On Burma's war-torn peripheries, men with guns are everywhere. They are part of the Burmese military junta, a dizzying array of separatist insurgent groups, and private armies linked to powerful smugglers, among many other organizations of violence. The relationships between these contending armed actors are remarkably complex and fluid across space and time. In some areas at some times, insurgents and the state are locked in a vicious struggle to the death. In others, both tacit and formal norms have emerged about the “red lines” above which insurgent violence will lead to a state crackdown. In other areas and at other times, the state and its ostensible rivals cooperate to protect and facilitate drug and timber smuggling, even as they also periodically clash. And in yet other contexts, former insurgents now operate as loose pro-state militias that are neither integrated into nor fully distinct from state power. There is extraordinary diversity in the rules of the game that structure interactions between state forces and non-state armed groups. Political orders develop, evolve, and collapse.
Though particularly dramatic, Burma is not a unique case. Intricate and often surprising relationships between states and non-state violent actors emerge and change within internal conflicts: mainstream politicians build armed wings, states collaborate with militias against common foes, police ignore private counterinsurgent armies, militaries tacitly share sovereignty with insurgent enemies, and warlords place their loyalists inside state security forces. While scholars of civil war have studied varying relationships between armed actors and civilian populations, their fundamental assumption about the relationship between contending armed actors is that both sides are locked in a straightforward struggle for a monopoly of violence.
As a result, political science lacks a conceptual language to describe varying political orders in civil war, much less to theorize their origins. The conventional approaches in the field seek to find the correlates of wins and losses in civil war or to explain fine-grained local variation in patterns of violence. They overlook the diverse interactions between states and insurgents that construct political authority and control. State and non-state actors have both cooperative and conflictual relationships that create dramatic variation in who rules—and how—in war. These wartime political orders in turn shape patterns of violence against civilians, governance and economics, and post-war politics.
An emphasis on the varying political relations between states and insurgents flows from the plausible, oft-repeated claim that civil wars represent competitive state building. The actual implications of this analogy have been ignored, but recent research on state formation and state building offers powerful insights into how political orders develop. The underpinnings of states differ dramatically, from centralized Leviathans to loose collections of local notables to alliances of central and peripheral armed actors. If insurgency and counterinsurgency resemble competitive state building, this fascinating political heterogeneity should be found in civil wars as well. The relationships between states and insurgents should be determined not solely by military violence but also, like the evolution of states, by political relationships and bargains.
To systematize how these deals, coalitions, and conflicts vary within civil war, I map out a conceptual typology of wartime political orders. The typology uses two dimensions—the distribution of territorial control and the level of state-insurgent cooperation—to characterize distinct political-military relationships between states and insurgents. I identify six wartime political orders, ranging from collusion and shared sovereignty to spheres of influence and tacit coexistence to clashing monopolies and guerrilla disorder. These orders vary within conflicts, across conflicts, and over time. They are scalable and portable to different contexts and levels of analysis, from struggles in peripheral villages to the ultimate outcomes of wars.
I offer a new dependent variable that provides a conceptual vocabulary for identifying political relationships in war. Perhaps more importantly, I also invite scholars of civil conflict to think creatively about the politics of violence. The typology is a starting point for more ambitious research agendas. Though I focus on mapping out the concepts rather than theorizing the causes of their variation, the implications of this exercise in concept formation challenge basic, frequently unstated, assumptions about civil war. States and insurgents are not simple-minded maximizers of monopoly but instead are optimizers of authority in complex, often counterintuitive, interaction with other armed actors.
This claim has relevance to policy debates. The range of orders during war is more varied than existing prescriptions realize and thus stable outcomes in conflict zones like Iraq and Afghanistan may not look anything like what conventional doctrines suggest. Informal bargains, collusive state-insurgent relationships, and shared sovereignty are often less costly and more enduring than trying to build strong states, an endeavor that integrally involves coercion, extraction, and centralization. The wartime political orders I identify point to important new avenues for conflict resolution.
My argument advances in five parts. First, I outline the current state of the field. The dominant research programs are valuable in many ways. However, they largely ignore the bargains, deals, and conflicts between states and insurgents that constitute political life during civil war. Scholars need to move beyond solely studying macro-level conflict outcomes and micro-level patterns of violence. Second, I use recent research on state formation to show that there is a wide array of relationships between governments and other actors. States are not engaged in an all-consuming quest for territorial authority, but instead are intertwined with other social and political forces that shape authority across time and space. Theoretically we should expect civil wars to be characterized by similar dynamics.
Third, I develop a conceptual typology of six wartime political orders. The different types of wartime political order vary according to the distribution of territorial control between states and non-state armed actors and the level of cooperation between these armed forces. There is enormous heterogeneity in how states and insurgents interact, ranging from total war to tight collusion. I interweave these different forms of political order with specific examples from contemporary South Asia to illuminate how this conceptualization provides new insights into a region torn by numerous forms of political violence.
Fourth, I outline theoretical implications of this conceptual typology. Future research needs to theorize variation in state and insurgent political interests, study the co-evolution of state and non-state political authority, incorporate other actors in the construction of order, and explore how governments deal with armed groups in contexts as diverse as inner-city policing, electoral campaigns, and international politics. Rather than a category unto itself, civil war should be situated within a broader set of processes that combine politics and violence.
I conclude by arguing that scholars need to be more wide-ranging in forming the basic concepts that underpin research. Fundamental questions about the construction and maintenance of political order during violent conflict are open for exploration. In terms of policy, it is clear that there is no single pathway to stabilization or set of best practices in counterinsurgency. Policymakers should instead be comfortable with ambiguity and innovation. Cutting deals with insurgents, propping up local power centers, and simply ignoring areas of resistance can all be more effective in forging stability than trying to bolster central state authority. Though the pursuit of a counterinsurgent Leviathan is appropriate under some circumstances, it should not be the default form of conflict resolution. As the experiences of Iraq and Afghanistan show, violence management and indirect rule can be easier and more stable than trying to impose a violence-monopolizing central state on rebel peripheries.
Concepts and Outcomes in Civil War Research
Recent years have seen an outpouring of new research on civil wars. Two major research agendas aim to explain the dynamics of civil war. First, macro-level studies try to explain why civil wars are won and lost. Second, micro-level studies focus on the patterns of violence that emerge during wars. This research is impressive, from understanding when international interventions succeed and fail, to specifying mechanisms through which militaries learn in counterinsurgency, to explaining spatial variation in sexual violence in war.
What existing research is missing is a way of describing the nature of authority, politics, and order in a particular area or war at any given point in time. How did the armed actors in Helmand province in Afghanistan interact during 2007? Has that interaction changed since and, if so, how? Unfortunately, scholars of civil war lack a conceptual vocabulary to describe the relations between armed actors. The absence of concepts addressing political arrangements in war is problematic because concept formation “lies at the heart of all social science endeavor.”Footnote 1 Much of the literature on insurgency and counterinsurgency de-prioritizes what Sartori refers to as “‘what is’ questions” in favor of “‘how much’ questions.”Footnote 2 The rest of this section shows how the dominant research on civil war dynamics misses the opportunity to examine the political relationships between states and insurgents.
