Determining the Dynamics of Genocide By Analyzing Patterns of Social Media Event Reports

Description

The genocide in Gaza has been called the first live-streamed genocide, and has been reported on from Palestine through text, images and videos that document events of genocidal violence, posted on social media by those who are experiencing it. These events (archived at, e.g., genocide.live: Databases for Palestine) can be converted to time series of the frequencies of particular categories of events to reveal patterns of behavior inherent in carrying out the genocide. These dynamical patterns fill a gap in narrating the story of the genocide between geopolitical decision-making and individual tragedies. Analyses of overall time series generated from events in genocide.live indicate that the genocide proceeds through a set of complicated, mostly linear dynamics dependent on a large number of degrees of freedom (PalUROP Final Report: Quantitative Data Analysis of Genocidal Dynamics in Palestine, 2026). However, certain categories of events, such as bombings and home displacements, show evidence of lower-dimensional, nonlinear dynamics with significant forecast skill (manuscript in preparation). Building on this previous PalUROP Project, we propose to investigate low-dimensional dynamics of the genocidal killing machine using a substantially larger data set (archivegenocide.com: Archive Genocide). Steps in the analysis include testing data fidelity, linear maps and correlations in space and time, nonlinear forecasting in dynamical phase space, connecting trajectories in dynamical phase space with mapped events in geographical space and time, inferring genocidal dynamics from patterns emerging from the analysis, and data visualization. Our goal in this project is to articulate examples of genocidal dynamics that connect militarized governance with documented tragedies.