The online game Pardus is helping researchers model complex social interactions that happen in real life.

The online game Pardus is helping researchers model complex social interactions that happen in real life.


Even online, you can't have more than 150 friends

Your starbase is almost complete. All you need is a few more tons of ore. You could take the afternoon to mine it from an asteroid field, but you’ve heard of a Ska’ari who trades ore for cheap. So you message your alliance, use your connections to set up a meeting, and hop in your spacecraft. It’s good to have friends, even if they are virtual.

An online science fiction game may not seem like the ideal place to study human behavior, but physicist Stefan Thurner has shown that the way people act in the virtual world isn’t so different from how they act in the real one. Thurner studies all sorts of complex systems at the Medical University of Vienna, so when one of his doctoral students just happened to create one of the most popular free browser-based games in Europe, Thurner suggested using the game, called Pardus, to study the spontaneous organization of people in a closed society. For almost three-and-a-half years, they monitored the interactions of roughly 7000 active players at one time within the game’s virtual world.

Unlike in real life, Pardus players’ moves are tracked and their interactions are recorded automatically by the game. “We have information about everything,” Thurner says. “We know who is where at what point in time, … who exchanges things or money with whom, who is friends with whom, … who hates someone else, who collaborates with whom in entrepreneurial activities, who is in a criminal gang with whom, etc. Even though the society is artificial, it’s a human society.”

Thurner was especially interested in testing the theory in anthropology that there is a limit to the number of face-to-face relationships a person can maintain at once. “One hundred and fifty is the number of people you can have meaningful relationships with,” at least when you’re talking about real-world interactions, says Robin Dunbar, an anthropologist at the University of Oxford in the United Kingdom who discovered the limit and was not involved in the new work. “This turns out to be correlated with core areas of the brain,” particularly “the frontal lobes and the temporal lobes.” In other words, our brains aren't large enough or interconnected enough to maintain an infinite number of personal relationships.

Some scientists theorize that the so-called Dunbar’s number could be larger for online relationships, because the time it takes to have a social interaction is reduced when all you have to do is send a quick message, as opposed to meeting up for coffee, a meal, or a full day of activities. Pardus offered an interesting way to study this, Thurner says, because players can form different kinds of relationships with each other. In the game, players choose an official faction to join, which determines which side a player is on. Within these factions, players are able to communicate and create formal groups known as alliances. Players can also declare other players as friends. Thurner and his colleagues then used their data on interpersonal interactions to divide these friendships into two groups: close friends, or players who had declared friendship and communicated through private messaging, and acquaintances, or players who had designated each other as friends but did not message.

Remarkably, though the game sets no constraints on the size of alliances, players organized themselves into social structures that matched those found in the real world. The largest alliance in Pardus was only 136 members, suggesting that the limit of 150 personal connections is an inherent part of our social psychology, Thurner and colleagues report this month in Scientific Reports.

That’s not the only way relationships in Pardus mirrored offline friendships. In real life, we’re more emotionally invested in those we consider our close friends, creating smaller groups of relationships within our 150-person circle that we spend the most time and energy tending. Correspondingly, Pardus players invested more time interacting with friends they messaged, concentrating their social attention on close friends rather than distributing it among a larger web of acquaintances. Thurner suggests that we may characterize people into different groups in order to keep connections straight in our heads, but that “no one knows how this mental map is organized.” In this study, “we may have seen the first insight into how this is managed in humans.”

“This is a good example of a paper that suggests a clear correspondence between how people behave in real life and virtual environments,” says social scientist James Ivory, who studies social and psychological aspects of people online at Virginia Polytechnic Institute and State University in Blacksburg. “People tend to behave like people, whether they’re in a prehistoric world, a business, a knitting group, or a video game. Instead of looking at behavior in video games as alien, what you basically have is a place where you can study people.”