Conflict Outcomes: Wins, Losses, and Draws
In cross-national studies of conflict outcomes, conflicts are frequently coded as wins, losses, and draws after they end, and as “ongoing” or “other/low activity” while they are still being fought. Conceptually, the assumption is that outcomes can be mapped onto a straightforward win-loss spectrum, and methodologically, scholars tend to rely on aggregated measures (i.e., at the country or conflict level, rather than within it) as co-variates of an aggregated dependent variable. These studies often take an entire campaign or state-insurgent dyad as a single observation, which means that the win/lose/draw approach has important uses but tells us little about wars in progress or variation across time and space during wars.Footnote 3 Rather than being able to systematically identify different relationships between the Taliban and Afghan police in Helmand and in Paktika in 2007, scholars label the entire Afghan war as a single outcome, if they study it at all.Footnote 4
Strikingly, the most politically ambiguous categories—draw, ongoing, and “other/low activity”—actually capture a very substantial proportion of conflict codings in large-N datasets. For instance, in an important study of civil war outcomes by David Cunningham et al., the category “low activity/other” characterizes the largest single category of outcomes—nearly 40 percent.Footnote 5 Joakim Kreutz finds that 32 percent of civil wars from 1946–89 and 48 percent from 1990–2005 terminated as “Other,” with a further 20 percent of post-1990 cases ending as “Ceasefire.”Footnote 6 Since 1945, approximately a third of the completed (excluding ongoing) conflicts examined in another major recent piece of research are coded as “draws.”Footnote 7 Since this type of conceptualization also cannot capture over-time and within-conflict variation, exclusively using it means that we lose a significant amount of information about the nature of wars.Footnote 8 Even in cases of completed conflict, there are many outcomes other than straightforward victory, defeat, or formal peace agreements. Scholars need to study more fine-grained relationships between states and insurgents.
Small-N studies of counterinsurgency outcomes face similar challenges. First, their goal is to explain clear victory and defeat in conflicts.Footnote 9 Wins and losses are the only outcomes of interest, which limits their ability to speak to the kinds of bargains and deals we observe in places like Burma. Second, while they take limited war and political considerations seriously, these studies generally examine cases of (relatively rare) foreign interventions. In some cases they also make blanket assumptions about the political interests of interveners rather than theorizing variation in resolve, interest, and goals.Footnote 10 Such studies thus miss a significant swath of both relevant conflicts and relevant outcomes.
Studies of macro-level conflict outcomes, whether large- or small-N, are important and answer key questions, but cannot easily capture the political relationships between state and non-state armed actors, either across wars or within them. This awaits conceptualization.
Local Outcomes: Control and Violence
Micro-level approaches, by contrast, tend to use fine-grained local metrics to describe the nature of a war across time and space. Two concepts are regularly deployed to make civil war analytically legible—violence and control. First, an extraordinary amount of detailed data and theory has been developed on local killing, sexual assaults, refugee flows, and numerous other forms of violence as dependent variables within and across wars.Footnote 11 These indicators are obviously extremely important for understanding dynamics of warfare, but they do not provide a conceptualization of who rules in a particular area or time period.
This is because violence can be observationally equivalent to a number of different political-military situations. It can occur in a situation where a state or insurgent has a monopoly of violence and is engaged in mass killing, or when a state is engaged in a competitive struggle to forge institutions against an insurgent, or when a state is cooperating with insurgents to target drug-running rivals while smuggling heroin across a porous border. There are many ways for violence to break out and many reasons for it to spread or de-escalate. Using violence alone to describe the political state of a conflict in an area is problematic because it does not necessarily tell us how power is mediated and negotiated between contending actors.
Embedded in some of these studies of violence, but conceptually distinct, is the use of territorial control as an indicator of the nature of a war in particular areas at particular times. This approach examines the military relationship between states and insurgents, and is used by both academicsFootnote 12 and counterinsurgent militaries relying on population control strategies.Footnote 13 Here the goal is to understand which side has presence where, and who controls which territories to what extent. Instead of focusing exclusively on violence it aims to understand who the key players are and what power they have. Control assumes a zero-sum realm in which a fixed amount of authority is divvied up between competing actors. Stathis Kalyvas argues that control is a function of military effectiveness and geography.Footnote 14 Political interests are explicitly taken as exogenous.Footnote 15
However, the same balance of control can involve completely different political relationships. Consider three situations in which each side controls the same amount of territory and has equal military power. In the first, the two sides are engaged in a brutal conflict for supremacy with indiscriminate shelling, massed offensives, and ethnic cleansing. In the second, again with the same balance of control, the two sides sometimes clash but have developed a set of understandings for limiting violence. Commanders regularly signal what kinds of conflict they are willing to accept, and what kinds will lead to escalation. In the third, once more holding the balance of control constant, the two sides collaborate to smuggle gems and timber, cooperating to hold off rivals even as they continue to disagree politically. There are different games being played in each of these situations despite the identical military balance. Relative power and capacity is obviously important, but only captures part of the story.
Who Rules in Civil Wars?
We do not have systematic concepts to describe political authority and order in civil war. There is existing research that frames civil conflicts as something other than a purely adversarial clash of wills between insurgents and the state. William Reno notes the existence of “shadow states,”Footnote 16 Kimberly Marten grapples with warlordism,Footnote 17 Kenneth Menkhaus and Leonard Wantchekon argue that there are different ways of building political order in and after conflict,Footnote 18 Martha Fischer and Beatrix Schmelzle compile cases of “hybrid political orders,”Footnote 19 Dipali Mukhopadhyay explores variation in how Afghan warlords cooperate with the state,Footnote 20 Paul Staniland suggests that political interests can trump state capacity in the deployment of coercion,Footnote 21 and Elizabeth Wood highlights a deal between the army and FMLN in El Salvador.Footnote 22 However, these works do not systematically map out the conceptual space they are empirically uncovering or the possible relations between states and non-state armed groups.
We can judge some wins and losses after the fact, measure the incidence of control and violence during the conflict, and point to puzzling anecdotes and examples, but we have no systematic concepts, no “data containers,”Footnote 23 to identify and systematize variation in who governs and how this governance is shared, negotiated, and contested across wars, and across space and time within wars. Conceptual stretching is rife in the pivotal “meso-level”Footnote 24 analytical space between the local and the national where states are forged or shattered, populations displaced, and new forms of governance built. Patterns of violence and variation in overall war outcomes are obviously important and worthy of serious study. But thinking more creatively about order in war allows an escape from using new data to answer old questions. Original concepts can push forward the underlying arena of inquiry.Footnote 25
War, State Formation, and Political Order
Insurgency and counterinsurgency is a contest over the shaping of political order in a contested area, a means of determining who rules, how much, and in what ways. It is thus conceptually very similar to state formation and many analysts view civil conflict as competitive state-building.Footnote 26 Though this analogy is widely accepted, its implications have not been carefully drawn out. Recent research tells us emphatically that there is no single form of the modern state, nor any dominant pathway of state building. Some states were built in unremitting blood and fire, others as fractious collections of elites, and others as stable but tacit deals between political forces; some states are highly centralized and others are highly decentralized; some states extend their infrastructural power to the most distant peripheries while others choose indirect rule or neglect. Even within countries, there is dramatic variation in the modalities of political order: indirect rule on the periphery can exist alongside centralization in the heartland.
