Sharing three square feet with six smelly strangers makes for an uncomfortable situation, but what about when six become ten? The distance between "group of happy party people" and "panicked mob" is almost equally short. Take what happened at that rave in East Croydon, London, as an example. The good news is that scientists are starting to study crowd management. After all, suffocating during your favorite band's set is a pretty shitty way to die.
The art behind managing crowded areas on festival grounds involves the carving of pedestrian routes, the placement of tents, the layout of camp sites, and the concert's overall programming. So far, these have relied on the experience and gut instincts of the festival organizers. But there is an increasing interest in figuring out whether digital models could provide better solutions.
According to Dorine Duiven, a PhD candidate at the Delft University of Technology, in the Netherlands, large crowds are predictable and can therefore be modeled by software. Duiven's team researched the possibility of using drones to observe pedestrian traffic and develop a system that would allow organizers to intervene when things look like they're going sour—all by using the Dutch music festival Lowlands as a case study.
“I research dynamic pedestrian traffic during large events," she told me. "I love festivals, and pedestrian systems are some of the hardest to make sense of. They consist of individuals moving in two-dimensional spaces with large behavioral components. That really fascinated me.” Her scientific field of choice is relatively young. While automotive traffic has been studied for more than 60 years, pedestrian-traffic research has only been around for 30.
Duiven says the reason behind this lack of interest is twofold: First, crowds in public spaces have only recently become large enough to cause trouble. Second, people seem to be developing an increasing preference for large-scale events—a trend that is emphasized by the exploding number of outdoor festivals and the growing world population.
So what Duiven and her colleagues are trying to do is capture the behavior of people in large groups and figure out how they move, how many people can be crammed into a space without losing their marbles, and what the best way would be to deal with them if that happens. To be able to do this, they used something called the Social Force model, which is a computer-simulated model of the movement of pedestrians, developed in 1995. Researchers all over the world use Social Force, but that doesn’t mean the model is perfect: “Sometimes weird stuff happens and the model destabilizes," Duiven says. "Then it starts predicting things that couldn’t happen in reality, such as a huge increase in crowd pressure or people moving at physically impossible speeds.”
When you think about it, that's not too strange. The model works by reducing individuals—agents—to a couple of basic characteristics. External factors are taken into account—like the width of walking routes—but the individual agents also have preferences. This means that each of them needs to be given a kind of personality—things like the concerts they attend or the speed with which they walk need to be established beforehand.
But there are certain characteristics that are harder to predict, like the assumption that women on average walk slower than men. Cultural influences have to be taken into account too: Most countries have a "right-hand bias," meaning that people have a tendency to avoid oncoming traffic by stepping to the right. But in the UK and Australia, it is the other way around.
These seem like small adjustments, Duiven notes, but when you look at the behavior of groups consisting of thousands of people, they add up. The small peculiarities involving human behavior and movement become extremely important. This could mean that the model should be able to anticipate the collective dodging movement when approached by some annoying dude wearing a "Free Hugs" sign.
Fortunately, anticipating these occurrences is getting easier and easier for scientists. Tallying next to a pedestrian route will soon be history; researchers can study the festival visitors in their natural habitat by analyzing camera images or following the signal and location of mobile phones.
Finally, Duiven says that festival organizers hardly make use of these models at the moment, which is quite a pity. “When you can accurately predict the bottlenecks in the system and consequently remove them, you should be able to provide a solution for a problem before the problem occurs."
Aside from the tragedy that occurs when people hand out their self-esteem one free hug at a time, bigger calamities can be modeled with the Social Force model. By simulating problematic situations, one could provide a better layout for emergency exits, fences, and escape routes, keeping the risk to a minimum and the entertainment to the max.
But as these models become more and more advanced, so will their usage become more widespread to prevent tragedies such as that of the 2010 Love Parade in Berlin—where hundreds of people got injured and 21 people suffocated to death. Duiven suspects that not all festival organizers are aware of the possibilities yet, but hopefully her work will take us one step closer to changing that.
More on the ways people have fun:
WATCH – Raving in the Black Sea