Office of the Director of National Intelligence

In the early stages of acquisition of strategic intelligence collection systems, we may distinguish between developing new capabilities or adding additional capacity. To inform this decision, we suggest the use of a routine analytic product: long time series of reported accuracy and confidence. We test this by simulating notional collection and reporting in situations of inadequate capability and capacity. Inadequate capability, particularly where possible alternatives are not acknowledged, is the most striking case with consistently high confidence and low accuracy. Inadequate capacity is more recognizable, with well-calibrated but widely varying confidence and low accuracy. The adequate collection is identified with well-calibrated, slightly varying confidence and high accuracy. Capability and capacity may be able to be traded with similar performance.

The new opportunities for social modeling are broad, and there will be an increased interest among government decision-makers in what can be learned. Yet the areas where the researcher's supply and the decision-maker's demand have the best match are limited and are not always in the areas favored by the academy. Translation and application of research findings to contemporary situations remains a challenge. Use-inspired research sets a direction, but phronetic social science clarifies some of the methodological changes that are required but not always welcome. Intelligence analysts have their own work to do, to frame situations as complex systems that are not reducible to immediate facts about individuals. In addition, social reality is often constituted by what people say and think, and the cognitive turn in social science can buttress a decision-maker's understanding that should no longer elevate material and behavioral phenomena over motives, ideology, narrative, social influence, beliefs, emotions, and the like. The literature on complexity should have a salubrious effect as well, by changing the expectations of decision-makers away from simple causes, prediction, gradualism, and repetition to a greater appreciation of threshold effects, contextual factors, interactions, path dependence, and uncertainty. In order to deliver on these perspectives and domains applied to specific cases, analysts will need to arm themselves with modeling. Yet models have their own dangers and have not always represented reflexive processes that, while confounding for the modeler, are realistic. Social science will improve intelligence and decision-making but only as it improves itself in train with human expertise and political judgment.

n/a