On September 13 2005, Blizzard – the developers of video game World of Warcraft – accidentally unleashed a plague. The hugely popular online role-playing game takes place on Azeroth, a virtual world with densely packed cities separated by stretches of open country, along with relatively unexplored areas like jungles and caves. Earlier that day, a new software update had granted millions of players access to Zul’Gurub, a new jungle-like area of the game world intended for those with relatively powerful characters.
The heart of this new section was a duel with a winged serpent called Hakkar, a powerful foe with the ability to infect player’s characters with a disease called Corrupted Blood, which would then be passed on to other nearby characters. It was designed to make fighting Hakkar slightly more difficult by slowly draining a player’s health, but there were unintended consequences.
World of Warcraft gives players the ability to fast travel – they can move instantly from remote areas like Zul’Gurub back to cities, for example, to stock up on supplies. It meant that powerful players who had been infected with the virus were able to carry it to mass population centres before they died or were healed.
The spread of the virus was accentuated by two factors – like the bubonic plague, new outbreaks were started by in-game pets, which could become carriers. Players often put them into something like suspended animation before or during big fights to protect them, but when they came out of this state they started new outbreaks. The game’s non-playable characters – shopkeepers and the like – are basically impossible to kill, but they could still carry the virus, so they quickly became super-spreaders.
Corrupted Blood soon developed into a full-blown in-game pandemic. As the skeletons piled up in the densely packed auction house in the capital city, one player realised that this could be more than just an amusing anecdote.
As well as being an avid gamer, Eric Lofgren is also an epidemiologist, and he realised that the way players reacted to Corrupted Blood could offer a valuable insight into how we can expect them to behave during the rapid spread of a real-world contagious disease.
That’s really crucial, because a lot of the models that scientists use to try and predict how a disease like Covid-19 will spread are built on assumptions about the way people will behave. But we’re pretty irrational creatures – few would have predicted, for instance, that fans barred from attending the recent Champions League football match between Borussia Dortmund and Paris Saint-Germain would gather outside the stadium anyway, or that Ukrainians would voice their fear at a quarantine centre being set up in their area by going to the quarantine centre and throwing stones at a bus carrying evacuees from Wuhan.
“Traditionally when we do computer-based simulations we know everything about the world,” says Lofgren, who published a 2007 paper on the Corrupted Blood outbreak with colleague Nina Hefferman. “The people in those simulations only act the way we tell them to act. Here we get the full view of human irrationality.”
There are echoes of Corrupted Blood in the way the coronavirus spread from remote to urban areas, and in some of the behaviour we’re seeing in the worst hit countries. For powerful characters, the disease was no more troubling than a common cold, so they just went about their daily lives – but ended up spreading it to areas where more vulnerable players quickly died from it.
Some tried to be “first responders,” Lofgren says, travelling to the epicentre of the epidemic and trying to heal players who were infected – but this often meant contracting the disease themselves and then spreading it – we’ve seen parallels of this with healthcare workers becoming sick and passing away due to a combination of the coronavirus and general exhaustion.
As news of the outbreak spread, some people logged on to the game to see what the fuss was about, and promptly became infected themselves. There were even isolated incidents of players deliberately trying to spread the virus – we haven’t seen that in the real world, thankfully, although NBA player Rudy Gobert was heavily criticised for deliberately touching microphones and recording devices at a press conference a couple of days before testing positive for Covid-19. Perhaps there’s another parallel. “You do get people who go to work even though they’re sick because economic circumstances demand it, or to not let down their team,” Lofgren says. “We are also seeing some people not taking it seriously, and wilfully ignoring the risk, which is parallel to intentionally spreading it.”
Although Lofgren doesn’t remember seeing much evidence of stockpiling (no runs on virtual pasta), Corrupted Blood did have a wide impact for several days. “The capital cities, which were very densely populated and the central social and economic hub of the game, became very hard to live in,” Lofgren says. “There were some fairly significant disruptions to the economy of the game.”
For Lofgren, it emphasised the importance of behaviour in the spread of epidemics. “People’s decisions about their own risk are extremely important,” he says. But despite the rich data on human behaviours collected during Corrupted Blood, little further research has been done – partly because of the expense and difficulty of getting developers on board who don’t to jeopardise the entertainment value of their products. “This was fun for people because it was this exciting emergent crisis but also it didn’t last very long,” Lofgren says. “If it had gone for a couple of weeks people would have been frustrated.”
However, behavioural economists have also used games to try and tease out human behaviour during epidemics in a more formal setting. In 2013, Frederick Chen, an economist at Wake Forest University in North Carolina, designed a 45-day online game that simulated the outbreak of disease.
Players received points for staying healthy, and lost them if they got sick, and at the end of the study they were given a cash reward linked to the number of points they’d won. Each day, participants were told whether they were healthy or infected and how many other people were infected, and they had to decide whether or not to protect themselves against the outbreak for the next day’s update.
The parallel is to self-isolation – there’s a cost associated with protecting yourself, but it’s smaller than the cost of getting the disease. In different phases of the experiment, Chen changed the cost of self-protection – how many points people would lose if they chose to inoculate themselves against infection in the next round. “The lower the cost, the more willing people were to self protect and the lower the disease prevalence,” he says. “If you make it cheaper and easier for people to self-protect, people will do so.”
But there were some people who almost never protected themselves, regardless of the cost or the prevalence of the disease. Others chose not to protect themselves until they were infected for the first time, after which they were more cautious (in this experiment, you could get infected twice).
The problem he identified is called ‘self-protection fatigue,’ and it happens because as the prevalence of a disease drops, people get more brazen – they start going out again and stop protecting themselves before the disease has been eradicated. “You implement these social-distancing measures and if they’re working nothing bad happen and people take their foot off the gas and that’s when bad things can happen, “ he says.
“One thing that very starkly stood out is that the disease in my game didn’t have to be a problem,” Chen says. “Everybody collectively could have eradicated the disease if they acted safe enough.” That never happened – and in fact, people performed worse than Chen had expected, so much so that he had to give some of his grant money back because he paid out far less in prizes than he had predicted.
Economists call these externalities – actions we take which have a negative impact on a third party but don’t impact the people making the decision. They’re why climate change is such a big problem, and why the environmental impact of palm oil isn’t reflected in the price of Nutella.
They were a factor during the Corrupted Blood epidemic too. In response to the outbreak, the developers made a concerted effort to try and halt its spread. Some players effectively went into self-isolation, restricting themselves to remote areas of the game. There was a failed attempt at a quarantine – players kept escaping (echoed in the mad dash out of Lombardy in the hours before the lockdown was imposed there), and an effort to ask infected users to ‘flag’ themselves as such to warn others to keep their distance.
But nothing worked – the powerful players whose characters weren’t threatened by the virus simply carried on as usual, unaffected by the disease even as they spread its havoc across the virtual world. In the end, the only way for the developers of World of Warcraft to stop the spread of Corrupted Blood was to take drastic, co-ordinated, worldwide action. They reset the server.
Amit Katwala is WIRED’s culture editor. He tweets from @amitkatwala
More great stories from WIRED
😓 Does alcohol kill coronavirus? The biggest myths, busted
📺 The best shows coming to Disney Plus UK
💩 Gender neutral toilets are a massive failure (so far)
🏙️ A huge Airbnb scam is taking over London