This paper describes the process and outcomes of using an agent-based model of Syria to forecast the counterfactual effect of different “treatment conditions” on the likelihood of civilian casualties in Syria between 2011 and 2014. Treatment conditions included unified western support for the Free Syrian Army, US retaliation against Assad after crossing the “red line” in 2013, the development of a democratizing bubble that is allowed to grow in an inclusive way, weakened jihadization of the conflict, and no intervention by Iran.
The overall purpose of this pilot project is to explore the contribution that agent-based modeling can make to understanding, anticipating, and thinking counterfactually about mass atrocity events. Phase I of this project proposed to devise model operationalizations of the concepts and variables central to the current thinking on atrocities, and then experimentally test these operationalizations within our existing country models. In this document we describe that process, along with examinations into whether these operationalizations and country models produce credible patterns and forecasts of atrocity occurrence.
For a detailed description of our ABM input data and modeling strategy, please see our Model Creation Process Documentation.
The key to Lustick Consulting’s Verification & Validation (V&V) process is iterative analysis of model inputs, theory, outputs, and empirical indicators. V&V should not be a static, one-time event, but rather an ongoing process of hypothesizing, testing, and confirmation or disconfirmation. In order to implement this goal, Option I of our V-SAFT project called for six quarterly Verification & Validation reports from August 2012 to December 2013 that included our system status, data inputs, and model forecasts. In addition, we performed a MESA Epistemological Decomposition of our Venezuela model in May 2012 to unpack the data and theory behind the model. Lastly, we held multiple sessions with two Subject Matter Experts–Dr. Allen Hicken and Dr. David Faris–who helped us better understand how our models relate to the real-world. Both experts found our approach compelling with relation to their countries (Egypt, Indonesia, the Philippines, and Thailand). This report describes the empirical portion of our process.
Miguel Garces, Ian Lustick, Thomas McCauley, Patrick O’Mahen
This paper was presented at the 2016 International Studies Association Annual Convention in Atlanta, GA.
What role do different forms of violence play in shaping insurgencies and counterinsurgencies? Traditional political science research has recognized the potential political consequences of violence, particularly when it comes to strategic bombing and nuclear weapons. It has also been recognized that among insurgencies there are “repertoires of violence,” which structure the tactics that can be deployed. Unlike the city-destroying capabilities of major powers, however, there has been little investigation into what political effects, if any, accrue from these small-scale choices of how insurgencies and counterinsurgencies choose to implement acts of violence. This paper will draw on our own previous work studying Improvised Explosive Devices (IEDs) for the Office of Naval Research “Sciences Addressing Asymmetric Explosive Threats” program in order to use agent-based modeling (ABM) to explore tactics as options within a repertoire of violence with possible political consequences. By drawing on a rich vein of government reports and media accounts surrounding the development and use of IEDs by insurgent groups around the world, we identify and describe a set of theoretical parameters that inform both a group’s decision and ability to deploy IEDs or any other specific tactic, including infantry engagement, mechanized forces, bombings, or airstrikes. Our findings suggest that although tactics can sometimes play an interesting role in the political dynamics of insurgency, it is more likely that these effects are outweighed by larger political circumstances.
Miguel Garces, Thomas McCauley, Ian Lustick
This document is meant to be a primer on Lustick Consulting’s model of Syria & Iraq (we call it “Syraq” for short). It is a spatio-temporal network model of both countries, designed to help explain and forecast political instability. This document will not be an introduction to how our model works, what kinds of past validation we’ve done, or the broader goals of the project under which this model has been funded. Instead, it will be about the Syraq model specifically, and how it can and cannot help us make sense of how that complex system operates. Also included is a brief report on general patterns appearing in our output data.
UPDATE: Syraq Validation Report
Ian Lustick, Patrick O’Mahen, Miguel Garces, Thomas McCauley
During military operations in Iraq and Afghanistan, Improvised Explosive Devices (IEDs) became the single most deadly weapon used by insurgents against the U.S. military and allies. In response, the U.S. Department of Defense has spent roughly $75 billion on vehicles and equipment designed to counter IEDs during the length of the Iraq and Afghanistan conflicts (Zoroya, 2013). The focus of the vast majority of this spending to counter IEDs has been on narrow technical approaches to counter the problem, including for example 22,000 Mine-Resistant Ambush-Protected vehicles (MRAPs) that alone cost $36 billion (Carey & Yossef, 2011). This emphasis on detecting IEDs in place or directly protecting soldiers, while understandable, ignores broader questions about the use of IEDs in insurgencies that are useful to both policymakers and field commanders. We use this paper to theorize about and develop methods to answer the question of under what circumstances would an insurgent group deploy IEDs instead of other available weapons. To bridge this gap, we build on Wood’s concept of the “repertoire of violence” (2006) as a foundation for constructing a theoretical model of the changing patterns in the use by insurgents of different weapon types. Next, we mine both scholarly and journalistic descriptions of IED use in numerous insurgencies to outline distinctive features of IED use in conflicts. We then develop a country-level Agent-Based Model (ABM) for studying the interactive and evolutionary dynamics produced by competition within an insurgency among different tactics available to insurgents.
