Rethinking Network Science Big and Modeling for Critical Infrastructure Protection, Analysis, and Development
From Santa Fe Institute Events Wiki
|Working Group Navigation|
organized by Cris Moore (SFI), Paul Hines (University of Vermont), and Matthew Koehler (The MITRE Corporation).
Our nation’s, and indeed all of human society’s, dependence on infrastructure continues to grow. Unfortunately, our ability to analyze, model, and understand our infrastructure is not keeping pace with this growing dependence. This creates a number of significant problems, inter alia: How do we prioritize limited investment capital? How do we prioritize protection? How do we forecast changes over time, as a population’s density changes, and as cultural/historical idiosyncrasies develop?
In general, it is straightforward to represent an infrastructure system, or even multiple interconnected infrastructure systems, as a network. However, simple applications of network science do not necessarily produce useful insight, and may be premised upon faulty assumptions of cause and effect, and how effects spread. This is particularly true for some sectors of our infrastructure such as the power grid. Here, although the “wires” of the power grid do create a network, the impact of changes that occur at one node do not flow cleanly from that node to adjacent nodes, as in epidemiological contagion.
This second workshop on these topics (the first being: Power Grids as Complex Networks: Formulating Problems for Useful Science and Science-Based Engineering, held at SFI in May 2012), will attempt to formulate the developmental necessities to create a network science that can handle the complexities of infrastructure networks, especially the power grid, and begin to tackle the related question of when a model of infrastructure is “simple enough” to be useful but not so simple as to be misleading. The basic question for this workshop is, generally: How do we rethink network science to increase its relevance to critical infrastructure analysis and protection while not increasing the detail level so much as to make the analytic system unwieldy?