Foreign Relations of the United States, 1969–1976, Volume II, Organization and Management of U.S. Foreign Policy, 1969–1972

206. Memorandum From the Consultant to the National Security Council (Marshall) to the President’s Assistant for National Security Affairs (Kissinger)¹

Washington, May 1, 1970.

SUBJECT

• Intelligence Inputs for Major Issues: A Substantive Evaluation and Proposals for Improvement

When we first talked, you indicated a concern for the quality of the intelligence product you receive on major issues such as the SS–9 program, trends in Soviet strategic forces, etc. During the last two months I have focused on the intelligence product on the Soviet military strategic weapons and forces and the question: How can you get a better product in the future?

The standard products of the intelligence community do not give you and your staff what you want and what you need. The NIE’s are of little use to top level decision makers and/or their staffs, even though that is their ostensible audience. Their real service is that of supplying an agreed intelligence input to the work of staffs several levels down in the bureaucracy, and as a starting point for the NIPP (now the DIPP, Defense Intelligence Projections for Planning).

Other products vary in usefulness. The new President’s Quarterly Report on Soviet Strategic Forces is factual, concise, and well presented. Some of the Caesar series² provide valuable background on Soviet leaders, the structure and functioning of the upper level bureaucracy. But most of the product, when it goes beyond the reporting of factual information, or immediate inferences from it, is not very impressive.

The analysis of hard data and factual reporting on Soviet forces is good; indeed, we now know a great deal more about today’s Soviet military posture and R&D programs than we knew about the 1960 Soviet force posture and programs in 1960. Intelligence on Soviet forces and programs is better today than in the past; but it can still be improved. Intelligence reporting and analysis can and should do a better job of assisting top level decision makers.

The weakest point I find is in the judgments of intelligence analysts and estimators about plausible or likely Soviet behavior, in particular their understanding of the decision processes that influence Soviet military posture. The explicit or implicit assumptions and hypotheses concerning the roots of Soviet behavior seem much too simplified, and rely too frequently upon a model of the Soviet government as a single unified actor pursuing an easily stated strategy.³

Presumably the governmental decision-making process there is just as complex as ours, involves the interaction of contending bureaucratic elements, and can attain only a limited measure of rationality. None of this shows through in the standard intelligence product, except in those paragraphs designed to protect against future developments falsifying the estimate or judgment. These include sentences listing the factors that may also influence future Soviet behavior: economic difficulties, bureaucratic conflicts, bloc political problems, etc. A form of defensive writing in the spirit of defense driving.

The fact that intelligence analysts’ judgments about likely Soviet behavior do not seem that much better than those of less involved persons is disappointing. In principle, they should be the real experts, and in some ways they are. But I have long felt that intelligence analysts have not devoted enough effort to studying past Soviet behavior with regard to military posture formation; have not sufficiently focused upon understanding the structure and objectives of the various organizations involved in the relevant decision-making processes.

In my view, if we are to understand past Soviet force posture decisions, or to improve our forecasts of alternative future force postures, we have to entertain more complicated hypotheses about the sources of Soviet behavior regarding military force posture formation.

Substantially improving the intellectual quality of the analysis of Soviet behavior is a longer term goal. I hope some effort can be made to push forward in this area. But let me return to the more immediate problem of getting you a better product.⁴

What Do You Really Need?

It is hard for me to answer that question completely. Only you, Larry Lynn, and others immediately concerned with specific decisions and problem areas can do it. However, I would suggest that on a few issues each year

• —where a great deal is at stake,
• —where there are contending views on which option to choose,
• —where major uncertainties almost certainly exist as to the future evolution of Soviet strategic forces,

you need a different sort of intelligence product than you now get.

For at least these few problems you need in-depth intelligence studies that surface the uncertainties, display and argue alternative explanatory hypotheses regarding past Soviet behavior and future projections. These studies should give you

• —as much separation of facts and judgments as is possible,
• —where major judgments are made, argument as to the basis of the judgment.

On these really important issues you should dig into the intelligence analysis as deeply as you can before making decisions. You have to understand what is behind it before you can accept it. The only alternative is boxscoring of experts. In general that is not a feasible procedure.

What Can Be Done?

A number of things can be done to get you better, more useful products. I believe you ought to use the following strategy:

• —Improved communication of your and your staff’s needs. You are not getting through loudly and clearly now.⁵ Clear and persistent statement of needs should lead to an improved product. Put the burden on Helms and the community to find the ways to satisfy you.
• —Initiate discussion with Helms aiming at a major review of the intelligence community’s support of yourself, the NSC decision-making process.
• —Develop new procedures to get non-standard products now for a few selected problems of highest importance to you.

Specifics of the strategy are covered below. Note that it is designed to get a better product for you, not to improve the structure and functioning of the intelligence community in the short run. It attempts to bypass, for the moment, the probable sources of the problem. You might prefer a more intrusive strategy that tries to influence the structure and functioning of the community at an early date. If so, see Tab A.

Better Communication of Your Needs

One general observation to begin with: Causes of product deficiencies lie on both sides of the producer-consumer interface. Top level needs have not been expressed clearly or persistently enough. There is little feedback or criticism of the intelligence product.

The community misperceives some of the needs of top level people, and a doctrine that limits their response. Moreover, the intelligence community does nothing that could be called research on customer needs. The organization of the interface between the two groups does not facilitate communication of customer needs, and discussion of how to match needs and producer capabilities.

I recommend the following:

1. Preparation of a statement to Helms expressing your needs (see Tab B for some things it should say), coordinated with guidance on the format and content of the Soviet Military NIE’s. (See memo on that matter from Larry Lynn and A. W. Marshall.)⁶ Probably it would cover some of the same ground, but discuss alternative ways of getting a more useful product as well. Repetition will not hurt. Do not underestimate the communication problem. Follow up with detailed discussion with Helms and others on how to get the new procedures going.

2. Use at least one person on your, or the NSC, staff full time as a communication link, mainly to CIA. Tom Latimer is coming on board. Consider setting some part of his time aside for this function. Another person might be added to the NSC staff solely to perform this communication function. I can suggest some names if you are interested. Principle characteristics should be a familiarity with all community organizations, and a personal disposition to consider intelligence as a service to consumers, not an activity with its own goals.

The communication function will be time-consuming, if done right. It is not the standard sort of liaison function that is needed. Frequent trips to CIA, and elsewhere, and extended meetings with people at all levels in the community will be required to keep their view of your needs correctly focused.

Helms as DCI and CIA are the key communication targets. CIA has the best current capability to respond; there is just no where else to go. But you should communicate your needs also to DIA (Defense) and INR (State).

Again I stress the difficulty of the communication problem. The procedures to obtain non-standard products, discussed in the section after the next one, are designed to assist the communication problem; indeed, these procedures should be thought of as part of the communication effort.

Steps to Initiate a Major Review of Intelligence Support of the NSC Decision Process

While short-run measures are taken to obtain more useful products, a more basic look can be taken at intelligence community organization and functioning in support of you and the NSC decision process. The timing for such a review may be good. For example:

1. The Office of National Estimates (ONE) is going through a transition period. It is at the end of an era. Abbott Smith, the head of ONE, will retire as soon as a successor can be picked. Helms and others lean toward bringing in someone from the outside; this is an opportunity to add some new blood and leadership. But equally at stake are: What sort of organization should ONE be? What should its role be? How should the NIE process operate?

You have a major stake in the choice of the successor and the DCI conception of what sort of an organization ONE is to be, what sort of a role ONE is to have in the future. (See Tab C.)

2. The Blue-Ribbon Panel will report on Department of Defense organization and management on 30 June 1970. It will focus in part on the need to change Defense intelligence organization and management. You have a major stake in what is decided, in particular as regards the future development of DIA. (See Tab D.)

I think you will want to rebuild the national intelligence process. At present it is foundering because of the decline in the ONE/BNE role and status, reflected in the virtual DDI monopoly of the intelligence role in the NSSM process, and other causes. One view of what to do about the national process is contained in Tab A. But what is really needed is a full-scale review of the current situation and recommendations for change. The review or study group should include representatives of the intelligence community, of NSC members, and of the NSC staff. It is very important that consumer representatives as well as intelligence representatives be involved in the review.

