Intelligence Tradecraft and Bayesian Models
Abstract
The US Intelligence Community (IC) has long been faced with the responsibility for assessing the relative likelihoods of multiple future scenarios—most of which are unlike those previously experienced. Intelligence analysis tradecraft currently involves a number of structured analytic methods: some used by both the IC and by the private sector, and others unique to the IC. These methods are designed to provide a framework that allows analysts to wrestle with complex causal chains and dependencies. However, policymakers and their private-sector counterparties increasingly require more detailed information about the most plausible future scenarios (not just the most likely), and their relative likelihoods. Increasingly, Bayesian models that provide such information are gaining favor. Bayesian models that utilize “all-source” intelligence derived from both structured and unstructured data might prove valuable in providing assessments of increased vulnerability to an event of national security import or even predict its occurrence, and assisting decision-makers in evaluating their options. This presentation provides an overview of issues related to using of Bayesian models in this Community, used to provide both situational awareness and enhanced predictive analytics. It also compares these methods and protocols to those of the financial community and discusses how each community might profit from adopting some of the methods of the other.
About the Presenter
Christina I. Ray Senior Managing Director for Market Intelligence, Omnis, Inc. McLean, VA cray@omnisinc.com