Social context congestion games

  • Authors:

  • Vittorio Bilò;Alessandro Celi;Michele Flammini;Vasco Gallotti

  • Affiliations:

  • Department of Mathematics, University of Salento, Lecce, Italy;Department of Computer Science, University of L'Aquila, L'Aquila, Italy;Department of Computer Science, University of L'Aquila, L'Aquila, Italy;Department of Computer Science, University of L'Aquila, L'Aquila, Italy

  • Venue:
  • SIROCCO'11 Proceedings of the 18th international conference on Structural information and communication complexity
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

We consider the social context games introduced in [2], where we are given a classical game, an undirected social graph expressing knowledge among the players and an aggregating function. The players and strategies are as in the underlying game, while the players costs are computed from their immediate costs, that is the original payoffs in the underlying game, according to the neighborhood in the social graph and to the aggregation function. More precisely, the perceived cost incurred by a player is the result of the aggregating function applied to the immediate costs of her neighbors and of the player herself. We investigate social context games in which the underlying games are linear congestion games and Shapley cost sharing games, while the aggregating functions are min, max and sum. In each of the six arising cases, we first completely characterize the class of the social network topologies guaranteeing the existence of pure Nash equilibria. We then provide optimal or asymptotically optimal bounds on the price of anarchy of 22 out of the 24 cases obtained by considering four social cost functions, namely, max and sum of the players' immediate and perceived costs. Finally, we extend some of our results to multicast games, a relevant subclass of the Shapley cost sharing ones.