Category Archives: Models

Gaming, Complexity, and Science

From September 13th to October 8th 2005, an epidemic ravaged, affecting an estimated 4 mio. people and littering the streets of hard-hit cities with corpses. Survivors fled to the countryside or abroad, further spreading the disease. Since it happened in a high-tech society, epidemiologists have used the data gathered to generate unique insights for epidemics.

Yet, chances are you have not even heard of that epidemic. Neither had I. Because it hasn’t really taken place. It was an outbreak of Corrupted Blood in the MMORPG (massively multiplayer online role-playing game) World of Warcraft [1].

The outbreak has fascinated scientists not shying away from the unconventional subject [2]. Yes, it all happened in an artificial and not very realistic system. Real-world disease victims don’t respawn, contrary to what zombie movies have us believe. In the game there is no immunity and no mutation of the disease. There is, however, one thing no model can offer: humans — millions of real humans with real motivations, fears, and behaviors. This and the amazing availability of data make such games a perfect medium to observe for science.

So far, most research into video games has focused on games’ influence on player psychology, education through games (e.g. gamified language learning in Duolingo), and making scientific tasks into games to crowdsource science (e.g. fold.it crowdsourcing problems in protein conformation). And It’s certainly a good thing that science is getting serious about a reality that, for better or for worse, will stay with us for good. It certainly is relevant to everyday lives — 60% of Americans play video games daily [3], and it’s unlikely that numbers for Europe lag far behind.

One side to gaming that has been neglected is how it can help along the science of complex systems. Simply put [2]:  “the epidemic of Corrupted Blood within World of Warcraft was the result of unintended interactions between different elements of the game”. The game is certainly complex at around 5.5 million lines of code at inception and 12 million players at peak. Every one of those 12 million is a self-motivated and realistic agent, not some simplified simulation. During the outbreak, players banded together, rushing into crisis areas to help those affected, spontaneously organizing groups of first responders. Others meanwhile sabotaged the efforts. Such simulations share many characteristics with highly detailed agent-based models in epidemiology [2], such as interconnectedness, transportation, and mass migration, along with spontaneous reactions. They describe nonlinear and chaotic interactions: The actions of any one player influence any other. Even without a crisis, players spontaneously organize into clans, divy up work and manage resource allocation, deciding how to spend shared game time and resources.

This self-organization is nowhere more apparent than in games’ trading systems. Virtual Economy is by now a widely discussed phenomenon. Since games rarely have centralized monetary systems controlling money supply, they run into issues with inflation [4]. As property and wealth in games often behave different than in real-life, strange effects are caused by seemingly innocuous changes. This gives game developers an economic sandbox with real-life subjects and variable conditions. One of the best examples of in-game economies is the game EVE Online. Economic activity is so essential to gameplay that the developer hired a full-time economist [5] and publishes economic data regularly. All the while, players find more creative ways to cooperatively manipulate the in-game economy in ethically dubious ways. Sound familiar?

How much more could we learn if games were used consciously as simulations or at least systematically evaluated? Could they help make real-world systems more robust? Of the (so far) three epidemics in World of Warcraft, two were accidents; one was actually planned as such. The economies of games rarely come about by conscious design. As it stands, Epidemiologists rarely get a chance to study the reactions of people faced with a bona fide epidemic. Economists rarely if ever get to design experimental economic systems in the real world. If all these researchers had access to complex systems involving millions of participants, would they be able to find and combat systemic problems more easily? Maybe we should start using the data we have available in these games. In the worst case scenario, the efforts would still make a fascinating read. And gamers are certainly more than happy to participate in experiments [6]. Let’s give them a chance. There’s a horde of willing participants out there.

Disclaimer:
The author is (sadly) not sponsored by Blizzard or any other game developers in the article. He still refuses to play MMORPGs, much to the dismay of his peers and apologizes to hardcore gamers for any mistakes or simplifications.

This article was inspired by a series of videos on scientific themes in video games, most notably:
Epidemic in World of Warcraft: https://www.youtube.com/watch?v=YNhP5b9YafM
Economics in MMORPGs: https://www.youtube.com/watch?v=sumZLwFXJqE

[1] https://uk.reuters.com/article/us-flu-virtual/online-blood-plague-offers-lessons-for-pandemics-idUKTRE53Q4HI20090427
[2] Lofgren, E. T., & Feff, N. H. (2007). Personal View The untapped potential of virtual game worlds to shed light on real world epidemics. The Lancet Infectious Diseases, 7 (September).
[3] http://www.theesa.com/about-esa/industry-facts/
[4] https://www.gamasutra.com/view/feature/134576/virtual_economic_theory_how_mmos_.php
[5] https://www.rockpapershotgun.com/2014/05/21/eve-economist-interview/
[6] https://www.sciencealert.com/a-wildly-popular-video-game-about-deadly-plagues-is-adding-anti-vaxxers-to-the-list