A Refined Study of the Complexity of Binary Networked Public Goods Games

Abstract: We study the complexity of several combinatorial problems in the model of binary networked public goods games. In this game, players are represented by vertices in a network, and the action of each player can be either investing or not investing in a public good. The payoff of each player is determined by the number of neighbors who invest and her own action. We study the complexity of computing action profiles that are Nash equilibrium, and those that provide the maximum utilitarian or egalitarian social welfare. We show that these problems are generally NP-hard but become polynomial-time solvable when the given network is restricted to some special domains, including networks with a constant bounded treewidth, and those whose critical clique graphs are forests.