Paper #: 14-11-043
We live in an era of increasing connectivity in human societies and in technology. These structural changes in the ways we interact with each other and with increasingly ubiquitous computational and communication devices have been formalized in research across several disciplines through the dynamics of complex informational networks. Complex networks are (mathematical) graphs, connecting nodes (people, computers) via edges (relationships, wires). While much progress in methods for network analysis has been achieved, the fundamental principles that drive network growth in human societies and in worldwide computer networks remain rather obscure. Mechanistic models for the origin of certain structural graph elements have now become common but the formal connection between large empirical studies of network evolution and fundamental concepts of information, learning and social theory remains only latent. To address these issues, I argue here that the most interesting aspect of the dynamics of informational networks in complex systems is that they are the physical manifestations of processes of evolution, inference and learning, from natural ecosystems, to cities and to online environments. I formalize the general problem of learning and computation in network environments in terms of average structural network changes and propose a conceptual framework to explain the transition from initially static, undifferentiated and information-poor environments to dynamical, richly diverse and interconnected systems. I illustrate these ideas empirically by providing examples from cities, and from global computer networks and webs of documents. I finish with an overview of expected changes to urban form and function and to computational hardware under likely technological scenarios.