Lognormal distribution in the digg online social network

Abstract

We analyse the number of votes, called the digg value, which measures the impact or popularity of submitted information in the Online Social Network Digg. Experiments over five years indicate that the digg value of a story on the first frontpage follows closely a lognormal distribution. While the law of proportionate effect explains lognormal behavior, the proportionality factor a in that law is assumed to have a constant mean, whereas experiments show that a decreases linearly with time. Our hypothesis, the probability that a user diggs (votes) on a story given that he observes a certain digg value m equals a×m, can explain observations, provided that the population of users that can digg on that story is close to a Gaussian.