The assumption implicitly built into studies of civil war that central states represent homogenizing, monopolizing Leviathans is unsustainable. As part of a corrective in recent decades, extensive scholarship has identified multiple trajectories of state building in Europe.Footnote 27 Miguel Centeno and Jeffrey Herbst specify how the intersection of international and domestic variables created distinct state-making trajectories in Latin America and Africa.Footnote 28 In west Africa, Reno shows how state power itself can be “outsourced” to commercial and mercenary networks and Catherine Boone highlights the sub-national variation in state penetration.Footnote 29 Barkey explores the incorporation of non-state bandits into nascent bureaucracies in the Ottoman empire.Footnote 30 Ayesha Jalal and James Mahoney trace out the impact of colonial legacies on patterns of authority and Dan Slater and Tuong Vu identify the social-coalitional underpinnings of varying post-colonial states.Footnote 31 These studies make it clear that states have been and often remain partially monopolized and characterized by seemingly “heterodox”Footnote 32—but in fact commonplace—relations with other actors.Footnote 33 Such bargains are integral to political order writ large.Footnote 34
These findings have clear relevance to civil war. The various forms of order within states—representing compromises, clashes, and bargains between political, social, and economic actors—should also found within conflicts. What Boone notes of western Africa is true of many parts of the globe: “localities and provinces have been incorporated into the modern state in highly variable ways and to varying extents.”Footnote 35 Karen Barkey identifies a model of politics in the Ottoman Empire in which the imperial center “has to work with peripheries, local elites and frontier groups to maintain compliance, resources, tribute and military cooperation and ensure political coherence and stability,”Footnote 36 that has resonance far beyond traditional empires, whether in northern Iraq or the eastern Congo. Charles Tilly's classic analysis of state formation, centered on shifting relations between specialists in violence, echoes how elites try to deploy and control violence in contexts as different as electoral violence in Kenya and paramilitarism in Colombia. Multi-faceted deals between states and other armed organizations are pervasive in both war and peace, resembling Joel Migdal's broader observation that “territories have hosted a diversity of rules of the game … social control has not been of a piece, but it has frequently been highly fragmented through a territory.”Footnote 37
If insurgency and counterinsurgency are forms of competitive state building, then we should think of them as a contest over the shaping of political order rather than purely a military conflict to be won, lost, or drawn. Like the development of states, orders in civil war represent particular norms, bargains, and balances of power across time and space. There are differences between patterns of state formation and insurgency—notably in scale and timeline—but the core dynamics of bargaining, coercion, and coalition share basic similarities. We should observe uneven, heterogeneous landscapes of authority and control, with pockets of government-insurgent collusion intermixed with areas of conflict, shifts in political relationships between rebels and counterinsurgents triggering escalation or accommodation, and diverse state-insurgent arrangements about how lethality is (or is not) controlled. The logic of violence management often dominates that of violence monopolization.
Wartime Political Orders
Making these broad insights analytically useful in the study of conflict requires being able to systematically identify varying relationships between states and insurgents. We need to create a way of differentiating between military-political arrangements in civil war. To do so I construct a conceptual typology of six distinct wartime political orders. Political order here refers to the structure and distribution of authority between armed organizations: who rules, where, and through what understandings.Footnote 38 “Wartime” indicates that there are contending armed forces that have been or currently are engaged in violent conflict. The monopoly of state violence is broken and a situation of “dual power” exists. Wartime political orders vary in their level of cooperation and in the existence of a segmented or fragmented distribution of control between armed organizations.
I study the simplest relationship, between an armed group and a state. There is no doubt that civilians, international organizations and external states, criminal networks, economic elites, and many other actors also shape wartime political orders. However, I begin with the most basic situation because it is where we should intuitively expect the most straightforward conflict. If we find substantial variation in the political relations between states and non-state armed groups, it should open the door to a much richer study of how politics work in war. I explore some of these other actors when outlining future research agendas later.
I build the typology of wartime political orders along two dimensions. The first is the distribution of territorial control. This reflects the presence and structure of armed actors in a particular territorial domain, which is important for shaping the types of relationships and arrangements that are possible.Footnote 39 Control provides a key background condition to understanding the dynamics of conflict and cooperation, whether in a specific village or the conflict as a whole, because it reflects the capacity of the actors and the structure of their competition. Drawing on Kalyvas, I distinguish between situations with a distinct, segmented distribution of control in which each side controls some territory, and situations with an overlapping, fragmented distribution of control in which both sides have presence throughout the area under contestation.Footnote 40 We can think of segmented sovereignty as analogous to a conventional military frontline, whereas fragmented sovereignty intermixes state and insurgent armed forces. In both segmented and fragmented distributions of control the state's monopoly on violence has broken down and multiple armed contenders for power exist. The difference between the two comes in how this division of power is structured.
As noted above, control and the balance of power alone do not tell us enough about the relationships between armed actors. I introduce a new dimension to broaden analysis: the level of cooperation between the state and an insurgent actor. It is here where political interests shape the interaction between organized specialists in violence. If state formation bears resemblance to civil war, this dimension is where we should see dynamics of cooperation and bargaining operating alongside violence and conflict. Armed organizations do not simply hurl force at one another until one side is annihilated; instead, they try to calibrate the deployment of military power according to goals and strategies. By looking at the level of cooperation, we can see how political orders within and across civil wars emerge and change. Insurgency and counterinsurgency are not free of Clausewitz, who reminds us that “if war is part of policy, policy will determine its character.”Footnote 41 Wartime orders are fundamentally shaped by the political relationships between the contenders.
There are different levels of possible cooperation between states and violent non-state actors, even holding the military balance constant. Cooperation can be active, passive, or nonexistent. Active cooperation involves clear coordinated action towards a shared objective, whether jointly ruling territory, attacking shared enemies, or colluding over illicit economies. Passive cooperation involves live-and-let-live bargains structured around norms of acceptable violence and the creation of “red lines” by state and non-state actors below which each side is willing to restrain violence. This is a world of managing escalation and limiting lethality. Nonexistent cooperation involves intense conflict and hostility, in which norms and expectations of violence are unpredictable and fluid. This is the realm of total war, but it is only one part of the broader world of political violence. There are numerous games being played within war that need to be understood to make sense of patterns of violence and authority. Political relationships should shape these orders, not just the relative balance of power, control, and doctrine; the exact same stock of guns and men can be used in very different ways depending on state-insurgent cooperation.
Table 1 pulls together these two dimensions in a typology of wartime political orders. Different combinations of territorial control and state-insurgent cooperation create six distinct political orders in conflict.Footnote 42 Under segmented control, we see shared sovereignty, spheres of influence, and clashing monopolies as insurgent-state cooperation decreases. Under fragmented control, we see collusion, tacit coexistence, and guerrilla disorder as insurgent-state cooperation decreases. States and violent non-state actors are engaged in bargains, probes, clashes, and arrangements over the distribution and nature of political authority.Footnote 43 Each type of wartime political order involves different patterns of interaction between states and insurgents. The nature of violence should vary across each type of order, as should the nature of political authority.
A typology of wartime political orders
I discuss the types of wartime political order within each level of insurgent-state cooperation and provide specific examples from civil wars in South Asia since 1947.Footnote 44 This region has experienced an extraordinary profusion of internal conflicts across types of regimes, state and insurgent war aims, economic conditions, conflict outcomes, and bases of mobilization. This is not intended as a comprehensive empirical treatment of the subject, but instead is simply aimed at mapping out how wartime political orders can vary across and within wars.Footnote 45 Examples bring concepts to life and an accumulation of supposed anomalies can in fact illuminate new insights.Footnote 46 Wartime political orders are not the domain of a few obscure outliers, but are instead hiding in plain sight within many conflicts. The cases also reveal how state formation has occurred alongside and even through these processes of cooperation and coercion as political authority is negotiated between state and non-state actors.
Active Cooperation
The state and insurgents can actively cooperate towards a shared goal, including reaping the benefits of illicit economies and destroying mutual enemies. Rather than targeting non-state armed actors, in these situations states can work alongside them. I distinguish between two forms of active cooperation: shared sovereignty and collusion. They differ according to the distribution of control between the state and insurgents, as shared sovereignty involves segmented territorial control while collusion is characterized by a fragmented, overlapping distribution of control.
