Occasional Paper Series No: 4 - Intelligence analysis: what is it good for? The problem with probabilistic forecasting
There are a number of challenges associated with trying to measure the value of intelligence analysis. One current solution that has gained popularity is to focus on predictive intelligence, and to use statistical techniques to test predictions against the actual course of events. I will to demonstrate that this approach is not only fundamentally flawed in terms of method it is also dangerous because it gives priority to idle speculation about unknowable futures. I want to show that intelligence analysis is best measured by its ability to give decision-makers the broadest set of options and that its value resides in the outcomes of the actual choices that are made. My arguments and conclusions are based on two case studies: the Battle of Kursk during 1943 and the 2012 Malian coup. This approach will assist senior managers and strategic direction setters across intelligence agencies whose output includes predictive intelligence. It will be especially helpful to officials grappling with the problem of how to measure the quality and value of their intelligence analysis.