against the authorities; the potential for the politicization of the situation is diluted by calling for direct, concrete action to solve practical issues and stylistically by aligning RWS with the Migration Agency.

In both cities, there appears to be a shared understanding that the very foundation of the problem stems from the Dublin Convention and that Swedish authorities are incapable of dealing with the gravity and the extent of the crisis. However, what follows from this initial shared ground appears to be two very different articulations. In Malmö, the observed difference between the response of the state and what it could potentially do fuels anger towards the state. A central piece of this articulation is the pronounced expectation that Sweden ought to provide a dignified welcome to all refugees and that the politicization of the conflict arises from the moral expectation and the observed lack of engagement of the authorities. In Stockholm, the lack of capacity of the state is recognized, but it is not connected to strong moral expectations what the state ought to do but instead linked to the category of ‘transit refugees’ that do not properly fall under the responsibility of the state. Hence, the problem formulations adopt a practical focus in solving concrete issues in the best possible manner. The different and shifting problem interpretations that emerged in Malmö and Stockholm show that these LONs do not have predefined interests that would guide their interpretation of the policy problem, but such interests are in fact results of (ongoing) conversations between the participants.

Interpreting LON-state relations

We will now discuss the ways in which the relationship to the state becomes articulated in the conversations of the LONs. In Malmö, the articulations of the problem in political terms – as a failure of the welfare state – and the LON as something that can bridge between the actual situation and the moral ideal – led to the dominant conceptualization of state-civil society relations in initially adversarial terms, i.e. aimed at changing the government’s policies:

Politicians must help. When there are people who come here but want to go to another country e.g. have relatives there they should not have to register. What is most important the Dublin Convention or human rights?

This post appears amidst a long sequence of criticism directed at the state. The juxtaposition in the end of the quote presents human rights and morality as oppositions to the rule of law. Whilst the LON acts as an adversary advocacy group bringing this moral dilemma into the public awareness it is the politicians that are invoked solve it.

In addition to the absence of the welfare state, also the conscious withdrawal of the public sector from the ideals of the moral welfare state provoke the participants. For example, the police as the neutral guardian of the law is attacked:

The police force many refugees, who come to Malmö C [central train station], to get to the Migration Agency’s accommodation facilities. These refugees really want to travel on to Norway and Finland, where they have relatives, family or friends. The police see it as if they give the refugees a chance not to be arrested by forcing them to the Migration Agency because the refugees would otherwise be arrested for they reside illegally in the country. This was told directly to us by the police in Malmö the day before yesterday. The police also gave a demonstration of how incredibly ill-informed they are, on many levels, about how the asylum process works.

Many such exclamations against the authorities convey both ill-will and frustration resulting from the inability of the participants to carry out their own actions. At the very end of the post, the police are represented as uninformed about the rules that apply to asylum policies and by implication, the participants as knowledgeable. Given the recognition of a clash between the service the LON produces and what the state does, the adversarial relationship, where the LON would lobby the state to change its policies has yielded to a supplementary frame of reference, where the direct communication with the state is severed and the LON assumes an independent service producing role. In this relationship, the state is ascribed negligence and ignorance while the LON is depicted as the well-informed and welfare providing party. Given the fact that the supplementary frame is rare in Sweden, this new position elicits self-reflection from the side of the participants:

Perhaps this can change Sweden’s stance. Why do we always expect the state to solve all problems.[?] Is it wrong of us [who are not the state] to offer our help?

This question, however, is questioned by pointing out that the authorities do not accept the LON’s contributions:

But they do not accept that individuals are responsible for such stuff.

A statement, which is confirmed by another participant:

Exactly, they are not satisfied with our performance.

In these articulations, we see how the positions of ‘being not accepted’ and ‘the authorities not being content’ are brought together blurring the boundary of what are legal restrictions – which welfare services LONs are allowed to produce – and what concerns the appraisal of the service by the state. The participant who initiated the thread, comments that RWTM is not helping the refugees in order to get ‘a pat on the back from the Migration Agency’ and asks ‘why is their opinion important?’ As the discussion continues the LON’s dependence on the authorities’ approval turns to address the different resources and competences they possess:

They have told us in Malmö C [central train station] that they do not need us because their work is better than ours and they have more knowledge than we do. Therefore one feels a little so so about what you express. The Migration Agency unfortunately does not help but it is we volunteers that help the most. The Migration Agency hands out sheets [of paper] and cannot provide migrants neither accommodation nor food. Because in Malmö it is jam-packed.

To which the initiator of the whole thread on civil society’s independence adds:

But it is true that they can offer more than us and that they have more knowledge than individuals. Then if they have enough of it to offer is a different matter.

