ABSTRACT
Intelligence analysis has essentially relied on a process of group consensus to come up with a prediction of some threat or other foreign development of national concern, both within individual agencies and the Intelligence Community (IC). Uncertainties, to be sure, make threat or other intelligence prediction difficult enough, because they risk making predictions inaccurate. Compared to uncertainties management, managing analytic flaws that cause errors merits priority concern, because it is simply a more practical and realistic way to tackle the prediction problem in intelligence. Transparency poses a special challenge for computerized statistical probability analysis, because this analysis depends on sophisticated mathematical calculations that only math and computer whizzes truly understand. To be most effective, analysts should try to conduct alternative hypothesis testing that both uses multiple hypotheses and judges them by various tough criteria. Notwithstanding its benefits, alternative hypothesis testing still must contend with a powerful rival in intelligence for imposing quality control on analyses.
