Christine Mathew June 4, 2015 Spatial Modeling Final Paper: Examining Epstein’s “Rebellion” Model Introduction: Rebellion is defined as political violence inflicted against a state by civilians (Boswell, 540). Literature based around the subject of rebellion often describes the phenomena of civil violence as a product of a series of external grievances in combination with political factors. Understanding rebellion and why people rise up and join a rebellion helps to reveal explain how conflicts have evolved over time (Humphreys, 436). Research about rebellion can also be useful in re-examining how conflict resolution strategies are conducted, and how they can be improved (Humphreys, 436). There have been many attempts in the literature to try and understand why some people stand up and demonstrate against a government while others remain compliant. It’s important to realize that not every person has a choice in whether or not they rebel, as coercive forces often times manipulate people into taking a position in a conflict (Grigoryan,171). However, there are both rational and non-rational reasons that are used to help explain why people rebel. On the rational side, one may choose to rebel in the hopes that doing so may lead to changes that would improve their condition (Gurr, 130). Non-rationally, there are psychological findings that suggest that the simple act of expressing one’s frustrations or grievances is “self-satisfying” (Gurr, 130). However, this still leaves researchers relatively unaware about the specific associated factors and internal thresholds that once surpassed, lead an agent deciding to rebel Joshua Epstein took a complex systems approach to addressing this gap in the literature by creating a model about rebellion, and the various elements associated with rebellion. Contrary to what one may assume, rebellion is rarely sparked out of public grievances alone, as there are many external systematic elements that play key roles. One theory suggests that rebellion is likely to occur when the discontented public has been granted the freedom to ignite action (Gurr, 144). For example, rebellion is more likely to ensue when a government’s military control has become diminished, as the lack of enforcement is seen as a weakness and an opportunity to take arms. It is commonly expressed among scholars that rebellion occurs not only when people feel justified to commit violence, but also when their external circumstances allow for rebellion to be feasible (Daly, 473). The objective of using Epstein’s rebellion model is to examine what local behaviors at an individual level drive the emergent pattern of rebellion across a landscape. A complex systems and simulation modeling approach is necessary to achieve this objective because individual decisions about whether or not to rebel is largely influenced by neighbor’s actions, and as rebellion rarely spans an entire geographic region it is useful to model how the shape of rebellion maneuvers over space and time (Daly, 474). In order to understand the behavior of rebellion, it is critical to discern what leads an individual to rebel or remain quiescent. I hope to achieve this objective by running a parameter sweep to conduct a sensitivity analysis, which will determine which parameters the model is most sensitive to, ultimately revealing what is influencing individual’s decisions to rebel. Methods: The purpose of this model is to examine the inner workings of civil violence, and how decentralized rebellion responds to repression attempts of a government authority. Specifically, this model can be utilized to research what factors spark rebellion at an individual level. The entities included in this model are agents and cops. Agents are characterized by their grievances, risk aversion, probability of arrest, and their vision. An agent’s grievance is comprised of two parts: hardships (which represents hardships ranging from physical challenges to socio-economic hardships) and perceived legitimacy of the overall regime. Agents also have a unique risk aversion level, which is significant overall in the sense that agents will rebel depending on whether or not they believe it is worth the risk. Cop characteristics include vision, which allows them to “see” what agents to arrest. Cops are able to move within the spaces visible to them, and arrest only the active agents within their visible range. The global variables in this model are the factor determining the arrest probability of an active agent, and the threshold at which the level of grievance must be great enough to trigger an agent to rebel. When examining the process of the model, it’s import to know what is going on within each actor at an individual level. Based on their randomly assigned levels of grievance and perceived legitimacy of the government, agents will either rebel or remain quiescent. If the agents are actively rebelling, they turn the color red, and if they are quiescent, they are green. As the shade of green per agent becomes