Stratification in a network model of information exchange

We propose a network model in which a node exchanges information with its neighbors and rewires its link from a neighbor to the neighbor's neighbor when it is given a greater chance to get a new item of information. In simulations, the degree distribution evolves briefly from an initial uniform assignment to be scale-free in the tail, and then freezes with a heavier tail, which corresponds to the monopolization of links by a few hubs. After freezing, the class of nodes in the middle of the degree distribution shrinks, and information levels become stratified. Stratification is controlled by the rewiring rate and the number of links per node. We suggest that increasing the connectedness of nodes is the best way to avoid stratification.