Seminar Announcement

A Characteristic Network Approach for Modeling Protest Durations

  • Speaker: Dr. Brian Goode
  • Discovery Analytics Center Virginia Tech
  • Date: Friday, Nov. 6, 2015
  • Time: 1:00pm - 2:00pm
  • Location: Room 325 (NVC)


Protests and civil unrest events carry high societal impact, and are examples of complex systems exhibiting collective behavior within a population. We find that protest behavior across countries show differences in the duration of periods characterized by protest and non-protest activity. We posit that these differences can be explained with a characteristic network that models the social structure of interactions within a country. This paper explains the dynamics governing protest and non-protest durations as a phase transition process of a characteristic network. The network is complete and described with only three parameters: network size, convergence rate, and edge strength. The likelihood of the network transitioning into or out of a protest state is a survival probability distribution (SPD). Using the Fokker- Planck equation, we calculate the SPD with a mean-field approximation of individual actions occurring within the complete network subject to random fluctuations in a double-well potential. The results show that the distributions produced by our model converge to the distributions of collective behavior durations in the protest data from the ICEWS database for many countries. We also show evidence that the distributions of protest and non-protest durations are asymmetric with respect to the double-well potential.

Speaker's Biography

Brian Goode is a research associate with the Discovery Analytics Center (DAC) at Virginia Tech (NCR - Ballston). Brian's research interests are to better understand complexity, uncertainty and what constitutes elegance in the models used to describe such phenomena. He received his doctorate in mechanical engineering from Virginia Tech in 2011, where he performed urban gunshot localization with acoustic beamforming and played hide-n-seek for the Office of Naval Research (ONR) using a differential game theoretic approach to navigation, guidance, and control of autonomous vehicles. As a member of DAC, Brian applies these same control theoretic and signal processing concepts to "big data" problems such as predicting the virality of an idea or "meme" in terms of user participation on a social network, investigating characteristic behavior of protest populations, and evaluating metaphorical usage of South American bloggers. Brian encourages experience through immersion having studied improv comedy, received an Australian hang gliding license, taught school in rural Thailand and bicycled across the continental United States.