[The guerrillas] had been fighting the army and many soldiers had been killed. They were poor lads who could barely carry such heavy equipment and were struggling in the marshes with water up to their waists against guys who knew the terrain inside out. The guerrillas launched an ambush. Afterwards, the paramilitaries came and told people that it was the fishermen’s fault that the soldiers had been killed, because the fishermen were supplying food to the guerrillas. They called us together at eight in the morning to tell us that. They took the fishermen and killed them in front of the whole village. 
1This interview extract illustrates many of the phenomena that are frequently encountered within societies where the proliferation of violence challenges the methods available to social scientists for analyzing states and political authority.  In recent years, numerous studies have called into question the simplistic analysis that associates such situations with state weakness or even failure. Sociologists, political scientists, and anthropologists working on the actual ways in which political power is exercised have argued convincingly that there is no zero-sum game between state power and the influence of non-state armed groups. Some scholars are interested in the study of non-state forms of authority, which do not constitute “anti-states”, but are often entangled in networks that cross boundaries, undermining the division between state and non-state.  Research into situations of civil war clearly shows that these are not realms of chaos and anomie, but that they cannot be dissociated from the emergence of political orders, even though these may be local, fluctuating, and unstable.  Violence provokes the formation of multiple types of authority, which coexist within a framework of collaboration, negotiation, competition, and conflict.  The concept of “twilight institutions”, as proposed by Christian Lund,  highlights the lack of clearly defined authority that results from such local and unstable arrangements, although the term is not exclusively intended to study armed groups. In such contexts, where localized and fragmented forms of authority proliferate, the idea of the state is not necessarily rejected by the actors. On the contrary, violence can be seen as a vector of social mobility, or even of political integration, within a state field that is recognized as a central arena of competition. Accordingly, the paradox noted by Timothy Raeymaekers, Ken Menkhaus, and Koen Vlassenroot is observable in a wide variety of conflicts:  despite multiple forms of non-state authority, the state is often the end in view for myriad violent initiatives. 
2This study is not intended to go back over these conclusions, which have been widely discussed and supported by recent work.  It focuses instead on analyzing a “borderline”  case in the light of the existing literature. Colombia can be cited, in many respects, as an example of “privatization of the state”:  in many parts of the country, authority has been exercised through a form of delegation, or “discharge”, facilitated by close contacts between the army, local political elites, and violence entrepreneurs. However, in parallel with the multiplying “hybrid”  ways of exercising authority, means of dealing with violence have developed within the state itself. Accordingly, criminal law and security policy continually implement new modes of intervention and new techniques of governance. This has led to an expansion of the political and administrative framework, to the development of new forms of action, and to the maintenance of the state’s position at the center of most armed activity. It has also shaped the image of the state as a body that is separate from society,  situated “above” it,  and responsible in particular for the task of classifying and defining types of violence.
3This analysis of the Colombian case will focus on one set of actors in the armed conflict, the paramilitary groups, who practiced various forms of violence between the early 1980s and the mid-2000s. Following their partial demobilization and the extradition of a significant number of their leaders to the United States in 2008, their former members are now divided between DDR programs (for disarmament, demobilization, and reintegration) and criminal activity, although these two pathways are not mutually exclusive.  The territorial domination exercised by these groups has also given rise to the emergence of new armed networks, which in turn replicate the paramilitaries’ command structures and social base.
4The starting point of this study is a dual interpretation of the concept of violence, which appears in my data as both a practice and a public problem. The analysis of violence carried out by paramilitary groups and the study of “paramilitarism” as it has been dealt with by various institutions do not represent two independent dimensions. On the contrary, their linkage provides us with a method that is well suited to studying the impact that the proliferation of violence has on the state. Taking account of both these dimensions leads to a rejection of the dichotomous view that the supporters and enemies of the established order must be systematically opposed.  Paramilitarism does not fit into either of these categories. Since the paramilitaries support the army as well as local political and economic elites in the fight against guerrillas, they should not be regarded as external to the political framework simply because of their criminal activities.  At the same time, however, the problematization of paramilitarism as a form of criminal activity—led by the courts or the institutions responsible for security policy—must be taken seriously, making it impossible to categorize these groups as mere auxiliary forces.  The existence of intense intra-state conflicts regarding the stance that should be adopted towards the paramilitaries places these groups in an ambiguous position: their power is based on their ability to control territories and to exert repressive violence against anti-establishment or marginal groups. At the same time, some manifestations of this violence lead them to be discredited as dangerous militias or as mere criminals, devoid of political motives.
5Being neither simple bandits nor auxiliary combatants, paramilitary groups illustrate the complex relationship between the state and violence in a context where violence is a common means of exercising political power. How can we make sense of this positioning? What political processes are at the root of this relationship between the state and non-state violence? How does the state deal with this violence and, in turn, how does an armed actor make use of the opportunities for action, constraints and resources that state action represents?
Map: Department of Magdalena
Map: Department of Magdalena
6I argue that paramilitarism is a form of relative dissidence, a range of tactics aimed not at opposing the political system but at acquiring mobility within it. Overturning the view of a state “besieged”  by proliferating violence, the study of paramilitarism reveals that the state is actually able to impose ways of dealing with violence that are endorsed by the armed groups themselves. These groups then position themselves strategically in the political arena, seeking support from within the state; they also model their appearance to conform to an ethos of combat, rooted in a particular history and institutionally disseminated. The concept of relative dissidence therefore reflects two separate realities. It certainly allows for the fact that violence represents a common form of social and political mobility: it facilitates territorial control, which has a particular bearing on electoral influence. However, this form of dissidence is relative or “normed” because it is located within the framework of the institutions that manage violence and define the scope for political action, even when it takes extremely brutal forms. The rest of this study examines these two dimensions in sequence.
7The analysis is based on some of the results of a field survey carried out in 2009 and 2011 in Colombia,  over a total period of almost a year, during which time about a hundred structured interviews were conducted and numerous observations were made. This research was accompanied by extensive consultation of legal archives, accessed through both judicial institutions and plaintiffs in civil proceedings.
The Practice and Benefits of Violence
8In 2009, when I arrived for the first time in the department of Magdalena, in the north of the country, the paramilitary groups had only recently been demobilized. New armed groups were being formed and were recruiting heavily from within the community of former fighters where I conducted my first investigations.  Because they had less control over local communities, however, I was able to gather a wealth of accounts of the ways political power was wielded in areas controlled by paramilitaries. These accounts are supported by judicial sources that magistrates were beginning to produce as they investigated the relationship between the paramilitaries, politicians, and state officials.
