Preface to a Theory of Intelligence
CIA HISTORICAL REVIEW PROGRAM
RELEASE IN FULL
22 SEPT 93
The Office of Training on 22 July 1974 established the Center for the Study of Intelligence, a framework within which discussion groups and researchers can pursue investigation of the theory and practice of the profession of intelligence. Fellows in Intelligence, appointed by the Director of Training for terms of various lengths depending on the scope of specific research projects, will deal with a broad range of issues having to do with the ways in which intelligence is gathered, processed, and used. Discussion groups sponsored by the Center, involving both Agency and non-Agency participation, will meet to discuss questions of long-range importance to the Agency, both substantive and procedural. The research product and reports on the work o f the discussion groups will be given appropriate distribution to inform and to stimulate further thought about the problems.
The aim, as with Studies in Intelligence, is to develop a professional literature of our discipline which will contribute to the growth of a systematic body of knowledge about intelligence. The accompanying article, examining the rationale underlying the establishment of the Center, echoes an article by Sherman Kent in the initial September 1955 issue of this publication entitled "The Need for an Intelligence Literature."
PREFACE TO A THEORY OF INTELLIGENCE
Lawrence T. Mitelman
It is an obvious but essential point that the question of why we collect and produce intelligence precedes the question of how. In a world of increasingly constrained resources, the clear answer to the question of why must be that the intelligence is in the most serious national interests. Those interests, as defined and made explicit by the political leadership of this country, are the only justification for the collection and further processing of information. We no longer live in an age tolerant of the "nice-to-know"—need-to-know has assumed an additional meaning for us. But the effort to develop standards to guide intelligence with reference to both the why and how is conceptual as well as practical. Theory, however assimilated and unexamined its assumptions may be, precedes practice. A theory, as I shall use the term rather broadly, is a well-coordinated conceptual system which requires a rigorous statement of premises, assumptions, and relationships to give order to a body of observed data or empirical situations. Clearly a theory has other characteristics as well, but at this early stage I would much prefer to emphasize the process, not the product; to stress the effort to make general statements about relations rather than the statements themselves.
Why a "theory of intelligence?" Or rather, several or even many theories of intelligence? Why the effort to be explicit about assumptions, variables, and relationships when ambiguity, particularly in a bureaucratic context, may be more acceptable and less conflict-producing? There is, after all, a well-researched and persuasive literature arguing the many values of ambiguity and ignorance. There is recent evidence from the world of international diplomacy that ambiguity, even in the most intricate of negotiations with seemingly unambiguous outcomes, has proved useful. Not to erect too many straw men, one must distinguish between conscious or intentional ambiguity and lack of clarity in one's own mind. The decision to be vague for a purpose differs fundamentally from vagueness that proceeds from uncertainty about purpose. Again, why precedes how, and theory — whether identified as such or not — precedes practice. One justification, then, for theorizing about intelligence is to encourage clarity of thought about assumptions and explicitness about purposes. The effort is far from frivolous. Consider, for example, the following passage from The Real CIA by Lyman Kirkpatrick:
"The role of the intelligence services in the future is, and should be, that of keeping the policy makers fully informed of anything happening (or about to happen) in the world that might affect the United States politically, economically, or militarily, either directly or indirectly. ... What our intelligence services must learn is what weapon any power on earth is planning to build. ... Until there are guarantees of a safe disarmament, the only margin of safety in national defense is 100 percent accuracy in our knowledge of other countries' weaponry." (p. 282f)
This passage, from a book which has been described as constituting "a resounding defense of CIA's role in our society," could probably command the assent of many at work in intelligence today. Examine for just a moment some of the crucial assumptions which underlie this view of the proper role and scope of intelligence activity. One need only ask what kinds of events abroad in the world do not affect the United States politically, economically, or militarily, either directly or indirectly, to know that this is an impossibly broad charge. The response will come: "we all know what he means." Perhaps. It may be that he means just what he has written — what then? The object is not to belittle a point of view, it is to highlight the assumptions implicit in that view. Is 100 percent accuracy in our knowledge of other countries' weaponry really necessary, or desirable, or attainable? Is, in fact, such accuracy "the only margin of safety" in national defense?
It is not necessary to belabor the issue, but to emphasize once again, the effort to theorize encourages explicitness about assumptions and relationships. There is a further point to be made. Without some theoretical apparatus, it is immensely difficult to establish standards of relevance or levels of priority. What among a flood of impressions and data is pertinent to an inquiry, and investigation? Why? What collection systems are preferred to what others and why? How have the increasing sophistication and reliability of technical collection affected the need for non-technical collection? Much is made of the distinction between capabilities and intentions. Is there a threshold of capability below which intentions do not matter? To address such questions successfully requires that one have at least some theoretical context at hand. Lacking it, one is consigned to almost random observation and decision. Observation and experience are far more productive when guided by intelligent hypothesis.
