[RAND's] ideological bent was disguised in statistics and equations, which allegedly made its analyses "rational" and "scientific." Abella writes:What Johnson says of RAND is typical of defense think-tanks. They tend to recruit academically minded Ph.D.s, whose understanding of the real world is off-center to begin with, and which remains almost completely innocent of the actual practice of statecraft and warfare. (See this post for a small sample.)
"If a subject could not be measured, ranged, or classified, it was of little consequence in systems analysis, for it was not rational. Numbers were all -- the human factor was a mere adjunct to the empirical."
In my opinion, Abella here confuses numerical with empirical. Most RAND analyses were formal, deductive, and mathematical but rarely based on concrete research into actually functioning societies. RAND never devoted itself to the ethnographic and linguistic knowledge necessary to do truly empirical research on societies that its administrators and researchers, in any case, thought they already understood....It is also important to note that RAND's analytical errors were not just those of commission -- excessive mathematical reductionism -- but also of omission. As Abella notes, "In spite of the collective brilliance of RAND there would be one area of science that would forever elude it, one whose absence would time and again expose the organization to peril: the knowledge of the human psyche."
Following the axioms of mathematical economics, RAND researchers tended to lump all human motives under what the Canadian political scientist C. B. Macpherson called "possessive individualism" and not to analyze them further. Therefore, they often misunderstood mass political movements, failing to appreciate the strength of organizations like the Vietcong and its resistance to the RAND-conceived Vietnam War strategy of "escalated" bombing of military and civilian targets.
In my experience (30 years' worth), the output of a defense think tank is seven parts theoretical, mathematical, and speculative (i.e., bull----) and one part empirical.