Navigating social spaces: armed mobilization and circular return in Eastern DR Congo

This article discusses the social mobility of combatants and introduces the notion of circular return to explain their pendular state of movement between civilian and combatant life. This phenomenon is widely observed in eastern Democratic Republic of Congo (DRC), where Congolese youth have been going in and out of armed groups for several decades now. While the notion of circular return has its the Congolese case shows ( . .. ) that combined with prioritising immediate security gains, it is the generalised politicisation of reintegration that most likely impeded coherent implementation processes (Perazzone 2017: 274).


Introduction
Armed groups and armed mobilization have gained increased attention in the study of civil war. Debates on the proliferation, incentives, strategies and discourses of armed groups have inspired different subfields of study including the political economy of war, governance and political order-making in conflictaffected areas, and rebel governance (Schlichte 2009a,b;Adam et al. 2017;Kasfir et al. 2017;Malejacq 2017;Bultmann 2018;Ka¨ihko¨2018;Krieger 2018). A growing number of studies also focus on those joining armed groups, on their individual sense making, navigation and social mobility, and on their return to their communities of origin (Richards 1996;Peters and Richards 1998;Weinstein 2006;Hoffman 2011;Debos 2016) Similarly, the disarmament, demobilization, and reintegration of combatants have been studied in detail, as well as strategies and interventions facilitating these processes (Cilliers 1995;Humphreys and Weinstein 2007;Theidon 2007;De Vries and Wiegink 2011;Muggah and O'Donnell 2015;Honwana 2017).
Failed reintegration and protracted security challenges are presented as key explanations why 'recycled rebels' (Vogel and Musamba 2016) feel entrapped in a constant cycle of mobilization, demobilization, and remobilization. Existing literature tends to analyse this cycle mostly from a security perspective. We know little about individual combatants' incentives and agency and about larger processes of socialization and social rupture as part of armed mobilization, this despite its determining impact on return processes to civilian life. Perazzone (2017) concludes for the disarmament, demobilization, and reintegration (DDR) literature in the Congo for instance that it 'tends to focus on disarmament and macroanalyses of the country's DDR and Security Sector Reform (SSR) as a whole', yet 'fine-grained empirical studies of ( . . . ) practical reintegration processes remain limited' (Perazzone 2017: 256).
Analysing the example of eastern DR Congo, this article wants to contribute to a better understanding of combatants' mobility and analyses the individual return to armed combat. It looks particularly at the mobility of rank-and-file combatants rather than their commanders, and at their capacity to navigate between different social spaces. In line with Moore (1973) and Bourdieu and Waquant (1992), we consider armed groups as such spaces, or as relatively autonomous domains of activity that are in constant interaction with their larger social environment but have their own set of rules, structure, relationships, and hierarchies defining and informing the behaviour of their members and the relationships between them. We introduce the notion of 'circular return' to explain combatants' navigation between such spaces of civilian and combatant life. We argue that a focus on this circular return of combatants helps us to move away from a macro-analysis of armed mobilization and to contribute to a better understanding of the persistence of armed group proliferation and mobility in eastern Congo. This mobility is inspired by a multitude of incentives, ideologies, factors, and processes that go beyond dominant arguments about failed reintegration processes, the protection narrative, or 'Big Men' mobilization efforts in search of an increase in their bargaining power (Utas 2012). (Ex-)combatants' readiness to return to combat is largely influenced by former experiences. These experiences explain not only the appeal of return but also the different difficulties ex-combatants face when trying to reintegrate into their home communities or elsewhere.
We stress that to understand this circular return between social spaces, (longterm) armed mobilization should be understood as a process of socialization. The presence of armed groups is to be considered a moment of social rupture as much as an experiment in constituting new social spaces, a new way of life, a new form of social and symbolic capital, and eventually a new identity. Ex-combatants' consequent capacity to navigate between different spaces largely defines their response to political or security dynamics, to mobilization and demobilization campaigns, or to their own individual challenges and ambitions. While Vigh (2009) defines social navigation (which includes a routinization of movement) as a conceptual tool to capture mobility within a social formation that itself is moving and changing, we define it as mobility between different social spaces or spheres of action. Joining armed groups are as much a way of making sense of changing realities as a response to security threats or mobilization efforts.
Young combatants' pendular movement in and out of armed groups is guided by a complex of motives, and individual and collective incentives. Armed groups provide an alternative social space, which simultaneously imposes and offers a new way of life, social, symbolic and economic capital, protection, forms of belonging, and a new identity to its members. As one combatant in eastern DRC confirms, I really liked the respect I enjoyed in the armed group. My family respected me too. I also had a good collaboration with the local authorities. Added to this are some advantages such as the ability to recommend someone to an organization for job opportunities (interview #22 with ex-combatant, Nzibira, July 2018). It helps to explain why, as our data show, that despite the hardship of recruitment, being a member of armed groups in some cases is considered a more attractive option than returning home: what I appreciated when being part of the armed group, was the respect, the composure training, and the new relationships we can build. I had no real concerns, I had money and I had a lot of relations ( . . . ) I did not like returning home, missing my own freedom when being in the armed group (interview #20 with ex-combatant, Nzibira, July 2018).
As social spaces, armed groups produce their own sets of rules, values, and resources that cannot easily be converted to civilian life. Armed mobilization particularly redefines the social and symbolic capital that helps to structure combatants' habitus (Bourdieu and Waquant 1992). Returning to their home communities, in many cases, results in a relative loss of the capacity to mobilize these capitals, as is illustrated by the marginalization and frustration often experienced. Return to combat, in this sense, is a reclaim to what was lost as a consequence of demobilization.