Shared sovereignty involves active cooperation between a state and its foes in which each side has bounded control over particular territory.Footnote 47 This is a negotiated form of political order in which the insurgent organization retains autonomy and standing structures of coercive capability. The state has not shattered its foe but instead the two sides have arranged a clear division of influence and authority that satisfies both in the pursuit of mutual gains. Violence between the forces is minimized and institutionalized mechanisms for achieving joint goals are devised, even with no monopoly of violence. The clear territorial line dividing the two contending forces allows a centralized and straightforward division of authority and power upon which sophisticated cooperation can be built. The central state and peripheral armed groups coordinate their behavior as in an alliance, without either side fully controlling the other. Over time a shared sovereignty political order may remain static, lead to full state incorporation, or break down into a different, more conflictual form of wartime order.
Shared sovereignty characterizes the relationship between the contemporary Burmese state and several insurgent armies among minority ethnic groups. In a spate of ceasefire deals since the late 1980s, the Burmese military regime (the tatmadaw) and its insurgent foes have created a set of sovereignty-sharing arrangements that have minimized violence on the periphery without creating a central monopoly. Smith argues that “the very existence of such co-operative schemes involving former battlefield foes decisively changed the military and political balance in much of the country”Footnote 48 even though “no political agreements were made”Footnote 49 about an ultimate solution to these wars. This can be highly formal cooperation: “Tatmadaw units are often disarmed at the entrance to territory patrolled by these ethnic armies; upon exiting the territory, they receive their arms again.”Footnote 50 In the Kachin and Wa areas, standing insurgent areas with territorial control and up to 35,000 soldiers have become Border Guard Forces, a label that masks their enduring autonomy. As Steinberg informs us, “the cease-fires allowed minority group armed forces to hold their weapons as long as they did not attack the tatmadaw.”Footnote 51 Sovereignty is explicitly and formally shared between the central state and its long-time foes—even as the broader political conflicts on Burma's periphery show little sign of being permanently solved.
Afghanistan witnessed similar arrangements in the period between 1989 and 1996 and since 2001. From 1989 to 1992, the Najibullah regime “moved away from a bureaucratic chain of command toward a system based on patronage, in which the state pays powerful leaders to supply troops from among their followings.”Footnote 52 From 1992 until 1996, even amidst extraordinary levels of violence, Kabul regimes actively co-opted and bargained with various non-state armed actors that controlled some territory, including groups that had previously waged war against the central state. Organizations like the Jombesh-yi Milli, Emirate of Herat, and the Hezb-i Wahdat would cooperate with the official leadership of the Afghan state at various points, creating a temporary wartime political order of shared sovereignty in their areas of control.Footnote 53 These alliances shifted back and forth over time, showing the importance of changing interests and strategies in civil war rather than just raw capacity.
Since the overthrow of the Taliban in 2001, the Karzai regime has engaged in broadly similar behavior with a variety of warlords and autonomous commanders in Afghanistan.Footnote 54 Giustozzi argues that the post-2001 Afghan state-building process has involved “direct cooptation of the military class by the central government.”Footnote 55 Central state authorities and armed groups have found an array of ways to cooperate despite enduring disagreement about ultimate political goals.Footnote 56 Shared sovereignty orders show that the existence of multiple armed groups alone does not cause particular levels or forms of violence; instead, shared political interests can create areas of order even without a monopoly of violence.
Collusion is a situation in which the state actively cooperates with non-state armed actors that are geographically intermeshed with its areas of operation. States provide logistics and protection, while insurgents offer intelligence and deniability. Collusion involves the coordinated pursuit of a shared goal, such as facilitating illicit smuggling, targeting common enemies, or divvying up gains from extortion. Because of the fragmented distribution of control, the operations of states and insurgents are interwoven: insurgents use military bases for shelter, state forces accompany insurgents on raids, and state-backed political candidates deploy insurgents as bodyguards and strongmen. Rather than the distinct divisions of authority and control under shared sovereignty, here the production of violence and influence is difficult to cleanly separate along territorial boundaries of state and non-state authority.
Common manifestations include “shadow state”Footnote 57 networks, former insurgents who have switched sides, parastatal organized criminal organizations, and alliances between counterinsurgents and insurgents against common foes. The chaos and grim banality of many war zones is accentuated by the diversity of armed groups operating alongside and in cooperation with official state forces. These actors remain outside of the ambit of the rational-bureaucratic state apparatus and can include fighters who fought against the state and even still oppose it. These alliances may break down or be transformed over time, but they can also be important and enduring elements of wartime order. The violence that accompanies these orders will be different than a purely adversarial clash of wills; it is coordinated to target mutual enemies and reflect the political bargains made between the relevant actors.
There are several cases in South Asia that clearly fit this concept. In northern and eastern Sri Lanka, the Sri Lankan government cooperated closely with “flipped” Tamil insurgent groups who continued to carry arms. In Kashmir Indian security forces colluded with flipped Kashmiri insurgent groups against common enemies; the rise of these so-called “renegades” had its roots in shifts in political interests that led to tighter cooperation.Footnote 58 In both cases, the state and non-state armed groups shared intelligence, collaborated in targeting common enemies, and became tightly intertwined. Control and the balance of power were endogenous to political conflict and deal-making.
In certain areas of Afghanistan during the anti-Soviet war, Afghan soldiers and police were known to actively cooperate with the mujahideen. For instance, a secret American diplomatic report from 1982 notes that “collusion between the mujahideen and the security forces is a major asset for the resistance. Successful guerrilla raids on police and military arms depots are frequently the result of ‘inside’ assistance, and intelligence provided to the mujahidin by military officers is indispensable.”Footnote 59 Many Afghan military units and soldiers tried to avoid full-fledged combat with the mujahideen and sought instead to assist them, even as both sides also engaged in lethal violence towards one another. Variation in violence was crucially linked to variation in collusion.
Other collusive arrangements are looser. Pakistani intelligence services are credibly believed to have sponsored the splinter group Muttahida Qaumi Movement-Haqiqi as part of the urban insurgency in Karachi in the 1990s.Footnote 60 Pakistan's security apparatus has continued to support armed groups on its territory, including the Haqqani network and Quetta Shura of the Afghan Taliban and portions of the Pakistani Taliban.Footnote 61 In the Indian Punjab in the early 1980s, parts of the Indian state and Congress party actively propped up the nascent Sikh insurgency of Jarnail Singh Bhindranwale to achieve their own political goals in an “unholy alliance.”Footnote 62 Northeast India sees various “local ethnic autocracies” in which the state buys off former or potential insurgent rivals.Footnote 63
Rather than the relentless imposition of state power against violent non-state actors, both state elites and local security forces have often chosen to work with them. A consequence can be the legitimization of unaccountable, deniable violence as states and their ostensible foes work together to shatter their enemies and to exploit opportunities for revenue extraction. Civil wars can combine cooperation and conflict. Though the specific manifestations of cooperation differ according to the distribution of control, the underlying political dynamics remain the same. Despite pious proclamations to the contrary by state authorities, “bandits and bureaucrats”Footnote 64 frequently operate hand in hand. This relationship shapes how violence is deployed and how political rule is experienced on the ground.
Passive Cooperation
The intermediate realm of passive cooperation exists when the state and insurgents agree to live-and-let-live bargains and tacit deals that create implicit but often incomplete and tenuous arrangements for the management of violence. Norms emerge to limit and discipline coercion but they are not formalized in the sense of shared sovereignty, and cooperation is of a negative rather than a positive form, unlike collusion. The state and non-state armed groups establish mechanisms to minimize the extent to which they kill one another without agreeing on a final deal to resolve a conflict. This negotiated form of political order involves processes of boundary formation, deterrence, limited war, coercion, and signaling. It is more fragile than shared sovereignty or collusion and is a common type of political order in protracted insurgencies. The specific manifestation varies by the distribution of control: we see spheres of influence when control is segmented and tacit coexistence when control is fragmented. Violence is contained and coordinated between a state and non-state armed group as a way of maintaining bargains and informal arrangements for cooperation.