9Although the creation of militias by landowners, political parties, and the military is a longstanding phenomenon in Colombia, the last two decades of the twentieth century were characterized by the significance of the drug economy and a desire for national power projection. In many regions of Colombia, those who came to be known as paramilitaries were the heirs of violent enterprises that emerged between the end of the 1970s and the beginning of the following decade, in the context of expanding drug production. These groups became politicized during the 1980s and embraced an anti-communist approach. This allowed them to form alliances with local elected officials, who were happy to be guaranteed “safe” elections in a context of active mobilization of leftist parties and of rural and urban social movements. The next decade was marked by the emergence of a form of national coordination of paramilitary militias, namely the United Self-Defense Forces of Colombia (AUC). This group was actually more of an organizational facade than a truly unified group. The creation of the AUC was also partly motivated by the expansionist ambitions of a network of paramilitary groups from Urabá and Córdoba (on the western part of the Caribbean coast). As the history of the paramilitary groups is rooted in specific regional contexts, this article will focus on the particular case of Magdalena rather than making general statements about the development of these militias across the country.
Guardians of a Violent Order
10The first groups of violent entrepreneurs  in Magdalena grew up around two clans: the Giraldo and Rojas families. Both had migrated from the interior of the country and arrived in the area in the 1960s. Despite their migrant origins, they used a rhetoric of autochthony against the traffickers who controlled the most profitable stages of the trafficking. Traffickers from the neighboring province of La Guajira were thus ousted by armed force, or their activity was limited to a few restricted areas. At the end of the 1970s, Hernán Giraldo and his offspring were in control of a large part of the northern slopes of the Sierra mountains, while Adán Rojas and his family controlled the western slopes.
11These violent enterprises were to be transformed during the 1980s. The arrival in the region of the guerrilla groups of the Revolutionary Armed Forces of Colombia (Farc) and the National Liberation Army (ELN) led them to take up the anti-communist cause. Their intention was to protect landowners from racketeering rebels, to limit the influence of the communist unions, which were perceived as being close to the guerrilla groups, and to suppress all forms of leftist political mobilization, which was threatening the dominant position of the incumbent political elites. From the end of the 1980s, the assassination of leading figures in politics and in the trade unions was part of this strategy.
12This politicization of a criminal milieu can be explained by two key factors: firstly, the guerrillas’ eagerness to take a part of the drug trafficking profits, in the form of the collection of a “revolutionary tax”, made these groups a threat to the entire trafficking community. Secondly, the politicization of these violence entrepreneurs can equally be explained by the alliances that they formed with elected officials and local economic stakeholders. These relationships, which initially developed through the process of sharing out the plentiful proceeds of the marijuana trade, were reconfigured to provide protection to the incumbent stakeholders when their positions were threatened by the presence of the guerrillas and related political mobilizations, as well as by trade union action.
13Despite paramilitary violence, the influence of the guerrilla groups continued to increase during the 1990s. The lack of impact of counter-insurgency violence facilitated the arrival of the AUC, whose forces were better trained and equipped with more sophisticated weapons. The majority of AUC officers and troops were from other areas of the coast or from the geographical department of Antioquia. Local groups were absorbed into their ranks. The Rojas family militia aligned itself with the AUC all the more readily because it had been severely weakened by Farc attacks. The Giraldo faction put up a resistance that was, however, rapidly defeated.
14AUC advances on the plains and urban areas of Magdalena resulted in the guerrillas withdrawing towards the mountains and to some districts of Santa Marta, the department’s administrative center. The guerrillas did not attempt to halt the paramilitary advance or to protect the communities whose areas they controlled: this can largely be explained by the fact that they had only recently settled in the region and were not strongly rooted. From that point, the paramilitaries set up a very strong system of social control based on information networks, roadblocks, and the use of threat.
15Paramilitary groups took control of territory in the first place by controlling population movement, as described by a trader from the village of Monterrubio (near Fundación):
As my daughter was sick, we had to take her to Santa Marta for treatment. When we went to the town, we would buy goods that we sold afterwards in the village, like clothes, radios, things like that. They didn’t like it, they told me, “You either go or you stay put, but stop all this coming and going“. I replied that I had a sick daughter and that I had to travel. “Even your own mother wouldn’t believe that story“, they told me. They sent me packing. [.. . ] The same thing happened to my friend, Juan Gamarra; he had a car and he used to transport goods for the people living in the countryside. They killed him after accusing him of taking food to the guerrillas. 
17In practice, there was an obstacle to controlling population movement: most of the AUC troops came from outside the region and did not know the inhabitants. They therefore had to set up some means of information management. Information administered by the police was therefore put to use to help with surveillance of the population. One of my interviewees described a partnership of this type, in which the paramilitaries demanded that every inhabitant of the village obtain from the police station a document proving their place of residence, since this information does not appear on Colombian identity cards. The paramilitary patrols could then distinguish between residents and outsiders.
18This process of surveillance also relied on the inhabitants themselves, who were encouraged to denounce any suspected contact between their neighbors and the rebels. This, of course, incited a number of wrongful denunciations, where paramilitary violence was exploited as a means of settling private conflicts. In the accounts we collected, interviewees described the fear aroused by those inhabitants who were close to the paramilitaries:
A lot of deaths were due to false information. For example, I did not get on well with my neighbor, the guy really annoyed me. So, if I’d been friendly with a paramilitary, I could’ve told him—this guy is a guerrilla. They would’ve come for him and killed him. They killed a lot of innocent people like that. 
20In the exercise of their power, the paramilitaries thus incorporated individuals, as participants in the social control system. It did not matter that some of these denunciations were malicious and self-serving: the role of informants was not only to provide information to the organization, but also to make the inhabitants feel that they were constantly under surveillance.
21The apprehension of violence is also a clear factor in its effectiveness. In other words, when paramilitary groups kill, they seek not only to end life, but to do so in such a way as to produce terror. The macabre staging of massacres demonstrates this use of fear as a resource. The first incursion into a village was often characterized by acts of barbarism and the way bodies were treated was a central element. A worker from the village of Santa Rita (Remolino) explained:
The village prostitutes were among the first to be killed. They [paramilitaries] said that they did not tolerate prostitution in areas under their control, and that the prostitutes were sullied [sucias] anyway because they had slept with guerrillas. It was horrific, they cut them up with a chainsaw in front of everyone and then threw their remains into the holes of the electric poles, they ripped the poles out of the ground and put them back afterwards. 
