When Crowds Try to Flee for Their Lives | Hacker News Reader

When Crowds Try to Flee for Their Lives

atlasobscura.com

76 points740223 comments

One of the best explanations I've read of this phenomenon was on reddit:

https://np.reddit.com/r/worldnews/comments/3pcvfb/saudi_arab...

This gives me a new appreciation for the panic bars (https://en.m.wikipedia.org/wiki/Crash_bar) I've been seeing my entire life. I only recently discovered (thanks to a well-hidden tag on one) that they are fire safety devices, but this article makes it clear how important they are in crowd safety for building-type venues.

Presumably not launching fireworks into the ceiling and limiting crowd sizes are the other steps that would be taken in this particular situation.

It really is sad that people have to relearn fire safety lessons over and over. (Note that none of this applies to outdoor crowds, which is obviously a much harder problem when scaled up)

"At first security guards thought patrons were trying to leave without paying and blocked exits, trapping victims inside." One other thing to improve is to have people pay up front. This would have prevented the guards from closing the exits. If you have a tab apparently sometimes people try to do a mass break out.

Another great thing is to always have people leave through the fire exits to get them familiar with them (I think they do this in the UK).

I'm surprised they didn't mention the fire at The Station club, where 100 people died in the fire because they couldn't get out in time. Mostly because they headed back to the door they came in (a behavior mentioned in the article)

https://en.wikipedia.org/wiki/The_Station_nightclub_fire

Some technical resources:

http://www.vfdb.de/download/TB_13_01_Crowd_densities.pdf

http://www.gkstill.com/Support/crowd-density/CrowdDensity-1....

http://www.gkstill.com/Support/crowd-flow/MovingDensity.html

And one very dramatic video:

https://www.youtube.com/watch?v=eu4DlwE0Bu8

Viewer discretion advised.

Absolutely mandatory reading on the subject:

http://www.newyorker.com/magazine/2011/02/07/crush-point

One thing that is used to prevent bad situations is to place wide round columns before the exits. This will direct the flow around the columns creating a little more time for people behind the column to go thought the exit and remove some pressure of the people in front of the columns.

So is there any action which, if taken by a minority of people, could avert disaster? How big should this minority be?

Nope. Once a crowd reaches a certain density you're pretty much just screwed. It's just physics at that point. You can't even get out (that's part of the problem). The best you can do is to never get yourself into a super-dense crowd situation in the first place.

As far as I can tell, all research on this topic assumes that all people behave in the same way. So it would be interesting to investigate how to change this.

People in a super-dense crowd behave in the same way because they have no choice. A super-dense crowd is essentially an incompressible fluid with people as the constituent particles. If everyone stood perfectly still everyone would be fine, but that is not a stable situation. One person tries to move, but they can't because they are packed in so tightly. There is nowhere to move to. So that person's movements are transmitted directly to their neighbors. Again, if all of those neighbors just ignored the fact that there was someone wriggling next to them and continued standing perfectly still, they might be able to contain the problem, but that's not what happens. Some of the neighbors of the initial mover respond to the movement by trying to move themselves. As long as the average number of people responding to movement by moving by moving themselves is >1, the movement will inexorably amplify and spread.

> People in a super-dense crowd behave in the same way because they have no choice.

Yes, it seems like all research focuses on the situation where it is already too late (perhaps because the fluid equations are easier to handle).

I'd like to see some simulations where:

1. Half of the people refuse to move.

2. Half of the people start moving in the opposite direction.

3. Half of the people stand still and try to hold hands with other people (of that group), effectively blocking other people.

> Half of the people refuse to move.

When someone shoves you, you can't "refuse" to move. Momentum has to be conserved somehow.

> Half of the people start moving in the opposite direction.

Which half?

If you can somehow split the crowd cleanly and get the two halves moving away from each other that might help, but this is usually not possible. And getting a randomly selected half to move in a different direction than the crowd as a whole is simply not possible, again because of physics. A dense crowd is essentially an incompressible fluid. All of the particles in a local area have to be moving in the same direction.

> Half of the people stand still and try to hold hands with other people (of that group), effectively blocking other people.

You really don't have a good mental model of what a dense crowd is like. There's no need to hold hands. You are already pressed up against neighbors on all sides of you. In a dense crowd you are going to remain stationary relative to your neighbors whether you like it or not. That is the root of the problem.

> You really don't have a good mental model of what a dense crowd is like.

And you are clinging to much to your crowd-as-a-fluid abstraction. Abstractions often don't hold up in the real world, especially if you actively try to break them. Further, yes, if you squash people hard enough they behave like a fluid, but what can we do at or before the state-transition?

> what can we do at or before the state-transition?

Get the hell out of there as fast as you can while you still can.

That's what everybody else is doing. The suggestion above to try and work your way to the edge of the crown seems sensible, even if that edge isn't in the direction of the exit. I've been in dense crowds but noticed that there is sometimes a significant gap between the crowd and the walls.

I think the main point is that you "get the hell out of there" where "there" is the crowd itself not whatever the crowd is trying to move away from.

The received wisdom (from Wertheimer) seems to be to try and get to the edge of the crowd:

'After you're pushed forward, like in a wave there's a lull. In that lull is your chance to move, and the way you move is on a diagonal, between pockets of people. There's always space between people. A couple of steps sideways, another wave surge, then another couple of steps in the next lull. You work your way out that way till you get to the [edge].'

source (amongst several): http://www.scmp.com/article/984081/how-survive-when-crowd-st...

Also worth highlighting, as soon as you detect any kind of wave like behaviour start working to get to the edge of the crowd, as that's the big warning sign things are likely to get out of control, since I rather suspect at some point moving in any direction isn't an option.

It would be quite challenging to get a significant portion of the people to stand still or even to run towards the fire.

Challenging indeed, but I wonder what percentage of a population you could get to do it, and whether it would make a significant enough impact to actually increase the chances of the entire crowd getting out alive.

After all, we have firefighters run into fires. And the odd random 'hero'. So it isn't impossible. I wonder, if it could be proven to actually improve your chance of survival, how many people could be incited to move toward the least dense space (toward the fire) assuming they wouldn't be slowing the escape by doing so (not already near an exit), and perhaps even pull back others from pushing at the rear of the crowd. If you could remove enough pressure to allow freer movement of the people near the exits to significantly increase the exit flow, it could perhaps make enough difference to save everyone, including the people who voluntarily moved to the back, who may or may not have made it by simply crushing at the door.

It would probably be highly dependent on conditions and how many people could be expected to take such an 'enlightened self interest' amid a panic. But it would be neat to see it modeled.

Something I haven't heard mentioned is that humans are bipedal, and require a certain amount of movement to stay upright. Push someone beyond the point where their stabilizer muscles can handle, and they have no option but to move a foot to brace themselves (or fall over). This would likely amplify any crowd motion.

> all people behave in the same way

because once the reptilian survival instinct gets triggered, there's no hope of anyone behaving rationally. It's similar to drowning in a body of water. Once you start to panic, your instincts will trigger you to move any and all muscles in attempt to save yourself. In water, behaving this way will cause you to exhaust yourself and sink. In a crowd, it will cause you to shove and compress your neighbors which will trigger their survival instincts. Just like atoms undergoing nuclear fission.

Reminds me of the https://en.wikipedia.org/wiki/Hillsborough_disaster

Reply on news.ycombinator.com