Toward a Unified Understanding of Casualty Distributions in Human Conflict

TL;DR

This paper unifies various findings on casualty distributions in human conflict, explaining differences, dataset effects, and providing a generative model that accounts for power-law behaviors and deviations, advancing a comprehensive understanding.

Contribution

It offers a unified framework that reconciles conflicting results, explains dataset influences, and introduces a generative theory for casualty distribution patterns in conflicts.

Findings

Reconciles different power-law exponents in conflict data.

Shows how dataset compilation affects distribution analysis.

Provides a generative model explaining power-law behaviors and deviations.

Abstract

We are able to unify various disparate claims and results in the literature, that stand in the way of a unified description and understanding of human conflict. First, we provide a reconciliation of the numerically different exponent values for fatality distributions across entire wars and within single wars. Second, we explain how ignoring the details of how conflict datasets are compiled, can generate falsely negative evaluations from power-law distribution fitting. Third, we explain how a generative theory of human conflict is able to provide a quantitative explanation of how most observed casualty distributions follow approximate power-laws and how and why they deviate from it. In particular, it provides a unified mechanistic interpretation of the origin of these power-law deviations in terms of dynamical processes within the conflict. Combined, our findings strengthen the notion that a unified framework can be used to understand and quantitatively describe human conflict.