23Moreover, the paramilitaries made an open display of their presence, without being denounced by any of the authorities. No one reported them, and they were uninhibited and ostentatious in their displays of power. This impunity is particularly striking in the many reported cases of selective killings that took place in broad daylight on central city streets. The very visibility of this violence reinforced the power of the paramilitaries. As one teacher from the town of Aracataca said:
In Aracataca at that time , there was a white 4x4 that was creating a lot of fear. Whenever we saw the vehicle, we knew that someone was going to die, because every time it appeared, it brought paramilitaries who were coming to kill someone. People used to call it the little dove of death [la palomita de la muerte]. 
25Interviewees all highlighted the general feeling of helplessness and the absence of protection, not because the state was absent, but because its representatives—both police and military—took no action against the acts of violence carried out by the paramilitaries. Violence appeared to be inevitable, since denouncing it was pointless.
26However, despite the armed paramilitary’s hold over the daily lives of individuals, the fragmentation of violence described here did not lead to the marginalization of state authority in zones controlled by the armed groups. As I will now demonstrate, the violence should not be interpreted as a challenge to state authority, but was instead part of the process of institutionalizing the collaboration and collusion that existed between armed actors and political leaders.
27The paramilitary’s territorial control resulted in significant electoral capital. In the areas controlled by these groups, voting took place under duress and was accordingly a matter for wheeling and dealing with local political actors. These collusive relationships allowed the latter to secure their election, or simply to remain in the running in a context in which violence was an essential resource in terms of political competition. In exchange, the paramilitaries gained access to local budgets, which were tapped using classic corruption techniques (notably via channels of public procurement). The following analysis briefly outlines methods used by the paramilitaries to control voting patterns, before focusing on how they exploited their electoral capital.
28Physical violence was not always necessary to control the electorate. In the most tightly controlled areas, domination over all aspects of everyday life, as described above, turned elections into a mere sham. Electoral registers and reports were completed in advance; in these cases, voters did not even have to tick the name of the nominated candidate.  In the words of one inhabitant of Chivolo:
When you went to vote, they had already voted for you, you signed and the returning officers gave you the certificate to show you had voted.  Or you went to vote and, as you moved up the queue, the officials ticked the name and you put the ballot paper in the ballot box, just to pretend you had really voted [.. . ]. If someone didn’t turn up to vote, the officials signed in their place and afterwards a paramilitary guy would bring their certificate to their house. 
30However, violence could be used, especially to quell any signs of resistance. Those individuals who held some form of power within a community were the most exposed. This was confirmed by one interviewee, who lived in Fundación:
We ran an agricultural co-operative with my uncle. They came to us to tell us that we and everyone else in the co-operative had to vote for a specific candidate. We told them we couldn’t tell people who to vote for. They said: “Everyone here must vote for José Gamarra, otherwise you’re dead“. That’s why they killed my uncle, because they said that if you’re not with them, you’re against them. 
32The methods used by the paramilitaries gave them a firm grip on the electoral process. In the 2000 and 2003 local elections, candidates supported by the AUC were elected to the post of governor, which carries significant powers in the Colombian political system. In the 2003 elections, Trino Luna, who was close to the paramilitaries, was even elected as the sole candidate. With more than 81% of the total vote, blank ballots were the only votes cast against him.
33Such situations, in both Colombia and other comparable cases, are most frequently interpreted as cases of “state capture”. According to this understanding, criminal groups outside the political order use resources such as violence or money to take over state institutions, preventing them from fulfilling their role to protect the general interest.  This type of analysis, which suffers heavily from its normative bias, prevents us from analyzing these relationships as social configurations that share in the exercise of political power. As suggested by Jean-Louis Briquet and Gilles Favarel-Garrigues,  it is more heuristic to analyze the circumstances of collaboration between violence entrepreneurs and professional politicians, the situations that encourage violence to flourish in the political arena, and the way that this violence transforms access to positions of power.
34The example of the 2002 legislative elections serves to clarify this point. According to his personal papers, the paramilitary leader Jorge Cuarenta  met with parliamentary candidates on 22 November 2001. The purpose of the meeting was to form “pairings” consisting of one candidate to the Senate and one candidate to the House of Representatives.  They divided the department into three electoral districts, based on the distribution of the population. Each pair was guaranteed the vote in one of these districts. According to the Supreme Court, in return for the guarantee of election, candidates to the Senate paid 350,000 dollars to the paramilitaries, and candidates to the House of Representatives paid 175,000 dollars.
35To ensure the election of his candidates, Jorge Cuarenta put pressure on the local representatives whose election he had secured in 2000. Thanks to this support, the candidates put forward by the paramilitary had spectacular results. In the southern municipalities of Magdalena, the paramilitary’s candidate won 32,543 votes against his competitor, who received only 513 votes. In the municipalities that formed part of these electoral districts, votes for a candidate were sometimes aggregated at more than 90%. This was the case, for example, in the municipality of Tenerife, where the candidate to the House, Jorge Caballero, who was supported by the paramilitaries, received 97% of the vote, a proportion that was previously unheard-of.
36The district-based method was not only useful in optimizing voting to ensure the victory of allied candidates. Jorge Cuarenta, who came up with this system of apportionment, was also seeking to position himself as an indispensable intermediary between the great electoral barons of the department and the voters, so as to diminish the power of the former. One politician gave the following analysis:
The aim was to send people to areas where they had no historical footing, where they did not know the mayors, the councilors, the electorate, the whole of the patronage machinery in fact! They sent guys from the North down to the towns in the South and vice versa. In the end, the candidate was elected, and by a strong majority! But he wasn’t familiar with his area. In the meantime, someone else had been elected to his former fiefdom. Once he was in the Senate, the politician was dependent on the paramilitaries, because they were the ones who had got him elected. 
38This system aimed to uproot elected members, in order to deprive them of the territorial foothold that formed part of their electoral capital. This uprooting did not mean that politicians found themselves at the mercy of the paramilitaries. While they may have had to make concessions to them, they also participated in establishing the system of power distribution. Above all, they remained in control of the resources that were integral to their relationship with public institutions.
39Since decentralization reforms were implemented in the 1980s and 1990s, local political figures and national representatives have played a key intermediary role between local levels and central institutions, which often have no other representatives in the area.  This increased the capacity of the elected officials to distribute jobs and financial resources, but, on a greater scale, also enabled them to control a wide range of public budgets. This was a key factor in their bargaining with the paramilitary militias.
