From scaling to microscopic mechanism of armed conflict

The control of armed conflict presents a major and ever-present challenge to society because of its widespread repercussions like humanitarian, economic, and political crises. Yet, our understanding of armed conflict is limited. One problem is that conflict involves a combination of events that occur over short and long time horizons and geographic scales. This variety in the temporal and spatial scales makes it difficult to understand conflict scientifically and especially to model it mathematically. As a result, there is no comprehensive and transparent approach for connecting the smallest with the largest scales. Another difficulity is that the definition of conflict is not fixed. While we often assume that armed conflicts that we read about in our textbooks such as World War I are defined neatly, this is not the case. Armed conflicts consists of many smaller events ranging from neighborhood violence to militarized engagements, but which specific events can be labeled as conflict or even belong together in a war is often decided qualitatively. These difficulties present an opportunity for explicitly incorporating multiple scales into a mathematical model of conflict. To approach this problem, we will rely on a systematic method that we previously proposed using a standard, comprehensive conflict data set, the Armed Conflict Location & Event Data Project. The highly resolved data set allows us to pinpoint events at kilometer scales and cluster them into conflict clusters spanning nation- states. We will also consider developmental indicators like poverty, governance, and infrastructure. The data will serve as a test for different types of dynamics and noise that we will incorporate into a family of models. By exploring how the output of the models changes as we include increasingly detailed information about conflict events and developmental indicators, we aim to explain how large-scale conflict patterns like wars can be systematically determined and defined from the way that individual actors behave.