Navigating social spaces 3
This is particularly visible in the case of eastern DR Congo, where since the early-1990s, local youth has joined rurally based armed groups, rebel movements, or local defence forces, both during and after the Congo Wars (1996-97 and 1998-2003). Attempts to disarm and demobilize these groups have had limited effect, and armed structures have shown strong resilience to peace deals and DDR efforts. Demobilized combatants often decide to return to their armed comrades or respond to opportunities provided by newly created ones, and to navigate between their position and role of armed actor and civilian. Being combatant provides the opportunity to use both roles simultaneously. Navigating in and between different social spaces has become a persistent condition of life of many combatants and is guided by a complex interplay between social and political dynamics, collective needs and grievances, and individual interests. Not everyone prefers to remain in this state of pendular mobility though. Those who received considerable advantages as part of reinsertion programmes usually tend to prefer not returning to the armed group they have been with. However, a new security challenge easily triggers next rounds of mobilization despite having received reinsertion kits. The case of the Raya Mutomboki armed group, which again started operating as a result of a direct security threat to local communities and caused a massive enrolment of new and former combatants, illustrates this complexity of interactions between civilian and combat life. While many of these 'new' recruits did not tend to return home once the security threat disappeared, they kept close links with their families and local society at large.
Other cases seem to confirm the difficulty of combatants in cutting the links with their former comrades and commanders. This can be because of strong pressures of rebel commanders, leading to 'forced circular return', or can be a result of shifts in the security context and the consequent need to protect the community, revealing a persistent state of readiness of ex-combatants. It can also be out of frustration with the limited benefits and results of reintegration into society and the difficulties to return to civilian life. Or it can be out of nostalgia with the material and social advantages of combatant life, explained by one combatant as 'kukula kwa bure', or the habit to 'eat well without too many efforts required'. In the Ruzizi Plain (South Kivu), for instance, combatants preferred to keep their combatant position and wanted 'to be like General Bede Rusagara', a very influential armed group leader who had been successful in making fortune out of combat and was considered a role model.
In this article, we investigate these different dynamics explaining circular return and mobility between social spaces of combatants in South Kivu, located in eastern DR Congo. The research draws on extensive and long-term fieldwork by the authors in the region, and on more recent interviews between April and August 2018 with combatants, ex-combatants, local authorities, civil society members, youth associations, and host communities in Buzi, Kalonge, and Bunyakiri (Kalehe territory), Nzibira (Walungu), Nzovu and Kigulube (Shadunda territory) the Ruzizi Plain, and Bukavu, all located in South Kivu. Data were gathered through ethnographic methods, including individual and focus interviews, observations, and informal conversations. In Bukavu and Kinshasa, also interviews were conducted in 2019 with governmental and non-governmental organizations involved in DDR processes. The article starts with a conceptual discussion on circular return and mobility between social spaces. It then presents a historical overview the dynamics of armed mobilization in eastern Congo. It is followed by an assessment of the different DDR programmes, which were developed in the Congo since the formal start of the peace processes in 2003 and explains the limited impact these programmes have had on armed mobilization. It finally presents an in-depth analysis of how combatants constantly navigate between different social fields.

Circular Return and Mobility between Social Spaces
Recently introduced concepts such as 'circular migration', 'circular mobilities', 'split return', or 'recycled refugees', all recognize the mobility of migrant or refugee populations and point at the repeated migration experiences between an origin and a destination, involving several departures and returns (Hugo 2013). While this circular migration as part of human mobility is not a new phenomenon, it only recently received wider recognition. Today, there is a growing literature (including policy documents and statements) that focuses on this form of mobility, yet little agreement exists on its definition. Schneider and Parusel (2015) for instance understand circular migration rather broadly as 'a flexible form of repetitive movement between different destinations'. Displacement literature equally recognizes the prevalence of a persistent condition of mobility of internally displaced people and refugees, described a 'pendular mobility' or 'circular return'. In conflict environments, pendular mobility points at mobility as a way of life or as a permanent strategy to seek for security and protection. It also includes the daily return to what are considered safer areas to spend the night, which is omnipresent in conflict-affected rural areas in eastern DR Congo.
Similar patterns of pendular mobility can be identified among members of armed groups. A constant cycle of mobilization-demobilization-remobilization as observed in eastern Congo, both point at the persistence of drivers of mobilization and the failures of demobilization and reintegration efforts. This cycle has been documented at large, and several reasons explaining it have been discerned (Utas 2005;Theidon 2007;Nussio 2011;Wiegink 2013), including the perseverance of unresolved conflict dynamics; the proliferation of a rising number of small armed groups, contributing to a fragmentation of the military landscape; the growing involvement of low-level political actors in armed mobilization leading to a democratization of militarized politics; failures of demobilization and reintegration efforts; and disruptive military responses (Vlassenroot and Verweijen 2017).
