Die Rolle von Auslöseereignissen in der Erklärung und Vorhersage der Eskalation und Ausbreitung innerstaatlicher Konflikte
Zusammenfassung der Projektergebnisse
The project focused on the analysis of trigger events in the causation of conflict intensity cascades. This research was embedded in the theoretical framework of power-law distributions, where structural risk potentials and initializing incidents interact to generate intensities of vastly different and often unpredictable magnitudes. During the course of the project, two important goals were pursued. First it had to be established whether the cascade approach, which derives from the concept of selforganized criticality, is superior vis-à-vis competing theories favouring stochastic explanations. This was done on the basis of an empirical analysis which involved 168 dyads of intrastate violent conflict and discriminated between frequency distributions conforming and not conforming to a power law. As only models rooted in self-organized criticality can provide a plausible and parsimonious explanations of both behaviours at once, their simultaneous and equal presence in the data is a powerful indicator of the viability of the cascade approach. According to the findings of the self-organized criticality school, even infinitesimally small incidents can trigger enormous avalanches. This proved to be a major challenge for the project, as the level of detail in the sources was mostly insufficient for reembedding initiating conflict events into the historical context from which they emerged. However, since trigger events arise from continuous dynamics as lightning emerges from a thunderstorm, the project was able to take an alternative route, analysing the conflictual background processes. This was done on the basis of the ICEWS dataset, which contains information on a large number of automatically identified and geo-referenced cooperative and hostile interaction events. The analysis involving more than 2 million ICEWS data points showed that a large share of arrests and demonstrations among newsworthy incidents are good predictors of the occurrence of power-law conflict events in the geographical neighbourhood in the following year. Establishing the background from which events in power-law distributed data emerge is an important step for understanding the scenarios in which trigger incidents – i.e. the first in a sequence of power-law events – appear.
