From networks of cities to systems of cities

Abstract

This chapter elaborates on a shift in urban theories from a static and structural view of "networks of cities" in a territory toward a systemic conceptualization and dynamic analysis of interdependencies and co-evolution in complex systems. The major concepts and methodological issues that were used in the process of theory building are reviewed during three major stages in the history of the science of cities, including examples taken mainly from the literature in Europe and United States. In a first step, the role of communication networks in the historical evolution of cities is intuitively described and the first measurements of cities spacing, size and functions lead to the construction of formal theories of networks of cities according to static models as central place theory and rank-size distribution. In a second step the dynamics of city networks is inferred from statistical analyses and cross sectional comparisons of the attributes of large numbers of cities at different points in time leading to constructing an evolutionary theory of systems of cities and including inter-urban communication and spatial diffusion of innovation. In a third step the growing availability of urban data, especially about urban interactions, and the increased power of computing algorithms enable to fully explain this evolutionary theory of systems of cities by including features of complex systems as bifurcations, fractals or scaling laws as well as theories of social sciences about networks, innovation and social change. This comprehensive theory is both exemplified on a huge variety of empirical data and tested through dedicated simulation models.

The role of complexity for digital twins of cities

“…In fact, most models of cities still treat cities as systems built from the top down. Complexity science changes this perspective and allows one to consider cities as multi-level systems 42,43 , which involve many bottom-up processes. This makes it possible to explain highly important signatures such as power laws and scaling 44 , as well as long-range correlations across the various networks 45 .…”