Modeling Urban Panic

Paul Torrens, at the Arizona State University School of Geographical Sciences, has a computer simulation that models urban panic:

"The goal of this project is to develop a reusable and behaviorally founded computer model of pedestrian movement and crowd behavior amid dense urban environments, to serve as a test-bed for experimentation," says Torrens. "The idea is to use the model to test hypotheses, real-world plans and strategies that are not very easy, or are impossible to test in practice."

Such as the following: 1) simulate how a crowd flees from a burning car toward a single evacuation point; 2) test out how a pathogen might be transmitted through a mobile pedestrian over a short period of time; 3) see how the existing urban grid facilitate or does not facilitate mass evacuation prior to a hurricane landfall or in the event of dirty bomb detonation; 4) design a mall which can compel customers to shop to the point of bankruptcy, to walk obliviously for miles and miles and miles, endlessly to the point of physical exhaustion and even death; 5) identify, if possible, the tell-tale signs of a peaceful crowd about to metamorphosize into a hellish mob; 6) determine how various urban typologies, such as plazas, parks, major arterial streets and banlieues, can be reconfigured in situ into a neutralizing force when crowds do become riotous; and 7) conversely, figure out how one could, through spatial manipulation, inflame a crowd, even a very small one, to set in motion a series of events that culminates into a full scale Revolution or just your average everyday Southeast Asian coup d'état -- regime change through landscape architecture.

Posted on January 14, 2008 at 12:09 PM • 40 Comments

Comments

vbnmJanuary 14, 2008 12:43 PM

news media footage of riots won't be detailed enough. he can put all that urban surveillance camera video footage to use estimating his model parameters. Another reason more cameras is a good thing.

Grand DelusionsJanuary 14, 2008 1:11 PM

"...conversely, figure out how one could, through spatial manipulation, inflame a crowd, even a very small one, to set in motion a series of events that culminates into a full scale Revolution or just your average everyday Southeast Asian coup d'état -- "

I think executing a 'full scale Revolution' might be a tad more complex than rearranging the neutral space in the town square.

Isn't this more easily done in Second Life?

Bryan FeirJanuary 14, 2008 1:20 PM

The last couple of entries sound like someone's been thinking about the 'psychetecture' idea from Dean Motter's 1980's comic book 'Mister X'...

Carlo GrazianiJanuary 14, 2008 1:23 PM

I like the "Terrorist Mall Designer" idea. I think he may already have struck near Minneapolis, though.

CharlyJanuary 14, 2008 1:24 PM

Sometimes, when watching some "physicohol" fights in night-clubs and the way they extend all over the room, I often tell myself that we should simulate it one day or another...

It will surely be possible with this project... ;-)

Cheers

SnarkJanuary 14, 2008 1:34 PM

4) design a mall which can compel customers to shop to the point of bankruptcy, to walk obliviously for miles and miles and miles, endlessly to the point of physical exhaustion and even death.

This is satire, right?

Rennie deGraafJanuary 14, 2008 1:49 PM

"...figure out how one could, through spatial manipulation, inflame a crowd, even a very small one, to set in motion a series of events that culminates into a full scale Revolution or just your average everyday Southeast Asian coup d'état -- regime change through landscape architecture."

It sounds like this guy takes Feng Shui to a whole new level.

WalterJanuary 14, 2008 1:57 PM

I always wonder how they test these sorts of algorithms. You can see if they sort of look realistic, but how do you know if they really are?

HarryJanuary 14, 2008 2:47 PM

@Snark - only the "and even death" part. For the most part malls are designed for minimum efficiency so the shopper passes as many storefronts as possible.

AlanJanuary 14, 2008 3:07 PM

#1 - Not everyone flees from a burning car. Some stay to watch.

#2 - This experiment is being done in transit malls around the country currently.

#3 - All of the Fox news views stay where they are sue to not believing the Librul Media.

#4 - The original idea for this was casinos. I expect that the ideas will be used for new malls. Old ones are too hard to retrofit.

#5 - Christmas.

#6 - Lots and lots of marbles. (Works well when seeing "Riverdance" as well.)

#7 - In this country? Not a whole lot. What are they going to do? Get up?


Timm MurrayJanuary 14, 2008 3:13 PM

Now Wil Wright just needs to come around and package it up into a game. SimRiot, just in time for Christmas 2009!

Matthew X. EconomouJanuary 14, 2008 3:47 PM

This reminds me of Dan Lockton's "Architectures of Control" blog (http://architectures.danlockton.co.uk/). A few months ago, he posted something about casino design and decoration that discussed how it was designed to keep one gambling.

wildcatherderJanuary 14, 2008 4:06 PM

Point 4 is dead giveaway that this is a tongue-in-cheek article. You guys are way too gullible.

renoXJanuary 14, 2008 4:22 PM

I remember an article about this kind of crowd simulation used for Muslim pelerinage: the crowd they have is huge there!
Of course they cannot rebuild the area but with some modest modification, they achieved very good result.

ared38January 14, 2008 5:16 PM

People are far too calm in this simulation. Those on the edges wouldn't wait patiently for space to be avalible. They would push and shove.

ForRealJanuary 14, 2008 5:29 PM

Computer models prove whatever you set out to prove with them. They are not otherwise worth comment.

CitadelJanuary 14, 2008 6:12 PM

@For Real: "Computer models prove whatever you set out to prove with them. They are not otherwise worth comment."

