Pierre-Olivier Montiglio

Applying ecological and evolutionary theory to video game data


It has been a while since I posted something. Since my last post, I moved to McGill University and started a new postdoc (funded through a NSERC fellowship). I figured I would present a project that I have been thinking about throughout the last few years. My goal is to get it going at full speed in the coming year. It involves applying an ecological approach to investigate and design massive multiplayer online video games (MMOGs). It would benefit many ecological disciplines directly by providing a way to assess how good we are at predicting and manipulating ecosystems. Hopefully, it would also benefit game designers, by providing some tools to handle the erratic and complex behavior of players.

Collection of video games for Nintendo consoles. Credits TarkusAB (wikipedia commons).

Now many of you are probably having a mental image of video games that looks a lot like this or this. The video game is a well-designed set of challenges that you tackle, sometimes through an avatar. Such video games are fun and very additive, but there are very different types of games out there. Games can now have an online component, allowing players to interact with each other in real time. Halo is a good example. Looking at its fan-supported encyclopedia shows the complexity and dynamism of the game’s content. These games are made up of interactions among hundreds of thousands or millions of players every day in real time. For example many, many players are constantly online in New Eden, the virtual world in which Eve-Online takes place. The scale of such interactions is comparable to the amount of interactions one would observe in some populations of free-ranging animals, or within an ecological community. Interactions among players are often occurring in a virtual world that has a clear spatial component and that is persistent (it continues to change and shift even when players are offline). Virtual worlds have their own dynamics, economies, and culture.


The behavior of players, and their interactions now make up a significant portion of the content of many video games. A challenge of game design is to develop a substrate that will allow players to interact and behave in diverse ways, ultimately impacting how much, how frequently, and how long players will stick to the game. Player behavior and the pattern of their interactions are very hard to predict and to control. As a result, designing MMOGs is major challenge. Many, many nascent game companies go bankrupt every month because their main project fails to recruit the critical mass of players that is required for the game content to emerge. Even when these games survive their first steps of development, they still require important and frequent updates (or patches), readjusting the rules governing interactions, or fixing a potential loophole that some players have come to exploit. These patches can themselves introduce substantial chaos. In one notable case, players’ unanticipated behavior and interactions drove a huge pandemic, killing thousands of characters and perturbing the game World of Warcraft for months. To tackle this huge challenge, game design is now informed more and more by analytics. Companies use GIS (geographic information systems) and statistical analyses to inform the development of their games, optimize the gameplay, and maximize player retention. However, the quantity of data available from video games poses the challenge of figuring out which questions or parameters should deserve the most attention.


The main idea behind my new project is that video game design could be informed by ecological and evolutionary theory. Ecologists deal with many of the same types of data analysis issues outlined above. In fact, they often tackle these challenges with much less data! Ecological and evolutionary theory can point out which type of data should be the most valuable to understand player behavior and interactions. Ecological and evolutionary theory can also point out which aspects of players’ behavior and interactions should be the most important in promoting stable or resilient game dynamics, or in maximizing player retention. Much like conservation, ecologically-informed game design could also enable us to build virtual worlds supporting a higher diversity of players. In turn, data on player behavior and interactions could be very valuable to test and refine ecological or evolutionary processes, which are often very hard to quantify and observe (or manipulate) in nature. Some economists have already started analyzing virtual economies, and some biologists are getting more and more interested in using video games to inform their research. I have multiple analyses on the way, so stay tuned! A student, Julien Céré also recently started his PhD on these ideas, in partnership with the company Behaviour Interactive. He aims to apply game theory and social network theory to analyze the behavior of players in specific multiplayer games. Although he is just starting, we have been getting a very good attention from the media. Keep an eye for his progress as he goes through his PhD!