This thesis explores why evidence for increasing returns is more abundant today than in the past. When a physical distance plays an important role in limiting social interactions in old days, social networks tend to be cliquish. Advances in transportation and communications systems have affected the structure of social networks by reducing the role of a physical distance. Recent development in graph theory suggests that this change is most likely to reshape social networks from cliquish networks to networks with more random connectivity. Given this as an assumption, we developed a model for adoption dynamics. The adoption dynamics of our model show an abrupt phase transition from a shared market to the winner-take-all market when we rewire the cliquish network with some random connections between nodes. Such a phase transition has been proposed as one explanation for why winner-take-all markets prevail today.