Spheres of influence are segmented areas of control in which the state and its armed-group foe agree to limit the boundary violations across each sphere.Footnote 65 State and insurgent leaders engage in low-level but recurrent communication over which types of violence and policy are acceptable and which will trigger an escalated response. Ceasefires, informal truces, and agreements about where state and insurgent forces will tread (and when) are intended to manage spirals of escalation. Priority is given to minimizing costs of repression and managing conflict rather than attempting to maximize a state or insurgent monopoly of violence. Spheres of influence political orders are a type of negotiated but tenuous indirect rule in which states set limits on insurgent expansion but accept that central authority is not enacted in insurgent-controlled areas. This is less explicit and formalized than shared sovereignty, reflecting a desire to minimize joint damage rather than achieve common goals.
Specific roads, valleys, or rivers become boundaries of authority; warlords, insurgents, and strongmen are granted discretion by state institutions over the meting out of punishment and distribution in particular areas; and communications between state and insurgent commanders are used to establish the boundaries of spheres of influence. These are often pockets of insurgent authority and control enmeshed within a broader state. Spheres of influence are vulnerable to shifts in control and political context, but they can be deeply rooted features of a political landscape. When they operate, violence is deployed along particular boundaries and capped at certain levels as part of a cooperative (though not harmonious) relationship.
In India's far northeast, an insurgency in Nagaland has endured since the 1950s. A variety of deals have been struck with Naga armed groups, even as the overall conflict has endured. For our purposes, the clearest example of spheres of influence has existed in the state since a set of ceasefires in 1997 between the Indian state and factions of the National Socialist Council of Nagalim (NSCN). The Indian Army largely operates within well-defined geographical areas and the state government only intervenes into certain political issues. The Naga insurgents, who dominate swathes of territory and have built up infrastructure, do not attack the Indian state and instead focus on their own intra-ethnic power struggles.
This is a classic spheres of influence wartime political order. There is not active cooperation towards some shared goal, like there is in the more formal arrangements in Burma aimed at economic benefit and attacking joint enemies, but instead a straightforward attempt to minimize violence by maintaining boundaries. Violence and authority in Nagaland are constructed and reproduced by this bargain, rather than simply the balance of military power (which has remained similar even as cooperation has shifted). An Indian journalist insightfully observed this dynamic during the joint management of a protest:
We were in the midst of the two “parallel governments” of Nagaland working together in perfect co-ordination—the Naga Army of the NSCN-IM and the IRB [Indian Reserve Battalion]. The state Home Commissioner and the SDO (Civil), SP and District Commissioner of Dimapur were discussing with the Kilonsers [NSCN cadres] means to control the mob. Only the darker shade of the Naga Army's uniform differentiated them from the IRB.Footnote 66
During the Soviet occupation of Afghanistan, spheres of influence were recurrent. Roy notes that the Afghan intelligence service, KHAD, was sometimes able to set up “mutual non-aggression” deals with mujahideen groups.Footnote 67 Even Ahmed Shah Massoud, the vaunted “Lion of Panjshir” in the anti-Soviet resistance, cut a number of bargains with the Soviets in the mid-1980s.Footnote 68 Later, under the Rabbani regime, the armed forces surrounding Kabul would alternate between clashing monopolies, shared sovereignty, and extended periods of spheres of influence, with desultory shelling accompanying renewed deal-making.Footnote 69 In Sri Lanka, from 2002 until 2006, this political order characterized relations between the Tamil Tigers and the Sri Lankan government. Despite no agreement on the future of Sri Lankan politics and profound underlying hostility, the two sides agreed to at least temporarily reduce violence.Footnote 70 During this time, the LTTE ran a parallel administration on its territory with limited access by government authorities and no real access by state security forces; this “convoluted arrangement”Footnote 71 was a clear example of a spheres of influence order.
Since the mid-1990s, the Pakistani central state has followed a spheres of influence policy in Karachi and Hyderabad, relinquishing de facto control of neighborhoods in the city to the MQM as a tool for managing violence.Footnote 72 Until 2009–2010, the Pakistani state left significant portions of its territory in the hands of even the hard-line factions of the Pakistani Taliban because it lacked the political interest and resolve to deploy its forces against these insurgents.Footnote 73 In all of these cases, state capacity has not been unrestrainedly unleashed against non-state armed groups to pursue a monopoly of violence. Instead, states can calculate that they are best served by leaving armed groups alone, and insurgents are often satisfied with maintaining de facto control over an area rather than single-mindedly assaulting the state.Footnote 74
Tacit coexistence involves the interweaving of state and non-state violent organizations in the context of fragmented, overlapping control. Rather than clearly delineated spheres of influence, this order involves careful attempts to limit the degree of active conflict and violence between states and non-state armed groups in intermixed daily life. Security forces do not go out at night while insurgents do not go out during the day, states accept that insurgents tax local farmers while insurgents do not target senior government officials, and neither side makes a total effort to destroy the other. This is particularly common when governments have local contacts with insurgents that allow them to communicate the limits above which it will become politically necessary to engage in increased repression, and vice versa.
Tacit coexistence is not an actively cooperative context of shared goals, but instead one of managed expediency, an acknowledgment that neither side has the power or will to crush the other and that some kind of mediated mutual survival is necessary. Civilians and armed groups respond to this environment is distinctive ways. Families may place one son in the insurgency and another in the police, local businessmen make sure to give kickbacks to both sides, and the armed actors themselves refrain from dramatic military actions that could trigger a disproportionate response. Much of guerrilla warfare is not a chaotic clash of armed organizations but instead a careful, if brutal, attempt by each side to define and probe the boundaries of interaction. Violence in a tacit coexistence order follows the implicit rules of engagement about what is and is not acceptable to each side; it reflects the logic of mutual violence management rather than that of mutual violence monopolization.
Before the escalation of central counterinsurgency in 2009, the Naxalite guerrillas in central India were able to quietly operate in many areas nearly unchallenged by administrators who feared the consequences of crackdowns and by politicians uninterested in bearing the costs of war.Footnote 75 Large parts of India came under Naxalite influence as state and political authorities simply looked on, largely uninterested in the rural interior and unwilling to accept risk to restore a state monopoly of power. In Assam, in India's Northeast, during the late 1980s the state was run by a political party sympathetic to the insurgent United Liberation Front of Asom (ULFA). The state administration restricted its activities and avoided repression for this period. As a result, the Indian central government in Delhi “watched with a mixture of outrage and impotency, unable and unwilling to dispatch the military machine of the Indian Army.”Footnote 76 Bureaucrats and politicians treaded very carefully, even if they did not sympathize with ULFA, by entering districts at times and in ways that they tried to ensure would not lead to an encounter with the insurgents, and by simply ignoring areas of clear insurgent control. In Zabul Province in contemporary Afghanistan, “government representatives can only survive by tacit agreements with the insurgents.”Footnote 77
Violence is costly and its consequences are unpredictable, which creates incentives for actors who politically disagree to nevertheless seek means of bounding violence. When states and insurgents are divided along segmented lines of control, they can forge spheres of influence in which live-and-let-live norms emerge. When states and insurgents are intertwined but at least minimally cooperative we observe tacit coexistence, a world in which violence is ever-present but not total. Passive cooperation is sensitive to political shifts but it can be enduring, especially on distant peripheries in which states lack the interest to ruthlessly crack down and insurgents are content with de facto local control.
