An Insight a Day

How to Control a Crowd

Here's a rather interesting video on how to control a crowd. It's longer than 20 minutes, but I highly recommend watching it.

There are generally two types of crowds: A "competitive" crowd and a "cooperative" crowd.

A competitive crowd is one where everyone is rushing for something, like a limited supply of 200 game consoles on release day or fighting to get good seats at a concert. These crowds usually have a higher risk of leading to a crowd crush.

A crowd crush happens because the people at the back want to push forward, but there's not much space for those at the front to move forward. As a result, the people near the front or center start getting crushed.

To make matters worse, once the crowd density reaches a high enough level, there's nothing anyone can do anymore. At some point, the crowd behaves more like fluid dynamics than anything else.

Fortunately, most crowds don't end up in a crowd crush, and this is because they're more "cooperative" than "competitive." Imagine that a fire broke out. You would think this would lead to chaos and panic, but in most cases, people behave cooperatively.

Yes, everyone is "competing" to survive, but it's not like it's limited to only 200 people, so there's more incentive to help others. Also, identity plays a crucial role in how well a crowd cooperates. The greater the shared identity, the greater the cooperation.

Consider a common anecdote—on a normal flight, people barely talk. They’ll sit shoulder to shoulder with their fellow traveler for four or five hours without even acknowledging their presence once.

But as soon as a flight gets delayed, that changes. People start talking, they start problem-solving, they get friendly, they’ll even share food.

If the flight gets canceled, this cooperation often extends further and those that were complete strangers just minutes before might decide to work together and share a rental car to drive to their destination instead.

Crowd psychologists have noticed the pervasiveness of this anecdote, and they think they have an explanation. It’s all about identity.

When ambling around the airport, there is little shared identity beyond traveler which is weak since it is shared by so many.

But when a flight gets delayed, the identity shifts and concentrates: now it’s distressed traveler—a unique identity shared only by the unlucky few.

What researchers have found is that, the greater the degree of shared identity, the greater the degree of cooperation among a crowd.


In the 2005 London underground bombings, researchers interviewing victims observed a through-line of perceived unity among the victims in the moment, and countless examples of selfless, cooperative behavior even at the risk of personal peril.

This is observed in almost any disaster—there is always a high degree of crowd cooperation, far beyond what might often be portrayed in fiction, that typically leads to rather orderly, crush-free evacuations.


But another curious anomaly in crowd crush statistics is that it is quite rare for them to occur during civil unrest—during the very activity that defines mob mentality.


In riots, though, there is little competitiveness—there’s not a particular something to be gained or lost, and therefore there is little incentive to push and shove. Not only that, but there is typically a strong shared identity—before tensions escalated, a group of people gathered together in support of a particular cause.

Therefore, riots feature two of the strongest predictors of lowered crowd crush potential, explaining how some of the least organized events manage to avoid the worst.