Home     About     Journalism     Contact


Eureka Magazine

02/08/12
 

The Human Herd

There is a daily event at the Games involving millions of competitors under acute stress — the race to get to the venues. New crowd management techniques are supposed to keep it civil

Keith Still was stuck. He was penned in by a crowd a few yards in front of the entrance to Wembley Stadium and the Freddie Mercury tribute concert was about to start.

He and his friends had been queueing for four hours and were tired and annoyed. But Professor Still — a mathematician and Harley Davidson fanatic — sensed an intriguing puzzle in front of him.

"We were directly in front of Gate C and we were unable to get in," he says. "I was watching the crowd come up a grass embankment on either side and go through the turnstiles. Yet we were not moving. My mates were getting angry but I was fascinated. The physics wasn't what you expected. It showed me that crowds are not like grains of sands. They are faster around the edges."

If the crowd at the concert (a 1992 Aids event) had behaved like the sand in an hourglass, Still would have been in the best place. Grains near the walls are slowed by friction, while grains directly above the hole slip through quickly.

But it wasn't like that and crowd members in front of the gate were not getting through. For the rest of the evening Still kept the puzzle in his mind, and the next day he began to read up about crowd behaviour. It was a turning point in his career. His investigations helped to kick start the discipline of crowd dynamics and made him one of world's leading experts in crowd behaviour.

In the past 20 years academics like Still, of Buckinghamshire New University, have combined physics and psychology to show that many of the assumptions made about the crowds at sports events, demonstrations and pop concerts were flawed.

They have demonstrated that obstacles can speed up crowds, that crowds are usually altruistic and that one of the least helpful things you can do in a crisis is tell large groups of people not to panic.

The findings have influenced the design on new stadia and changed the way police manage public order. And they are now playing a central role in the £1.15 billion operation to keep spectators safe at the Olympics and Paralympics.

By September 9 eight million people will have queued, jostled and trudged through one of the world's most crowded cities on the way to Olympic venues and back again. Many will walk miles from Tube and bus stations, guided by an army of 6,500 volunteer stewards across newly constructed bridges and through freshly designed walkways. It is the biggest public event of its kind in British history and good crowd management (experts hate the phrase crowd control) is the difference between decorum and disaster.

On disasters, recent history is all too eloquent. In April 1989, 96 Liverpool fans died at the Hillsborough Stadium in Sheffield when supporters were herded through a tunnel into an overcrowded pen too small to contain them. Adults and children suffocated as they stood, unable to breathe in the crushing weight of bodies around them. And two years ago 21 music fans died at the Love Parade festival in Duisburg, Germany, when the venue became dangerously overcrowded.

It is partly in response to these disasters that the science of crowd dynamics has flourished. Soon after Still moved into the field he decided that most models being used by his peers were wrong. He dispensed with hourglass analogies and drew inspiration instead from a TV programme he happened to see on anti-chaos — the study of complex systems that show signs of regular patterns of behaviour.

Chaos theory explores how simple actions can lead to disorderly events and is usually expressed by the cliche of a butterfly's wings in South America triggering hurricanes in Europe. But there is also an opposite phenomenon: anti-chaos, the idea that chaotic events may spontaneously crystallise into order. The actions of individuals in a crowd are hard to predict, but the action of thousands of people in a crowd may lead to predictable patterns of movement.

Working on that premise, Still set to work on computer models that predict the behaviour of people in realistic timeframes. His virtual crowd members investigate the space around them and decide if there's any thing they need to avoid such as other people, walls or roads. Each is given a destination to reach and a maximum speed before being set loose in a virtual building, city or sports centre.

The first software package, called Legion, handled crowds of 500,000 with remarkable accuracy.When asked to estimate the time it would take to clear the pitch at Wembley Stadium after a Bon Jovi concert, the software said 14 minutes, 30 seconds. On the night, the actual time was 15 minutes.

For 20 years, Legion, its successor Myriad II and other as yet unnamed "fifth generation" modelling tools have continued to shed light on crowd behaviour. Still has used them to make the annual Hajj pilgrimage to Mecca in Saudi Arabia safer, and to predict crowd behaviour during last year's royal wedding.

Early on he identified a critical point in crowd density where fast moving hoards become shuf-fling, impatient and often grumpy masses: as crowds build up, the number of people passing any point increases until the density reaches four people in one square metre.

That's the point where it becomes difficult to move without bumping into somebody. The flow rate drops dramatically, triggering delays and congestion. To keep people moving, it is vital to keep the density below that magic number by manipulating the built environment to stop bottlenecks and managing the crowd "upstream" to limit numbers entering a tight space.

Still's models have also shown that barriers can speed up crowd flow. Every person in a crowd has a "wake" or a space behind that someone else can move into. If another person overtakes and moves in from the side, the wake is disrupted and the flow slows.

That's where a barrier comes in. If you are walking alongside a handrail in a stadium tunnel or Tube station exit, there are fewer opportunities for people to move in sideways and crowd flow is faster. A carefully designed central barrier in the middle of the opening created by a fire door, for example, can speed up the movement of a crowd by 75 per cent.

Still was not working in isolation. Complementary work by Dirk Helbing at the Dresden University of Technology found that crowds spontaneously split into lanes, allowing people to move more efficiently in both directions — but that if someone tries to go faster and overtake their lane, the whole crowd slows.

