On an evening in January A.D. 532, pandemonium broke out in the Constantinople Hippodrome, a U-shaped chariot racetrack surrounded by stadium stands. Two factions, the Greens and Blues—the predecessors of today’s soccer hooligans—broke into a fight. When the rest of the spectators dashed to escape, many became trapped by the rushing crowd, couldn’t reach the exits, and were trampled and killed. That incident was the start of the Nika riots that almost ended the rule of Eastern Roman emperor Justinian the Great.
Fifteen hundred years later, not much has changed. We still have stampedes, and still can’t get out of enclosed spaces properly. Since 2009, there have been fatal stadium stampedes in Morocco, Cote d’Ivoire, Bangkok, and Egypt. In their study of escape panic, Swiss physicist and sociologist Dirk Helbing concluded together with his colleagues that, “physical interactions in [a] jammed crowd add up and cause dangerous pressures… which can bend steel barriers or push down brick walls.”
Some animals evolved to clump together when threatened because it increased their chances of survival. “Predators have the ability to focus and concentrate on individual prey,” says Ralph Tollrian, a professor in Germany who has spent his career studying the predator confusion effect. “When they handle one prey, they can’t hunt the next.” Birds and fish form groups that move chaotically in the presence of a predator, giving it “cognitive overload,” says Randy Olson, who builds computer models of predator and prey behavior at Michigan State University. The predator’s cognitive overload can be so strong that it may give up on its pursuit entirely. “A confused predator can sometimes become frustrated and not hunt at all,” Tollrian says.
While we may not be able to unlearn our instincts, we might circumvent them if we better understand the nature of escape panic. Since studying panicking humans is difficult, scientists are turning to an unexpected source of inspiration: ants.
“Humans and ants are hugely different animals,” says physicist Ernesto Altshuler at the University of Havana, Cuba, who studied how ants escape in emergency situations. “But when you are in panic, humans behave in a very elementary way, and we may look a little bit like ants.” In a 2005 paper, Altshuler describes an experiment in which he placed black Cuban ants in a circular white petri dish with two exits placed symmetrically on opposite sides of the dish. When he dropped an ant-repelling fluid into the dish’s center, more ants consistently stampeded toward one of the exits than the other, replicating the asymmetrical behavior that people can exhibit.
Here, in plain sight, was stampeding behavior common to people and ants. In fact, the correspondence goes further than that. Nirajan Shiwakoti, who studies crowd dynamics at Monash University in Melbourne, Australia, showed in 2011 that the mathematics describing ant escape are similar to those describing human escape. “Ants naturally form collective traffic and follow physical paths in ways that resemble human crowd movement,” he says.
Shiwakoti recognized an opportunity: He could make an animal model, not for human disease, but for human panic. So he began to experiment with the location and nature of exits that his ants would face, and discovered something surprising. Creatively obstructing the flow of the panicked ant crowd sped its escape.
Shiwakoti experimented with different exit scenarios using square petri dishes that had exits located in the middle of a side and in the corner. Shiwakoti found that an exit located in the middle of the wall with no obstacles in front of it was the least efficient set up. He measured the efficiency of an exit by the amount of time it took the ants to get through it. On average, it took 50 ants 18 seconds to get through an unobstructed mid-side exit. Adding a column in front of the mid-side exit reduced that time to 14 seconds. A corner exit with a column in front had an escape time of less than 11 seconds. But the best escape time was achieved with a corner exit without a column in front—less than 9.5 seconds.
“If you have an exit in the middle of the wall, you can imagine coming from the left side, the right side, and straight,” Shiwakoti says. These different streams of ants, or people, have to merge at the exit and take turns to pass. But people are impatient, and start pushing and shoving. Columns help structure the flow. “The column gives you some channels on the left and on the right, and this reduces the conflict at the exit.” The reason the corner exit is so efficient, Shiwakoti says, is because it has an intrinsic ability to structure the flow. “If the exit is in the corner, then people are probably only coming from left and right, so you have a more uniform flow.”
Shiwakoti and his team are experimenting with placing barriers in front of the Melbourne football stadium exits that lead to the train station. The preliminary results look promising. “Just by having small architectural changes in the layout, or the train stations, or stadiums, you can have massive improvement in terms of evacuation rate,” Shiwakoti says. Perhaps we shouldn’t be surprised at the unexpected lessons we’re learning. Ants have been learning how to deal with congestion for millions of years. They might just show us the way out.
Conor Myhrvold writes about science, technology, and nature. His other works have appeared in Nature, Scientific American, MIT Technology Review, Ars Technica, and Fast Company Labs.
This article was originally published in our “Symmetry” issue in May, 2014.