This technical report put together by the Rand Corporation is remarkable for its understanding of arcane art of intelligence analysis and its thoughtful ideas on how to improve that art. It usefully divides intelligence challenges between `puzzles' that are solvable if only enough information can be acquired and `mysteries' that can be resolved only through intuition and target knowledge. This distinction has been one that co-author Gregory Treverton has long advocated. The book also divides the processes of analysis into three levels: 1) technical processing analysis; 2) single source analysis; and 3) all sources analysis. Perhaps most significantly the authors deplore the trend within the Intelligence Community (IC) to create `generalist' analysts rather than analysts with specific subject expertise. The logic behind this trend of course is that generalists can be moved about as needed from crises to crises as required. The folly of this concept has been repeatedly demonstrated most recently by the NCTC failure to "connect the dots" in the most recent terrorist incident (the Christmas Day underpants bomber).
The authors provide what appears to be an effective set of improvements to the analytic tradecraft, but their dissection of analytic processes, to this reader at least, comprise the most important part of the report.

