In September of 2005, players in Blizzard’s incredibly popular MMORPG, World of Warcraft, started dropping dead from a new disease accidentally introduced by Blizzard in an update to the game. The problem started when Blizzard’s programmers added a new instance, which is a separate dungeon disconnected from the outside world that players can enter to attempt quests. The new instance, Zul’Gurub, contained the god of blood, Hakkar. Players would enter the instance in groups of 20 in an attempt to kill Hakkar. One of Hakkar’s abilities was a spell that he could use on players called Corrupted Blood. The spell dealt a large amount of damage (800+) to any player within its vicinity when it is cast, but also lingers on the players afterwards. After the initial damage the spell does an additional 200 damage to players every 2 seconds for a short time.

Normally, the disease was confined the instance and only the players that entered Zul’Gurub and attempted to fight Hakkar would catch the disease. Eventually, some players figured out how to bring the disease out of the instance and into the general population. Once in the general game world, the disease would spread from player to player in densely populated cities. Since the disease was intended only to effect higher level players that could access the instance, it was lethal to most all players lower level players nearly instantly.

How does this relate to Networks? Well, last year, at the Games For Health conference in Baltimore, epidemiologist Nina H. Fefferman, Ph.D spoke on the difficulties in modeling disease origins and control, and how examining MMO populations could solve some of the problems inherent with more traditional models. The three traditional models are: compartmental models, network models, and agent-based models. Each has its own strengths and weaknesses, but the main problem is that all of them make a significant behavioral assumption. This is a problem, as people are notoriously hard to predict. As Fefferman pointed out, people may stay home from work during an epidemic, or they may not; they may self-quarantine for fear of exposing health workers or they may deliberately expose as many people as possible. “During the anthrax scare, people acted irrationally, asking for smallpox vaccines. You can’t convince people on these things.”

Games may be able to overcome these problems by allowing Epidemiologists to directly observe a disease spreading through a population and actual human beings response to it. When Corrupted Blood first appeared in World of Warcraft, Blizzard attempted to institute a voluntarily quarantine to stem the disease, but it failed. Some players didn’t take the threat seriously. Malicious players used the pandemonium as a prime environment for griefing (griefing is when one player goes out of their way to deliberately ruin the experience of another player). Eventually, after several days of failing to cure the problem, Blizzard was forced to do a hard reset of the game servers.

Interestingly, and importantly if games are going to be used as models for real world epidemics, player behavior was closely aligned with real-world behavior. Healers volunteered in city centers. Lower-level characters warned others away at city limits. Nervous players fled the cities. Suspicion and fear was ran rampant. Though one trait was particularly enlightening: curiosity, something epidemiologists did not generally build into their models. Some players attempted to enter infected areas to witness the chaos, then rush out before contracting the disease themselves. This behavior has real-world parallels, particularly in the case of journalists, who must rush towards a problem to cover it, then rush back out. Even something that is generally considered a video-game behaviour, griefing, has real-world parallels. There was the infamous “AIDS patient zero,” who slept with as many people as possible to spread the disease. In World of Warcraft this was mimicked by people deliberately bringing the disease back from Zul’Gurub and into populated areas once the method for doing so was discovered. Finally, the incident in World of Warcraft also raised questions: would groups of real people behave like some in game guilds, who often kicked out longstanding members who infected guild-mates due to a perceived “lack of responsibility”?

Fefferman has been in contact with Blizzard and other MMO developers, hoping to introduce new scripted diseases into more online games, to watch them spread.

Sources:

GFH: The Real Life Lessons Of WoW’s Corrupted Blood
Virtual plague spreading like wildfire in World of Warcraft