Nat. Commun. 8, 1888 (2017)

The theoretical machinery used to study complex networks has found extensive use in evolutionary game theory, particularly when it comes to understanding the dynamics of cooperation and the emergence of complex societies.

Most contact networks have a heterogeneous structure: they are made up of many connected clusters of varying size, as opposed to the homogeneous lattice structure commonly seen in models describing (say) magnetic systems. Provided the distribution of the connected clusters in a contact network is scale free — so that the degree of their connectivity follows a power law distribution — it has been shown theoretically that cooperation is favoured. Unfortunately, this prediction has not been confirmed in recent experiments, suggesting something deeper must be at play.

By mapping the nodes of real complex networks onto a metric space, Kaj-Kolja Kleineberg was able to make a more useful representation, in which the popularity of a node and its degree of similarity to other nodes was made explicit. Using this representation, he confirmed that heterogeneity does not always favour cooperation. It can, in fact, even hinder it.