40The case of the hospitals clearly illustrates this point. The hospitals are managed independently and are in possession of very large budgets in comparison with other public bodies. As a result, the appointment of hospital directors was a strategic issue for both the most influential local elected officials and the paramilitary leaders. Carlos Tijeras, commander of the William Rivas Front of the AUC, explained:
We had decided that the directors of the hospitals should be approved [by the AUC]. We looked at their resumes and interviewed the candidates. That’s how we managed to cohabit. They [the politicians] kept their positions in the administration and were in charge of the whole meritocracy show [.. . ]. They were committed to paying a tax on contracts, which was taken as law throughout the area. 
42Influence was exerted on the management of public hospitals alongside a selective violence intended to secure the networks that would facilitate the misappropriation of funds. For example, in April 2004, the director of the hospital in the municipality of Zona Bananera, José Alejandro Lacera, was murdered by paramilitaries. Carlos Tijeras later acknowledged responsibility for this crime during the legal proceedings against him. The year before, Zully Esther Codina, a union leader and bookkeeper at the Santa Marta central hospital who was also a journalist for local community media, was murdered as she left her home: according to sources close to her, she had found documents proving embezzlement in the hospital and was about to make them public. 
43But the channels of corruption were sometimes more sophisticated. One important enabler was the contracting-out of public services. The awarding of contracts to companies that were “friendly” to the paramilitaries, or were controlled outright by front-men, became a means of appropriating public funds. Several examples serve to illustrate this process.
44In Magdalena, eleven municipalities privatized local tax collection  to profit the regional women’s foundation, Mujeres de la Provincia, a legal front for the AUC. The same foundation also secured a contract with the regional electricity supplier for payment collection, which could have yielded 20% of the amounts paid.  In the department of Sucre, also located on the Caribbean coast, a system for misappropriating funds was established in the town of Coveñas, which had become wealthy from taxes paid by the national oil company (Ecopetrol). Here businesses served as proxies and signed contracts (for the improvement of the garbage collection network, for the purchase of uniforms for municipal employees, or for the construction of local roads) that were never actually fulfilled. For a commission, the businesses passed the money on to the paramilitaries, who then paid local political leaders. This system would have raised more than 5,000,000 dollars for the local paramilitary leaders in 2003 alone. 
45These corruption strategies, and on a broader scale the collusive relations between violence entrepreneurs and professional politicians, do not indicate that practices associated with the exercise of power had changed significantly in this region, or elsewhere in Colombia. At most, the presence of the paramilitaries contributed to the centralization of resource-sharing networks and helped to concentrate the capacity for violence in the hands of a few key players in the area. However, these resource exploitation networks remained volatile, particularly because of their clandestine nature. In fact, even though they may control the flow of information in their territories, such politico-criminal groupings are powerless when a national authority—such as the judiciary or a body with administrative oversight—intervenes. This illustrates the limitations on building local power through collusive alliances. Unless we fall for a normative and nominalist definition of the state, such practices cannot be considered as contributing to its weakening or failure. Rather, they indicate the importance of collaborative relationships and the appropriation of public resources, which, far from sidelining the state, make it the central arena for resource sharing.
46However, relations between the state and paramilitary groups cannot be reduced purely to these collaborative relationships, as the anthropologist Aldo Civico suggests in a book that presents itself as an “ethnography of the Colombian death squads”,  because, parallel to this proliferation of “hybrid” forms of authority, ways of using violence developed within the state structures themselves. These methods would largely determine paramilitary strategies, since they formed the framework within which violence appeared as a means of political action.
The Transformations of the Paramilitary Problem
47If this situation can be described as one of relative dissidence, this is largely because paramilitary activities did not take place within a state vacuum but, on the contrary, in a space that was saturated with political definitions of violence. Broadly speaking, the paramilitary trajectory can be described as a dual process of exclusion and the search for recognition. Firstly, we can describe it as a process of exclusion because, while the paramilitary were originally thought to be useful in assisting in the counter-insurgency effort, they were increasingly categorized as criminals because of their links to drug trafficking. In their search for recognition, they countered this description with the claim that they should be acknowledged as “self-defense groups”. Fundamentally, the history of the “paramilitary problem” revolves around legitimizing and delegitimizing the use of private violence against the enemies of the regime. Let us briefly return to these two processes.
From Military Auxiliaries to Mob Army
48Army intervention is often apparent in the origins of the first paramilitary groups. The military, who provided weapons or collaborated in the establishment of militias, were acting in accordance with legislation of the 1960s that allowed the creation of “self-defense committees” (juntas de autodefensa), which were formed from local populations and whose main purpose was to provide information to the army and to act as a first response unit in the event of a rebel attack. The creation of these committees was not only enshrined in legislation; it also featured in officer training manuals. These manuals define the self-defense committee as “a military-type organization made up of civilian personnel selected from within the combat zone and trained and equipped to carry out maneuvers against the guerrilla groups threatening the area and to act in co-operation with the army during fighting”. 
49A former army officer described the organization of these auxiliary forces in the following terms:
The philosophy was that, since there were not enough police and army forces to protect the villages, it was necessary to look for local inhabitants who were willing to help with security [.. . ]. They were trained in the use of weapons and organized into groups of ten to fifteen people. In their village there would be an army intelligence officer, who pretended to be a civilian and who commanded the self-defense committee. Whenever the news came in that the guerrillas were approaching, the committee would take up arms and station themselves ready to retaliate, until the army arrived. 
51These relationships of complicity met with harsh criticism from the beginning of the 1980s. In 1983, Colombia’s Inspector General (the procurador general de la nación, responsible for overseeing the administration of public offices) made public a list of army men responsible for acts of violence against civilians committed in collaboration with paramilitaries. At the same time, the leader of the social democrat faction of the Liberal Party claimed that paramilitaries protected by the military had murdered local villagers whom they accused of collaborating with the guerrillas. Throughout the 1980s, in addition to these denunciations, emerging human rights organizations accused the state of using the paramilitaries to wage a “dirty war” against the opposition. However, all these protests were countered by a unified reaction from the armed forces, which prevented the progress of any investigation.
