Geographic patterns of diffusion in the 2011 London riots

Surprisingly little research has examined the localised diffusion of riots within cities. In this paper, we investigate such patterns during the 2011 London riots, and consider how they changed as police numbers increased. Understanding how offences spread in space and time can provide insights regarding the mechanisms of contagion, and of the risk of events spreading between contiguous areas. Using spatial-temporal grids of varying resolution, and a Monte Carlo simulation, we compare observed patterns with those expected assuming the timing and location of events are independent. In particular, we differentiate between four space-time signatures: "flashpoints" of disorder which appear out of nowhere, "containment" whereby already affected areas experience further events, "escalation" whereby rioting continues in affected areas and spreads to those nearby, and "relocation" whereby the disorder moves from one locality to those adjacent. During the first half of the disorder, fewer counts of relocation diffusion were observed than expected, but patterns of containment, escalation, and flashpoints were all more prominent. For the second half of the disorder, when police capacity increased roughly three-fold, observed patterns did not differ from expectation. Our results show support for theories of spatial contagion, and suggest that there was a degree of coordination amongst rioters. They also show that police activity did not just suppress rioting, but dampened the influence of contagion, without displacement.