There is a famous scene in the 1979 British comedy Monty Python’s Life of Brian where the main character, a Judean everyman named Brian joins a subversive group called the “People’s Front of Judea” only to realize that it is competing against several other groups with similar names (“Judean People’s Front” or “Judean Popular People’s Front”). “Splitters” is how the members of the first group scornfully refer to their competitors. This is to say that the splintering of extremist groups, both armed and nonarmed ones, is such a common occurrence that it has become part of popular culture. Recent research has put a number on this phenomenon: about 30% of nonstate armed groups form by splitting from existing armed groups; the trend appears to be rising and, as a result, civil wars are becoming more fragmented.

Nevertheless, the effects of armed group fragmentation remain poorly understood. Does splitting harm or help the rebel cause? Does it end up radicalize or moderate the rebels? Does it contribute to shorter or longer conflicts? These are the core questions explored by Evan Perkoski in his study of rebel group fragmentation.

A widespread view posits that splintering hurts political movements, including nonstate armed actors, thus potentially speeding up their demise. Not necessarily, answers Perkoski. His main finding is that rebel group fragmentation is not always a negative outcome, seen from a rebel’s perspective. In fact, the outcome depends on the modalities of the split in the first place. The book’s central insight is that understanding what the fate of splinter groups is, whether they survive or even thrive, and whether they radicalize or moderate, requires a detailed account of why and how they broke away in the first place. Existing research has tended to blackbox this process, stressing instead factors such as repressive intervention and hierarchical leadership structures.

Perkoski opens this black box and makes two key claims, while also marshaling plausible empirical evidence in support of both of them. First, he claims that variation in the rate of survival of splinter groups is a function of the number of disagreements that motivated the split. More specifically, if the original disagreement was about a single, overarching dispute (what he describes as a unidimensional split), then the emerging group is likely to develop aligned internal preferences, be cohesive and resilient, and hence survive longer as a result. Second, the variation in radicalization of splinter groups is a function of the type of disagreements that motivated the split. Unlike ideological and leadership disputes, disputes about strategy are likely to motivate and attract hardliners, thus turning the new group into a more radical entity. Groups emerging out of both a unidimensional and a strategic dispute are a counterinsurgent’s worst nightmare, as they are likely to prove more resilient and radical.

The supporting evidence includes two large case study comparisons. The first pits the Irish National Liberation Army (INLA) against the Real Irish Republican Army (RIRA); and the second Al Qaeda in Iraq (AQI) against ISIS. These case studies are supplemented by an analysis of 300 randomly selected groups. Perkoski finds that older groups are more prone to splintering; and that splinter groups are equally likely to survive compared to parent groups, but that those formed out of unidimensional disputes greatly outlast those that emerged out of multidimensional splinters; they even outlast the parent groups that emerged from. Overall, and controlling for other factors, unidimensional splitters are nearly 25% more likely to survive compared to multidimensional ones. He also finds that within unidimensional splits, strategic disagreements produce both longer-lived and more radicalized groups, as expressed in higher rates of violence: these groups kill more people per individual attack. Overall, this is a clear and well-executed study on a question that is arguably very hard to study—and study well. The main concerns are related to conceptual and measurement issues. On the conceptual side, it is very difficult to differentiate between unidimensional and multidimensional disputes, as it is difficult to distinguish between ideological, leadership, and strategic disputes. On the operational side, it is equally challenging to collect the granular data needed and it is even more difficult to interpret it. Historians, whose job is to execute this task for a single case, constantly disagree about interpretations.

The book offers both policy and research contributions. On the policy side, states should not assume that a “divide and conquer” approach to conflict is a good idea; indeed, it often backfires. Moreover, when negotiating with rebels, states should try to make sure not to create incentives for splits; they should also refrain from negotiating with highly factionalized groups since they are also likely to fractionalize. At the very least, states ought to be highly informed about the internal dynamics of their nonstate foes. On the research side, this book reminds us that the observable behavior of groups masks numerous less-observable processes operating below the surface and, thus, calls for more and better research about the internal dynamics of nonstate armed groups. It also suggests several important extensions, most notably about how parent groups behave after a split. I would also add the need to revisit the obverse of splits, namely alliances and mergers in light of the internal dynamics highlighted in the book.

This review is a good opportunity to call for more synthesis. There are basically three ways to approach internal armed conflicts: one is to analyze intergroup dynamics; the second one is to focus on the dynamics between armed groups and civilian populations; the last one consists of examining internal group dynamics. Divided Not Conquered clearly belongs to the last category. Perkoski tries hard to hold the other two dimensions constant, but this is inherently difficult—it is a general problem in the literature and one that hampers synthesis. This explains why the takeaways tend to be on the modest side—“Findings offer little in the way of precise guidance” he warns (p. 183). Perkoski deserves credit for not over-claiming. However, we can do better by thinking more carefully, and more ambitiously, about how to bring the three dimensions together.