52Private violence by the paramilitaries was discredited not in the name of human rights but of state security. Indeed, as early as 1984 a violent clash erupted between a section of the drug trafficking milieu and the state, with the first seeking to influence the criminal policy of the second. Although the paramilitaries were not central actors in this conflict, they often took sides with their allies in the drugs world. These alliances led them to assassinate elected officials and judges, thus overstepping their primary role as anti-subversive militias. At the same time, police and intelligence officers were attempting to take the drug problem into their own hands in order be recognized as having a specific role, distinct from the military,  who were at that time the dominant security professionals. They came down heavily against all drug-related activity, establishing a new priority in terms of threats faced. They claimed that although the most pressing threat until the beginning of the 1980s had been the communist revolution, the emergence of powerful drug traffickers had become an even greater threat to the stability of the state. 
53Accordingly, while the militias, since the beginning of the 1980s, had been deploying high levels of violence against any form of activity that they associated with the political left, it was not because of this brutality but because of their connections with drug trafficking that they were perceived as a threat. A presidential adviser who was in office at the time recalled that,
It’s sad to say, but until they killed officers of the law and leaders of the traditional political parties, no one got upset about things. The assassination of members of the UP [Patriotic Union or Unión Patriótica, a left-wing party supported by the Farc guerrillas] was seen as a settling of accounts between armed actors. They were associated with the Farc, and no one in Bogotá was worried about guerrillas dying. People used to say: “Whoever lives by the sword, shall perish by the sword“ [él que a hierro mata, a hierro muere]. 
55These concerns gradually led to a reinterpretation of paramilitarism as a danger, not because of the intrinsic characteristics of these groups but because of their association with the major threat of the time, namely violence linked to drug trafficking. This explains the rapid change in the way the paramilitaries were viewed. In 1989, a series of decrees repealed the legislation that had been the basis for the creation of the “self-defense committees”. At the same time, the Colombian Supreme Court ruled that the arming of civilians was unconstitutional. Finally, the government created the Special Task Force (Cuerpo Especial Armado, or CEA), an elite police task force to fight “private justice groups”, as the paramilitaries were now known within government circles.
56These measures would prove to be ineffective. No effort was made to combat the collusion between military and paramilitary forces. It was not until the second half of the 1990s that criminal investigations against members of the military began to threaten the total control held by the army over the way that the counter-insurgency endeavor was run. It may, then, appear that the criminalization of paramilitary groups was purely a facade. This is not the case, however: as the following pages show, the process of criminalization determined the behavior of these groups, as they sought recognition as armed groups with a political basis. 
57The inherent association of these groups with the drug economy is grounds for political disqualification. In the face of this, the paramilitaries attempted to promote a different interpretation. They would present themselves not as a “mafia army” but as “self-defense” groups, having taken up arms in response to guerrilla aggression and the absence of the state. This contention was based not only on a reconstruction of the paramilitaries’ history but also on a transformation of their social structures, in an attempt to categorize themselves as armed groups with a political basis. I will highlight the significance of this category in the Colombian context, before describing the efforts made by the paramilitaries to be viewed in this way.
58The paramilitaries’ claim to be recognized as having a political basis drew on a history of political participation and social mobility through violence. As Luis Martinez described in relation to Algeria, Colombian history is characterized by various forms of activism that place value on violence as a means of access to power. These “imaginaries of war”  go back to the birth of Colombia as an independent state in the nineteenth century, when civil war was a common means of gaining control of the emerging state, of safeguarding the autonomy of the provinces, and of influencing taxation policy.  These imaginaries have been transformed and adapted in modern times, as successive governments have defined the guerillas as subjects of peace negotiations, with a claim to political legitimacy. They are also nourished by the local implementation of international humanitarian law, which defines belligerent parties according to their organizational characteristics—such as the control of a territory or responsible use of command. Furthermore, these imaginaries materialize through the legal category of political offense. Defined by law as a “crime of conviction”,  the political offense is a crime “inspired by an ideal of justice” that is nevertheless characterized by the violation of “the constitutional and legal order”.  The Constitutional Court of Colombia, while recognizing that political criminals choose “wrongful or disproportionate means”, clearly differentiates them from “those who pursue disorder to an inherently perverse and selfish aim”. 
59This is not to say that the sources of violence are to be found in either social, legal, or ideological bases. It can, however, be stated that the representation of the armed actor as a political actor draws on a history of armed political activism. Moreover, in a conflict punctuated by recurrent cycles of negotiation, an armed group’s access to a political status meant that its members might be considered legitimate bargaining counterparties by government authorities. During negotiations over demobilization, the status of political offense would grant them favorable judicial treatment.  However, looking beyond the diverse local dynamics that influenced the formation of the first paramilitary groups, most of them did not align with accepted understandings of war, especially as defined by comparison with the rebels. Members of these paramilitary groups were almost always trained in arms in criminal circles.  They did not have military training and generally operated in small, loosely structured groups.
60Various proponents of paramilitarism sought to construct and promote a different image. First among these was Carlos Castaño. Although he had been a member of the mob, he embraced counter-insurgency thanks to his brother, Fidel, in the 1980s.  After Fidel’s death, Carlos began to form a paramilitary group with a political identity and a national mission. This was what he claimed in the first interview he gave to a journalist:
We are not a private security firm, nor a family business, as some people describe us. We are a national counter-insurgency organization; we aim to create a self-defense presence wherever there is a guerrilla presence [.. . ]. Because every day the state, with its armed forces, proves incapable of controlling the advance of the guerrillas. So we must walk alongside our enemy. 
62This group, which in 1994 named itself the Self-Defence Forces of Córdoba and Urabá (Auto-defensas campesinas de Córdoba y Urabá, or ACCU), modeled itself on the army and the guerrillas. The aim was to project the image of a politically motivated armed organization with a responsible leadership.  The United Self-Defense Forces of Colombia were formed on the basis of this same model in 1997. This brought the amorphous paramilitary together under one banner as an organization with unified command, a national strategy, and a single agenda.  This led to a proliferation of supposed positions of authority within the group. A “joint staff”, a “political steering group”, and even the post of “Inspector General” emerged. In the Colombian context, in which relations between the state and armed groups are deeply marked by successive negotiation cycles, the creation of a unified and structured group was essential to the process of changing their image from that of local militia to national political actor.
63It is important to understand the link between these transformations and the way the paramilitaries actually function. For example, the quest for a political image involves certain stylistic transformations, such as the adoption of military uniforms and an anthem. In Magdalena, the integration of local groups into the AUC found expression in such changes, as a former paramilitary described:
Question: When did you start wearing camouflage?
Response: As soon as we became AUC. Our name was changed. We were told to call ourselves Tayrona Resistance Front. We were also taught the anthem of the self-defense committees. From then on, we always wore military clothes. Before that, we dressed like any Christian person [in civilian clothes]. 