To argue the need for theorizing and eventually some theories about intelligence is not to urge the need for dogma or binding doctrine. It is to urge, and very strongly, the need for a dialogue on doctrine, for the venting of complementary or even competing theoretical constructs. There has been too little of that, and we are the poorer for it. The literature on intelligence is, for the most part, anecdotal or case-oriented and little given to theorizing about the nature of the processes involved in intelligence work. What we lack, even given the serious and sustained work of the editors and many contributors to Studies in Intelligence, is a cumulative, critical literature, a literature from which people may learn and against which they can react. It was just a very few years before the founding of OSS that John Maynard Keynes wrote his epochal General Theory of Employment, Interest, and Money. In the years since, economic theory has burgeoned, spawning in the process a welter of derivative and counter theories and, also in the process, sharpening, honing, growing in policy usefulness. By contrast, since the publication of Sherman Kent's Strategic Intelligence for American World Policy there has been almost nothing of comparable intellectual merit or persuasiveness written about intelligence. Admittedly, intelligence officers have had much to do since OSS days, while the generally overt character of data for economists has permitted many to work in that field. But is there no one among the many gifted people who have worked at and thought about intelligence over the last three decades who is inclined to advance some general propositions about the processes and purposes of intelligence? With a flippancy I think he might enjoy, I could have called this paper, "Is Sherman Kent Enough?"
The theoretical explorations I suggest could take many forms, depending upon the disciplinary interests of the researcher. Two brief examples, from economic theory and decision analysis, hopefully, will give some sense of how one might proceed and what the policy relevant results might be.
In the economic example, consider the intelligence community (or, more modestly, the CIA) as a firm engaged in producing a product. By definition, production is held to be the transformation of one commodity into another. There is no requirement that the commodities be tangible. Orchestras as well as factories produce. The question arises, then, "What does CIA produce?" Surely not paper, or reports, or even advice. It can be argued that what CIA and other intelligence organizations produce is certainty. That is, it is the task of intelligence to reduce the uncertainty attending situations and options of interest to policy makers. Assume that the product is produced by some combination of two factors, capital and labor, one or the other of which may be fixed. Capital includes the cost of technical collection systems as well as the equipment needed to interpret and process data. Labor includes not only the labor of human collection but also the labor required to process information gathered by technical means. Assume further that capital is fixed — given anticipated budgetary constraints, this may not be too pessimistic an assumption. What would happen then if labor were increased in production? According to economic theory, the relative contribution of labor in the production process declines as more labor is used. This is no more than a statement of the "Law of Diminishing Returns." What relevance does this analysis have for the management and allocation of resources for intelligence gathering and production? At best, given this grossly oversimplified statement, the analysis points to the possibility that greater infusions of labor into the process will not necessarily results in proportionate increases in production. There may be some more optimal mix of the two factors. Too, diminishing returns and the corollary, increasing marginal costs, suggest that we will be paying ever more for a product constantly decreasing in cost effectiveness. One policy prescription which might emerge is that collection and production must become vastly more selective. Returning to Mr. Kirkpatrick's assertion that "the only margin of safety is 100 percent accuracy in our knowledge of other countries' weaponry," it can be argued that given the above, the cost of even approaching 100 percent accuracy may become prohibitive. At what level less than 100 percent do we establish the "margin of safety" and why? These are issues and questions the theorist about intelligence could profitably address from one disciplinary perspective.
The above illustration is drawn from a field, economics, which falls within what has recently been labeled the "analytic paradigm" of decision theory. According to this paradigm a decision maker knows his goals, the constraints he is subject to, the resources at his disposal, and has some ranking of available options which will yield maximum positive results. Given this formal design, the problems of aggregating individual choices and preferences become enormously complex. The analyst of foreign affairs who is necessarily concerned with explaining and predicting the behavior of foreign states when working, consciously or not, within the rules established by the analytic paradigm is impelled toward a view of affairs which is frequently misleadingly rational and coherent.
Increasingly in the literature of the social sciences there are alternatives to the analytic paradigm being discussed, a new set of conceptual lenses which, proceeding from different assumptions, yield fruitful new insights into problems of decision making. This developing new paradigm stresses procedures, not outcomes. It asserts that the decision maker, in the words of a recent paper on the subject, "strains to avoid direct outcome calculations and thus to eliminate the impact of uncertainty" and "is sensitive only to a limited range of highly pertinent information." Interpreting the behavior of foreign states through this model, particularly during times of crises, could produce starkly different results from the analytic paradigm and different missions for intelligence in the process.
It would be pointless to ask which of these two ways of inquiring into decision making is correct. Rather we should ask how might they complement each other in intelligence work to convey a richer understanding of foreign state behavior so that intelligence may better serve the needs of policy.
These two examples are merely that. The would-be theorist about intelligence is by no means limited to the world of the economist or decision analyst. It is far more likely that other disciplines, more abundantly represented in the Agency's ranks, will contribute to this effort. There will be, in the parlance of the political scientist, islands of theory awaiting some grand concept to connect them. But without the islands, it is unlikely that the concept will have the power to explain and predict that may one day be possible. What is essential is that the experience in the workings of intelligence gained in the last three decades not be permitted to persist only as an undifferentiated mass of data and impressions. If we are to learn, if we are to grow professionally from one generation to the next, then we must begin now to look for the patterns and general lessons of our profession.