Dynamic Katz centrality reveals transient vs. stable key players in drug networks

A new arXiv paper from researchers Daniel Catlin, Giulia Berlusconi, and David J.B. Lloyd tackles a persistent problem in criminal network analysis: most centrality measures treat networks as static snapshots, ignoring how roles shift over time. Using data from a two-year judicial investigation into a drug trafficking and distribution network, they applied dynamic Katz centrality to track how actors' influence evolved month by month. To account for the notorious incompleteness of criminal justice data, they introduced a novel robustness test—adding new edges via Bernoulli random trials to simulate missing links—and measured how much node rankings changed under varying levels of uncertainty.

The results show a clear split between actors who maintained central positions throughout the entire investigation and those who were crucial only for brief windows. Dynamic Katz centrality proved more effective than traditional static measures at differentiating individual contributions, even among highly central nodes, and remained useful when data were incomplete. For organized crime researchers, this offers a more nuanced view of criminal collaboration; for law enforcement, it suggests a way to identify consistent power brokers versus dispensable operatives, potentially sharpening disruption strategies. The paper is available on arXiv (2509.08028) and represents a practical analytical upgrade for temporal network analysis in high-stakes settings.