Integrating Nuclear Proliferation Modeling, Inference, Technologies, and Assessments: An Application to State-Specific Proliferation Risk Assessment Case Studies

Year
2011
Author(s)
G. A. Coles - Pacific Northwest National Laboratory
Alan Brothers - Pacific Northwest National Laboratory
Angela Dalton - Pacific Northwest National Laboratory
Amanda White - Pacific Northwest National Laboratory
Stephen Walsh - Pacific Northwest National Laboratory
Abstract
In this paper, we delineate an innovative approach to nuclear proliferation risk assessment in support of strengthening international security. This approach integrates Bayesian network (BN) model development, original graphical BN model diagnostics, novel tools for expert elicitation and for evidence attachment and evaluation, and empirical case analyses into a single compact assessment system. We describe each system component and illustrate their synergistic integration in six state-specific proliferation propensity assessment studies. Past proliferation risk assessment research has focused primarily on model development, identification of data collection needs, or assessment techniques without mitigating the fragmentation and isolation among these disparate pieces or integrating them to solve the assessment puzzle. In addition, statistical inferences and risk forecasting are often produced in the absence of procedural transparency. In addition to bringing the multiple components into one coherent framework, our approach dissects the analysis and inference into tractable mathematical steps and gives the whole process greater explicitness and clarity. The tool repertoire advanced through our research also contributes substantially to the enhancement of proliferation assessment technologies. Supporting international security through modeling nuclear proliferation risks and vulnerabilities poses persistent policy and scientific challenges. Our current research marks a promising step toward ameliorating these challenges. Future research may seek to address related modeling and methodological barriers by refining Bayesian models with more rigorous validation procedures, optimizing expert elicitation procedures and tools, exploring modeling proliferation from a “risk” centric perspective to a vulnerability-focused perspective, and generating new research insight from innovative modeling approaches.