A stochastic game approach for competition over popularity in social networks (Q367440)

The author considers competition between individuals who create contents and want to distribute the content through a content provider. An individual can pay the content provider to receive a preferential treatment to the content in order to accelerate its dissemination. Each of the contents is characterized by some given potential popularity. An individual decides whether to invest in advertisement or not, depending on the potential popularity of the content and its competing contents. This decision problem is formulated as a stochastic game with a finite state and action space. In Section 2 the problem is modeled as a continuous-time Markov game and in Section 3 the standard uniformization approach is used to transform the problem into an equivalent discrete-time Markov game. A characterization of the equilibrium policy is obtained. In Section 4 the author begins by considering that the dissemination utility is linear and then considers that also the acceleration costs are linear. Section 5 concerns the case of no state information, that is, it is assumed that the players do not observe the state, and they know either the initial state or its distribution or its expectation. Next it is shown that the stochastic game is equivalent to a differential game. In Section 6 the author considers the game obtained for an infinite horizon and computes the equilibrium. Section 7 is devoted to concluding remarks where the author comments on the relation between the differential game and the original stochastic game, and mentions several issues for future research, like coupling between various social networks and spatial considerations.