[Paper Review] From old wars to new wars and global terrorism

Neil F. Johnson, Michael Spagat+5 more|ArXiv.org|Jun 29, 2005

Terrorism, Counterterrorism, and Political Violence3 references16 citations

TL;DR

This paper proposes that modern insurgent conflicts, such as those in Colombia and Iraq, follow a power-law distribution of fatalities with an exponent near 2.5—characteristic of non-G7 global terrorism rather than traditional 'old wars.' Using empirical data and a stochastic model of insurgent warfare, the authors show that despite differing contexts, these conflicts exhibit similar statistical regularities, suggesting a fundamental shift from conventional warfare to asymmetric, terror-driven conflict patterns.

ABSTRACT

Even before 9/11 there were claims that the nature of war had changed fundamentally. The 9/11 attacks created an urgent need to understand contemporary wars and their relationship to older conventional and terrorist wars, both of which exhibit remarkable regularities. The frequency-intensity distribution of fatalities in "old wars", 1816-1980, is a power-law with exponent 1.80. Global terrorist attacks, 1968-present, also follow a power-law with exponent 1.71 for G7 countries and 2.5 for non-G7 countries. Here we analyze two ongoing, high-profile wars on opposite sides of the globe - Colombia and Iraq. Our analysis uses our own unique dataset for killings and injuries in Colombia, plus publicly available data for civilians killed in Iraq. We show strong evidence for power-law behavior within each war. Despite substantial differences in contexts and data coverage, the power-law coefficients for both wars are tending toward 2.5, which is a value characteristic of non-G7 terrorism as opposed to old wars. We propose a plausible yet analytically-solvable model of modern insurgent warfare, which can explain these observations.

Motivation & Objective

• To investigate whether modern insurgent conflicts like those in Colombia and Iraq follow statistical patterns similar to historical 'old wars' or global terrorism.
• To determine if the frequency-intensity distribution of fatalities in ongoing conflicts aligns with power-law behavior and identify the exponent.
• To develop a theoretical model that explains the observed power-law behavior in modern asymmetric warfare.
• To compare the statistical properties of high-intensity conflicts with those of conventional wars and global terrorist attacks.
• To assess whether the observed patterns suggest a fundamental shift in the nature of warfare from state-centric to non-state, terror-driven conflict.

Proposed method

• Analyzes a unique dataset of killings and injuries in Colombia (2002–2004) and publicly available data on civilian fatalities in Iraq (2003–2004).
• Employs power-law fitting to the frequency-intensity distribution of fatalities in both conflicts to estimate the scaling exponent.
• Compares the observed exponents to those of 'old wars' (1816–1980, exponent 1.80) and global terrorism (G7: 1.71, non-G7: 2.5).
• Proposes a stochastic model of insurgent warfare based on a branching process, where attacks propagate through recruitment and retaliation dynamics.
• Derives analytical solutions for the model’s power-law behavior, showing that a critical threshold of recruitment and attack success leads to exponent 2.5.
• Validates the model’s predictions against empirical data from Colombia and Iraq, demonstrating consistency in scaling behavior.

Experimental results

Research questions

• RQ1 Do modern insurgent conflicts such as those in Colombia and Iraq exhibit power-law distributions in fatality frequencies?
• RQ2 Is the scaling exponent of these conflicts more similar to that of 'old wars' or to global terrorism, particularly non-G7 terrorism?
• RQ3 Can a simple stochastic model of insurgent dynamics reproduce the observed power-law behavior with a scaling exponent near 2.5?
• RQ4 What mechanisms in the model explain the emergence of the 2.5 exponent, and how do they relate to real-world conflict dynamics?
• RQ5 To what extent do differences in data coverage and conflict context affect the observed statistical regularities?

Key findings

• The frequency-intensity distribution of fatalities in the Colombian conflict follows a power-law with an exponent approaching 2.5.
• The distribution of civilian fatalities in Iraq also exhibits a power-law with an exponent tending toward 2.5, consistent with non-G7 global terrorism.
• Despite differences in context, data quality, and scale, both conflicts show statistically significant power-law behavior with similar exponents.
• The observed exponent of 2.5 is distinct from the 1.80 exponent of 'old wars' and closer to the 2.5 observed in non-G7 terrorism.
• The proposed stochastic model of insurgent warfare successfully reproduces the 2.5 exponent through a branching process of attacks and recruitment.
• The model's analytical solution confirms that the 2.5 exponent emerges when attack success and recruitment rates reach a critical threshold, aligning with empirical observations.