A rich literature on ex-combatants and their mobility in post-conflict settings has documented cases including El Salvador, Sierra Leone, Mozambique, Liberia, Colombia, Burundi, and the DR Congo (Theidon 2007;Christensen and Utas 2008;Wiegink 2013;Fririksdo´ttir 2018). Most analyses label the pendular mobility of combatants, or the continuous mobility between combatant and civilian Navigating social spaces 5 spaces, in terms of remobilization. It is commonly agreed that combatants, despite demobilization and reintegration efforts, often risk being remobilized and to return to the bush and rejoin their former comrades, or adhere to alternative armed groups. Alternatively, combatants can respond to calls from politicians or other Big Men who strategically rely on ex-combatants and other marginalized youth for their own (political) campaigns, a phenomenon described by Christinsen and Utas (2008) as 'politricks'. And in still other cases, ex-combatants are considered responsible for persistent levels of insecurity because of their involvement in organized crime and crime networks after the war has ended or their demobilisation.
All of these scenarios incited the need for DDR programmes, which are assessed a crucial part of peace-building strategies. For Wiegink (2013: 3), who studied the case of RENAMO in Mozambique, these programmes start from two misleading assumptions. One is that ex-combatants would naturally want to return "home", without taking into consideration what home is and how it has changed during the war. The second assumption is that armed groups are seen mainly as military structures that can easily be dismantled, ignoring the fact that they also entail a web of social relationships. As the case of Sierra Leone illustrates, war-time mobilization networks tend to continue after the termination of conflict (Utas 2005). The same can be said about the networks of solidarity and support between excombatants. The example of RENAMO shows that ex-combatants do not simply fade away but continue to be an important identity in post-war politics and society (Wiegink 2013).
These cases illustrate how armed structures remain an option to those who have left them even after the formal end of war. While the return to armed struggle is a 'profoundly multi-layered and social process' (Wiegink 2013: 10), so is the individual decision to take part in it. The narrative of remobilization, however, obscures the complexities explaining it. It is mainly understood in terms of a return to violence and combat, itself the result of 'an interaction between entrepreneurs of violence, military affinities, intermediaries, and selective incentives' (Themner 2013). More importantly, it pays little attention to the agency of those being mobilized, the larger social processes and context leading to a return to armed combat, the effects of long-term membership of armed groups, and the socialization process this membership entails. Vigh (2009) introduces the notion of social navigation as an analytical starting point to understand the intersection between agency, social forces, and change. This concept helps to unravel the interactivity between the way social agents wove within social formations and the way these social formations move and change over time themselves. This 'motion within motion' tells us about 'the relationship between the environment people move in and how the environment itself moves them, before, after and during the act' (Vigh 2009: 425). In uncertain conditions, mobility thus cannot be understood without looking into the shifts in the larger social context itself. It is not only a reaction against but also conditioned by these shifts and is expressing an attempt to 'disentangle from confining structures, plot an escape and move towards better positions' (Vigh 2009: 419). Joining armed groups, in this sense, thus could be read as a search for a better position caused by social flux and change. As we will illustrate below, also a return to armed groups can be considered as a search for an environment that feels more predictable, feels more 'at home', and gives a better access to different forms of capital than the home community itself, which is experienced as a strange place because of changes happened during absence. Circular return, in this sense, is not only a search for opportunity but also a reaction to an increased sentiment of estrangement.
This merging of two different forms of movement (one of social formations and one of agents within these changing social formations) partly builds on Bourdieu's understanding of the interplay between agency and social force (Bourdieu and Waquant 1992). Vigh distances himself, however, from Bourdieu's notion of the social field for it 'implies a stability and demarcation of social structure that corresponds poorly to the reality of changeable and emergent social environments' (Vigh 2009: 427). We agree with Vigh's critique on Bourdieu's perspective on movement and the slowness of habituation; and rather than being limited to vertical movement within an existing field, movement indeed can also happen between different fields or social spaces, as combatants' circular return tends to confirm. Vigh over-emphasizes the volatility and change in social formations in conflict settings though and thus neglects their resilience to what he identifies as determining processes of entropy and instability. As the case of eastern Congo reveals, societies seem to be much more stable in conflict environments and speed of change less dramatic than assumed, despite turmoil and high levels of insecurity. Hoffmann and Verweijen (2018) argue that in conflict-affected eastern Congo, existing structures do not merely disintegrate but are constantly being reproduced by new actors and in new social spaces. Registers giving meaning and structuring these spaces do not disintegrate or disappear in war contexts but are constantly being reproduced by those resisting them. The same can be said about matrices of perception and practices of government. Despite the reduction or even disappearance of state authorities, these practices show a remarkable level of resilience.
In line with this, Schlichte (2009a: 17) points at the constant interactions between armed groups and the larger social environment and describes armed groups as figurations, or 'smaller social settings, groups and less structured collectives, and as ensembles of interdependent individuals ( . . . ) linked by asymmetric power balances, as they exchange favors or commodities, as they maintain emotional ties, and even as they fight'. These balances are rather precarious and volatile because of acts of consent and contestation but also the persistent action of actors.
These figurations resemble what Moore describes as (emerging) semiautonomous social fields. In our analysis, we rely on Moore's concept to understand armed groups and combatants' mobility between these groups and society. Similar to Bourdieu, for Moore, semi-autonomous fields are social spaces that have 'rule-making capacities, and the means to induce or coerce compliance; but it is simultaneously set in a larger social matrix which can, and does, affect and invade it, sometimes at the invitation of the persons inside it, sometimes at its own distance ' (1973: 722). We consider armed groups as such fields and want to understand how rules and regimens-in interaction with larger society from where they originate-create structure, generate power to obtain compliance and at the same time create a form of belonging. This attachment between the combatant and the armed group helps to construct identity, place, solidarity, etc. With Shariff (2008), we agree that the concept of social field can be considered both a unit of analysis and a normative order and thus has an analytical as well as an empirical value.
