Lee, Edward D.; Bryan C. Daniels; Christopher R. Myers; David C. Krakauer and Jessica C. Flack
Large-scale armed conflict is a characteristic feature of modern societies. The statistics of conflict show remarkable regularities like power law distributions of fatalities and duration, but lack a unifying framework. We explore a large, detailed data set of 105 armed conflict reports spanning 20 years across nearly 104 kilometers. By systematically clustering spatiotemporally proximate events into conflict avalanches, we show that the number of conflict reports, fatalities, duration, and geographic extent satisfy consistent power law scaling relations. The temporal evolution of conflicts measured by these scaling variables display emergent symmetry, collapsing onto a universal dynamical profile over a range of scales. The measured exponents and dynamical profiles describe a system distinct from prevailing explanations of conflict growth such as forest fire models. Our findings suggest that armed conflicts are dominated by a low-dimensional process that scales with physical dimensions in a surprisingly unified and predictable way.