65These transformations were also achieved by integrating individuals with military know-how into the group. The guerrillas of the Popular Liberation Army (Ejército popular de liberación, or EPL), some of whom joined Castaño’s group after their demobilization in 1991, brought with them this type of organizational model, but the ACCU also benefited from the input of former soldiers. These helped to transform the paramilitaries, bringing a military ethos that these professional men of violence had lacked.
66This military ethos was embraced by the ranks as well as by officers and non-commissioned officers. In terms of understanding the introduction of organizational models and know-how, however, we need to look most closely at the officers and NCOs. The case of Carlos Mauricio García is a good example. Originally from the city of Medellín, Second Lieutenant García had been conducting anti-guerrilla operations in eastern Antioquia. He was accused by investigators from the Inspector General’s Office of forcibly recruiting civilians to serve as “guides”: he was said to have kept them in arbitrary captivity and to have forced them to reveal strategic information about guerrilla activity, thus exposing them to reprisals by the rebels. After receiving an official reprimand, he left the army. He then joined the Castaño brothers’ group, taking the battle name of “Doblecero”. 
67García was then sent to lead a group tasked with expelling the rebels from Medellín, the country’s second largest city. From May 1997, the “Metro Bloc” (Bloque Metro, or metropolitan) attacked rural localities around the city, before trying to gain a foothold in the working-class neighborhoods of the suburbs.  Doblecero, then, found himself in an urban environment that featured a wide range of armed actors.  Established criminal bands and smaller gangs competed for the profits of drug trafficking, racketeering, and various illegal activities (such as prostitution, gambling, and loan-sharking). Rather than alienating these powerful actors, Doblecero sought to integrate these gangs within the Metro Bloc. However, the former second lieutenant hoped to promote a discipline and set of values that were foreign to the world of gangs. He wanted to construct chains of command and regulate individual behavior. Their use of drugs and ostentatious displays of wealth went against the counter-insurgency values he was seeking to defend. As he said in an interview with the anthropologist Aldo Civico:
When we entered Medellín, [the gang members] had no political or military structure, they were marijuana smokers and hitmen. They had no interest whatsoever in political or organizational work. We in the Metro Bloc began working with these bands to see what progress we could make with them. After politicizing them, we tasked all the gangs with fighting the guerrillas, setting them against the Farc and the ELN. 
69The effects of the transgressions generated by the attempts at politicization were apparent down to the level of individuals. To some extent at least, the gang members were indeed motivated by the same symbolic and material rewards that so disgusted Doblecero. The gangs were marked by their past association with drug trafficking, with all that this implied in terms of projecting social appearances of success, displaying their machismo through ostentatious consumption and their bandit image. A former gang member from the Comuna 8 district explained:
Doblecero wanted us to be soldiers. But I didn’t join up just to get up at five o’clock and parrot “Yes sir, no sir“. I didn’t join up to act like a poof [marica] and go around singing the self-defense forces anthem. And as for social work, that’s for social workers. 
71An analysis of the tensions arising from these attempts to transform the Medellín gangs into soldiers fighting the counter-insurgency cause is beyond the scope of this study. It should, however, be noted that they reflect a process of self-affirmation of identity through which the paramilitaries attempted to conform to a model that was largely at odds with the social dynamics of the groups’ creation and function. This process highlights the continuing centrality of institutions in shaping the forms taken by political action, including violence, which very largely results from the way in which armed groups model themselves on categories constructed by the state, in a continual search for recognition by the state.
72* * *
73At first glance, the situation in Colombia at the end of the last century and the beginning of this one is characterized by significant fragmentation of violence. Between the 1980s and the mid-2000s, vast areas of the country were controlled by armed actors—guerillas and paramilitaries—who often represented the only form of authority. This has led some to describe these figures as “warlords”,  which is, however, a very loosely defined term.  The challenge of this research has been to look beyond the image of apparent fragmentation in order to show how the practices of armed groups and their relationships with the political order preserve and strengthen the central position of the state. Relations between non-state armed groups and the state have been described in terms of relative dissidence: the use of this concept underlines two key features of the Colombian case, but these could also be taken as the basis for comparative studies. Firstly, despite the fact that paramilitary groups help to maintain a conservative social order that is largely compatible with the regime, paramilitary violence does not fall within a continuum associated with legitimate use of coercion. The paramilitaries find themselves banished outside the state framework, labeled as criminals, considered by some to pose a threat to state stability, or, at least, to the state monopoly of violence.
74Secondly, while the paramilitaries cannot be defined simply as auxiliary soldiers because of their criminal status, the complexity of their position becomes apparent if we consider the efforts that they make in order to obtain political and legal recognition from the state. Violence thus produces an arena of social mobility, which lies within a framework of state intervention and whose effectiveness largely depends on the support or recognition of established institutional stakeholders. Similarly, the framing of violence by institutions that hold the monopoly over its categorization is a joint product of the institutions’ efforts and the rapidity with which the armed groups adopt the same categorizations. By transforming their social structures, but also by shaping the behavior and ethos of their members, the militias reaffirm their commitment to an institutional structure that has been able to create zones in which violence is a valid strategy for social mobility and political affirmation.
75This case study opens up two angles of research. The first focuses on the way in which my conclusions contribute to the debate around the concept of “government by discharge”, to use the Weberian expression as reworked and popularized by Beatrice Hibou.  Thus, while plans for the delegation of state functions are put forward by political and administrative leaders, we can equally observe a willingness on the part of the armed actors to take over state functions in the territories that they control. Beyond the issue of intentionality, we frequently see an institutionalization of relations between armed actors and the state, as illustrated by the case of electoral alliances in Magdalena.
76The second research perspective leads us to focus on the ways in which violence is characterized, based on categories that are firmly rooted in a history of social mobility achieved through the use of arms, but which nonetheless contribute to the reproduction of state authority. Thus, the counterpart of the fragmentation of the monopoly on violence is the fact that activities relating to the categorization of violence are firmly anchored within the state framework. This does not mean that the state monopoly on the categorization of violence is unequivocal or centralized, but that here too the state still plays a central role. 
Interview, Santa Marta, 2009.
This difficulty has been highlighted by, for example, Fernando Coronil and Julie Skurski (eds), States of Violence, Ann Arbor, University of Michigan Press, 2005.
