ABSTRACT
This essay begins by probing some of the more important puzzles that arise from, or that remain unanswered by, the most recent scholarship on US, Soviet, and Cuban intelligence in the missile crisis, and proceeds to consider the question, 'How should we assess intelligence performance?' As the Cuban missile crisis experience demonstrates, it is difficult to gauge intelligence performance both meaningfully and rigorously. Three mistakes commonly frustrate the enterprise: selection bias; privileging hindsight; and thinking of intelligence performance in terms of a 'rate of success'. It is more fruitful to evaluate intelligence performance over an extended period of time and against a standard of reasonable expected performance determined in part by past performance and in part by current technical 222capacity. Even more importantly, however, an evaluation of intelligence performance must take into account the policy side of the intelligence- policy interaction. The essay concludes with a discussion ofthe 'theory' and practice of intelligence assessment, and ventures a few procedural recommendations.
