Kalyvas, Stathis N. 2006. The Logic of Violence in Civil War. Cambridge University Press: New York.

lThis book is the product of a Doctoral project that blossomed into 10 years of sustained interest in the dynamics of violence in civil wars. As such, it can be extraordinarily technical and advances its argument in painstaking details, with a slew of examples provided for every point. However, the argument itself is quite interesting, and so is worth knowing and thinking about. The detailed summary of the argument appears on pages 11-14, so this will be abbreviated. Kalyvas begins by demonstrating the extremely high level of violence in civil wars, esp. for the civilian population. Multiple narratives have been developed for this phenomenon -- civil wars are like diseases or forces of nature (they erupt, for example) that reveal the true Hobbesian face of humanity. Or, alternatively, civil wars are an expression of a total political polarization of a society that subsumes all other differences and pits two sides against each other in a battle of annihilation. Or perhaps civil wars are incredibly violent because, in many cases, the challenge faced by the combatants is not how to destroy their enemies, but how to go about finding them in the first place. It is this last explanation that feels most plausible to Kalyvas, and it's where he builds his theory.

If the real difficulty in an insurgency is finding the enemy and its supporters, then both sides have a couple of options. They can use selective violence to kill or capture the other side's informants, collaborators, militias, and leadership. But this requires a robust intelligence network that provides inputs on who to target. Alternatively, they can employ indiscriminate violence, which virtually always fails because it only drives people to the other camp or causes them to suffer mutely (if there is no way to avoid the violence by, for instance, becoming an informer, then it doesn't make sense to incur the costs of doing so to no benefit.) So selective violence is the way to go, and most combatants generally come about to this approach. (It should be noted here that Kalyvas excludes genocide and ethnic cleansing from his model, because the complete destruction and NOT subjugation of a population is the goal in that case.)

Unfortunately, selective violence is very hard to pull off well, as it requires a lot of information, which in turn requires the freedom of movement or control required to gather that information. In areas completely controlled by one side, the level of violence is rather low, because identifying and eliminating enemy elements is much easier. However, as an area becomes contested, the level of violence spikes, because the two sides are engaged in a cycle of reprisal and counter-reprisal aimed at the destruction of informer networks and the cooperation of the population. However, as the two sides reach parity, the level of violence against civilians decreases, as both sides are unwilling to alienate the population (and thus potentially lose the area) and the informer networks are mutually deterred.

That analysis adopts a pretty standard model for COIN -- there are actors, there is the population, the actors try to influence the population, and whoever can achieve both popular legitimacy and military supremacy wins. Where Kalyvas gets interesting is when he retells this story from the perspective of the individual non-combatant living in a civil war. Running against mythology, Kalyvas asserts that the overriding concern for each member of the population is to stay alive, and demonstrates the point by noting that control of the population militarily is an excellent indicator of its cooperativeness from a mind-bogglingly diverse set of locales. Because control and cooperation are related, he is then able to split the conflict into 5 levels of control (insurgent controlled, insurgent contested, contested, incumbent contested, incumbent controlled) and describe the decision-making procedures in each one.

What he finds is that rational actors in contested zones will use the combatants to settle local grievances and engage in reprisals against one another in a manner demonstrated by the two block quotes below.

Intimacy: "... intimate violence in civil war is, as I suggested, often related to interpersonal and local disputes (it is "right under our feet, so close to us ...") rather than impersonal abstract hatred. Yet this statement should be interpreted less as a mere endorsement of the Hobbesian view of human nature as fundamentally violent under conditions of insecurity, and more as an observation about the power of the practice of denunciation as spurred by interpersonal competition. For intimate violence signals less a process of politicization of individual life and more a process of pervasive privitization of politics; less a transgression of social ties and more their full, though pervasive, expression. The evidence of malicious denunciation in symmetric and concentrated environments further undermines the view of intimate violence as the exclusive outcome of deep divisions -- though such violence can eventually transform interpersonal hatreds into impersonal ones." (362-363)

Alliance: "Alliance entails a process of convergence of interests via a transaction between supralocal and local actors, whereby the former supply the latter with external muscle, thus allowing them to win decisive advantage over local rivals; in exchange, supralocal actors are able to tap into local networks and generate mobilization. A great deal of action in civil war is, therefore, simultaneously decentralized and linked to the wider conflict. Thus civil war is (also) a process that connects the collective actors' quest for power and the local actors' quest for local advantage. Put otherwise, violence can be a selective benefit that produces local mobilization via alliance." (383)

He then applies the theoretical model to an incredibly detailed micro-analysis of a specific area of Greece and catalogs the local experience with violence during the Greek Civil War using detailed judicial archives, personal interviews, and military records. It's really quite amazing, and, of course, the findings validate the theory. In fact, looking at the model provides a really useful way to approach aggregation of local inputs in COIN, and should be thought about carefully, as many theories sound very good on paper, but are difficult, if not impossible, to put into practice.

SEE ALSO: Wickham-Crowley (1992) pg. 260-261, for a how this process proceeded in Latin American revolution/counter-revolutionary contexts.