The Role of Trigger Events in Explaining and Predicting the Escalation and Diffusion of Intrastate Conflicts

升级动态和空间扩散是冲突的普遍方面：随着时间的推移，所有冲突的强度和广度都在演变。然而，事实上，升级和扩散是两维空间中更一般的传播过程的例子：这一空间要么代表地理表面，要么代表涉及暴力手段及其后果的冲突强度的二维衡量标准。冲突的历史相当于它在这样一个双层、二维的状态空间中的轨迹。对升级和扩散的研究发现，结构性基线风险和引发触发事件的相互作用是这类过程的核心。在复杂网络中，无限小的触发器实现了现有的势，因此可以引起级联或几乎任何大小的雪崩链式反应通过给定的网络传播。越来越多的证据表明，国家内部暴力冲突的强度和范围遵循“幂定律”：虽然事件的规模和发生的频率之间存在高度规则的关系，但事件在一段时间内的分布高度不规则，因此几乎不可能进行可靠的预测。虽然对结构性风险的评估属于冲突研究的标准范畴，但仍然缺乏对触发事件的系统分析，以及导致这些事件的过程和由此产生的轨迹。该项目旨在加深我们对如何解释国家内部冲突的强度和范围的波动和经常出乎意料的变化的理解。通过将重点放在触发事件上，该项目推动了国家内暴力冲突扩散和升级的理论统一和经验主义证实的模式的发展。该项目分五个步骤进行。第一步是筛选触发事件，调查这些事件发生的地点和时间。第二步使用定性新闻分析，以重建冲突行为者的通信和活动以及冲突-外部中断，从而导致特定的启动事件。第三步是对经验化的启动事件进行系统的比较，得出触发事件的类型学。第四步着眼于在启动事件发生后发生了哪些后续事件，调查我们是否可以发现某种类型的触发因素和冲突的进一步发展之间的规律性。在第五步也是最后一步中，对发现进行二进制分类器评估，以评估触发事件类型的预测性能。

Escalation dynamics and spatial diffusion are universal aspects of conflicts: Over time, all conflicts evolve in their intensity and extension. In fact, however, escalation and diffusion are instances of a more general spreading process in a two-dimensional space: A space representing either a geographical surface or a two-dimensional measure of conflict intensity involving the means of violence and its consequences. A conflict’s history is equivalent to its trajectory through such a two-layered, two-dimensional state space. Research into both escalation and diffusion has found that an interaction of structural baseline risk and initiating trigger events is at the heart of such processes. In complex networks, infinitesimal triggers actualize an existing potential and can thus cause cascades or avalanching chain reactions of virtually any size propagating through a given network. Increasing evidence shows that the intensity and extension of intrastate violent conflicts follow a “power law”: While there is a highly regular relationship between the size of an event and the frequency with which it occurs, the events are distributed highly irregularly over time, making reliable predictions almost impossible. Whereas the assessment of structural risk belongs to the standard repertoire of conflict research, a systematic analysis of trigger events, with the processes leading up to them and the trajectories which emerge from them, is still lacking. The project aims to add to our understanding of what explains the fluctuating and often unexpected variation in the intensity and extension of intrastate conflicts. By zeroing in on trigger events, the project advances the development of a theoretically unified and empirically substantiated model of the diffusion and escalation of intrastate violent conflicts. The project proceeds in five steps. The first step consists in a screening of trigger events, investigating where and when such events took place. The second step uses qualitative news analysis in order to reconstruct the conflict actors’ communications and activities as well as conflict-external discontinuities, leading to a specific initiating event. The third step consists in a systematic comparison of the empirically contextualized initiating events, resulting in a typology of trigger events. The fourth step looks at which consequent incidents ensued after the initiating event occurred, investigating whether we can detect regularities between triggers of a certain type and the further progression of the conflict. In the fifth and final step, the findings are subjected to a binary classifier evaluation in order to assess the predictive performance of the trigger event types.