Ian Lustick, Miguel Garces, Thomas McCauley, Patrick O’Mahen
Holland (2000) highlights how the study of board games can yield profound but underexploited insights into complex social phenomena. We build on this insight, treating games as small complex adaptive systems, with emergent properties susceptible to the kinds of analysis appropriate for much larger, highly complex arenas of interaction. Building on this foundation, we use Agent-Based Model (ABM) simulations as “virtual board games” featuring thousands of agents treatable as either players or game elements. Here we do so to study how insurgents use Improvised Explosive Devices (IEDs) and other available modes of attack and evaluate how a player’s counter-strategies can minimize the impact of IEDs. To develop this complex game, we use our Virtual Strategic Analysis and Forecasting Tool (V-SAFT). V-SAFT is a pioneering ABM simulation platform for building theoretically grounded, realistic models that simulate politics and conflict in real-world countries (Reichert, et al., 2014). This paper contributes to theorization of the analytic potential of board games. Also, we advance efforts to bring simulation, gaming, course of action analysis, and training into fruitful contact with one another within the general domain of the study of political violence and counterinsurgency.
Miguel Garces, Ian S. Lustick, Brian P. Levey
This paper was presented at the 2015 Advances in Cross-Cultural Decision Making Conference in Las Vegas, NV.
Social science-based efforts to achieve success forecasting events of interest, including domestic political crises in foreign countries, have advanced considerably in recent years, though controversy exists over claims of success, the validity of methods, and the credibility of codings. One issue that challenges every big data effort is the verifiable operationalization of concepts. That entails devising and deploying understandable and clearly codable proxies, or indicators, that do not distort the effective meaning of the variable being operationalized. An important example of these efforts is the W-ICEWS project funded originally by DARPA and then OSD/ONR. Organized as a team effort by Lockheed-Martin, ATL, the project has sought to improve its performance on a variable known as “Domestic Political Crisis.” Difficulties defining that variable have suggested the possibility of narrowing its focus to “Destabilizing Protest”–the occurrence of mainly non-violent unrest that threatens reigning institutions of authority seriously enough to warrant high-level attention from US policy makers. In this paper we report on efforts to develop this variable by combining publicly available assessments of regime fragility with event data measuring mass protest from the W-ICEWS project. We then test the viability of this new variable by making in-sample and out-of-sample forecasts using a multivariate model of social indicators.
If you are interested in the data used for this paper, please contact Miguel Garces at firstname.lastname@example.org.
Matthew Reichert, Miguel Garces, Ian S. Lustick
This paper was presented at the 2015 Annual Meeting for the International Studies Association in New Orleans, LA.
A significant theoretical innovation in the civil war literature has been the recognition and explicit theorizing of endogenous processes that interact authority, identity, and violence. However, this has also posed a hefty methodological challenge. Developing creative ways to implement or approximate quasi–experimental designs offers a promising solution – but at the cost of sacrificing attention to external validity. In this paper, we propose agent-based modeling as mediator between these two groups of research. Agent-based simulations explicitly model endogenous processes, while providing a ‘sandbox’ in which researchers interested in causal inference can test counterfactual claims. We demonstrate this role by modeling the emergence, disintegration, and co-evolution of insurgent orders in a dynamic agent-based space. We test hypotheses drawn from Kalyvas and Kocher’s “endogenous cleavages thesis,” and use the results to generate scope conditions for potential quasi-experimental designs.
If you are interested in replicating this experiment, please contact the authors for replication resources.
Ian S. Lustick
“Social radar” is a metaphor: in other words, a model. A model projects something known about a “source” onto something less well known—a target. We see the moon (the target) differently when we hear the metaphor: “The moon is a ghostly galleon.” In effect, the poet asks the reader to use “ghostly galleon” as a model for the moon. How we visualize the moon when we use that model results from our individual expectations, emotions, and images associated with the idea of a “ghostly galleon.” Under the influence of the metaphor, i.e., when employing “ghostly galleon” as a model for the moon, we see the moon not as an astronomical object but as a spooky vision, floating silently, mysteriously, perhaps ominously, through the clouded darkness. Taking “social radar” seriously as a metaphor, and therefore as a model, can illuminate many aspects of this vision for forecasting.