lostpunk5545
Bluelighter
- Joined
- May 20, 2003
- Messages
- 10,324
World of Warcraft plague and real world implications.
This is one of the coolest articles I've read in a long time. It's about the possibility of using a virtual plague in WoW to model the way people respond to the same situations in real life.
But first some background info. for those that are unfamiliar with the Corrupted Blood plague from WoW (as was I until I read the article):
http://news.bbc.co.uk/1/hi/technology/4272418.stm
http://en.wikipedia.org/wiki/Corrupted_Blood
http://blog.wired.com/gadgets/2005/09/warcraft_plague.html?entry_id=1230071
This is one of the coolest articles I've read in a long time. It's about the possibility of using a virtual plague in WoW to model the way people respond to the same situations in real life.
But first some background info. for those that are unfamiliar with the Corrupted Blood plague from WoW (as was I until I read the article):
http://news.bbc.co.uk/1/hi/technology/4272418.stm
http://en.wikipedia.org/wiki/Corrupted_Blood
http://blog.wired.com/gadgets/2005/09/warcraft_plague.html?entry_id=1230071
Virtual outbreaks, real world ramifications
* 24 February 2007
* From New Scientist Print Edition. Subscribe and get 4 free issues.
* Kim Coppola
UNTIL that fateful day, 20 September 2005, the disease was unknown. But once the outbreak began, its effects were devastating. For many, the infection was lethal. Dying victims oozed contaminated blood through their skin and anyone coming into contact with it quickly succumbed.
It all began when a group of adventurers encountered a deadly virus while exploring a labyrinth of caves. When surviving infected explorers made it back to civilisation, they initiated the outbreak. Travellers spread the word, and the disease, only boosting the carnage and the outbreak to epidemic scale. In the busiest towns, bodies piled high.
This is a true story. But there's a good reason you may not have heard about it. The disease took place inside a computer game called World of Warcraft, a virtual world where upwards of 7.5 million people log on to live fantasy lives where they kill monsters and collect treasure.
The disease, christened the "Blood Plague", was made by Blizzard, the game's developers. It was supposed to be just a bit of a fun - and to remain within the caves. When it spread, catching Blizzard programmers by surprise, it inspired scientists to look at virtual worlds as a new place to conduct serious research on people's behaviour. In the future, subjects as diverse as government decisions on how to contain a disease and theories of social behaviour could be based on knowledge gained through experiments in the virtual world (see "Online obedience").
Of course, computers have long been used to model the real-world spread of disease. For example, at the Discrete Simulation Sciences office of the Los Alamos National Laboratory in New Mexico, researchers use computer simulations based on statistical models to examine how diseases such as smallpox might spread in the US, and how they might be affected by various vaccination and quarantine strategies. For their Episims project, which is designed to model outbreaks in Portland, Oregon, researchers gathered census and transport records, and used the information to simulate the behaviour of every Portland citizen at 10-minute intervals, including where they travel and who they interact with in a normal day. They were then able to simulate how a disease might spread through this population.
The problem is the model only captures people's routine behaviour. Epidemics can lead to all sorts of out-of-the-ordinary activities, and panic can cause people to behave in completely different ways. The Episims model does not extensively modify behaviour once the disease begins to spread. "The model is very good at examining where disease will go once we know how people will behave, but if that [behaviour] changes, the model falls down," says Nina Fefferman, who studies disease control at Tufts University School of Medicine in Boston. In the Episims model, if an individual is sick enough, they stay home; if they have a sick child, one parent stays home. Otherwise, business continues as usual. Their model does not take into account the complex and unpredictable human responses to a disease outbreak, such as deliberately avoiding places where there are infected people. This is a crucial drawback, says Fefferman. Human behaviour is a fundamental component of how a disease spreads. Ignore its complexity and the results of your simulation could be wildly out.
"We know from the statistical modelling that having clumps of similar behaviours can greatly affect the outcome. So if you know 30 per cent of the people are going to run away, and 30 per cent of them are highly susceptible to the disease, it makes a huge difference to the outcome of the disease whether or not that's the same 30 per cent," she says.
Out of control
That's why the World of Warcraft plague got Fefferman so excited. Here was a chance to gather accurate information about how each of thousands of individuals might behave in an outbreak, a study that would be virtually impossible in the real world. As soon as she heard about it she phoned Blizzard. "I wanted to know exactly what happened," she says.
Press reports at the time described how the disease spread out of Blizzard's control. They had intended for the disease to remain within the caves. When the plague spread to the towns, they didn't know how to stop it. Even quarantine didn't work. In the end, to stem the epidemic, Blizzard had to take many areas of the game off-line. Fefferman reasoned that people would have behaved in a similar way to how they would during a real-world outbreak. She wanted to see if the outbreak held lessons for real-world disease control.
Key to Fefferman's idea is that in virtual worlds each character is controlled by a real person, rather than being just a statistic in a simulation, which means you don't need to try to predict their behaviour; you can observe it. "The real advantage of virtual-world modelling over statistical modelling is that for statistical modelling to work at all, you need to know how people will behave in advance," Fefferman says. In the virtual world you can look at the behaviour in response to the disease. For example, infected characters might flock to healing areas just as real-world victims would head for hospitals.