Boundaries between these fields, in our case between the armed group and the wider society, are processual and depend on the capacity to generate rules and coerce or induce compliance and to offer belonging, protection, security, and material benefits. In our analysis, we consider circular return then as the mobility of combatants between the semi-autonomous field of an armed group and their larger social context, and being driven by a search for a less fluid and more predictable and reliable social environment. This mobility is determined both by what emerging fields can provide and by the larger social environment where they stem from. It is the effect of a constant interaction of combatants with the same social fields.

Mobilization and Military Fragmentation in Eastern Congo
More than 15 years after the formal conclusion of the Second Congo War, eastern Congo's military landscape remains highly fragmented and extremely volatile, with new armed groups constantly being instituted, others disintegrating, and still others joining forces or fragmenting into different factions. Strikingly, over the last 25 years, the number of armed actors is showing a steady increase, this despite an inclusive peace process introduced in 2003. Patterns of mobilization have also shifted over time. Foreign-supported large-scale rebel movements and Kinshasabacked coalitions of rural-based nationalist self-defence groups, which characterized the Congo Wars (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003), today have been replaced by over 120 largely locally rooted and small-scale armed actors (Stearns and Vogel 2015;Vlassenroot and Verweijen 2017).
Different dynamics explain the persistence of armed resistance and mobilization in eastern Congo (Vlassenroot and Verweijen 2017). A first factor is the continued presence of localized and unresolved dynamics of conflict and insecurity, mainly centred around land access and other resources, citizenship, local power, identity politics, and the presence of foreign rebel groups. These dynamics explain continued claims by armed actors of self-defence and protection, and over communal rights and the right of self-rule. Such issues of contention are strongly defined and shaped by the existing governance framework of the Congolese state and the consequent practices of rule (Hoffmann and Vlassenroot 2014). A second set of factors includes the different connections gradually being established between armed groups and politico-military elites since the formal end of the Congo Wars as part of new and highly militarized politico-military and economic competition. This competition is mainly connected to electoral processes, political representation, army reform and the access to the spoils of resource exploitation (Stearns et al. 2013). And finally, as will be documented in detail in the next section, some of the effects of peace-building and state-building strategies have contributed considerably to armed mobilization. Attempts to reverse it through rebel-military integration, negotiations providing a share of peace-dividends, badly designed military responses and the lack of progress in dealing with local conflict issues, all have set in motion new claims to self-rule and to protection (Erickson Baaz and Verweijen 2013;Vlassenroot and Verweijen 2017).
While Congo's eastern parts have a long history of armed resistance (Vlassenroot 2013), recent dynamics of armed mobilization find their origins in the democratization process, announced by Zairian President Mobutu in 1990. This process provoked intense competition between the ruling elite and emerging opposition forces. In the then eastern Zaire, competition centred around citizenship and land access and political mobilization of existing ethnic antagonism, which eventually triggered the creation of a number of communal self-defence groups such as the Batiri (Masisi), the Katuku (Walikale-Kalehe), and the Ngilima (Beni-Lubero). While connecting customary leaderships and agendas to new local political networks, these groups also expressed a deep mistrust in Mobutu's political order. They were a symptom of a collapsing political system and order, explaining increased political fragmentation, a deepening economic crisis, and the breakdown of social services, including education. Faced with growing marginalization, lack of opportunities and supported by their communities, these armed groups presented an attractive alternative to growing numbers of rural youth. Their connection to politics and their social embeddness, gradually turned these groups into alternative structures of protection and political and social control, with ethnic affiliation as a key marker of inclusion and exclusion (Hoffmann 2014).
The Congo Wars only further confirmed the development of these alternative social spaces, attracting increasing numbers of (mainly male) youth. The first Congo War, caused by the regionally supported insurgency in 1996 of the Alliance des Forces Dé mocratiques pour la Libé ration du Congo-Zaïre (AFDL), provoked the armed mobilization of rural-based armed responses to what was experienced as a foreign occupation force. While in most cases these groups were not able, despite their nationalistic discourse, to transcend local agendas and build up any serious military capacity, they increasingly tried to impose their own rule in those areas under their control. During the second Congo War, which was triggered by the rebellion in 1998 of the Rwandan-backed Rassemblement Congolais pour la Dé mocratie (RCD) against President Laurent-De´sire´Kabila, existing armed groups revived and a next generation of groups was instituted, again out of patriotic sentiment. Some of these groups started to develop ties to political leaderships, business elites, and other big men, connecting them to local networks of profit and control (Stearns et al. 2013). At the same time, their increased capacities to generate resources and to impose their rule and induce or coerce compliance through the establishment of administrations de brousse (bush administrations), gradually detached them from, and reconfigured their ties with their social environments of origin.
