Distributed social systems

TLDR: A convergence of ideas taking place at the boundary between distributed computer networks and human social networks --- the former consisting of computing devices linked by an underlying communication medium, and the latter consisting of people and organizations in society connected by ties that represent friendship, interaction, and influence is considered.

Abstract: The study of large-scale networks has emerged over the past several years as a theme that spans many disciplines, ranging from computing and information science to the social and biological sciences. Indeed, a shared interest in network structure is arguably one of the forces that is helping draw many of these disciplines closer together. As one aspect of this broader theme, we consider a convergence of ideas taking place at the boundary between distributed computer networks and human social networks --- the former consisting of computing devices linked by an underlying communication medium, and the latter consisting of people and organizations in society connected by ties that represent friendship, interaction, and influence.Distributed computing systems have long been intertwined with the social networks that link their user populations. Recent developments, however, have added further dimensions to this relationship: the growth of blogging, social networking services, and other forms of social media on the Internet have made large-scale social networks more transparent to the general public than ever before. They have also raised new research challenges at the interface of computer science and the social sciences --- challenges in which the study of distributed computing has the potential to provide considerable insight.We discuss three related areas that illustrate the issues at this interface. The first is centered around the small-world phenomenon --- the premise that most pairs of individuals in a social network are linked by very short paths (or "six degrees of separation") [36]. In earlier work, we proposed that the social-psychology experiments providing the first empirical evidence for the phenomenon [25, 35] were related in fundamental ways to the problem of decentralized routing [14], and this theme has been pursued in a number of subsequent papers (e.g. [5, 8, 15, 17, 24, 29, 31, 32]). This line of research has helped to abstract some of the general principles underlying random graphs in which decentralized routing and search are feasible --- structures in which local information is sufficient to reach designated targets in the network. In the process, close connections have been developed to research in the design of decentralized peer-to-peer systems [3, 20, 21, 22, 23], and some of the patterns suggested by the basic models of small-world networks have been borne out to a striking extent by empirical studies of social network structure [2, 19].As a second area, we consider cascading behavior and the diffusion of information in networks. Rumors, fads, innovations, social movements, and diseases spread through human social networks [9, 28, 30, 33] in much the way that information propagates through a distributed system. And as with small-world networks, the analogies between the computational and social versions of these phenomena turn out to be deep rather than superficial. One of the oldest connections here was the pioneering work on epidemic algorithms presented by Demers et al. at PODC 1987 [6], in which probabilistic rules for information dissemination in distributed systems are modeled on aspects of biological epidemics (see [12] for a recent overview of this topic). Recent work has exploited similar analogies in the development of viral marketing strategies to promote new innovations by word-of-mouth effects [7, 13, 18, 27], in the growth of on-line communities and social networking sites [4], and in the analysis of information cascades among weblogs [1, 10].Finally, we consider game-theoretic models for these types of search and diffusion processes. The use of game theory to analyze networks of interacting strategic agents has become an active area of research in computer science (see e.g. [11, 26, 34]); in the present context we can ask how the introduction of economic incentives affects the performance of decentralized search or information diffusion algorithms. In particular, if the intermediaries on a path from a query to an answer require compensation for their participation in the search, then the dynamics of the system depend crucially on both the structure of the network and on the rarity of the answer; the resulting analysis leads to natural questions related to strategic behavior in branching processes [16].