Working Papers
Why Do Rebel Groups Ally and Fight? A Network Model of Wartime Cooperation and Conflict
This article develops a relational theory of wartime cooperation and conflict and tests it on the Syrian civil war, reconstructing the inter-rebel network as a two-layer object, one layer for cooperation and one for conflict, from georeferenced event data on 77 armed groups active between 2011 and 2024. Building on accounts centered on identity, power, and territory, the theory derives four hypotheses, including a "frenemy" hypothesis that treats the coincidence of cooperation and conflict on the same dyad as a third, analytically distinct relational type rather than a residual overlap between the other two. Five complementary network models, cross-sectional and temporal exponential random graph models, a multilayer ERGM, a four-category multinomial logit, and an additive and multiplicative effects model, are estimated jointly, with agreement across their non-overlapping assumptions serving as the inferential warrant. Geographic proximity is the strongest and most consistent predictor of whether any tie exists; power operates as a node-level activity effect, with the most resourced groups forming disproportionately many ties of every kind; and ideology suppresses conflict without driving alliance, an effect concentrated early in the war. The frenemy pattern, rather than reflecting a distinct dyadic tendency, proves to be a by-product of activity concentrated on a small number of powerful hubs.
From News to Networks: An LLM-Based Pipeline for Extracting Rebel Alliance Structure, Identity, and Power.
This project constructs an original data pipeline to extract directed inter-rebel network data from the Global Database of Events, Language, and Tone (GDELT) for the Syrian civil war (2011-2024). Existing conflict event datasets treat actor dyads as undirected and omit perpetrator-level identity and power characteristics, making it impossible to reconstruct inter-rebel alliance and rivalry networks over time. This pipeline addresses that gap by retrieving source articles linked in GDELT event records and passing them to a large language model coding protocol that extracts three variable categories: relational ties (cooperation or rivalry, directed), identity composition (ethnicity, religion, ideology), and power characteristics (membership, territory, funding, relative capability). The resulting dataset will be used to test how alliance structure shapes rebel governance outcomes in Syria and will be made publicly available as a generalizable tool for researchers studying fragmented insurgencies in any conflict context.
To Which Victor Go the Spoils? Predicting "The Day After" Regime Change.
A growing amount of research has analyzed when national movements and insurgents overthrow the existing government, but which groups within those movements and insurgencies actually come to power after regime change, and why? We are finalizing the construction of a massive dataset that will help us answer these questions.
The Problem of Peace: Network Dynamics of Rebel Groups and Governance Institutions. A Comparative Network Analysis of HTS and SDF
Why did two armed groups operating within the same Syrian civil war, Hayat Tahrir al-Sham (HTS) and the Syrian Democratic Forces (SDF), produce similar institutional governance arrangements despite being ideologically worlds apart? HTS is a Salafi-jihadi organization rooted in global jihadism; the SDF is organized around Kurdish-leftist democratic confederalism. Standard ideological explanations would predict radically different governance forms. Instead, both groups constructed formal administrative institutions with partial civilian inclusion and multi-actor participation mechanisms - one through the Syrian Salvation Government, the other through the Autonomous Administration of North and East Syria. This article argues that this convergence is explained by the relational architecture of wartime alliances: specifically, the identity composition and power distribution of inter-rebel networks. Drawing on a temporally disaggregated dataset of Syrian inter-rebel cooperation and rivalry ties from 2011 to 2024, the article demonstrates that HTS's dense early cooperation network and selective rivalry management, and the SDF's federated multi-ethnic coalition structure, both generated relational pressures toward institutional accommodation. These findings advance a relational theory of rebel governance that locates the origins of formal governing institutions not in elite preferences or ideology alone, but in the wartime network structures that determine which commitments are credible before the fighting ends.
Peace Processes and Terrorist Tactics: Negotiation Timing, Battle Deaths, and Terrorism Onset
Peace negotiations between governments and non-state armed groups are widely assumed to reduce violence. Even when it has been shown that peace processes increase violence, it was through the exclusion of other groups (Arı et al., 2025; Blaydes & De Maio, 2010; Höglund, 2011; Pospieszna & DeRouen, 2016; Prem et al., 2022; Sener, 2025; Stedman, 1997; Sundberg & Melander, 2013). This article challenges that assumption by examining whether peace processes alter the form of violence rather than eliminating it. Specifically, whether negotiations prompt rebel groups to shift from guerrilla warfare toward terrorism. Drawing on the case of the Kurdistan Workers' Party (PKK) in Türkiye from April 1989 to December 2019, I construct a monthly panel combining terrorism and battle event data. Terrorism is operationalized using three conceptually distinct target-type measures. Logistic regression and OLS models disaggregate the effects of conflict intensity and negotiation timing on terrorism onset and intensity. Results show that active negotiations are associated with significantly higher probabilities of terrorism onset across all three operationalizations, even controlling for prior conflict intensity. Battle deaths robustly predict whether terrorism occurs in a given month but not how much terrorism occurs once it begins. These findings are robust to the choice of target-type definition and suggest that peace processes may create strategic incentives for non-state actors to adopt terroristic tactics, with important implications for conflict management and counterterrorism policy.
Schools of Authority: Rebel-Run Education Systems in Civil Wars.
Education is one of the most intriguing public services that rebels manage during civil wars. Unlike other public goods, insurgent-run education constitutes a long-term investment, with high organisational costs and delayed and uncertain returns. Nevertheless, several insurgent organisations provided education in areas under their control, establishing alternative institutions to manage education systems. Such institutions offered certification, developed curricula, oversaw teachers, and provided vocational training. Existing scholarship shows that secessionist groups often use education to build legitimacy, while others emphasise schooling as a recruitment tool, especially when combined with health services. Building on this work, this article examines why non-state armed groups establish education systems and how they adapt these systems during conflict. Drawing on a wide range of first- and second-hand sources related to Syrian rebel groups, including propaganda magazines, internal documents, and annual school-level reports produced by non-governmental organisations and United Nations agencies, the article investigates why and how Syrian rebel groups establish and institutionalise education systems during the armed conflict.
Rebel Group Constellations: A Methodological Approach to the Study of Rebel Alliances.
Fostering Constructive Conversations: AI-Mediated Dialogue for Intergroup Understanding.