The territories of Kalehe (South Kivu) and Walikale (North Kivu) offer an interesting illustration of a search for legitimacy in their control over territories, resources, and populations. In these territories, since the early-1990s, armed groups have caused a major impact on the local political and social order. An outburst of ethnic violence in neighbouring Masisi (North Kivu) in 1993, which was triggered by Mobutu's democratization process and intensified struggles over land and political representation, eventually also affected Walikale and Kalehe and set in motion a process of armed mobilization of rural youth which has continued until today. During the Congo Wars, these groups merged into a well-organized politico-military Mayi-Mayi resistance structure first against the Rwandan-backed AFDL rebel movement and since 1998 against the RCD. Headed by Padiri Bulenda and massively supported by local youth, it was more than just a resistance movement. It claimed the right to rule and presented itself as a political project based on registers and symbols of stateness (Hoffmann 2014). The movement was able to gain popular support by its reference to two issues of existential importance to local communities: marginalization and security. As we argued elsewhere, 'while the former concern revolves around the historical marginalisation of local communities in politics and governance, the latter frames local communities as in need of protection. These issues give meaning to armed groups' bids to local authority and legitimise their engagement in a wide range of governmental practices normally ascribed to the state, such as taxation and the provision of justice' (Vlassenroot et al. 2016). For its recruits, Padiri's armed group also evolved into an attractive and alternative social space giving access to social and economic opportunities, and power and providing social services in return. In the end, the armed group increased its rule-making capacities and means to induce its rule through the mobilization of different sets of social, symbolic, and political registers, while keeping (and redefining) a connectedness to the larger social environment and an overlap with other social spaces (Hoffmann 2015).
The peace process, which formally started in 2003, had little effect on the local politics of mobilization. Unresolved local conflicts, continuing security concerns, and communal tensions explain the refusal of many rural armed groups to integrate in the newly created Congolese army or to return to society. In addition, former combatants who did integrate in the Congolese army kept their former communal or territorial claims or were getting increasingly frustrated by the lack of recognition and their growing marginalization in the new military environment. New groups emerged and replaced disarmed and demobilized ones, while others disintegrated into smaller-armed actors. Failed reform of the security sector and DDR contributed to their proliferation as did new security threats produced by a lack of army protection and the continued presence of foreign-armed groups on Congolese territories. As a result, armed groups have gradually evolved into dominant power brokers, which are deeply concerned with ruling territory, people, and resources. They have become part and parcel of power dynamics, have colluded with local and national political and customary leaders, and have developed different techniques and strategies to impose or sustain their authority (Vlassenroot et al. 2016; Hoffmann and Verweijen 2019).
As part of a 'democratization of militarized politics' (Vlassenroot and Verweijen 2017), local elites started relying on armed structures to reinforce their power base and to forge access to or reposition themselves in the national political arena. At the same time, being connected to political elites for armed groups enhanced increased political leverage, authority, and access to resources (Vlassenroot and Verweijen 2017). As a member of parliament confirms, 'armed groups do not exist without the string pullers. They are at the local, provincial and national level; even sometimes regional. Certain economic and political operators derive their economic and political power and strength from these armed groups' (interview, national deputy, elected representative of Masisi, Kinshasa, June 2018).
This situation is also acknowledged by state authorities, who confirm politicomilitary elites' dominant position in eastern Congo's military landscape: 'we know there are people who activate these armed groups. We even know some names. But we cannot worry them so as not to create a general crisis in the country' (Focus Group, Interior Ministry Advisers, Kinshasa, June 2018).
The particular position of armed groups in political and security dynamics is confirmed by more recent developments. Since the conclusion of Congo's electoral process and the instalment of Felix Tshisekedi as the new Congolese president in July 2019, armed groups are forced to revise their position. Many of them have recently expressed their willingness to demobilize their forces and have sent some of their combatants to assembly points in the anticipation of their reintegration. Yet, other groups have joined forces with the Congolese army, which has started tracking down foreign-armed groups based in Congo. Whether these dynamics will result in a redesign of eastern Congo's military landscape and a reduction in the number of armed actors remains uncertain, given that links with political networks are largely intact, sources of contention unaddressed and new reintegration efforts not very promising, explaining armed groups' ambiguous stance (interview # 26, Bukavu, October 2019).

Disarming, Demobilizing, and Reintegrating Combatants
The disarmament, demobilization, and reinsertion of combatants, from the start of the peace process in 2003, have been a crucial part of peace-building, state-building, and stabilization efforts in the DRC. Recent estimates by the United Nations Organization Stabilization Mission in the DR Congo (MONUSCO) indicate that so far more than 124,000 combatants have gone through different formal demobilization programmes (Monusco 2019). While this as such can be seen as a considerable achievement, DDR efforts have not been able to stop the proliferation of armed groups nor reduce the level of violence in large parts of eastern DR Congo. Even more, as argued above, the number of armed actors has only further Navigating social spaces 11 increased and the mobilization of combatants further continued. Several factors explain this lack of real success of DDR efforts, already in 2007 depicted by Boshoff (2007) as a 'never-ending story'. Based on interviews with former combatants, Richards (2016) adds that while armed fragmentation may encourage desertion and participation in formal DDR, it can also have an opposite effect and provoke forced recruitment or voluntary remobilization. For Perazzone, the Congolese case shows ( . . . ) that combined with prioritising immediate security gains, it is the generalised politicisation of reintegration that most likely impeded coherent implementation processes (Perazzone 2017: 274).
Underlying assumptions about what should work, a lack of contextual reading of mobilization processes and security challenges and a rather technical approach to demobilization help to explain the limited impact of DDR in the DR Congo. Besides the flaws in the DDR approach itself, Vogel and Musamba (2016: 3) conclude that four factors affect patterns of continued mobilization in Congo: 'local security dilemmas between and within communities; resistance by elites with political agendas; perverse incentives for commanders; and social processes among rank-and-file ex-combatants'.
