An Empirical Investigation of Preferential Attachment Among Influential Members of a Large Artificial Community

One among the many questions in social network analysis is how links form among members of Internet-mediated social network (ISN), where most members are usually anonymous, while link formation (i.e., interactions) between members are facilitated only by non-personal communication technologies. Researchers offer preferential attachment (PA) as a possible mechanism that can explain the behaviour of link formation, not only for real-world communities, but for artificial communities, such as ISNs, as well. PA suggests that members choose to be linked with members characterized with many links who are considered “central” to the community. This is because it is believed that central members can be relied to as a channel, if not the source themselves, of information, of wealth, or of any other kind of currency that the community is using. In this paper, the link formation process of members of one large ISN was examined to look for empirical evidences of PA among members who were clustered together according to the order of magnitude of their number of links at the global level. Members whose initial number of links that totals only up to ten thousand exhibit the opposite of PA, while members whose initial number of links that sums greater than ten thousand exhibit PA. This means that the lower bound for initial links for PA, at least for this particular ISN, is 10,000. Additionally, for those members whose link formation follow the PA mechanism, the order of magnitude of the rate of increase in their number of links is proportional to the order of magnitude of their initial number of links.