Helbing also discovered the oscillating school gate phenomenon. Imagine two streams of parents trying to get through an entrance from opposite sides at the end of the school day, some arriving late to pick up their children, others leaving having picked them up already.

As parents move through the entrance into the school, those on the opposite side must wait. But after a few seconds, the pressure of the incoming parents drops, allowing parents on the other side to break through in a stream. Once that happens, parents entering the school must wait, and pressure builds up on that side. A few seconds later, the movement switches again. A few more seconds, another switch.

Other patterns crop up naturally in crowds, such as the way crowds form arches as they clog up narrow exits, and the way they move more quickly around smooth corners than angular ones. Many of these patterns can be explained by basic physics, but research has also highlighted the importance of group psychology.

People tend to use the exits and routes they are most familiar with, and they tend to herd like cattle and sheep, switching off the decision-making parts of their brains. Families are much slower-moving than average in crowds, going at the speed of the slowest child, and groups of any kind pause to re-form after moving through an exit, which in turns creates another obstacle.

"Models are best to see how systems will break," Still says. "We see how a dumb crowd that is not aware of anything but getting to their destination and the environment around behaves. We keep it simple. The purpose is not to have an intelligent crowd that comes to harm, but to find out how a dumb crowd would come to harm."

In real life, the best way to prevent a crowd becoming restless or coming to harm is through effective communication. That can be through social networking services such as Twitter, oldfashioned loud hailers, LED displays and most importantly of all, stewards. Stewards keep crowds informed about delays, coordinate emergency evacuations and spot potential problems. And to do that they need the trust of the crowd. "Imagine you go to a venue and there's a wall of security guys with berets and the word 'security' printed on their jackets," Still says. "How do you feel? Now imagine they have no hats, but instead pastel-coloured T-shirts with the words 'crowd safety' printed on them. Now you respond very differently."

Locog, the Olympics' organiser, may have had problems recruiting enough security guards, but its army of 70,000 volunteer stewards is arguably more important for keeping crowds happy. Their purple and red outfits were carefully chosen to send a message to spectators. "Some of the research we have done identi-fied that when stewards wear black or dark blue, they are perceived as more of a policing element," says Professor Chris Kemp, also of Buckinghamshire New University. "When they wear tabards or bright fluorescent tops, they are seen as supportive of crowds."

Locog is reluctant to talk about crowd management but planners have been looking at a host of possible crisis situations, from fires and overcrowding to terror threats and stampedes of any kind.

The good news — and it is surprisingly new news — is that crowds rarely panic. "People use the word panic quite a lot, but when you ask them what they mean they say people were frightened and fleeing, but behaving in an orderly way," says Dr John Drury, of Sussex University, who studies crowd reactions to disasters.

His studies into the 2005 London Underground bombings found little evidence of panic. For the 20 minutes before emergency services arrived, people behaved impeccably. Strangers were tying tourniquets, reassuring the injured and sharing water. "In emergencies you get a sense of common fate which helps people redefine themselves in collective terms and that's why co-operative behaviour is so common," he says. "It helps them construct a situation where they share something."

Once again, communication is the key to helping crowds stay calm and engendering a sense of community.

In the early 1990s researchers Guylène Proulx and Jonathan Sime staged two evacuations of an underground station in Newcastle. In one, an alarm sounded and a voice told people to get out. In the second, travelers were told there was a fire and where it was. When the crowd was given detailed information about the emergency they evacuated more quickly. "Some planners advise communication in code to stop panic," Drury says. "But the best predictor of a survivor in a fire is the time taken to recognise the danger. The real issue isn't people over-reacting but people under-reacting." And the worst thing to do in an emergency is to shout, in the style of Corporal Jones, "Don't Panic!"

"Telling people not to panic can be counter productive -it can quite easily be taken by the crowd as evidence that there is something to panic about," Drury adds.

At Leeds University Business School, Rose Challenger has been looking at past disasters, including Hilsborough, to find lessons for events such as Olympics. At Hilsborough the authorities were so focused on the threat from hooliganism that they failed to protect spectators from the risks of overcrowding. The towering perimeter fences that prevented pitch invasions also contributed to the deaths of so many fans.

"We need to make sure that everyone is not just focusing on one issue at the expense of others," Challenger says. Which raises the probably unanswerable questions of which threats may have been exaggerated, and which ignored just because they happen not to have been realised in the past.

"Are there any areas where someone believes it 'could not happen here'? Have we learned lessons from the past? Has training and education in crowd management been sufficient? Is the thinking and action of different agencies co-ordinated? Are roles and responsibilities clear?"

Disasters are rarely caused by single issues. They are usually the result of a confluence of failures — poor training, poor communication, poor technology and a lack of vision.

Many Olympic Park stewards will be familiar to ticket holders by the time this magazine appears.

These are the ones recruited to cover the "last mile" — the walking routes between transport hubs and venues. The Last Milers will tell crowds if delays are expected. But if not, the hundreds of thousands of people streaming back from the Games each night should be oblivious to how they have been manipulated and monitored.

The idea is that the wisdom of crowds and kindness of strangers should prevail over a mindless herd mentality, but if so it won't be by chance.


Copyright David Derbyshire 2011