The assumptions we hold about men and women tend to replicate themselves online.

[optional image description]
When joining the multi-player game Pardus, players choose a gender. The circles above indicate male players; the squares represent female. The nodes are connected by six different types of links between players, representing their friendship or enmity status, exchange of messages, trading activity, aggressive acts (attacks), and their placing of head-money (bounties) on others as a means of punishment.

Pardus is a Massively Multiplayer Online Game. It creates a space-based universe in which players interact with each other, cooperating and fighting and developing societies and economies. The game currently features over 300,000 players from around the world. And each of those players' actions is recorded -- without the players being strictly aware of the recording.

Which means that Pardus is both a game and a huge, rich data set.

The researchers Michael Szell and Stefan Thurner, of the Medical University of Vienna, took that data set and analyzed it. They focused on social interactions within the game -- and in particular on the way the players' genders related to each other within its framework.

Szell and Thurner found what many other researchers have suggested: that the gender assumptions so familiar from the analog world tend to replicate themselves in the digital.

Their key findings:

•Females are less risk-taking but wealthier.
•Females attract positive behavior.
•Females show homophily (birds-of-a-feather-type clustering).
•Males are heterophiles (showing diversity in their social groups).
•Females have more communication partners.
•Females organize in clusters.
•Males prefer well connected communication partners.
•Females reciprocate friendships.
•Males respond quickly to female friendship initiatives.
•Males respond slowly to female enmity initiatives.

None of that, for better or worse, is terribly surprising. There's a relatively small body of research focused on gender differences in digital social settings -- and most of that research points to the digital replication of gender roles, rather than the reversal of them. In the classic sense of technology as simply a collection of tools that extend humanity's core humanity ... that makes sense. A recent analysis of 2 billion mobile phone calls and almost half a billion text messages -- broken down by the ages and genders of phone service subscribers -- found that the men's and women's communications patterns mimic the genders' reproduction strategies. Women, the analysis proposed, are more focused on opposite-sex relationships than men are during the period of their lives when they are reproductively active. After they turn 40 or so, they tend to have more communications with fellow females.

For the Pardus analysis, the most interesting question is how gender as a construct, rather than a biological fact, plays out in its digital world. The obvious caveat to Szell and Thurner's findings is that gender as delineated in the game may not equate to a player's physical gender. (The researchers estimate that some 15 percent of Pardus' players play with an avatar of the opposite gender.) That caveat, however, may not be much of a caveat: The expectation of gender might be as determinative as anything else. The online world may give us the freedom to be -- or, at least, to act as -- whoever we want to be. But the assumptions of the atom-based world have a role to play, as well.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.