Choosing Sides: The Dynamics of Coalitions

2006年秋天，逊尼派部落转向伊拉克的基地组织，并与美国结盟。这一点--加上美国军队的激增和美国战略的改变--扭转了叛乱的进程。联盟决策和联盟动态往往是决定国内和国家间战争结果的关键因素。然而，过去20年来在战争起因和进行方面所做的理论工作，对联盟几乎没有什么可说的。这项工作有非常重要的见解。但它几乎只专注于两位演员的模特。在这些文献的基础上，这个项目试图将从两个参与者的战争模型中得出的见解扩展到制定和分析联盟。这些问题包括第三方关于是否以及如何干预消耗战的决定；第三方在代理人战争中的角色；分而治之的一般模式，以及在联盟接近目标时往往会分裂的向心力。该项目的一个主要成果将是为估计第三方干预内战的影响的经验工作提供更明确的概念基础。展望该项目的更广泛影响，联合国难民事务高级专员办事处最近报告称，2014年有近6000万流离失所者，其中一半是儿童(《纽约时报》，2015年6月18日)。内战是一个主要原因。更清楚地了解联盟动态可能是朝着减轻这些影响迈出的一小步。联盟决策和联盟动态往往是决定国内和州际战争结果的关键因素。然而，自Fearon的《理性主义战争解释》(1995)以来的20年里，关于战争原因和行为的正式博弈论工作主要集中在两个角色的游戏上，通常是两个角色的讨价还价游戏。这项工作在很大程度上对联盟问题保持沉默。在这些文献的基础上，本项目试图将双行动者战争模型的见解扩展到制定和分析冲突阴影下的模型联盟动态。更具体地说，这个项目的目标是识别和分析联盟形成中广泛的、规范的问题，然后将它们形式化为“主力”模型。初步工作已经分析了第三方在决定是否在消耗战中选边站队时面临的问题。两个参与者的消耗战通常被用来模拟许多不同类型的冲突，包括内战。被认为是全付费、第二价格的拍卖，消耗战已经被推广到允许n个玩家竞争n-k个奖品(Bulow和Klemper 1999)。然而，令人惊讶的是，之前没有人分析过一个非常自然的问题，即当第三方对谁赢得消耗战有偏好时，它会做什么，并可以通过支持一方或另一方来影响结果。本项目将首先推广和推广消耗战选边模型分析的初步结果。然后，该项目将制定其他联盟问题，如分而治之(或统治)，以及在联盟接近目标时往往会分裂的向心力。

In the fall of 2006 Sunni tribes turned on Al Qaeda in Iraq and allied with the United States. This - along with the surge in American troops and a change in American strategy - reversed the course of the insurgency. Alignment decisions and coalition dynamics are often key factors in determining the outcome of civil and interstate wars. Yet theoretical work done over the last two decades on the causes and conduct war of has very little to say about coalitions. This work has made very important insights. But it has almost exclusively focused on two-actor models. Building on that literature, this project attempts to extend the insights derived from two-actor models of war to the formulation and analysis coalitions. These include a third party's decision of whether and how to intervene in a war of attrition; the role of third parties in proxy wars; a general model of divide and conquer, and the centripetal forces that tend to break coalitions apart as they near their goals. A main result of this project will be a clearer conceptual foundation for empirical efforts to estimate the effects of third-party intervention in civil wars. Looking to the broader impacts of the project, the United Nations High Commission for Refugees recently reported that there were nearly 60 million displaced people in 2014, half of them children (New York Times, 18 June 2015). Civil war was a major cause. A clearer understanding of coalition dynamics may be a small step toward mitigating these effects.Alignment decisions and coalition dynamics are often key factors in determining the outcome of civil and interstate wars. Yet the formal, game-theoretic work done on the causes and conduct of war in the twenty years since Fearon's "Rationalist Explanations of War" (1995) has largely focused on two-actor games, often on two-actor bargaining games. This work is largely silent on coalitional issues. Building on that literature, this project attempts to extend the insights derived from two-actor models of war to the formulation and analysis of models coalitional dynamics in the shadow of conflict. More specifically, the goal of this project is to identify and analyze broad, canonical problems in coalition formation and then formalize them as "workhorse" models. Preliminary work has already analyzed the problem facing a third party in deciding whether to take sides in a war of attrition. Two-actor wars of attrition are commonly used to model many different kinds of conflict, including civil wars. Thought of as all-pay, second-price auctions, wars of attrition have been generalized to allow for n players competing for n-k prizes (Bulow and Klemper 1999). Surprisingly, however, no one has previously analyzed the very natural question of what a third party would do when it has preferences over who wins the war of attrition and can influence that outcome by supporting one side or the other. This project will first extend and generalize the preliminary results of the analysis of the model of choosing sides in a war of attrition. The project will then formulate other coalitional problems like divide and conquer (or rule), and the centripetal forces that tend to break coalitions apart as they near their goals.