Multirelational organization of large-scale social networks in an online world

Citation: Michael Szell, Renaud Lambiotte, Stefan Thurner (2010) Multirelational organization of large-scale social networks in an online world. Proceedings of the National Academy of Sciences (RSS)
DOI (original publisher): 10.1073/pnas.1004008107
Semantic Scholar (metadata): 10.1073/pnas.1004008107
Sci-Hub (fulltext): 10.1073/pnas.1004008107
Internet Archive Scholar (search for fulltext): Multirelational organization of large-scale social networks in an online world
Tagged:

Summary

Szell's paper presents a summary of a series of quantitative analyses of a social networks inside a large multiplayer game (MMORPG) which is kept anonymous but which sounds not unlike a major game like Eva Online.

The initial half of the paper seems to be a general argument for the use of data from online communities in the study of social networks. The paper details the power of online data gathering over survey-based approaches to networks due to its longitudinal nature, its lack of bias and precision, it's completeness, and some of the nuance that are hard to capture in other ways.

In the game presented, there are a series of different types of relationships that players can declare or enact. The authors divide these into two groups of positive network ties which include friendship, communication, and trade and negative network ties which include enmity, attack and bounty. The contribution of the paper primary lies in the consideration of these overlapping networks within the larger undirected network. The paper reminds us that within any large society, there will be many different overlapping networks and they will interact with each other in ways that may have important effects.

The authors detail each of the positive "network-network interactions" through correlations and with largely descriptive and largely predictable results. For example, friends communicate more often with each other and individuals rarely attack or are enemies with their friends.

The primary empirical contribution of the paper seems to be a test of the structural balance hypotheses on a large scale in a real community. The structural balance theory claims, in its weaker form, that a triad where a friend of a friend is an enemy is unstable and will be underrepresented in networks. The stronger form predicts that forms where an enemy of an enemy is an enemy will also be unstable. Empirically, the authors find that, indeed, there two forms are underrepresented in the dataset from the online game.

The authors suggest that their study represent an important step in measuring the "multidimensionality" of human networks.