Can you prove that with a compuer model?

AlanJanuary 14, 2008 6:34 PM

I wonder if his "urban panic simulation" was designed to model the current housing market.

iglooJanuary 14, 2008 7:00 PM

It's already been used in a movie plot: V For Vendetta has some similar mass manipulation scenarios.

Niyaz PKJanuary 14, 2008 7:40 PM

That is really interesting. I wonder how much hardware resources will be needed for a full fledged simulation of a sufficiently large crowd.
And how accurate are there models then? How could we test for accuracy? May be we could try to simulate a real life incident in the part and see what the results turn out to be.
Mob can be really unpredictable. They do random things at random times. Even if it was perfectly random, we could have modelled it. But it is not. :)

PaeniteoJanuary 15, 2008 2:39 AM

You forgot a scenario: Simulate "normal" behaviour and arrest everyone else...

"Citizen, according to our simulation, you should be 3 blocks away from here. Explain yourself."

supersnailJanuary 15, 2008 3:04 AM

I think this model fails to take into account cultural, experience and generational differences.

cultural:-
Compare and contrast
"three mile island":-
nuclear plant trashed but safety systems worked well enough to contain damage and very little radation was leaked result blocked highways, panic, panic in Congrees as entire civil Nuclear programme was halted.
Chernobyl:-
Worst civil nuclear disaster ever. People styed at home. Only moved when told to. No mass panic of any kind. Firefighters went into plant to render it safe even though they were told beforehand they would receive fatal doses of radation.

Experience:-
People with military training react differently to civilians in dangerous situations. Civilians follow the crowd, trained personel avoid beiong caught up in panicing crowd.

Generational:-
Cautious middle aged people try to get out of there as quickly as possible teenage males move closer to get a better look.

RogerJanuary 15, 2008 3:30 AM

As interesting as this model is, it is not quite as novel as one might think at first; spatial crowd dynamics has been a formal study for long enough -- at least a decade and a half, probably longer -- that there are already established commercial software packages for doing this (in fact, entire companies that make their livings do nothing but crowd simulation), and even professorial chairs in the field.

Current state of the art isn't even in the ballpark of the degree of sophistication that could foment a revolution. In fact, the ones I have seen don't simulate emotions beyond calm/panicked, nevermind analysing the effects of culture, community and so on.

geekWithA.45January 15, 2008 8:25 AM

While you can deconstruct certain non deterministic phenomena such as mobs, revolutions, and compulsive shopping, the result is rarely a recipe that you can actually use to construct the phenomena at will.

At the very best, it'll yield something that a Go player would call "influence".

Alan PorterJanuary 15, 2008 8:59 AM

I've personally seen #4 in Hong Kong at the shopping center on Victoria Peak (www.thepeak.com.hk). I had such a hard time finding an exit, I seriously wondered how many people would be trapped if there were a fire in the building. But then again, I bet you could get some real bargains as the place went up in flames.

Alan

AndrewJanuary 15, 2008 9:29 AM

Every time I've been in a crowd and Bad Stuff has happened, I've sent my untrained friends to the nearest safe telephone and taken my trained friends to see if we can help. Usually we have.

I wonder how the crowd model accounts for both dispatched and ad hoc responders?

Pull a gun in California, and people will scream and run away from you.

Pull a gun in Utah, and people will pull their own guns and run towards you.

Pull a gun in Florida, your mileage will vary. Except at Disney World.

J.D. AbolinsJanuary 15, 2008 10:50 AM

Andrew's comment about the possibility of different responses to a "bad event" depending upon location, culture, and peoples' training is an interesting one.

The video and photos of the simulation I've seen briefly appear to assume walking or running humans. In many real life situations, there may be an assortment of complicators. I am thinking of people in wheelchairs, people walking with assistance of Zimmer frames/walkers, blind people using mobility canes or dog guides, etc.

Also, I wonder about the prospect of people falling down, getting trampled, and affecting the flow of the crowds.

I realise that the program is a simulation and cannot cover every possibility. At the same time, it is prudent to remember it is a simulation and, while possibly helpful, is not a blueprint for how real people will move in a real life situation.

Matthew SkalaJanuary 15, 2008 2:22 PM

The list of potential uses, including making people shop to their deaths, was added by the "Pruned" Web log author; it's not part of the goals quoted from the project's own Web site.

jayhJanuary 16, 2008 9:33 AM

There is so much that I cannot imagine being reliably modelled, I really doubt the usefulness of this except as a very broad concept model. Since small changes in these uncontrollable details would drastically affect how things were played out, and systems cannot simultaneously be designed for all, and there would not be enough info to real-time try to model an existing situation, I'm not sure how much could be learnd


*Temperature/humidity (current and the last few hours)
*Music on the radio
*recent events (police/community conflicts, major sports events etc)
*alcohol and other intoxicants
*momentary mistakes/misjudgements of authority officials.
*opportunistic actions of hostile individual actors (smashing a store window, attacking a cop)
*weapons available
*Whether crowd consists of identifiable groups (racial etc) that could polarize into conflict or unite against authority elements
*gender and age makeup of the crowd
*Visibility of news media

realistJanuary 18, 2008 8:38 AM

Based on historical fact, it would be more constructive to model of how officials panic so citizens could prepare and take appropriate action.

realistJanuary 18, 2008 8:38 AM

Based on historical fact, it would be more constructive to model of how officials panic so citizens could prepare and take appropriate action.

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc..