Coordination and Responsibility
APPROVED FOR RELEASE
CIA HISTORICAL REVIEW PROGRAM
22 SEPT 93
COORDINATION AND RESPONSIBILITY
R. J. Smith
In discussing the coordination of national intelligence it seems to me essential to recognize at the outset that coordination is certainly here to stay and probably will continue to be conducted pretty much along present lines. No amount of talk will either make it go away or alter its basic nature. This is so not because those people presently responsible for coordinating national intelligence are insensitive to visions of an ideal world where gentleman scholars would discuss world problems broadly and then retire to write individual appreciations. It is so primarily because national intelligence has become an integral part of the complex machinery for planning and policymaking of the US Government and has thereby acquired responsibilities not previously held by intelligence.
In the earlier and possibly more light-hearted years of CIA it was always a matter of some speculation as to who the users of national intelligence really were. We had a distribution list with names on it, but we had little evidence as to what happened once the estimates were delivered. We were in the position of shooting arrows into the air - some of them elegantly shaped and still bearing the tool marks of individual craftsmen - and having them land we knew not where. There was some fretting over this uncertainty, but it was balanced to a degree by an accompanying freedom in how we directed our effort. Coordination in those days varied. in its difficulty and its intensiveness almost with the moods and states of health of the participants. On one occasion, a coordination meeting would become almost a pro forma operation. On another, it might be the scene of sharply personal bickering and bad feeling, illuminated with sparks of verbal wit and showered with forensic displays.
Over the past five years this has changed. The broadening development of the centralized planning and policymaking mechanism has brought sharp changes in all governmental activities involved with problems of national security. National intelligence has been affected along with the rest. At the same time, national intelligence has gained strikingly in prestige and authority, partly as a consequence of its new responsibilities in policy and planning but also as a result of growing maturity and technical improvement throughout the entire intelligence community.
We no longer are in any doubt as to what use is made of national estimates. In a majority of cases, the customer (the National Security Council, one of its major members such as the White House, or one of its subordinate components such as the Planning Board) has given us specifications for the task and has set a date for its completion. If our customer discovers new specifications to be included, alterations are made before the estimate is completed; if he discovers his need has greater or less urgency than originally thought, the timing is adjusted. In all those cases where the policy and planning mechanism has originated the request, we know from the outset that the finished estimate will become the basis for a review of US policy toward the area or problem under consideration. We know this will be true also of a substantial number of other estimates which have been initiated through other auspices, including our own.
It is not new for intelligence to serve as a basis for policy. To greater or less degree, this has always been so and has provided intelligence with its reason for being. What is new is that this relationship has been formalized and institutionalized in such fashion as to make the connection far more direct and effective than ever before. Recognition throughout the intelligence community of the immediacy of this connection has profoundly affected both the estimates themselves and their coordination.
The present day national estimate bears only an indistinct resemblance to one of its remote ancestors, the literary or scholarly essay. In the days of our youth the resemblance was more apparent than it is today, and it continues to be considerably more apparent in British national intelligence papers, known as "appreciations." (It may not be significant but it is at least interesting that for us the word "appreciation" carries connotations of artistic endeavor and to the British the word "estimate" conveys a mechanical totting up, not unlike the estimate the plumber provides before beginning work.) It is inevitable and proper that some readers, bringing to bear primarily the standards for literary or scholarly essays, should criticize the national estimates for general lack of reader appeal. It is perhaps also inevitable but considerably less proper that they should simultaneously place the blame for this condition entirely on the process of coordination.
National estimates are not scholarly essays. They are primarily work papers for planners and policymakers. This does not mean that these papers need be unreadable, or that they cannot be more readable than they sometimes are, but it does mean that they must be the embodiment of precise writing. Anyone who has ever tried to write really precisely - so precisely that several different groups of planners can get exactly the same content from a statement of fact or a judgment - knows that in order to reach such precision one must boil off nearly all the esters of personal flavor and strive for a flat objectivity. Also, in this connection, one must bear in mind that the planners and policymakers in question are high level and have neither the time nor the necessity to master enormous quantities of detail. They need only that amount of detail necessary to support the handful of key estimative judgments to be made about the situation before them.
Having said this much, let us look more narrowly at the impact of coordination upon these national estimates. First of all, let there be no mistake about the necessity for coordination. Many criticisms of the present coordinated estimates represent an attempt, in one guise or another, to squirm away from this necessity. It may be true that one individual, or a small group of talented individuals, could on many occasions write estimates with sharper edges than coordinated estimates, but the difficulty is that such estimates would not meet the need of the White House and the National Security Council. What the highest levels of the national government most emphatically do not need is a batch of estimates on the same subject by separate intelligence organizations, each paper out of key with the other in exposition, emphasis, and conclusion. This situation would merely pass responsibility for the ultimate intelligence judgment on to the policymakers. What they require instead is a single document which contains the collective judgment of the intelligence community, an estimate which delineates the areas of general intelligence agreement and identifies where necessary the points of major substantive dissent, an estimate to which all the chief intelligence officers of the national government will concur. Looked at from this perspective, the coordination process becomes the heart of the matter, not an unnecessary evil. Its characteristic defects and its burdens become problems to be worked with and to be eased, not avoided. In fact, looked at from this angle, one can even recognize that the coordination process has benefits and merits in its own right.
Knowing as they do that the finished national estimate will become the basis for a policy which will vitally affect the mission and responsibilities of their department, the representatives of the various intelligence agencies take the coordinating sessions seriously. As their departments' spokesmen, they have a deep and responsible interest in seeing that the final estimate does not ignore information available to their department or does not arrive at judgments contrary to the views of their departmental intelligence specialists and chiefs. At the same time, they must avoid damaging the prestige and integrity of their department by pushing departmental views in defiance of contrary evidence or by failing, to inform their department of the extent to which its view stands in isolation from the rest of the community.
The CIA responsibility in this process is different in kind but equally great. In the first place, the draft discussed by the coordination meeting is a CIA draft based on written contributions from the several departmental agencies. These contributions, frequently longer individually than the finished estimate, are rich in detail and analysis and provide a broad base for the estimate. The CIA drafters synthesize these departmental papers into a single estimate, making such augmentations or changes in analysis or emphasis as they think the objective situation requires. When this draft, well-tested within CIA, is placed before the coordination meeting, it has its own inner cohesion and strength. Like all well-constructed and ramified pieces of writing, its built-in inertia makes it hard to move very far. It responds gently to nudges but resists hard shoves. Moreover, it has the support and protection of the CIA representatives, including the chairman, who, though ready to accept suggested improvements and useful additions or corrections, are quick to challenge estimative changes unsupported by sound evidence or objective reasoning. The national estimate which emerges from this intensive coordination has been thoroughly stretched and tested but most times has not been altered fundamentally. On those occasions when. deep-reaching changes have been made, the CIA representatives have become convinced that these changes would produce stronger, sounder estimates.
A common complaint about coordinated intelligence - or coordinated anything for that matter - is that it merely represents the lowest common denominator of opinion. In the light of the discussion above, the only accurate rebuttal to this charge as it applies to national estimates is that it is not true. It is true that some degree of compromise is nearly always involved in the effort to reach full agreement. Short of going to war, no method other than compromise would appear to be available for reaching written agreement on really complicated matters. This is all the more true in the realms of judgment and future projection where national estimates must necessarily operate. Intelligent and responsible compromise is an essential tool in the coordination process, but, by definition, intelligence compromise does not include adding buckets of water to sound judgments merely to obtain agreed positions. The avenue which enables us to avoid this undesirable result is the dissent.
Beeping in mind that the primary mission of national intelligence is to provide the White House and the NSC with agreed estimates, it ought to be apparent that a national estimate laden with dissents would not fit the requirement. By the same token, however, an estimate which glossed over, or compromised out of existence, legitimate and fundamental divergences would not meet the requirement. One does not want to confront the President or the Secretary of Defense at every turn with unresolved differences which force him to make his own choice. At the same time, one does not want to paper over substantial divergences and let him believe no differences of view exist.
One must realize, however, that dissents are not easily contrived. First, the actual substantive difference must be isolated and the dissenter convinced that his is the dissenting and not the majority view. Then he must accustom himself to the notion of standing naked and alone in a footnote with his peers arrayed against him in the main text. Each of these stages is invariably accompanied by surges of new conviction on the part of the dissenter that his position is the right one, after all, and that one more try will convert the rest of the group. In short, the trickiest and most vexing problems in coordination revolve around the point at which the quest for agreement should be abandoned and a clearly defined dissent should be prepared. But to say it is hard is not to say it cannot be done. To prevent enforced coordination, statements of dissent are employed now as often as the skill of the CIA coordinators can bring them about. Growing maturity among the intelligence community will probably make this an easier result to obtain as time goes on.
Another common complaint about coordination is that it takes so much time the estimates are no longer fresh when they are produced. In actual fact, this criticism has less validity than almost any other. No one involved in producing national estimates would deny it takes time. Papers involving special research problems or new techniques have taken as long as ten months. Routine estimates commonly take six to eight weeks. On the other hand, the IAC machinery has produced a coordinated national estimate in five hours and has on several occasions produced them in 36, 48, or 72 hours. At first glance, in a world where the daily newspaper is regularly scooped by television, six to eight weeks, let alone ten months, seems an unconscionable amount of time. Even five or forty eight hours seems long. Viewed from the perspective of operational or current intelligence, it probably is a long time. Viewed from the perspective of planning national strategy, it is not. A number of our estimates project forward five years because it is necessary for some kinds of policy planning to look five years ahead. Nearly all the estimates project at least a year ahead. Against this time span, the time taken to produce them does not seem long. To put it another way, an estimate which could not withstand the passing of a mere eight weeks could scarcely serve as the basis for planning a year or five years ahead.
But whatever view one has about the right length of time to spend producing a coordinated national estimate, the remarkable fact is that the coordination itself - the time spent in meetings resolving differences in views and obtaining an agreed text - takes only a small fraction of the total time spent. A study of twenty-four planned and routine national estimates, the longest taking 285 days to produce and the shortest 62 days, discloses that the average time actually required for coordination meetings was under ten percent. The remainder was spent in the preparation of terms of reference, research, the preparation of agency contributions, and the writing and reviewing of the draft within CIA. Even this low percentage figure does not tell the full story because it includes estimates on such matters as Soviet gross capabilities, where weeks of meetings were held to work over the complicated evidence underlying detailed strength figures and capabilities estimates. A more representative figure for coordination meetings would be between one and three days, most commonly two.
Is one led inevitably by this discussion to the conclusion that the necessary art of coordinating national estimates is in a perfect state? The answer is certainly no. As in all goodsized meetings, both within government and without, progress in coordination sessions is frequently slow and uncertain. Too frequently, those who know the least talk the most. Even worse, on some occasions one of the participants may be virtually devoid of substantive grasp. Sometimes, persons with a fair understanding of the substance under discussion come so rigidly instructed regarding a certain point that discussion of it is futile. Almost always, there is a tendency among the participants to commit that fundamental but all-too-human semantic error, that of identifying the word inexorably with the thought: Thought A can only be expressed by Word A.
What is the remedy for this state of affairs? What can be done, particularly when much of the difficulty is inherent in the method? Can we overcome the fundamental inefficiency of the committee meeting, that peculiarly American contribution to the arts of governing? Well, certainly not, but we can exploit fully our growing technique in running meetings, extracting from them their maximum value as the creators of new perspectives and holding to a minimum their nonproductive aspects. Can we elevate semantic understanding and sophistication to such a level as to remove this most frequent barrier to agreement? Again, no, at least not all at once, but we can recognize this shortcoming in ourselves and thus contribute to greater flexibility in achieving a solution.
In short, the path to improvement of the coordination process lies not through the imposition of ideal solutions but through gradual, slow advance by small adjustments here and there. We can obtain better quality of representation at the coordination meetings. There is, in fact, perceptible progress in this respect over the past several years. The advantages of sending representatives with substantive understanding and empowering them with a fair degree of latitude in negotiation are already apparent to most of the IAC agencies. We can achieve a higher degree of group responsibility and freedom from partisan attitudes as maturity increases. Moreover, we can adopt various innovations in procedure as they seem desirable. We could, just for example, ask the IAC agencies to send representatives to participate with us in the drafting sessions on certain occasions in order to speed the process and facilitate agreement. But whatever we do, we cannot - as I hope I have made clear - do away with the coordination process. It is the heart of national intelligence. To make it tick strongly and surely is our problem.