In the nature of things, the national process, if it involves inputs from several components of the intelligence community, is an adversary process. Special attention will have to be paid to designing a process that works well. More attention than in the past must be given to structuring the incentive systems in the adversary process.

I suggest you begin discussion with Helms about the design and procedures for a review of the national intelligence process. The aims of the review would be an assessment of its current operation in support of your office, and the NSC decision process; and recommendations for future redesign of the national intelligence process.

New Procedures to Get Non-Standard Products

You need not only to communicate your needs, and hope for a good response, but to develop procedures to get what you want now.

I recommend that you:

1. Limit efforts to improve the NIE’s. Neither the process that produces them nor the performance of ONE/BNE can be changed in the short run. The solution to the NIE problem is part of the review effort.

2. Push for procedures to produce in-depth intelligence studies on a small number of selected intelligence problems each year. Selection of problem areas to be yours, perhaps in some cases in conjunction with the Secretaries of State and Defense.⁷ (See Tab B.)

The essential features of these procedures should be:

• —Involvement of top level decision makers and/or their staffs in the selection of study areas, drafting of terms of reference and the goals of the study.
• —Provisions for monitoring of the study as it proceeds and continued guidance and feedback from upper level people to all levels in the intelligence community.

Joint decision of upper level representatives and intelligence working level people concerning modification of study efforts to accommodate data and analysis problems and in-course redirection of study.

It will be very important that it not seem that the White House is writing its own intelligence estimates. The objective should clearly be to obtain from the intelligence community relevant facts, judgments, etc.

3. Continue a study of the SS–9 system initiated 1 April 1970 (see Tab E). It is an attempt to produce a non-standard product; one you or Larry Lynn might give a good grade. Projected completion is end September. It will take only a day or two per month to follow it and hopefully keep it going in the right direction. CIA has started a good effort in this study. It should be a good test of their current capabilities to explore some more complex, organizational behavior hypotheses in addition to the standard ones.

I plan to continue to manage this effort as I visit Washington periodically in the course of other work.⁸

Constant attention will have to be given to see that the procedures that are developed continue to function. The intelligence bureaucracy at all levels may resist these methods of operation. No fixed set of procedures may work all of the time. The recent study of the Israeli-Arab military balance, while not a typical intelligence study, may be a good model from which to draw some lessons.

In the case of that study, the keys to success appear to have been:

• —Study confined to fact finding, technical study, policy implications played down;
• —Full-time involvement of a NSC staff representative (in this case a consultant);
• —Lots of feedback of specific questions as study progressed;
• —No strong bureaucratic stance of State or DOD/Military Services.

By contrast, a 1968 attempt to evaluate the Israeli-Arab military balance produced only badly split views expressing policy preferences.

In any case, almost everyone’s experience is that the most rewarding and fruitful way of working with the intelligence community is one in which top level people deal more directly with the working level people than is usual. Both the people and the intelligence input at the bottom are better than the standard product.

Postscript

Many changes and improvements in intelligence community performance I would like to see are not easily effected by you. For example, as mentioned earlier I believe that major improvement in the analysis of Soviet decision-making processes is possible. But progress is slow and difficult to stimulate from the outside on this and many other areas of possible improvement. Nonetheless, I have appended at Tab F a short sketch of a number of areas that I feel the community should be doing more about. They mainly concern [what the] R&D community could do on the intelligence analysis, estimating, and projection processes. Two substantive studies are also briefly described.

1. Source: National Archives, Nixon Presidential Materials, NSC Files, Name Files, Box 825, Marshall, Andrew, Vol. I, 1969–1971. Top Secret; Sensitive. Sent for action. The tabs are attached but not printed.
2. In the margin Kissinger wrote: “What’s the Caesar series?”
3. Nixon drew parallel lines in the margin next to this paragraph.
4. Kissinger wrote an illegible word in the margin next to this and the previous sentence.
5. In the margin next to this sentence Kissinger wrote: “On what issues.”
6. Not found.
7. Kissinger put a checkmark in the margin next to this paragraph.
8. Kissinger initialed the “Approve” option.