Demobilization of armed combatants in the DR Congo started in 2004 with the Community Disarmament and Resettlement operational programme in Ituri. This programme, which followed negotiations between militia leaderships, aimed at pacifying and stabilizing the district of Ituri that between 1999 and 2002 had witnessed one of the most violent episodes of the Congolese conflict (Vlassenroot and Raeymaekers 2004). At its conclusion in April 2005, more than two-thirds of the estimated 15,000 combatants were demobilized voluntarily and some militia leaders were appointed to senior posts within the Congolese army. Reinsertion assistance was offered to former combatants, yet with limited success.
A national-level DDR programme was announced in 2003 following the signing of the Final Act of the Sun City Peace Agreement that initiated the creation of a new integrated army and included a number of arrangements about the disarmament and army or civilian reintegration of signatory armed actors. The United Nations Organization Stabilization Mission in the DR Congo (MONUSCO) supports through its disarmament, demobilization, repatriation, reintegration, and resettlement (DDR/RR) section the DDR priorities and policies defined by Congolese government. In total, 330,000 combatants were listed by the former belligerents to be either integrated or demobilized, yet there is general agreement that this figure was highly exaggerated to increase their share of the peace dividend. In 2004, the formal National Disarmament and Reinsertion Commission (CONADER) was created and, in collaboration with the Military Integration Structure) and under the supervision of the Ministry of Defense led Inter-Ministerial Committee on DDR, had to organize the disarmament, identification, and categorization of combatants (Vogel and Musamba 2016). CONADER was funded by the World Bank, as part of the Multi-Donor Reintegration Programme, which was launched in 2004 and was covering seven 12 Koen Vlassenroot et al. different countries in the Great Lakes Region. The CONADER-led National DDR Programme (PNDDR) only started in 2005 and aimed at demobilization and disarmament and at socio-economic reintegration of combatants. The programme did not operate in isolation but was connected to larger efforts to consolidate peace and political and social stability and promote economic recovery. Its approach, however, caused frustration and disappointment with former combatants. Reintegration was limited to the provision of a 'reinsertion kit' and did not engage communities of origin in the accommodation of combatants' return. Incentives for DDR were also the same for all former combatants, regardless of their ranks in armed groups. Irregularities in the distribution of benefits and in the payment of allowances to demobilized combatants and the exclusion of those located in remote areas further contributed to a lack of confidence in the DDR process (Vogel and Musamba 2016). CONADER's approach eventually also added to the perception that violence paid off. As was the case for the different military reintegration efforts and local peace negotiations with armed groups, DDR was considered as an opportunity to get access to economic benefits. It contributed to the development of a local economy of armed mobilization and a cycle of constant recycling of combatants for economic gain.
In 2007, CONADER was replaced by the Execution Unit-Disarmament, Demobilization and Reinsertion (UE-PNDDR). One year later, the second DDR phase was launched, which lasted until September 2011. It was followed, in 2014, by the third DDR phase, which wanted to deal with some of the pitfalls of previous phases and invest in reintegration efforts. This phase took a year to get implemented due to a lack of sufficient funding caused by donor reluctance. It started with a community awareness campaign and recognized the need for communal socio-economic development (Carayannis and Pangburn 2018).As part of its civilian protection and stabilization mandate and in support of the national DDR III programme, also MONUSCO introduced a second-generation DDR strategy called Community Violence Reduction (CVR). In doing so, MONUSCO wanted to move away from its previous strategy that almost exclusively focused on security challenges and military actors. It was replaced by a community-based strategy that wanted to reduce the level of violence and the proliferation of armed actors in society through a more integrated approach. CVR thus was presented as a paradigmatic shift towards a community mobilization-, engagement-, and participation-driven approach (Monusco 2019).
Nevertheless, the DDR III phase was not able to reverse the dynamics of armed mobilization either. By the end of 2018, in total, about 5,000 combatants, mostly originating from the Kivu provinces (internal UEPN-DDR document on file with the authors), were either demobilized or integrated into the Congolese army. Besides direct issues related to the programme, such as the transferring of combatants from their homes and the harsh living conditions in the relocation centres, armed groups also distrusted the programme because of their opposition to the government army and Kabila's regime in general (Vogel and Musamba 2016). At the same time, armed groups became more outspoken in their discourse towards DDR. The joint strategy of demobilization and military operations created a fear Navigating social spaces 13 of being completely sidelined as military actors and this without local security dilemmas being resolved. It exemplifies that despite the adapted agenda, security objectives were still prioritized. At the same time, DDR III has provided local political elites with new incentives to position themselves, further politicising demobilization. Recent demobilization dynamics again point at its complexities and ambiguities and illustrate that the decision to leave the bush is guided by larger socio-political dynamics rather than the prospects of reintegration offered by welldesigned technical approaches.

Armed Groups as Social Spaces
In this final section, we analyse combatants' mobility between social spaces and their circular return to armed combat. As mentioned above, in eastern Congo motives behind and practices of armed mobilization have shifted considerably over time. While in the early-1990s, armed mobilization in rural areas was mostly an effort guided by customary authorities trying to guarantee the protection of the community and of communal land against their neighbours and during the Congo Wars nationalist ideologies added to these motives, today armed recruitment is determined by a complex interplay of motives, sentiments, and interests. As our interviews indicate, these motives exceed dominant political and economic rhetoric and often include individual incentives. Collective and individual incentives constantly interact, can strengthen, neutralize, or even contradict each other. At the collective level, issues including the unaddressed need for protection, land access and political participation have guided their proliferation. As a former armed group spokesman testified, 'the armed groups have real political demands which have never been resolved. Often these revendications are sidelined and only a demobilization is offered. It can't solve anything. This is why it is often the rank-and-file combatants who are demobilized. The group leaders stay or return to the forest' (interview, Kinshasa, June 2018).
