Fast and Robust Flocking of Protesters on Street Networks

TL;DR

This paper investigates how protesters scattered across a city can rapidly form large, mobile, and robust groups using simple, local decision rules. By modeling the city as a street network and protesters as biased random walkers guided by logit-based rules, the authors systematically explore tactics and identify the alignment rule as the key driver of flocking, with additional rules improving robustness when combined. Through extensive simulations on Paris’ street network, they show that a tactic prioritizing alignment while incorporating small contributions from Follow and Attraction achieves significantly higher gathering and exploration than baselines, and that such groups are resilient to adversarial breakups. The work provides a reusable framework and metrics for evaluating flocking on networks, with implications for understanding collective movement in urban environments and for designing robust, decentralized gathering strategies.

Abstract

We present a simple model of protesters scattered throughout a city who want to gather into large and mobile groups. This model relies on random walkers on a street network that follow tactics built from a set of basic rules. Our goal is to identify the most important rules for fast and robust flocking of walkers. We explore a wide set of tactics and show the central importance of a specific rule based on alignment. Other rules alone perform poorly, but our experiments show that combining alignment with them enhances flocking, and that obtained groups are then remarkably robust.

Figures (8)

• Figure 1: We expect walkers scattered on a street network (left) to quickly gather (center) and then flock (right).
• Figure 2: A piece of the discretized street network around Place de la Nation in Paris.
• Figure 3: Space-time diagrams of three strict tactics on a line using only (from left to right) the Random, Attraction, and Alignment rule. The fraction of walkers is the relative size of groups with respect to the maximum number of walkers in a group in a given run.
• Figure 4: Gathering and coverage scores of all tactics. Each dot corresponds to the average last step value from ten runs of a tactic. The horizontal axis gives the gathering score, the vertical one gives the coverage score. From left to right and from top to bottom: tactics based mostly on the Random, Alignment, Attraction, Propulsion and Follow rule, respectively. On each of these plots, the red dot is for the strict tactic, that exclusively uses the corresponding rule. The bottom-right plot corresponds to tactics with no prevailing rule.
• Figure 5: Plots showing the evolution of gathering, coverage and sprawling scores for the strict tactics and the best tactic. Gathering score is in log-log scale for readability.

Theorems & Definitions (3)

• Definition 3.1
• Definition 3.2
• Definition 3.3