The discussion shows how the supplementary frame emerged out of the conflict with the state, but also how it was questioned by pointing out the greater competence of the state. The recognition of the ‘superior’ competence of the public authorities indicates the beginning of the co-optation that slowly takes a hold of the RWTM. By October, regular coordination meetings with the authorities lead to new regulations of RWTM’s work. This is not a welcome development for all participants:

XX [name of individual] why have rwtm [Refugees Welcome to Malmö] completely bowed down to the municipality which is obviously unable to deliver? Is it not time that you have a new conversation and try to hold your ground a little bit regarding this food catering faux pas? Buns and fruit are tasty but these are things my 2 year-old gets for snack and hardly any food. For only a few dozen cinnamon buns there is [can be made] lentil soup, carrot soup etc. for an entire block. So insanely bad economy if nothing else. If you have a so-called kitchen/cafe, you should hell make use of it otherwise you might as well put up a vending machine and give people tokens so you “do not have to” volunteer… Who the heck came up with this very stupid idea?

The post accuses RWTM for dropping its autonomous supplementary position and subsuming to the expectations of the state. In the response, the seeming subjugation is denied, but also justified by the benefits brought about by closer cooperation with the authorities:

We have not bowed down to anything and we have meetings several times a week with different units of the City of Malmö also about this issue. It occupies much of our waking hours since the day one opened on Posthusplatsen [place in central Malmö where refugees are catered for]. But the clock unfortunately has not more than 24 hours, and officials are working “only” 8-17, Monday to Friday… That you follow the law on food service in the premises of the municipality is hardly bowing down. We voice our criticism daily and clearly to those who have the possibility to change the dominant circumstances. We have not had access to these channels before. Now we do.

We end the story of RWTM with this quote emphasizing the importance of law and regulations that should prevail in the premises of the governing authority and how RWTM claims to be able to influence from within better than from without. Much like Selznick (1949) saw, formal co-optation embraces the challenging movement by bringing it into the governing authority. In the beginning, RWTM was mobilized to challenge the state on a dignified reception of the refugees; it ends with serving cinnamon buns, sandwiches, and fruit in tandem with the state.

In Stockholm, where the pragmatic problem formulation did not put a blame on the state, complementary relations between the LON and the state predominated. A discussion thread that calls for the RWS to help out at the Migration Agency office simply spells out the concrete needs the state agency is perceived to have:

Needed every weekday: Migration Agency in Solna [municipality in the county of Stockholm], Pyramidvägen 2. Open to 15. Needs: Small Festis [soft drink] Sandwiches (vegetarian and/or halal) Chocolate for the kids Reply to this comment (do not make a new comment) if you go there with things so that we know [about it].

This post indicates a shared problem formulation where the state together with the participants, in a complementary manner, produce services to the refugees. Note also how the post posits a certain permanence in this alliance: ‘Needed every weekday’. The replies to this call for help likewise focus on practicalities and mutuality:

Hey! [I] was just at the Migration Agency and left fruit yogurt Festis etc. In the room there were now about 50 persons. A relative of an asylum seeker named N [anonymized by the authors] helped me carry [things] from the car, He said, You can ask for him if You come here today. Requested items are sandwiches right now, in addition to fruit drinks etc.

Now we have left about 40 sandwiches, bananas, apples, Festis and sweets. It was all distributed quickly. Now, there were about 50 people waiting. […]

This discussion thread begins by pointing out a temporary deficit in the services the state produces (‘…there were now about 50 persons.’) without articulating it at all in political terms. Instead, the deficit is treated purely as a matter of coordination between the LON and the state. The temporary articulation of the problem also highlights the immediate effects of the LON’s welfare service provision.

At times, voices of criticism emerge even in Stockholm. For example, in September and early October 2015, RWS runs a temporary housing for ‘transit refugees’. The LON supplements the state as these refugees reside in Sweden illegally, not seeking asylum but preferring to continue their journey to Norway or Finland. Eventually, the City of Stockholm steps in to take over, although this is formally beyond the city’s administrative mandate. The participants readily acknowledge the contribution of the city, but also point out the continued importance of their own engagement in the service provision:

Hey! We excitedly await further information on the “accommodation of the city [of Stockholm]”! There is still a need for other accommodation facilities, Sickla/Nobelberget [accommodation run by LONs] has not always had enough places, so it is still needed. And the accommodation of the city [of Stockholm] surely needs volunteers!

