Reviewing the potentials of MMOGs as research environments: A case study from the strategy game Travian

Abstract

Massively Multiplayer Online Games (MMOGs) provide many opportunities for scientists. Previous research ranges from personality trait prediction to alternative cancer treatments. However, there is an ongoing debate on whether these virtual worlds are able to represent real world scenarios. The mapping of online and offline findings is key to answering this question. Our work contributes to this discussion by providing an overview of the findings from network-based team and leadership research and by matching them with concrete results from our MMOG case study. One major finding is that team size matters. We show that high diversity in the type of teams is a major challenge, especially when combined with the immense amount of data in MMOGs. In our work, we discuss these issues and show that a well-grounded understanding of the data and the game environment makes it possible to overcome these limitations. Besides the team size, the aggregation periods play an important role. Regarding MMOGs as research environments, we show that it is important to pay close attention to the specific game-related contexts, the incentive structures, and the downside risks. Methodologically, we apply support and communication networks to show the influence of certain group-based measures (e.g., density, transitivity) as well as leadership-centered characteristics (e.g., k-core, group centrality, betweenness centralization) on team performance. Apart from our findings on centralization in communication networks, we are able to demonstrate that our results confirm the theoretical predictions which suggest that the behavioral patterns observed in MMOG teams are comparable to those observed in offline work teams.

Fig 1. Teams in Travian are formed by the founder of the alliance, who initially holds all the power.

To grow their alliance, the founder invites players to join the team. Here the founder can grant special rights to the new alliance members. These rights give these players the de facto power to fill different roles within the alliance. As players expand their villages, they earn additional points in the game ranking. The ranking of the alliances derives, in turn from the sum of the points of their individual members.

Fig 2. Splitting the sample of all teams (with 2 to 60 members) into four specific subsamples was a prerequisite for comparing the network patterns of teams of different sizes.

Fig 3. Teams within Travian constantly form and dissolve.

In the first phase of the game, the number of alliance formations exceeds the number of dissolutions, which finds its peak at the beginning of the middle phase of the game. At this point a massive reorganization process takes place. In the later phases of the game, the number of team dissolutions exceeds the number of new formations, so the number of active teams steadily decreases.

Fig 4. A schematic representation of the observation periods taken from the mid game phase, starting with day 110 and ending with day 250.

Our modified moving window approach (advancing 30 days at a time) integrated all communication and support relationships that comprised a full 60-day period.

Fig 5. Support networks.

Correlation of each measure with team performance for the groups of small-, mid-, and large-size teams as well as all teams.

Fig 6. Communication networks.

Correlation of each measure with team performance for the groups of small-, mid-, large-size teams as well as all teams.