Krasnow Institute > Monday Seminars > Abstracts

Computational Policy Analysis, Robustness, and Resiliency:
Dealing with Deep Uncertainty, Incomplete Models, and Black Swans

Steven Bankes

There has long been a desire to bring the power and rigor of scientific analysis to public and corporate decision making. In spite of some important successes, “scientific policy analysis” has had difficulties, and the use of computer models has made these problems more acute. Specious analysis can result from incorrect implicit assumptions, exemplified by misleading risk calculations made prior to the financial meltdown. And suspicion of black box scientific analyses leads to rejection of rigorous work and failure to exploit what could be an important asset.

This talk will survey an approach to applying computational modeling to policy problems that are complex and deeply uncertain. Treating models as tools for inference rather than as means for prediction allows implications of knowledge they embody to be derived even when that knowledge is incomplete.  In particular, robust decision methods such as robust adaptive planning seek policies whose performance would be satisfactory across a broad range of future possibilities. These methods typically produce choice sets that can be characterized by differing vulnerabilities, and can be used to support mixed-initiative and iterative planning methods where humans and machines collaborate in identifying suitable decision options. The identified failure modes of robust candidates can be used to suggest adaptive extensions. And robustness analysis combined with adaptive mechanisms can provide an approach to devising resilient solutions.

This talk will both describe this approach to computational policy analysis and provide examples of successful applications.

Back to Top

The Krasnow Institute for Advanced Study
Mail Stop 2A1, George Mason University, Fairfax, VA 22030
Phone: (703) 993-4333 Fax: (703) 993-4325
Email: krasnow-webmaster@gmu.edu