The ironic style – ‘We excitedly await’ – of the post produces a double acknowledgement, that of the city taking over what the participants did and that of the participants themselves being still unparalleled in competence. By pointing out the insufficiency of the measures now taken by the city, the post articulates a relationship of dependency where the city still needs these participants to run the services in the facilities it provides. We see here the clear dominance of the complementarity between RWS and the public authority. Although RWS tries to articulate its superior power position, it does not call into question the need to communicate with the governing authority. Clear attempts to co-opt RWS were absent.

Empirical coda

As an empirical coda, in August 2018, we looked at the LONs mission statements and recent activities on the Facebook pages, as well as the news section of their homepages. The two LONs had formally registered as legal entities but had evolved in different manners. In Malmö, the LON had abandoned the political position and focusing even more strongly on welfare service delivery. The complementary lens, initially shunned, had eventually become all-encompassing, but the overall amount of activity seems to have petered out. The homepage was passive, showing activities only up until early 2017, and then solely focused on welfare service delivery of for example clothes and other similar items. An alternative Facebook group called Refugees Welcome Malmö (without the to) with an outspoken political focus had been created in the fall of 2017, but its Facebook page attracted just over 460 ‘likes’, in comparison to RWTM’s over 9000 members. Because the former is a page reporting only likes and followers, and the latter is a group with members, the comparison should be taken as illustrative only. In Stockholm, the mission statement of RWS in August 2018 pointed to a combination of welfare service provision and political advocacy in relation to migration policy. Their Facebook page was active, and the organization seemed to be thriving. In the news section of the homepage, activities were recently updated, showing both language training and soccer exercise for newly arrived immigrants, as well as support of protests against deportations of refugees who had been declined asylum.

Discussion and conclusion

Our empirical analysis indicates that in the absence of a formal organizational structure (Ahrne and Brunsson 2011; Selznick 1949), individuals in the LONs negotiated online, on a Facebook page, how to approach the refugee crisis, specifying the problem they should address, and organizing their activities. In Malmö, the confrontational and politicized policy problematizations led to a combination of adversarial and supplementary articulations accusing the (welfare) state for not carrying out its responsibilities. This gave the service production a clear political voice, questioning and competing with the actions of the state. In Stockholm, the pragmatic problematizations enabled a view of the state as a cooperative partner in need of more resources, which the LON would provide for. As the crisis prolonged and the state’s policies changed, the LON assumed a more adversarial stance. In both cases, the initial polyphony in the LON abated as the logistical tasks acquired more prominence (see also Bekkers et al. 2011). However, this logistical coordination was neither coordinated by the state nor did it necessarily contribute to the production of public goods as envisioned by the state. Instead, the very individualization of the personal action frames (Bennett and Segerberg 2012, 2013), and the lack of established organizational heuristics for decision-making (Cyert and March 1963; March and Simon 1958), made the LONs’ interpretation of the policy problem and relationship to the state unpredictable and contingent on how the online conversations evolved.

Our empirical analysis indicates that LONs are not easily governable, yet their development is to some extent dependent on the local institutional context. The findings point to two central conundrums related to LONs and governance: 1) How and by whom can the individuals participating in the LONs be governed? 2) How do LONs affect the network governance?

To answer the first question, we may think about the governance of the individuals converging in the LONs as a result of negotiations between activists. The LONs were to some extent managed by social media account administrators (Kaun and Uldam 2017), but more than anything else our empirical analysis shows that the LONs were interactively organized by the participants to the online conversations. This means there was no direct governance, by anyone, of these individual activists. The power in the LON resided in the online conversation itself, which evolved through a slow polyphonic and collaborative governance process. In these online interactions between activists, there was no organizationally constituted formal power structure. Instead, our analysis shows that it was primarily through formal co-optation (Selznick 1949) of individuals participating in the online conversations that the LONs could potentially be governed, but this came at the price of the LON losing the digital enthusiasm (Gerbaudo 2016) that had fuelled it in the first place. This happened in Malmö, where the co-optation of the LON to align itself with the objectives of the state eventually seem to have led to its demise. In contrast, in Stockholm, the LON survived the crisis years of 2015 and 2016 but gradually constituted itself more and more outside of the state’s policy objectives, focusing more on protest than on service delivery. In sum, the individuals in the LON could potentially be governed by the state by co-opting participants to the online conversation, but this may eventually harm the very existence of the LON as a whole.