At the individual level, being a member of an armed group should be understood as a search for access to opportunities and a way of making sense of a changing social environment; a response to individually or collectively experienced levels of insecurity and efforts of mobilization; and a response to what was lost during demobilization and what is perceived as a consequent remarginalization. Armed groups provide an alternative to civilian life, an alternative social space, and domain of activity. Overtime, combatants in eastern Congo have expressed a strong navigating capacity between such spaces and their wider social environment.
For many combatants, armed groups today are considered a refuge to deal with joblessness and a space of political and economic opportunity. These groups attract not only the rural poor but also teachers, students, traders, etc. As one 14 Koen Vlassenroot et al. respondent told us, 'armed groups are instrumentalised as spaces of transit for those in search of responses to their needs' (interview # 1 with civil society leader, Kalehe, July 2018). This account illustrates to which extend individual agendas have gradually become a key factor in the decision to (re)join armed groups and point at the recognition of armed groups as distinct social realities, providing a response to specific needs and offering a more predictable and reliable environment.
However, joining an armed group is more than just a search for opportunity. It is also about a search for belonging and identity, a response to a drastically changing security and social context. Security concerns continue to be of crucial importance, as is illustrated by the case of the massive mobilization of local youth in 2011 in Bunyakiri: 'the 2011 FDLR massacre of Kamananga which left many people dead in the Batembo community, and the rumor of the return of Tutsi refugees in 2012, caused a radicalisation of young people and their (re-)mobilisation into armed groups' (interview, Bunyakiri civil society leader, June 2018).
Yet, these are not the only driver of combatant mobility, as the following sections reveal.
Particularly those having been mobilized for a long time or from a young age, see their presence in armed groups as a learning process having shaped their personalities and identity. Joining an armed group is considered as entering a new social space, with its particular rules, hierarchies, and new codes of conduct.
'Suffering was there, but armed groups have also been a school to me. We have learnt to be prudent before acting, we have learnt how to live in a group with people from different ethnic backgrounds. The suffering was a school to us' (interview # 2 with ex-combatant, Kalehe, July 2018). This is also echoed by a combatant who told us: 'we have learnt to stay cool, to keep our self-control' (interview #3 with ex-combatant, Kalehe, July 2018).
Being part of an armed group, however, does not include a rupture with other social spaces. In fact, combatants in most cases remain connected to their home communities. This illustrates how social spaces overlap and armed groups operate in constant interaction with their larger social environment. This is the case for armed groups' leaderships, who maintain direct contact with customary chiefs, political representatives, and/or other community leaders, and also for their rankand-file combatants, who deploy different communication techniques to stay in touch with their families. Where mobile phone coverage is limited, 'ad hoc messengers', such as motorbike drivers, traders, or friends, are asked to pass messages, usually informing their families about their health.And during military operations close to their home villages, even physical visits are being made, this often without Navigating social spaces 15 any harassment from local authorities or members of the home community. Some combatants also regularly send money to, or are involved in business activities with their families. In other cases, armed group members trade with businessmen from their own community or have their own businesses, including ambulant trade, motor taxis, and houses for rent. Keeping links with the larger social environment is not without risk though as these might impose security threats to family member including regular visits from state security services.
Even if these accounts suggest that joining an armed group induces a process of socialization and adaptation to discipline and new codes of conduct, excombatants all confirm that combat life is tough and living conditions in the bush are harsh. For some, these conditions have been a cause to demobilize: 'I have suffered a lot while being in the militia because life was difficult. We ate rotten beans and maize paste for several years. It is the reason why I demobilised.' (interview # 4 with ex-combatant, South Kivu, July 2018).
Also operational and structural issues directly related to the armed group or its command triggered combatants to demobilize. In some cases, military leadership was considered being incompetent: 'to be directed by someone who is not educated and does not know how to manage combatants is disappointing. It is what I experienced in the Nyatura group with some incompetent commanders' (interview # 5 with ex-combatant, Kalehe, July 2018).
Others had to deal with authoritarian rule and commanders showing little will to reconcile in case of disciplinary issues. With this conduct being guided by military principles, there was limited space for negotiation. As one respondent told us, 'monoko ya Mokonzi nde re`glement -the discourse of the commander constitute the rules' (interview # 6 with ex-combatant, Kalehe, July 2018).
Frustrations are not limited to leadership issues though. Some complained that there was just not enough 'movement' (ambiance) within the armed group, which was partly due to a lack of capacity of their commander to initiate and make things moving.Pointing to a discrepancy between the rationale and interests of the combatant and of the commander, one combatant said 16 Koen Vlassenroot et al.
'siwezi bakia kwa commandant "Pasteur", il faut kwenda kwenye kuko ka movement -I could not continue in an armed group led by a "priest" (who does not offer opportunities to the "movement" or the access resources, to pillage or to abduct people), you have to go where the action is going, where the "movement is happening"' (interview # 8 with ex-combatant, South Kivu, July 2018).
In other words, the armed group is considered as an opportunity for selfrealization, while the 'ambiance' represents the search for different forms of capital. Yet, the same structures also impose limits to combatants' capacity to socially navigate in the combatant space. When combatants' agency becomes too restricted, the only option left is to try to leave the armed group and move to a different social space. This is not without risk, as it might be considered an act of desertion with the consequent punishment if caught.