No Cooperation
Total civil war occurs when the state and its opponents do not cooperate along any dimension: there are no norms about levels and patterns of violence, no tacit understandings about when which forces can go where, no accepted (if tenuous) division of sovereignty, and no attempts to communicate the limits of escalation. Unpredictable violence and unclear lines of authority and control characterize the interactions between fighters. Non-cooperation is particularly common in the early days of a conflict before the contending forces have a chance to forge deals. Sometimes non-cooperation is a brief burst of extreme unrestrained violence amidst a longer pattern of bargaining or the endgame of clashing armies after a long process of escalation. More rarely, it characterizes the entire nature of the conflict. This relationship can also result from the breakdown of more cooperative wartime political orders into a spiral of escalatory violence. Political interests shape how capacity is deployed, but given the lack of political cooperation, outcomes in these orders are substantially a function of military power.
Clashing monopolies is an order characterized by violent competition between the state and an armed actor that each control distinct territory. Clashing monopolies pit specialists in violence against one another across clearly defined battle lines. The boundaries between state and non-state forces are rigid and easy to identify. The contest is to determine who can inflict the most pain on the other in hopes of shattering the internal organization and fighting power of its competitor. Politics shapes war aims, but this contest is a primarily military struggle with an emphasis on logistics, manpower, and arms. It is similar to a conventional war in which each side tries to break through into territory held by the other.
Clashing monopolies characterized the war between the Liberation Tigers of Tamil Eelam (LTTE) and Sri Lankan state during most of the 1990s and between 2006 and 2009. Two territorially-segmented armed forces faced off across conventional battle lines in a war over whether the LTTE would survive. During these periods (especially that leading to the LTTE's destruction in 2009) there was a full application of military force, and relative capabilities determined the contours of control and authority.Footnote 78 Burma's “four cuts” military offensives against the Communist Party of Burma and ethnic insurgents would, for limited periods of time, lead to a political order of clashing monopolies before de-escalation and bargaining.Footnote 79
This was the wartime political order as the United Front/Northern Alliance desperately tried to hold off the quasi-conventional forces of the Taliban in Afghanistan from 1995 until 2001. Though not very sophisticated forces, these contending actors had territorial control and fought along loose front lines.Footnote 80 There was some bargaining in the initial period of the Taliban's rise (and there was always local deal-cutting with mid-level commanders), but by 1997 the war had escalated to one in which little quarter was taken or given among top leaders. Many Northern Alliance commanders, like Ahmad Shah Massoud, were fighting for their freedom and lives: “surrender would lead to annihilation.”Footnote 81 The Taliban leadership knew that they would similarly be granted little reprieve in the brutal battlegrounds of Afghanistan. The war aims and political interests of the contenders determined how violence was used in the conflict.
Guerrilla disorder is a situation of fluid violence in which there are few clear norms or rules about the infliction of lethal violence when insurgent and state forces are intertwined in the same physical spaces. Violence is an embedded part of political, economic, and social life, lacking clear or institutionalized rules for the management of lethality. Every opportunity to impose costs is taken by each state and both sides unleash their full insurgent and counterinsurgent capabilities. States, civilians, and insurgents are locked into the production of unrestrained violence. Ironically, this disorder is a distinct form of order unto itself. We should expect some degree of guerrilla disorder in all wars, but the extent and duration vary across space and time. In some conflicts, guerrilla disorder lasts for many years, while in others the armed actors shift their relationships towards greater cooperation. In other situations, cooperative relationships break down and guerrilla disorder emerges in its wake.
Guerrilla disorder has characterized several major South Asian insurgencies. In Kashmir during the 1990s, the Indian state colluded with some insurgent groups, but the other organizations were targeted in a brutal and sustained fashion. The Hizbul Mujahideen in particular was the object of massive counterinsurgency operations, rather than deal-making, between 1990 and 1999.Footnote 82 As noted previously, during 1990–1992 in Assam, the Indian state brought its might to bear against United Liberation Front of Asom (ULFA) with large-scale military offensives when ULFA's influence had risen to unacceptable levels.Footnote 83 In some areas of interior India guerrilla disorder has emerged as the central state and Naxal insurgents are locked into intense conflict after long years of tacit coexistence and spheres of influence.Footnote 84
Sri Lanka experienced guerrilla disorder during the two left-nationalist Jantha Viamukthi Peramuna (JVP) rebellions of 1971 and 1987–90, in which it was clear that the struggle was to the death for both the insurgents and the state, which pursued a “scorched earth policy against the JVP.”Footnote 85 Extraordinary atrocities and unrestrained violence occurred, as tens of thousands died in the Sinhalese south. Pakistan's counterinsurgency against Bengali guerrillas was similarly characterized by intense escalating violence rather than tacit bargains and the pursuit of norms for the containment of violence.Footnote 86 Situations of no cooperation lead to unrestrained violence that most closely approximates intuitive understandings of civil war. However, this remains only one (tragic) part of a broader range of variation.
Methodological Benefits of Studying Wartime Political Orders
This typology of wartime political orders provides a supple, fine-grained conceptualization of what actually happens in war zones. Such an approach is methodologically helpful for three reasons. First, it allows us to examine multiple relationships within a broader conflict rather than focusing on a single dyad. The state may have different relations with one insurgent group than with another, which cannot be easily measured in an overall conflict coding. For instance, India has cut deals with some insurgent groups while ruthlessly targeting others during conflicts in its Northeast region. Being able to capture this variation is consistent with the state formation literature, which does not ask who wins, but instead how authority is imposed and negotiated among different political players.
Second, this conceptualization allows us to measure the dynamics of a conflict at any given point in time. In year two a conflict may be characterized by a different form of political order than in year seven than at termination ten years later. Wartime political orders may be short-lived or deeply enduring, and my typology allows us to identify and trace them over time regardless of their duration. This fine-grained dependent variable can measure shifts in trajectories of conflict, rather than coding a whole campaign as a single un-variegated observation, ignoring a contemporary conflict to avoid censoring data, or coding any form of unfinished conflict “ongoing” or “other.” Wartime orders provide a valuable lens through which to systematically analyze the shifting tides of war.
To do this in real time, scholars and analysts can use interviews, patterns of force deployment, and the statements of key political actors to identify which rules of the game are being played in a particular area at a particular time. In Iraq in 2006, for example, it was clear at the time that there was a shared sovereignty order between the state and militias in Kurdish areas, and a guerrilla disorder situation in Anbar province. In contemporary Pakistan, the use of the military, the published and spoken statements of political elites and analysts defining their interests, and intelligence intercepts make it reasonably apparent which groups the Pakistani state is cooperating with in some form and which it is in opposition to. We can at least loosely understand which orders exist in different areas of Pakistan's northwest. These relationships can change and my approach provides a lens through which to describe escalation and de-escalation.
Third, any conflict or region of a conflict at any time can be characterized as involving one of these forms of order. Scholars can spatially disaggregate conflicts while still being able to speak to larger questions of political authority.Footnote 87 Observers have often noticed dramatic variation in the nature of wars from valley to valley, village to village, and region to region. Rather than having to code a conflict with a single value or rely on local violence statistics, we can explore this spatial variation in how insurgents and armies deal with one another. We can also scale this typology: in a particular area of a conflict we may see shared sovereignty but in the conflict as a whole the pattern may be spheres of influence. Political orders can be “nested” within one another. Though using the typology requires being very explicit about the level of analysis, relevant actors, and time period to which it is being applied, its flexibility is an advantage.
The Politics of Violence
I focus on concept formation rather than theorizing the variation in wartime political orders. That is a task for future research. Embedded, however, in the existence of the orders I identify are new areas for research. Four research agendas emerge from this study: the political interests of state and insurgents; the co-evolution of states and insurgents; other actors involved in the creation of order; and links to broader dynamics of politics and violence.
