Thursday, August 30, 2007

The Disease Vector, Virtually

I always find it interesting when events in the world of gaming somehow breach into the mainstream press. Even though gaming is far more widely covered now, stories still often have that "we've discovered a lost civilization" kind of wonder to them.

This time, it's the "corrupted blood" disease in World of Warcraft, which happened, um, two years ago.

The impetus for this sudden media interest is a research paper published in the journal Lancet Infectious Diseases. Written by Nina Fefferman and Eric Lofgren. The Untapped Potential of Virtual Game Worlds to Shed Light on Real World Epidemics explores how the disease was transmitted inside the World of Warcraft, and how studying an online population might improve our planning for real-world epidemics.

To see a few of these mainstream accounts, just go here (BBC) or here (Reuters).

DQ reader Kadunta took the time to read the paper in question, and here is his excellent summary:
As it was, the authors wrote that the Corrupted Blood disease spread far too efficiently for any real-world pathogen, but that could be corrected by tuning the parameters (e.g., how easily the disease is transmitted, how high the mortality rate is). Still, they noted how the players spread the disease by instantaneous transportation back to the populated areas from the battlefield with Hakkar, the source of the disease, before dying or being cured of the disease, and mentioned how this resembles the way many epidemics in the past have spread one area to another.

In brief, the main point of the article was to suggest the idea of integrating epidemics into MMO games as essential plot events and using the collected data (with permissions from the players) in epidemiological research.

The authors write that although there are two ambitious systems--TRANSIMS (Los Alamos National Laboratory) and EpiSims (Virginia Bioinformatics Institute, Virginia Tech)--that attempt to incorporate human behaviour in computer simulations, predicting human behaviour is far from easy. Still, in the physical reality, it is nigh impossible to set up a controlled experiment due to physical, financial and moral issues. In the virtual world, it's easier (but still not without problems), although the sizes of the experiments would have to be downscaled from pure computer simulations due to the limited number of players and such resources.

As an example of human behaviour with the outbreak the authors mention the "first responders," healers who went into the infected areas in an attempt to cure other player characters, and when they moved onto other areas after being infected, possibly ended up spreading the epidemic. Another example was the failure of the quarantine rules to keep the disease from spreading when players disregarded the rules, as was mentioned also in the Reuters story.

The authors admit that the players wouldn't necessarily act in the simulations like they would in the real life, but the players still have the commitment to the community formed within the game world that would cause the in-game player reactions to approximate real-life reactions. Even so, the validity of results gained in simulated outbreaks in MMO games would still have to be externally confirmed, e.g. by comparing the results with historical real-world outbreaks, before applying them to real-world scenarios.

An interesting side note (or at least for me): they mentioned that the player characters' pets appeared to have been the dominant vector (not much unlike I recall the bubonic plague to have spread, where the rats and their fleas spread the disease). The players were penalized for letting their pets die, so once they had been infected, the players dismissed their pets, effectively placing them into stasis and retrieved them once they had reached a safer place--like a populated city for instance--at which point the pets would then start (re)infecting characters and pets around them. Also, the shopkeepers, soldiers and other such NPCs, beefed-up to stand up to players attacking them, weren't in danger of immediately dying of the disease--but they could still be infected and infect their customers, prolonging the outbreak.

As my personal opinion, a MMO game focused on society, such as
A Tale In the Desert (Wiki entry here), might be a good basis for such experiments. Actually, the Wikipedia entry also mentions ATitD had (or has?) something called "Lung Spore Disease" in the third Telling, but many players didn't take it well. How much the disease resembled an epidemy, I don't know. Also, the purportedly stronger community spirit in ATitD would most likely affect the outcome of the simulation: it's possible that the quarantine rules, for instance, would be obeyed better than in WoW.

"Community spirit" brings up an interesting rabbit hole related to all this, and one that I'd never considered before: does the medical community prepare for griefers? If there's anything online worlds have taught us, it's that there will always be griefers. Always. And even in the real world, in the face of a horrific epidemic, there will be at least a few people trying to actively spread the disease.

In other words, epidemic planning cannot be considered complete without coming up with a strategy to contain BigDick69.

Even with the behavioral distortions that manifest themselves in any online world, though, the idea of seeding an epidemic solely for the purpose of collecting data on how the disease spreads is remarkably clever.

Site Meter