It explains why leaving an armed group should be considered an act of resistance (against the armed group hierarchy) at the same time a claim to their own mobility and agency. If successful, it often includes a moment of temporary demobilization, either by choice or by lack of other options. Demobilization is in many cases a strategy of transition, or the creation of a bridge to another armed group, in the expectation of entering new social spaces offering better conditions or benefits. As some combatants told us, 'after having left my group as part of a campaign to demobilise child soldiers, I joined the Nyatura thinking I would gain from it (but in vain)' (interview # 10 with ex-combatant, Kalehe, July 2018).
These accounts indeed indicate that combatants easily shift armed groups (including the regular army) and demobilization is to be seen as a passage to find new opportunities elsewhere: 'those returning do not easily reintegrate into the old armed group because they have left for some reason, they had left some traces' (interview # 11 with ex-combatant, Kalehe, July 2018).
In some cases, the return to combat is inspired by new security threats but often it is also guided by individual interests: 'I joined the Nyatura because in the regular army I only had the grade of captain without an occupation or salary. You feel more relaxed when joining a group which has some of your friends or people you know' (interview # 12 with ex-combatant, Kalehe, July 2018).
In other words, not only economic opportunity but also the existence of social networks guides combatants' mobility.
These accounts also help to explain combatants' mobility and capacity to navigate, and their readiness to return to combat despite being demobilized.

Navigating social spaces 17
Demobilization is often considered as a failure of self-realization. Either through spontaneous demobilization or entering a formal DDR programme, a lot is expected from their social reintegration. Yet, in most cases, it is leading to a loss of what has been reached during the membership of armed groups and even leading to what Utas describes a remarginalization. Utas illustrates how in the case of Liberia, this remarginalization is caused by extreme poverty and lack of economic opportunities and is maintaining a war continuum facilitating the return to armed groups as a strategy to deal with individual social and economic insecurity (Utas 2005). Also, in eastern Congo, a return to combat is often an obvious consequence when the new advantages of civilian life do not respond to the expectations and do not compensate what has been lost. Those having gone through a demobilization programme and returning to combat life complain about two major issues. First, they express frustration with what is offered with the reintegration kit, which is to be considered an incentive to return to civilian life. In the words of one respondent, Some of the friends who had left the armed group returned to combat because they have expectations of receiving money like the demobilised but these expectations are not met, so they join armed groups (Stefano, Kigulube July 28, 2018).
Second, ex-combatants are faced with several difficulties in finding back their place in society. Former combatants are easily targeted in case of criminal acts happening in society, are faced with limited access to livelihoods or opportunities in comparison to their former comrades who did not enrol, and often feel marginalized or even rejected by their home society. In Kalehe, a civil society leader told us how local society easily takes revenge on former combatants in case of theft: there are already six demobilised combatants who were killed by the population after having been cached while stealing goods. It is not their fault, he said, as they had no other option due the failure of the DDR3 mission (interview # 13, Kalehe, July 2018).
Other respondents informed us that once having deposit their weaponry, legal cases against ex-combatants are introduced by civilians in response to acts they committed as member of an armed group: I remember a case in Mpofi, where an ex-combatant was sent to court. He had stolen pigs when being a member of an armed group and members of the villages had decided to hunt him down. When a community is not being prepared for the return of ex-combatants, they risk facing the settling of scores (interview # 14 with member of civil society, Bukavu, July 2018).
While these issues inform us about some of the constraints of demobilization efforts, at the same they point at the limits of combatants' navigation capacities between different social spaces. Being part of an armed groups creates new forms of belonging and identity and access to symbolic, economic, and social capital, yet also redefines the connections with other social spaces that complicate their return 18 Koen Vlassenroot et al. and integration. Keeping a legitimate presence in these spaces and navigate between their identities of (demobilized) combatant and civilian tends to be more challenging than expected.

Conclusion
In eastern Congo, armed mobilization and the constant recycling of combatants have become a persistent condition, illustrating the existence of a war continuum. Combatants are in a continuous state of mobility between civilian and combatant life and have expressed a strong capacity to navigate between different social spaces. This social navigation cannot be fully captured when reducing the analysis of armed mobilization to security dynamics, which indeed ignores combatants' agency and larger processes of socialization and social rupture through armed mobilization. Armed groups are relatively autonomous social spaces with their own rules and procedures yet connected to their larger environment. They are spheres of action and spaces of opportunity giving access to different forms of capital. Incentives to join these groups have evolved over time and represent a constant interplay between collective and individual ones. Our analysis shows that these individual incentives go beyond the search for economic opportunity. Even more, previous experiences in armed groups turn to be a decisive factor in their potential return to combat after having gone through a civilian reintegration programme. When these programmes do not compensate for what has been lost through their demobilization or instigate a remarginalization of excombattants, they tend to contribute to what we have described as a process of circular return. This circular return points at combattants' navigation capacity between social spaces. It can be considered as an act of resistance against their commanders in case of hardship or lack of 'movement', or against local society in case of failed reintegration. The constantly navigation between different spaces expresses the formation of a 'split presence' aimed at keeping their legitimate presence in these spaces. So far, the different DDR approaches in the DRC have largely ignored this navigation capacity of combatants and the socialization effects, including the production of new identities, or armed mobilization. As a consequence, rather than contributing to a reduction in the number of armed groups, these approaches tend to contribute to the circular return of their members.