Last August, in a Shiite neighborhood in eastern Baghdad, a suicide car bomb went off near a police station across the street from an open air bus station. Ten minutes later, as people crowded in the station to watch the rescue across the street, another suicide bomber drove his car into the station itself. The carnage was widespread but far from over. Twenty minutes later, as the victims of the first two blasts were removed to Kindi hospital only 200 yards away from the terminal, a third suicide car bomb went off at the hospital's side entrance.
Flash forward 13 days to a bridge over the Tigris packed with Shiite pilgrims marching to a shrine for a religious festival. A rumor that there is a suicide bomber in the crowd created a panic that generated a stampede. 965 people die (in the crush or drowned in the river).What's disturbing (at least to me) is how this type of "fear management" could be scaled to radically increase the effects generated by simple and somewhat random terrorist bombings. One area of potential application is the planned disruption of crowds fleeing a city. As we saw in the evacuations from Katrina and Rita, our ability to manage mass evacuations are far from adequate even in a benign environment. In an evacuation due to a dirty bomb or even minor biotoxin release, the outcome could be much worse. In either case, if we had a thinking enemy that has planned a series of events that leverage a rudimentary knowledge of crowd dynamics (as exhibited in the bus terminal incident) to disrupt the evacuation, we would likely have a catastrophe. In this situation (as with most system disruption) the crowd would be used to do damage to itself.
As part of my research on the topic of crowd behavior, I found an excellent paper (Helbing, et. al.) published in Nature magazine on crowd panic. It offers a good framework for understanding crowd panic:
- People want to move faster than normal.
- Individuals start pushing (physical contact).
- Moving through bottlenecks becomes uncoordinated.
- Clogging and arching occur at exits.
- Jams build up.
- The physical interactions at the jams build up pressures (physical).
- Escape is impaired by fallen or injured people which serves to build up the pressure to even greater levels.
- The funneling and jamming will be unlikely to dissipate since people tend towards mass behavior and overlook alternative exits.
- Acceleration: If the velocity of movement and the number of participants is under pressure to accelerate, movement will slow down. Acceleration is influence by prior expectations, rumor, and individual acts of panic. Media management is important here.
- Funnels: The crowd will congregate on a limited number of known exits which will prove to have insufficient capacity. Funneling is influenced by en-route restrictions (security cordons, disruption of transportation networks, etc.), the availability/restriction of exits, and herd instinct.
- Jams: The press of people at the exit creates impediments to movement (an emergent effect that can take many forms). Follow-up attacks on the arc of the crowd at an exit can increase the propagation of impediments and the potential for stampedes.