| Geographic research matters more than ever for the twenty-first century. We humans now bear the responsibility for creating our geographies of spatial technologies, and for their effects. Yet we also finally have the capability to study the effects of site, situation, and agent behavior under controlled experimental conditions within a laboratory. We could study systems before, but not via repeatable controlled experiments. We could perform controlled experiments before, but not at the level of complex geographic systems. Until now, we could not specify the sites, situations, objectives, and strategies of heterogeneous mobile individuals on richly structured network landscapes in order to systematically explore the effects of their interactions with one another and with their landscapes.; Chapter 2 extends relational small-worlds (Watts and Strogatz 1998) to spatial graphs. Spatial small-worlds weight the reassignment of small-world shortcuts according to distance and according to the existing connectivity of target nodes. A contraction factor for each shortcut reflects its advantage with respect to the base network.; This dissertation formalizes geographic landscapes as interlocking layers of spatial graphs. GeoGraphs can represent anything from isotropic planes or natural landscapes through the most advanced spatial technology transportation and telecommunication networks. Chapter 3 describes the implementation of the GeoGraph library for the Swarm agent-based simulation system (Minar et al. 1996), and introduces an epidemiological model with mobile agents on networks.; Chapter 4 harnesses the spatial small-worlds from Chapter 2 and the GeoGraphs from Chapter 3 to explore systematically the effects of long-run globalization processes that shrivel the world's geography via spatial technology shortcuts. Four sectors of economic agents play a locational game on spatial small-world GeoGraph landscapes. Statistical regressions analyze eleven hundred simulations spanning ten spatial small-world landscape structures, eleven contraction factors, and ten agent random number seeds.; Results show that both relative and absolute geographic characteristics become more important as spatial technologies improve. Yet geographic structure matters more than spatial technology improvements such as speed. Path-dependence simulations show that prior settlements have a strong constraining effect on future settlement patterns even under the weakest possible conditions where agents have perfect information, maximize globally, and move costlessly. |