See, for example, Laurent Gayer and Christophe Jaffrelot (eds), Milices armées d’Asie du Sud, Paris, Presses de Sciences Po, 2008; Diane Davis and Anthony Pereira (eds), Irregular Armed Forces and Their Role in Politics and State Formation, Cambridge, Cambridge University Press, 2003; or, for a more anthropological approach, Thomas Blom Hansen and Finn Stepputat (eds), Sovereign Bodies: Citizens, Migrants, and States in the Postcolonial World, Princeton, Princeton University Press, 2009.
Adam Baczko, “Juger en situation de guerre civile: les cours de justice Taleban en Afghanistan (2001-2013)”, Politix, 104, 2013, 25-46; Roland Marchal, “Les frontières de la paix et de la guerre”, Politix, 58, 2002, 39-59; Paul Richards, “New War: An Ethnographic Approach” in Paul Richards (ed.), No Peace, No War: An Anthropology of Contemporary Armed Conflicts, Athens/Oxford, Ohio University Press/J. Currey, 2005, pp. 1-21.
Laurent Gayer, Karachi: Ordered Disorder and the Struggle for the City, London, Hurst, 2014.
Christian Lund, “Twilight Institutions: Public Authority and Local Politics in Africa”, Development & Change, 37(4), 2006, 685-705.
Timothy Raeymaekers, Ken Menkhaus, and Koen Vlassenroot, “State and Non-State Regulation in African Protracted Crises: Governance without Government”, Afrika Focus, 21(2), 2008, 7-22.
Richard Banégas, “La politique du ‘gbonhi’: Mobilisations patriotiques, violence milicienne et carrières militantes en Côte-d’Ivoire”, Genèses, 81, 2011, 25-44; Marielle Debos, Le métier des armes au Tchad. Le gouvernement de l’entre-guerres, Paris, Karthala, 2013.
See also the recent reflections on the construction of the concept of the weak state in and by international institutions: Sonja Grimm, Nicolas Lemay-Hébert, and Olivier Nay, “‘Fragile States’: A Political Concept”, Third World Quarterly, 35(2), 2014, 197-209.
On selecting cases for a qualitative study, see especially Bent Flyvbjerg, “Five Misunderstandings about Case-Study Research”, Qualitative Inquiry, 12(2), 2006, 219-45. On their analysis and comparative value, cf. Christian Lund, “Of What Is This a Case? Analytical Movements in Qualitative Social Science Research”, Human Organization, 73(3), 2014, 224-34.
Béatrice Hibou (ed.), La privatisation des États, Paris, Karthala, 2000.
Following the concept proposed by Rivke Jaffe, “The Hybrid State: Crime and Citizenship in Urban Jamaica”, American Ethnologist, 40(4), 2013, 734-48.
Timothy Mitchell, “The Limits of the State: Beyond Statist Approaches and Their Critics”, American Political Science Review, 85(1), 1991, 77-96.
James Ferguson and Akhil Gupta, “Spatializing States: Toward an Ethnography of Neoliberal Governmentality”, American Ethnologist, 29(4), 2002, 981-1002.
Sophie Daviaud, “Démobilisation des paramilitaires et recomposition des violences en Colombie” in Nathalie Duclos (ed.), L’adieu aux armes? Parcours d’anciens combattants, Paris, Karthala, 2010, pp. 143-73; Enzo Nussio, La vida después de la desmovilización. Percepciones, emociones y estrategias de exparamilitares en Colombia, Bogotá, Editorial Universidad de Los Andes, 2012.
For a critique of this binary opposition, see David Keen, “War and Peace: What’s the Difference?”, International Peacekeeping, 7(4), 2000, 1-22.
As in a normative view of the relationship between crime and politics; for a critique of such a view, see Jean-Louis Briquet and Gilles Favarel-Garrigues (eds), Milieux criminels et pouvoir politique. Les ressorts illicites de l’État, Paris, Karthala, 2008.
As in a large section of the literature in the field. See, for example, William Avilés, Global Capitalism, Democracy, and Civil-Military Relations in Colombia, Albany, SUNY Press, 2007; Raul Zelik, Paramilitarismo. Violencia y transformación social, política y económica en Colombia, Bogotá, Siglo del Hombre, 2015; Aldo Civico, The Para-State: An Ethnography of Colombia’s Death Squads, Oakland, University of California Press, 2015 (see also my review of this latter in Revue française de science politique, 66(6), December 2016, 1034-5).
Eduardo Pizarro and Ana María Bejarano, “From ‘Restricted’ to ‘Besieged’: The Changing Nature of the Limits to Democracy in Colombia” in Frances Hagopian and Scott Mainwaring (eds), The Third Wave of Democratization in Latin America: Advances and Setbacks, Cambridge, Cambridge University Press, 2005, pp. 235-59.
For a more detailed analysis, please see my published doctoral thesis: Jacobo Grajales, Gouverner dans la violence. Le paramilitarisme en Colombie, Paris, Karthala, 2016.
This period is discussed in Jacobo Grajales, “Violence Entrepreneurs, Law and Authority in Colombia”, Development and Change, 47(6), 2016, 1294-315.
As explained by Vadim Volkov, Violent Entrepreneurs: The Use of Force in the Making of Russian Capitalism, Ithaca, Cornell University Press, 2002. See also the edited volume by Briquet and Favarel-Garrigues (eds), Milieux criminels et pouvoir politique.
Interview, Fundación, 2009.
Interview, Santa Marta, 2009.
Interview, Cienaga, 2009.
Interview, Santa Marta, 2009.
In Colombia, ballot papers display all the candidate names alongside a photograph of each one and their party logo. To vote, the voter ticks the box of their chosen candidate.
Public officials in particular are required to have one of these certificates, as proof that they have voted.
Interview, Santa Marta, 2009.
Interview, Santa Marta, 2009.
See for example Claudia López (ed.), Y refundaron la patria. De cómo mafiosos y políticos reconfiguraron el estado colombiano, Bogotá, Mondadori, 2010; Luis Jorge Garay and Eduardo Salcedo (eds), Drug Trafficking, Corruption and States: How Illicit Networks Shaped Institutions in Colombia, Guatemala and Mexico, Bloomington, iUniverse, 2015.
Briquet and Favarel-Garrigues (eds), Milieux criminels et pouvoir politique.
Source: Trial of Senator Luis Eduardo Vives Lacouture. Supreme Court of Justice, criminal appeals court, 26 470, 1 August 2008.
Colombia has a bicameral system, comprising a Senate and a House of Representatives. Members of both chambers are elected on the same day, every four years, several months before presidential elections take place. Senators are elected on a national basis, whereas the candidates to the House of Representatives are elected by district. A “pairing” (fórmula) takes place when one candidate to the Senate and one candidate to the House present themselves to the electorate as part of a team. This mechanism, which has no institutional basis, allows electoral alliances to be formed.
