he pervasiveness of the Internet has lead research and applications to focus more and more on their users. Online social networks such as Facebook provide users with the ability to maintain an unprecedented number of social connections. Recommendation systems exploit the opinions of other users to suggest movies or products based on our similarity with them. This shift from machines to users motivates the emergence of novel applications and research challenges. In this paper, we embrace the social aspects of the Web 2.0 by considering a novel problem. We build a distributed social market that combines interest-based social networks with explicit networks like Facebook. Our Social Market (SM) allows users to identify and build connections to other users that can provide interesting goods, or information. At the same time, it backs up these connections with trust, by associating them with paths of trusted users that connect new acquaintances through the explicit network. This convergence of implicit and explicit networks yields \taps, a novel gossip protocol that can be applied in applications devoted to commercial transactions, or to add robustness to standard gossip applications like dissemination or recommendation systems.