|
|
Keynote Speech
The not so happily ever after - On the use and usefulness of social simulations
|
Dr. Seyed M. Mussavi Rizi |
Email: Abstract
In their quest to develop social simulations with high exploratory, predictive or explanatory powers, modelers tell stories of developing, verification, calibration and perhaps validation of social simulations, but seldom tales of making them useful save hinting at possibilities for happily ever after unions of users and social simulations. This build-and-forget practice has pushed to the sidelines development of tools and procedures that make social simulations useful. In this talk, I will pick up the story of social simulations and their users where it is often left off and outline an approach to solving the usefulness problem based on a simulation model of the Syrian conflict.
In the first part of the talk I will discuss the properties of toy models and empirical models as prevalent types of social simulations. Toy modelers subscribe to simplicity as the key criteria for useful simulations, hoping to produce new knowledge by relaxing the assumptions of dominant analytical models. To them model users are modelers and experts who speak different dialects of the science language. On the other hand, empirical modelers subscribe to representational correspondence to reality as the yardstick of quality, incurring time and money premiums to enhance the fidelity of simulation inputs and outputs to reality. In theory they aspire to build simulations that solve other people’s problems. Yet, curiously they have not developed interfaces that make simulations accessible to users, be it the common man or a sophisticated decision maker.
In the second part of the talk I will describe a nascent effort to make simulations a consumer-centric enterprise by discussing a simulation of the conflict in Syria that represents households, political alliances, and militias and their fighter units, and generates not only refugee and IDP flows, militia sizes, and political control over and force presence in neighborhoods, but rich narratives of the ebbs and flows of the conflict. The challenge here is not merely visualizing outputs, but designing workflows that enable users to interact with the simulation itself in ways that make exploring the outputs intuitive.
Curriculum Vitae
Seyed M. Mussavi Rizi is a partner at Scensei GmbH, a Swiss analytics firm. He has a Ph.D. in computational social science from George Mason University along with graduate degrees in economics from Tufts University and international relations from the Fletcher School. Dr. Mussavi Rizi has worked extensively on algorithms for fusing social networks, geospatial and remote sensing data into simulations, resulting in social simulations for technology evaluation, economic development, counternarcotics, humanitarian response to war induced displacement, and effective programming for long-term refugees. His recent work include simulations of online behaviors and cognitive architectures to capture pathways of influencing online and offline behavior, and designing workflows for analysts and decision makers who wish to use simulations for decision support.