To answer the second question, it is important to consider how the LON’s way of organizing through online conversations affects its possibilities to coordinate with other organizational actors in the network. LONs enter into contact with other actors through the participants’ prior individual contacts, or through contacts that emerge amidst service delivery. This potentially means a very heterogeneous, but also unstable, network with high levels of institutional uncertainty (Koppenjan and Klijn 2004). Resilience policy requires both expert as well as local, contextual, knowledge (Walker and Cooper 2011). LONs as ad hoc ventures may or may not have access to or possess such expertise or local knowledge. Furthermore, research on crisis management highlights the importance of a reliable quality of the information exchanged between partners in a governance network (Larsson 2017). It is through such information flowing through recognized communication channels that the governance network may implement the state’s policy objective in times of crisis. Yet in the case of LONs, whatever information is obtained, is distributed via social media, subjecting it to the same contingent logic of online conversations that in our empirical analysis resulted in a variety of policy interpretations and an evolving relationship to the state. Any attempts at meta-governance via information are thus extremely difficult in this scenario, as no established and reliable channels of communication exist with the LONs. This means that LONs’ engagement in the policy game (Koppenjan and Klijn 2004) of service production is not predictable or steady. Indeed, the only successful external governance of the LONs in our study was that of co-optation, however, this governance mechanism hampered the ability of the LON to engage activists to engage activists in the welfare delivery process – its perhaps most pronounced public innovation.

Resilience policy in crisis situations builds upon a paradox of an autonomous and pluralist, yet also already established, governance network of civil society organizations with a shared understanding of the policy problem at hand. In our cases, the access of the participating individuals to the human rights discourse – and willingness to confront the state and enter a legally grey area by helping the transit refugees – was crucial to the LONs and to the overall crisis management. In principle, this is precisely what a resilient society should do: claim autonomy from the state and profess a plurality of ideas under a democratic aegis, even in times of crisis (Sundelius 2005; MSB 2014; Dahlberg 2015). The LONs provided an effective, but unpredictable, medium for organizing welfare service delivery but, they did so by introducing a polyphony of different ideas about the state (and its institutions) to the service delivery process. However, the LONs did not act in an institutional vacuum (Lounsbury and Ventresca 2003; Stinchcombe 1965), and the relevance of the state in the online discussions in both cities, primarily in the hesitance towards the supplementary frame, echoed the historical tradition of Swedish civil society’s close and dialectical (Trägårdh 2010) relationship with the state. Accordingly, in Sweden, or similar contexts with a tradition of close state–civil society relations, the possibility of a more pro-active approach from the state towards the LONs as a way of meta-governance is worth further investigation (see also Sørensen and Torfing 2017).

In sum, the promise and threat of LONs lies in their high degree of self-sufficiency from state institutions and other partners in the governance network, combined with their proven efficiency to deliver services. The willingness of individuals to engage and participate in the delivery of public goods through social media platforms, points to the relevance for the state to consider new and innovative ways to engage with, or potentially co-opt, such social online initiatives. Here a precarious balance needs to be struck; the state needs to sustain the ability of the LON to engage a mass of individuals for the sake of delivering welfare services, and at the same time to some extent guide the LONs towards its desired policy outcome, perhaps building on locally institutionalized norms on state-civil society relations.

The study contains four limitations that are important foci for future research. Firstly, our focus on the online discussions of the LONs has rendered us an understanding of how policy positions evolve interactively online, but has not allowed us to explore other data sources, such as interviews, that would provide other types of insights into this process. Secondly, our study has left the state’s view, both at a national and a local level, unexplored. Glimpses of it can be extracted from the Facebook discussions, where posts explicate what the Migration Agency, the Police, or the local city authorities have communicated with the LON. Thirdly, more data is needed to understand how other actors in the governance network, such as established civil society organizations, respond to and are affected by the emergence of LONs. Fourthly, given the easily politicised nature of the LONs, future research needs to explore how co-creation of public values is affected by LONs as service providers.

Acknowledgments

This study has been enabled by encouragement from our colleagues at the Academy of Public Administration at Södertörn University, as well as funding from Riksbankens Jublieumsfond, grant number SGO14-1174:1. We are grateful for this support. We would also like to thank two anonymous reviewers for insightful and helpful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by Riksbankens Jubileumsfond [grant number SGO14-1174:1].

Notes on contributors

Jaakko Turunen

Jaakko Turunen has a PhD in Political Science from Uppsala University, and is a researcher and lecturer at Södertörn University, Stockholm, Sweden. His research focuses on language and politics, civil society, and social media’s role in political communication. His previous research has focused on Central and Eastern Europe and he is an editor of the Finnish Review of East European Studies.

Noomi Weinryb

Noomi Weinryb has a PhD in Business Studies from Uppsala University, and is an associate senior lecturer at Södertörn University, Stockholm, Sweden. Her research is centred on issues of accountability and organizing, comparing nonprofit organizations to public administration. She also makes cross-national comparisons across welfare regimes.