Interview, Santa Marta, June 2011.
The impact of decentralization on the composition of political staff is described by Francisco Gutierrez Sanín, Lo que el viento se llevó? Los partidos politicos y la democracia en Colombia, 1958-2002, Bogotá, Norma, 2007.
Tadeo Martínez, “La red ‘anticorruption’ de Jorge 40”, Semana.com, 2008.
Interviews with the husband and a friend of the victim, Santa Marta, 2009.
In Colombia, this tax collection is carried out directly by local bodies.
Tadeo Martínez and César Molinares, “Sucedió en la república independiente de ‘La Sombrerona’”, Verdad Abierta, 16 December 2008. Interviews with these two journalists were carried out in 2009 and 2011 in the cities of Bogotá and Barranquilla. Judicial investigations into these misappropriations of funds, and into the cases of corruption in the hospitals of Magdalena, led to a series of arrests in October 2016. Eighteen former mayors and hospital directors were placed under house arrest and are currently awaiting sentence.
This information was originally revealed by the press: “Así se robaron al Golfo de Morrosquillo”, Verdad Abierta, 7 September 2010. The mayor and nine town councilors from Coveñas were convicted for associating with criminals and misappropriating funds: High court of Sincelejo, decision dated 19 November 2015, 2010-00029.
Civico, The Para-State.
Instrucciones Generales para Operaciones Contraguerrillas (General Instructions for Counter-Guerilla Operations), Ayudantía general del comando del ejército, 1979, p. 188.
Interview, Bogotá, 2011.
On the role of the military in the fight against drugs, see Winifred Tate, “Human Rights Law and Military Aid Delivery: A Case Study of the Leahy Law”, PoLAR: Political and Legal Anthropology Review, 34(2), 2011, 337-54.
Jacobo Grajales, “Faire la ‘guerre contre la drogue’: Relations asymétriques et adoption d’un régime répressif”, Cultures & Conflits, 101, 2016, 181-98.
Interview, Bogotá, 2011.
In relation to the sociology of public issues, this idea allows us to “highlight dimensions that are prioritized in a given form of framework, but most importantly to draw attention to dimensions that are either excluded or marginalized”. See Claude Gilbert and Emmanuel Henry, “La définition des problèmes publics: Entre publicité et discrétion”, Revue française de sociologie, 53(1), 2012, 35-59, especially 42.
Luis Martinez, La guerre civile en algerie, Paris, Karthala, 1998, 23-34.
Fernando López-Alves, State Formation and Democracy in Latin America, 1810-1900, Durham, NC, Duke University Press, 2000.
The term was coined by the German philosopher Gustav Radbruch. For an analysis of the legal and philosophical basis of political offense in Colombia, see Iván Orozco Abad, Combatientes, rebeldes y terroristas. Guerra y derecho en Colombia, Bogotá, Temis, 1992.
Colombia, Constitutional Court, decision C-009-95, 17 January 1995.
This is why, in negotiations with the government between 2003 and 2006, the paramilitary leaders sought the status of political criminal. They were blocked by opposition from the judicial authorities, meaning that they were treated as ordinary criminals. For further information, please see Jacobo Grajales, “Quand les juges s’en mêlent: Le rôle de la justice dans la démobilisation des groupes paramilitaires en Colombie”, Critique internationale, 70, 2016, 117-36.
To the best of my knowledge, there is as yet no detailed sociological study on the origins of the members of paramilitary groups. There are a number of publications that provide some information. See for example Ana María Arjona and Stathis Kalyvas, “Recruitment into Armed Groups in Colombia: A Survey of Demobilized Fighters” in Yvan Guichaoua (ed.), Understanding Collective Political Violence, Basingstoke, Palgrave Macmillan, 2011, pp. 143-73; Francisco Gutiérrez Sanín, “Telling the Difference: Guerrillas and Paramilitaries in the Colombian War”, Politics & Society, 36(1), 2008, 3-34. Online
For a study of the brothers, see Maria Teresa Ronderos, Guerras recicladas. Una historia periodística del paramilitarismo en Colombia, Bogotá, Aguilar, 2014.
Germán Castro Caycedo, En secreto, Bogotá, Planeta, 1996, 227.
Fernando Cubides, Burocracias armadas, Bogotá, Norma, 2005.
Fernando Cubides, “Los paramilitares y su estrategia” in María Victoria Llorente and Malcolm D. Deas (eds), Reconocer la guerra para construir la paz, Bogotá, Norma-Cerec, 1999, pp. 151-99.
Interview, Santa Marta, 2009.
On García, alias Rodrigo Franco or Doblecero, see his interviews with the anthropologist Aldo Civico, in ‘No divulgar hasta que los implicados estén muertos’: Las guerras de ‘Doblecero’, Bogotá, Intermedio, 2009, especially 173 onwards.
The information relating to this armed group was taken from the archives of the trial conducted from 2004 to 2008 against Diego Fernando Murillo Bejarano, a key figure in Colombian paramilitarism, on counts of homicide, abduction, abduction by force, and associating with criminals: Third Criminal Court of the Specialized Circuit of Medellín. Rad. 2006-0241; Attorney General’s Office, Medellín Sectional. Rad. 2302.
For a full discussion of the dynamics of violence in Medellín, see Gerard Martin’s publication, which develops his doctoral thesis: Medellín Tragedia y Resurrección. Mafia, Ciudad y Estado, 1975-2012, Bogotá, Planeta, 2012.
Civico, ‘No divulgar’, 90-1.
Interview, Medellín, 2011.
Gustavo Duncan, Los señores de la guerra, Bogotá, Planeta, 2005.
A good example of the aporia that arise from this loose definition can be seen in Kimberly Marten, Warlords: Strong-Arm Brokers in Weak States, Ithaca, Cornell University Press, 2012.
Hibou, La Privatisation des États.
The first versions of this text were presented at CERI in June 2016 and at the European Centre for Sociology and Political Science (Centre européen de sociologie et de science politique, CESSP) in February 2017. I would especially like to thank Jean-François Bayart, Laurent Gayer, Arthur Quesnay, Grégory Daho, Bertrand Réau, Antoine Vauchez, and Yves Déloye for their comments, suggestions, and questions. I would also like to thank the editing team and anonymous reviewers at the Revue française de science politique for their attentive and constructive reading.