Social networks: Big and lively or small and stable?
By Kurt Kleiner Simple mathematical principles may govern the stability of many types of social network, say scientists who analysed the connections between collaborating scientists, and between cellphone users. Tamás Vicsek, a physicist at Eötvös University in Budapest, and colleagues decided to see if they could find similarities between two distinct social networks. They analysed the links between 30,000 scientists who had co-authored papers over 11 years, and cellphone calls connecting 4 million users over the course of a year, based on data from an undisclosed cellphone company First, they identified “communities” within these overall networks – collections of individuals who were strongly linked. In the case of the scientists, a community was formed by those who co-authored papers together. Co-authoring a paper establishes a link, and mutual links between more than two co-authors forms a community. But these links fade over time unless further papers are co-authored. In the case of cellphone uses, links were formed by the calls made between different people. The researchers then looked at how these communities changed over time. They were especially interested in what made some groups persist while others fell apart. In both cases, they found that communities of more than 20 people survived for different reasons to and smaller communities. Groups of more than 20 people had the best chance of surviving if they were dynamic – regularly replacing old members with newer ones, sometimes to the point where eventually none of the original members remained. Meanwhile, smaller groups, of less than 20 people, were more likely to survive they did not replace members frequently. These smaller communities were also more dependent on key individuals who maintained frequent contact with multiple members. Finally, the researchers found that the survival of each size of group could be predicted by levels of contact with members outside of the group. In general, the more contact members had outside the community, the more likely the community was to disappear. Vicsek says the availability of electronic records made analysing social networks much easier, and believes the results should apply to other social ties. John Scott, a sociologist at Essex University in Colchester who studies social networks, says that the principles described in the paper will be familiar sociologists, who have been studying social networks seriously since the 1960s. “The conclusions that they’ve come to are things that will be well-recognised within sociology generally, even if they haven’t been charted out in formal network terms,” Scott says. “What they’re doing is supplying a technique or method that enables us to perhaps say some of these things with greater precision than before.” Journal reference: Nature (vol 446,