darker, the level of grievance associated with that agent becomes higher. Cops, denoted as blue triangles, are able to move about the landscape, and depending on how far their vision is set to, they can spot agents who are either active or quiescent. Cops arrest and imprison any active agents within the scope of their vision. Grey agents are imprisoned agents, and their sentence is determined by the jail term parameter. Jailed agents leave prison with the same level of grievance that they had upon arrest. All actions take place within one discrete time step, which is not representative of any specific length of time. Jail terms are reduced at the very end of each tick. The spatial resolution is 33 by 33. One basic principle taken into account is that agents will determine whether or not they should rebel depending on their grievances and how they perceive government legitimacy. The model works off the notion that if grievances are high, and government legitimacy is low, then there may be rebellion. However, the model also adds the element of each agent’s calculated risk, by taking into account the theory that states that the risk of getting arrested is lower with a large population of active agents and a small number of cops, so agents are more compelled to rebel in this situation (Epstein, 7243). The emergent result from the traits is the spread of rebellion, a pattern that stems from the localized decisions made from individuals to rebel. The agents feel less risk of being imprisoned when there are other active agents around them, creating a positive feedback loop where participation in the rebellion is reinforced. Out of individual acts of dissent emerges a widespread rebellion throughout the landscape, which is measured by the amount of agents switching their quiescent state to an active one. The emergent property is also sensitive to and triggered by a decrease in government legitimacy in the model. The agents were explicitly programmed to decide to rebel or not, and based on their risk aversion values, the agents made decisions about whether or not the possibility of being imprisoned is worth rebelling. The individuals are assumed to sense and consider the possibility of being imprisoned, and use this for further evaluation when making the decision to rebel or not. As the local ratio of cops to agents lessens, agents sense that the probability of getting arrested also lessens, which encourages rebellion. Stochasticity is present in the random values of grievance (composed of hardship and perceived government legitimacy) and risk aversion assigned to each agent. The data collected from conducting a parameter sweep included how many agents were active when cops had vision ranging from 1-10, government legitimacy from 100-0%, a stable cop density at 5%, and a maximum jail term spanning 0-50 time steps. A parameter sweep testing the number of active agents when all of the aforementioned parameters are set was run in order to conduct a sensitivity analysis. I evaluated the model based off of surrounding literature describing the phenomena that may be encountered surrounding the subject of rebellion in respects to the shape it takes based off of decisions made at a local level. Though most of the literature focused on large political factors and their effects on rebellion, this model shows that individual decisions to rebel are also influenced by the actions of neighboring people. Results: The results of the sensitivity analysis indicate that rebellion is largely shaped by the amount of government legitimacy perceived by the agents. Grievance alone is not enough to spark rebellion, but when grievance is high and government legitimacy is low, rebellion will ensue (Fig. 1). However, if agents have a high level of grievance but government legitimacy is also high, the chances of rebellion are low. As one can see in Fig.1, government legitimacy hits zero and a few time steps pass before agents begin to rebel, and when they do, the rebellion sparks quickly. This means that there must also be other factors at play. In order to determine if the maximum jail term was having a significant effect on agents’ decision to rebel, the jail terms were also included in the sensitivity analysis. The results (Fig. 2) show that maximum jail time does not quell rebellion nor does it incite rebellion in any way. Another parameter measured in the sensitivity analysis was the amount of active agents when cops vision was altered to a maximum of 10 (Fig.3 - this figure zooms in towards the end of the model simulation run, where rebellion peaks, in order to give a closer look at the data). As one can see, the amount of cop vision does not have much influence on individual agent decisions to rebel. When examining Figure 4, it is clear that there is a spike in rebellion when agent population increases. This aligns with the theory that if the local ratio of agents to cops is high, agents will be more likely to rebel because there is less of a chance of them being arrested (Epstein, 7243). For example, a person in a crowd of ten people is more likely to rebel if 1 cop was around, than if there were only three potential rebels and 4 cops. The finals results conclude that local decisions to rebel are largely sensitive to the amount of government legitimacy there is, as well as how many agents are around, which diminishes their chance of being arrested, and increases the likelihood of rebelling. As each period of time passes, cops arrest active agents within their vision. Figure 1 Figure 2 Figure 3 Figure 4 Discussion: The results of the model suggest that simple local behaviors, for example one’s individual grievances, perception of government legitimacy, and personal risk aversion, can lead to the larger phenomena of rebellion. Rebellion can be considered a complex system however, because the shape of rebellion is also influenced by and can adapt to external factors, such as the overall public’s perception of government legitimacy, the number of available cops or the power authoritative forces, and the number of people participating in the rebellion. A positive feedback loop where participation in the rebellion encourages others to also join in ultimately reinforces the movement of civil upheaval. Stochastic elements include the individual level of grievance, risk aversion, and personal perception of the government. Though this model is not completely deterministic, rebellion always ensued when government legitimacy was non-existent, or had a value of 0. Conclusion Examining Epstein’s model on rebellion helped me understand that marginalization or collective societal grievances alone do not result in outbursts of rebellion towards a central authority. Instead, people are largely influenced by the actions of others, and by collective attitudes felt towards the central authority. I was surprised to learn that neither top down repressive policies, nor a single influential rebellious person, were able to instigate rebellion. Instead, the entire process was fueled from the bottom-up. Other elements that I hope to see studied in future models of rebellion include the likelihood of rebellion in geographic regions that have a history associated with rebellion versus in a place where rebellion is rare or has never happened. The model revealed that people are more likely to rebel if the risk of doing so were low and if a large amount of people were also rebelling, so I am also curious if the model could include a parameter that could account for people seeing on the media or receiving information from asocial media about how distant neighbors are behaving in a rebellion. Maximum jail time did not deter rebellion in any fashion, so this model could contribute to a larger discussion on the ineffectiveness of specific government control tactics, and how other methods should be explored for conflict resolution. Understanding the nature of rebellion and how it manifests from a local level could both benefit authoritative forces, and the frustrated and fired up masses fighting to have their voices heard. References Boswell, Terry. “Dependency and Rebellion: A Cross-National Analysis.” American Sociological Review, 55.4 (1990): 540-59. Daly, Sarah Zuckerman. “Organizational Legacies of Violence: Conditions Favoring Insurgency Onset in Colombia, 1964-1984.” Journal of Peace Research, 49.3 (2012): 473-91. DeNardo, J. (1985) Power in Numbers: The Political Strategy of Protest and Rebellion (Princeton University Press, Princeton). Epstein, Joshua M. “Modeling Civil Violence: An Agent-Based Computational Approach.” Proceedings of the National Academy of Sciences of the United States of America, 99.3 (2002): 7243-50. Grigoryan, Arman. “Concessions or Coercion? How Governments Respond to Restive Ethnic Minorities.” International Security, 39.4 (2015): 170-207. Gurr, Ted Robert. “Sources of Rebellion in Western Societies: Some Quantitative Evidence.” Annals of the American Academy of Political and Social Science, 391 (1970): 128-44. Humphreys, Marcartan. “Who Fights? The Determinants of Participation in Civil War.” American Journal of Political Science, 52.2 (2008): 436-55. Kuran, Timur. “Sparks and Prairie Fires: A Theory of Unanticipated Political Revolution.” Public Choice, 61.1 (1989): 41-74. Parkinson, Sara Elizabeth. “Organizing Rebellion: Rethinking High-Risk Mobilization and Social Networks in War.” American Political Science Review, 107.3 (2103): 418-32. Olson, M. (1965/1971) The Logic of Collective Action: Public Goods and the Theory of Groups (Harvard University Press, Cambridge, MA). Wilensky, U. (2004). NetLogo Rebellion model. http://ccl.northwestern.edu/netlogo/models/Rebellion. Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL. Wilensky, U. (1999). NetLogo. http://ccl.northwestern.edu/netlogo/. Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL. Terry Boswell, William J. Dixon, “Dependency and Rebellion: A Cross-National Analysis”, American Sociological Review 55, no.4 (August 1990): 540.
© Copyright 2026 Paperzz