What causes rebellion? Janet Lewis’s ambitious book, How Insurgency Begins: Rebel Group Formation in Uganda and Beyond, shines new light on this important question by offering an unusual glimpse inside the black box of the origins of insurgency. The book has important implications for literatures on state-building and armed conflict, showing how assumed knowledge on the key correlates of insurgency may be better understood as factors contributing to escalation in violence rather than as independent causes of conflict. Instead of investigating the causes of rebellion per se, Lewis seeks to explain why only some rebel groups survive past their initial formation.

Lewis’s argument moves in three parts. She first identifies a new stage of rebellion that exists after rebel groups form but before they become viable threats to the state. By overlooking this preliminary stage, Lewis contends that scholarship on rebellion suffers from selection bias, limiting inferential analyses and policy interventions. Second, including this embryonic stage shows that many factors associated with rebellion—including material resources (like natural resources or foreign investors), terrain, and ethnic grievances—are not that important to initial formation. Lewis focuses on ethnic grievances, using her new approach to show that, although rebel groups form in both ethnically homogeneous and heterogenous areas, they survive in ethnically homogeneous areas. Ethnic grievance therefore does not motivate rebellion, although it may be adopted post hoc as a justificatory narrative. In this sense, Lewis favors “feasibility” over “motivation” as an explanation for rebellion, concluding that “rebel groups form due to largely idiosyncratic, individual-level factors” (p. 60).

Finally, Lewis proposes kinship-based information networks as a previously unidentified mechanism linking ethnic homogeneity to rebel group survival. Rebels need the dense and cohesive kinship networks found in ethnically homogeneous regions to quickly spread positive rumors about themselves, thereby discouraging ordinary citizens from reporting the rebels to state authorities in the early days of the insurgency when the rebels are most vulnerable. This buys rebels time to organize and plan, so that by the time the state becomes aware of their presence they have sufficient capacity and agility to withstand attacks or the capture of a commander (p. 38). In contrast, in ethnically heterogeneous communities, these networks are thin and fragmented, such that ordinary people are likely to report on rebels before the rebels can persuade them not to.

Perhaps the most significant contribution comes from Lewis’s methodological approach. Along the lines of works such as Virginia Fortna’s Does Peacekeeping Work? (2008) and Hendrik Spruyt’s The Sovereign State and Its Competitors (1994), Lewis’s intervention deftly illustrates the pitfalls of selecting on the dependent variable, revealing a long-standing blind spot that has limited cross-national and quantitative studies on why insurgency begins. As Lewis notes, it is methodologically challenging to identify the origins of rebellion—an illegal activity that often carries with it the heaviest of state-imposed punishments if caught. Aspiring rebels are rarely forthcoming about their identities, goals, and strategies to violently challenge the state. This means that existing studies of rebellion often focus on rebellion-building activities among groups that have already gained public notoriety (p. 10). Lewis’s field research, employing interviews from across Uganda, within-country quantitative analysis, and archival research, paints a complex picture of insurgency that is therefore in equal parts unusual and valuable.

In addition to the implications highlighted in chapter 7, Lewis’s findings also offer insights into the complex relationship between intelligence and coercion. Although this dynamic is elaborated in more detail elsewhere (e.g., Stathis Kalyvas, The Logic of Violence in Civil War, 2006), Lewis’s careful empirical work reveals the extent to which intelligence and coercion complement rather than substitute for one another and have a nonlinear relationship. A rebel group lacking coercion may require intelligence for its very survival, but so does the state, despite its comparatively overwhelming capacity for coercion. Surveillance increases the value of coercion, and more so when coercion is limited.

The book’s ambition and originality necessarily result in some limitations. A few are of note to the general scholarly reader. The first centers on the definition of rebel group “viability.” Lewis notes, “It is difficult to specify a precise number of troops and months on a base without introducing some arbitrary judgement” (p. 34, n26), but the extent of this difficulty is belied when one of the threshold conditions is initially given as 100 (rebel) troops (p. 34) and later as 200 individuals (p. 127). Fortunately, the book’s findings are not sensitive to these different thresholds (p. 127). However, this discrepancy highlights an underlying issue: rebel “viability” describes something qualitative about a group’s capacity to threaten the state—whether linked to legitimacy, military control, service provision, sovereignty, and so on. But what this quality is remains unclear. Without such a grounding, the threshold conditions are necessarily somewhat arbitrary and may obscure key theoretical claims about rebels that underpin the intervention.

The second set of questions concern Lewis’s proposed mechanism for rebel group survival: control of local information networks. Rather than showing direct evidence of this proposition, Lewis focuses on its possible consequences; for example, rebels’ limited use of violence in their embryonic stage. We are told—both by Lewis and her interviewees—that limited rebel violence should be interpreted as evidence that rebels initially prioritize cultivating popularity and informational networks among civilians. However, this does little to dispel alternative interpretations; for instance, that rebels might lack the resources or know-how to launch larger attacks, or they do not engage in violence because they do not yet suffer from internal factions that could muddy their initial goals (see, e.g., Cunningham et al., “Shirts Today, Skins Tomorrow: Dual Contests and the Effects of Fragmentation in Self-Determination Disputes,” Journal of Conflict Resolution, 56(1), 2012). Although suggestive, the evidence does not quite seal the deal.

Then there is the functioning of the mechanism. Lewis offers a helpful theoretical account of how kinship networks could enable rumors to spread quickly. However, to show that the significance of ethnicity is attributable primarily to the swift exchange of information among kin (rather than through shared ideology, political agenda, or grievances, for example) would require disentangling the two. Yet, throughout the book, ethnicity is perfectly linked to the informational networks that Lewis argues are actually at work. The closest interrogation of this linkage is a set of paired comparisons in chapter 5, in which Lewis compares successful rebel groups from homogeneous areas to failed groups in heterogenous areas. This shows the plausibility but not the probability of her intervention; the cases are all “most likely” to prove her point. These methodological moves leave open the possibility of reverse causality. For instance, it is possible—and I venture plausible—that people are more likely to spread rumors when they are already disposed to support rebels.

Third, how kinship networks actually facilitate communication remains elusive, and key concepts of “information” and “rumors” remain thinly developed, leaving important issues opaque. For example, Lewis posits that civilians respond to no news about rebels in the same way that they respond to bad news; that is, by outing rebels (pp. 48–51). But this interpretation is puzzling: if secrecy of rebels’ whereabouts and their identity is a key condition for rebel group survival, then presumably fragmented networks would have benefits as well, limiting the spread of information and making it completely easier for rebels to control their narrative among fewer people. It remains unclear why ethnic networks spread positive rumors more than negative rumors and why they are more likely to spread unverifiable rumors than real information about rebels (which would presumably place rebels at high risk of exposure). These ambiguities also make it difficult to assess how advances in communication technologies, such as social media and encrypted modes of communication, might affect the argument.

These lingering questions highlight the ambition of Lewis’s inquiry. Although this book may not deliver a full-fledged theory as promised in its opening pages, its methodological intervention, conceptual work, and empirics make unique and valuable contributions to studies of state formation and violent conflict, initiating not one but several new areas of inquiry for scholars in the field.