Unfortunately, though, Fefferman was not about to get anything useful from the Blood Plague outbreak. Blizzard had been so busy trying to stem the spread of the disease, they hadn't thought to keep track of the statistics. So Fefferman will have to look elsewhere. First she needs to find a cooperative company willing to modify its game's code so that characters can be infected involuntarily. Fefferman would then seed a disease and wait to see what happened. She has spent 18 months devising a way to use virtual worlds to model the spread of disease. The games company would need to provide her with detailed information on the outbreak, including every character's movements, who they met, when and where they were infected and so on. She would also like to be able to interview players after the fact to establish why they behaved the way they did. Best of all, Fefferman says, would be to be able to secretly introduce diseases and factors affecting its spread that are beyond the control or knowledge of the players. For example, some characters might be silent carriers - infected but symptomless - while others might be immune.
Virtual-world modelling would not spell the end for statistical models, Fefferman says. They would work in tandem, solving different kinds of questions. Behavioural information from virtual-world modelling could be fed into powerful statistical modelling software to refine our assumptions about how diseases really spread.
So far, though, games companies have given Fefferman's plan the cold shoulder. For example, Linden Lab, which created the virtual world Second Life, turned down a recent request from Fefferman to model epidemics in the game. They didn't want to do things to their virtual residents without their consent. "They were very nice in suggesting that I could run a 'voluntary participation experiment' within Second Life, but that would defeat my purpose in the study," says Fefferman. Voluntary participation would create self-selection bias - only those who consented to it could catch the disease.
It's not all bad news. According to numerous blog and forum postings, World of Warcraft players enjoyed the Blood Plague outbreak and considered it part of the spectacle of the game. Blizzard has told Fefferman it may consider a planned outbreak at some point in the future. "They have promised that if they do incorporate something major like that in a controlled way, they'll call me," says Fefferman.
So does the virtual world hold the key to the future of epidemiology? Dimitri Williams, who studies the social and economic impacts of virtual worlds at the University of Illinois at Urbana-Champaign, thinks not. "You can't model behaviour based on games where the human behaviours don't map to the real world," he says. In virtual worlds there is nothing at stake - die, and you are reborn at the click of a mouse. There is no life-or-death incentive to avoid disease.
That's certainly true of the World of Warcraft outbreak, in which some players spread the disease on purpose. Blizzard's programmers planned only for those visiting those specific caves to succumb to the bloody death, yet some ingenious players discovered they could form a chain of infection. They used their virtual pets - characters can acquire them in this game - to help carry the disease out of the caves and take it to more populated areas, where the plague then spread out of control. For Williams this highlights the basic flaw in virtual-world disease modelling. "No one in the real world tries to spread diseases for fun, for example, so it just doesn't work."
Fefferman disagrees. "There are different risks. World of Warcraft has resurrection, so dying isn't as bad as it seems. But if your pet dies, that's it. It costs time, and virtual money, to get it back." Characters buy virtual pets in exchange for treasure they've found or in-game cash they've earned, which both take time to acquire. These pets can't be reincarnated for free, so if a player's pet dies, there's a real loss associated with it. And it's not just about virtual pets. People do invest their time, their reputation, and even real-world money in virtual worlds, so many players do feel they have something to lose. One player so firmly identified with the death of his virtual self that he compared Blizzard's slow response to the US government's mismanagement of the aftermath of hurricane Katrina. This might be at the extreme end of the scale, but Fefferman believes that most players feel there's enough at stake in the virtual world for them to care about not getting infected, so lessons learned there can be extended into the real world.
No one is saying, though, that the situations are exactly comparable. Of course there are differences. For most players, the World of Warcraft plague had little real-world consequence. Most enjoyed the event, which they viewed as a kind of unscripted virtual theatre. After two days Blizzard was able to contain the outbreak, and they reprogrammed the game so that no player could catch the Blood Plague again.
Nevertheless, virtual-world modelling could give valuable insights into how human behaviour makes a difference to the spread of disease, dramatically improving models and the accuracy of simulations. Virtual worlds might even throw up better strategies for combating disease outbreaks. World of Warcraft's virtual plague has kick-started a whole new approach to disease modelling, and its descendants may yet help save real lives.
Kim Coppola is a writer based in London
From issue 2592 of New Scientist magazine, 24 February 2007, page 39-41
Online Obedience
Is it possible to draw real-world conclusions from studies of the virtual world? Mel Slater of the Technical University of Catalonia in Barcelona, Spain, wanted to find out.
He decided to re-enact a famous experiment by Harvard psychologist Stanley Milgram, who showed that people would reluctantly obey a figure of authority and administer pain to an innocent subject even as the subject protested in agony. Slater asked volunteers to apply increasing "virtual" electric shocks to a computer-generated character they saw on a screen. They knew the female subject was not real; nonetheless, as she begged for the shocks to stop, and later seemed to fall unconscious, the participants exhibited genuine signs of stress, as if the shocks were being applied to a real person. The implication seems to be that if people identify strongly enough with virtual characters, conclusions from virtual worlds may well extend into the real.