State and Insurgent Political Interests
We cannot make general assumptions about the war aims and resolve of states or insurgents. States are often content not to devote their full resources to internal war. Deployed state capacity is endogenous to political relationships, which is why we see governments like India's and Pakistan's—which can rely on vast, ruthless security forces—willing to cut deals, bring insurgents into state governments, and look the other way at collusion between politicians, militants, and organized crime. Similarly, insurgents often shield and hide their strength as part of political-military strategy. They may be willing to cooperate with state power, creating stability even without a monopoly of violence. Bargaining, escalation, and calibration are pervasive in the behavior of both states and insurgents. This should not be surprising, since similar dynamics are common in international politics. Coercive diplomacy, deterrence, and limited war involve calibrated threats and applications of force rather than the pursuit of brute force.Footnote 88 Both within and between states, war is a means to achieving political goals, not an end in itself.
Thus the interests of contending armed actors may not be correlated with standard explanatory variables in the civil conflict literature like per capita GDP, regime type, or state capacity, or insurgent variables like group size and external sponsorship. Existing research has subordinated agency to structure by too often assuming fixed interests and then assessing how varying capacity shapes outcomes. We also need to theorize and endogenize the interests of actors in order creation: why, for instance, is Pakistan willing to crack down on some militant groups while tolerating others? These are fundamentally decisions about political interests rather than solely about the balance of power, control, or doctrine.Footnote 89 Exploring why states and insurgents do and do not arrive at deals for managing violence can reinvigorate the political foundations of research on civil war by showing how goals and strategies shape coercion.
Co-evolution of States and Insurgents
A second implication of this study is that the lines between states and their foes are often tenuous and fluid. Though here I have focused on dyadic relations between states and insurgents, the evidence shows that even this simple boundary can be problematized. States can be born of or entangled with non-state actors, and the bargains and relationships that emerge during and after conflicts profoundly influence political trajectories. The Burmese army, now a ruthlessly statist organization, was built atop non-state armed groups that had emerged during Japanese occupation in World War II.Footnote 90 In the post-Soviet Caucasus, new state structures were sometimes forged by coalitions of armed groups.Footnote 91 The consolidation of Indonesia's state came through the meshing of insurgent groups and preexisting governance structures, as did the consolidation of the North Vietnamese state.Footnote 92 States can even use insurgents to reap benefits. Wedeen, for instance, suggests that the Yemeni regime of Ali Abullah Saleh intentionally used violent disorder as a mechanism of control and alliance with the United States.Footnote 93 These examples are a tiny slice of the possible ways that armed groups constitute or affect state power, but they suggest how interesting and important these relationships can be.Footnote 94
Civil war, organizational formation, and political survival are thus intriguingly intertwined. A huge amount of research in civil war is explicitly state-centric,Footnote 95 but there is a strong case to be made for “de-centering” the state by examining the co-evolution of states and insurgents.Footnote 96 This means taking insurgents, warlords, and other organized violent actors seriously as autonomous actors. These organizations are not obscure speed bumps on the teleological highway to something called the modern state, but instead constitutive of who rules, where, and how much. This claim reflects insights from state formation research: states do not emerge simply from the decisive shattering of other actors, but instead through their co-optation, coercion, and incorporation. It also echoes the rich theoretical and empirical literature on contentious politics, which emphasizes the complex interactions of states and those making demands on them.Footnote 97 These processes may leave non-state violent actors with power rather than creating a monopoly of violence. We should not think of this outcome as illegible chaos or “incomplete” state building, but instead as its own distinctive form of order.Footnote 98
Other Actors in Wartime Order
The typology I have outlined examines relations between states and violent non-state actors. This is a useful starting point but it is obviously a massive simplification that needs to be deepened and nuanced in future. A more comprehensive understanding should incorporate other actors beyond governments and armed non-state groups in explaining how authority is created and distributed in war. Promising work has already been done on some of these other actors that can be built into broader studies of how they interact with states and insurgents. Civilians may be able to lobby, threaten, or influence armed groups (whether state or non-state), giving them a key role in shaping political authority.Footnote 99 This is not a novel claim. Yet it is crucial to remember that not only are insurgents and states competing for civilian loyalties, but that the armed groups may also be cooperating with one another. Civilians thus face the danger of being “frozen out” by collusive deals between armed actors, creating an even more complex environment within which to pursue their interests.Footnote 100 Rather than assuming a straightforward competition over the loyalty and compliance of civilians, new research should incorporate the ways that cooperation between groups structures the opportunities for civilian participation and influence.
External states and international organizations often have power that they can apply to bolster or undermine warring actors, or to even carve out political space for themselves. Conflicts in Lebanon would have had a very different complexion without Israeli and Syrian involvement, for instance, while the UN and NATO are key players in places like Kosovo and Cambodia.Footnote 101 It would be particularly interesting to explore how local armed actors try to capture and use international influences as tools in their domestic struggles, and thus to flesh out the varying relationships not just between armed groups but between armed groups and international actors. Criminal networks and economic elites are also important in civil war.Footnote 102 Economic players—like businessmen and smugglers—may be key brokers in putting together and maintaining these arrangements and their interests could shape the political objectives of contending actors beyond simply providing resources. A wide variety of actors can affect the distribution of authority and control amidst violence.Footnote 103 Getting inside these multi-faceted interactions is a clear next step for research.
Violence Management beyond Civil War
Finally, we can draw links between civil wars and non-war situations in which states and other armed actors try to manage and control violence. Collusion, cooperation, and deal-making are pervasive in numerous contexts.Footnote 104 The state is not locked into pure conflict with other producers of violence because it can manage or cooperate with them. In the Robert Taylor Homes, for instance, the Chicago Police Department would occasionally launch “gang suppression” offensives, but generally tried to contain instead of eliminate violence.Footnote 105 In north India, security forces are often willing to look the other way, even to collude, as caste and communal provocateurs mobilize violence against their enemies. The police have a wide variety of relations with thugs, criminals, landlords, businessmen, and politicians.Footnote 106 This interweaving of violence, daily life, and politics has resonance in contexts as different as elections in the Philippines,Footnote 107 imperial expansion in Asia,Footnote 108 “composite states” in early modern Europe,Footnote 109 the fusing of parties and paramilitaries in Weimar Germany,Footnote 110 and organized crime in Italy.Footnote 111 In each case, non-state violence was not automatically opposed by an impartial state, but instead emerged in a relationship with state authority. States are not neutral arbiters of social contention seeking to protect a monopoly of violence in the greater good; any theory or policy doctrine that assumes they are will have troubled encounters with reality.
Armed organizations, both state and non-state, regularly attempt to arrive at understandings about order and authority. Violence is not simply used to mount or repress total military challenges to state power, but instead is a flexible tool in pursuit of numerous political, economic, and social goals.Footnote 112 Exploring the deployment of violence across otherwise-different contexts is a way of breaking down barriers between related topics too often studied in isolation. As Cramer argues, violence can be integral to “normal” politics rather than simply an aberrant spasm of “development in reverse.”Footnote 113 We need to understand how these processes work rather than barricading off civil wars as a realm of unique dysfunction and abnormality.
Conclusion
I started from a basic question: if insurgency and counterinsurgency bear a “family resemblance” to competitive state-building, what does state formation tell us about the politics of civil wars?Footnote 114 I have argued that, rather than a total war of violence-monopolizing organizations, civil conflicts are instead frequently characterized by bargains, deals, and norms that structure patterns of violence, from collusion to spheres of influence to guerrilla disorder. This exercise in concept formation helps us make analytical sense of the puzzling array of relationships between states and insurgents that reveal themselves when we delve deeply into conflicts. The distribution of control and level of cooperation between states and non-state armed groups vary considerably, creating different forms of wartime order. Politics do not end when the first bullet is fired.
Implications for scholarship
Exploring how these orders work requires more concepts and more puzzles. Research into the politics of civil war can become deeper by deploying a wide variety of methodological tools to refine concepts, develop theory, and test hypotheses. Detailed fieldwork and archival evidence can uncover intriguing new puzzles and comparisons by delving into the past and present of war-torn countries. Quantitative data collection and analysis on increasingly fine-grained measures of conflict dynamics and termination can establish patterns and generalizations. Formal theory can unpack how and why strategic interactions between states and insurgents lead to different forms of political order. Ethnography can yield powerful insights into how insurgents, government officials, and civilians interpret and understand the varying politics of war zones. Concepts, theory, and evidence should be accumulated and challenged from all directions. Answering “how much?” questions is important, but intellectual progress requires an equal emphasis on asking original “what is?” questions.
Pushing the intellectual frontier forward also requires a broadening of how civil conflict is studied and understood. There is overlap between civil war and other fields and disciplines that provides both new questions and new answers. Scholars of political violence should more aggressively engage with research on economic development, state formation, political organization, international politics, and voting behavior, among many others. A subfield that speaks only to itself will lose intellectual vibrancy, but one that situates civil war as related to and embedded within broader political, social, and economic processes will spur the most promising new research.Footnote 115
Policy implications
The claims advanced here matter for policy. In current discussions of counterinsurgency and state building, the focus is on services, security, good governance, counterinsurgency “best practices,” and the creation of a legitimate monopoly of violence.Footnote 116 David Lake, for instance, argues that “building legitimate states by providing effective social services is a prerequisite for success” during “a global counterinsurgency campaign that encourages integration into the American-led international order.”Footnote 117 Providing security and building institutions is deemed essential to political and economic progress in the face of rebellion or instability; such an approach is often called “classical” counterinsurgency. This is sometimes the right policy. State-insurgent cooperation can be difficult to sustain and is fragile to spiraling into breakdown, while some regimes and some insurgents may be unwilling to bargain or cut deals. The natural policy prescription is thus deploying “government in a box”Footnote 118 to create an institutionalized environment for security and goods provision. These policy recommendations bear powerful resemblance to the “three great oughts” that guided and misguided American counterinsurgency during the Cold War: “security, good government, and progress.”Footnote 119
However, assuming that mass populations and local elites desire to be ruled by a powerful central state is problematic.Footnote 120 The people being “provided security” frequently have absolutely no interest in strange men with guns appearing in their villages to tell them what to do. State building amidst civil war is often bloody and cruel. While problems of instability are sometimes caused by insufficient state capacity, sometimes they are driven instead by fears of excessive state power. On the streets of Kashmir and in the jungles of eastern Burma, many people oppose the creation of a more powerful state (legitimacy, whatever that is supposed to refer to, is not in the cards). Population security and state building sound good in the metropole but often look very different on the periphery; seemingly technocratic policies like governance and service provision are in fact short-hand for state coercion, homogenization, surveillance, and extraction.Footnote 121 Leviathans are far more attractive to those who will run them than to those who will be in their gun sights.Footnote 122 Colin Jackson has powerfully shown that military efforts to “do politics”—through security and service provision—frequently misdiagnose the core political stakes of conflicts.Footnote 123 Attempts to impose the statist visions of influential counterinsurgency and military intervention doctrines can actually make things worse, leading to violence, resistance, and “durable disorder.”Footnote 124 As Ariel Ahram writes, “many efforts to strengthen states and eliminate militias have proven quixotic, if not counterproductive.”Footnote 125
Policymakers should carefully assess the benefits of both state building and variants of indirect rule, informal governance, and wartime order.Footnote 126 Violence management provides a potentially valuable alternative to violence monopolization, and the cooperative wartime orders I identify may provide useful pathways to stability. As Stephen Krasner notes in summarizing James Scott, “we must strip off the blinders that Hobbes has placed upon us.”Footnote 127 Order is not synonymous with a robust state or monopoly of violence. An accurate understanding of the historical reality of state formation, with its pervasive accommodations between, and interlacing of, state and non-state violence, can allow for more subtle and useful policies to minimize the terrible costs of conflict. We need to rebuild the intellectual foundations of theory and doctrine on counterinsurgency and state building.Footnote 128
Iraq clearly shows the relevance of this focus on multi-faceted, often informal, arrangements between states and non-state armed groups. The US occupation initially aimed to create a central state but this effort at escaping the Hobbesian dilemma disastrously failed. The failure to do “nation-building” well was not the result simply of poor planning but also caused by the deep difficulty of imposing a particular configuration of state power. Rather than capacity-building leading to legitimacy and good governance, a tenuous stabilization occurred instead precisely through the processes of collusion, spheres of influence, and shared sovereignty I have highlighted in my typology of wartime political orders. Militias and insurgents continue to exist, but their cooperation with state authorities and military forces has created a new set of orders.Footnote 129 In Kurdish northern Iraq shared sovereignty dominates, while in Anbar Province collusion and spheres of influence have largely replaced guerilla disorder since 2007. In Shiite areas, there is a mix of collusion, spheres of influence, and tacit coexistence between the state and paramilitaries.
Though they do not resemble a classical state monopoly, these arrangements have played a pivotal role in reducing violence. Stabilization in Iraq has not occurred by competently providing goods and services, eliminating all other armed actors, or winning the unconditional loyalty of citizens to the state, but it has had the virtue of at least minimal success. Analytically, the concepts offered in this article provide a useful fine-grained lens for understanding Iraq's political dynamics. Changing political relationships between central authority and peripheral armed actors have shaped the emerging order. Trying to forge formal state institutions has actually heightened distributional conflict, while informal norms and arrangements have created more space for live-and-let-live bargains and mutually beneficial collusion.Footnote 130 Over time, formal federalism or power sharing may emerge alongside something like a monopoly of violence, but in the short-run wartime orders provide broad stability while aggressive monopolization efforts are likely instead to be destabilizing.
My core themes are clearly important for the future of Afghanistan. As Thomas Barfield persuasively argues, state centralization after the overthrow of the Taliban has been ineffective.Footnote 131 There has been a disjuncture between the goal of state building and the reality of pervasive spheres of influence, shared sovereignty, collusion, and guerrilla disorder, which vary across space and time within Afghanistan. Formal political arrangements do not reflect the actual relationships between the central state and coercive power on the ground. Bringing the formal and informal into alignment is crucial. Stability thus may require the opposite of further attempts at state building and violence monopolization.Footnote 132 Crafting local deals and allowing Afghans to construct workable bargains is therefore the best option.Footnote 133 This process will certainly be messy, morally challenging, and enduringly tenuous, but creating political orders that accommodate multiple state and non-state armed actors is essential to Afghanistan's future.Footnote 134
Cooperative wartime orders may sometimes be second-best solutions compared to the (distinctly elusive) ideal of an inclusive, effective, and humane counterinsurgent state, but they are nevertheless solutions. Militarized state building is clearly not the only path to workable order in counterinsurgency; the brutal costs required to impose state dominance and maintain social control often make the game not worth the candle. The threat from weak and failed states has been dramatically exaggerated, and there is little need for expansive “global counterinsurgency.”Footnote 135 Under some circumstances less international intervention may actually lead to more stable political arrangements and state structures.Footnote 136 Policymakers, scholars, and citizens need to understand the real and existing processes of coercion, co-optation, and cooperation that constitute the politics of violence rather than embracing technocratic platitudes about good governance, security provision, and development. There are many ways to structure states and many ways to establish order. Understanding these illuminates the possibilities for political flexibility, creativity, and imagination in dealing with conflict.
