174. Memorandum From the Director of Central Intelligence (Bush) to the Chairman of the President’s Foreign Intelligence Advisory Board (Cherne)1


  • Recommendations of Team B—Soviet Strategic Objectives

1. I am forwarding to you the attached series of B Team recommendations on how national intelligence estimates should be produced, as we agreed in a previous discussion. We must, obviously, always examine suggestions such as these from experienced observers of the process with the aim of improving the process further. Few of this B Team’s observations are, however, entirely new to us, and the prob-lems they address are under scrutiny. The Team’s recommendations appear, moreover, insensitive to the costs and penalties of implementing them.

2. In considering the attached recommendations, it is useful for us to remember that the methods for producing national intelligence estimates have evolved over the past 25 years in response to the changing interests and styles of administrations, the organizational methods of a number of Directors of Central Intelligence, and the international situation itself. At present, national estimates on Soviet strategic programs and capabilities are produced by a method which centralizes the supervisory responsibilities in a National Intelligence Officer on the DCI’s staff, but decentralizes the analytical and drafting responsibilities to teams of analysts from the various intelligence agencies. This method is designed to ensure that significant analyses and judgments from all elements of the Intelligence Community are reflected at all stages in the process and that no single staff or agency determines the results. The process encourages the exposure of divergent views. NIO management is charged with ensuring that significant differences are illuminated and that consensus judgments due to bureaucratic pressures are avoided.

3. The present production method is consistent with certain principles which have guided the preparation of the NIEs throughout their history, regardless of organizational adjustments:

[Page 830]

a. A national intelligence estimate is the DCI’s responsibility in accordance with his statutory duties. The main text represents his best judgment.

b. A national estimate involves the participation of the agencies of the Intelligence Community, whose representatives on the National Foreign Intelligence Board have the right and duty to introduce into the estimate abstentions or opinions which diverge substantially from those expressed in the main text.

c. A national estimate is designed to address major topics of concern to US planners and policymakers, and hence its content and its producers cannot be isolated from the process it is designed to support; at the same time, it is not a mechanism for critiquing or recommending policy.

4. The current method of producing drafts is only one of a number of methods which could be employed. Any change, however, should be consistent with the style and needs of a particular administration and a particular DCI, while preserving the principles above.

5. Turning to the specific points in the B Team’s recommendations, I believe a number of observations need to be recorded:

a. Mirror Imaging. The B Team’s charge that “soft” factors affecting Soviet motivation do not receive “thorough” analytical attention is simply not true. What is obviously true is that the B Team’s analysis of these “soft” factors differs from that of at least some in the Intelligence Community. For example, along with much evidence of the Soviet drive to acquire military preponderance, there is voluminous evidence that the Soviets have a high respect for the technical and industrial might behind US military programs. As for the need to perceive Soviet objectives in terms of Soviet concepts, an effort has been made to judge Soviet policy on the basis of a large number of factors and influences, of which doctrine is one. In this year’s NIE 11–3/8,2 partly stimulated by the competitive analysis experiment, a special effort was made to describe Soviet objectives and military doctrine in Soviet terms so as not to leave any impression that the USSR had been judged only in US terms. This practice should be continued, but not to the extent that every specific estimate need be prefaced by a long exposition of the Soviets’ doctrine and ultimate objectives.

b. Net Assessments. I agree that those net assessments which are the result of a quantitative analytical process should be so identified. In NIE 11–3/8–76, we have largely eliminated net assessments which are [Page 831] not a result of such a process, partly because new evidence has given rise to greater uncertainty and partly because the significance of operational factors was well illustrated by the B Team on Soviet air defense. The NIE calls attention to the fact that a full net assessment would be required to take adequate account of such factors and that the estimate is not such a net assessment.

i. While accepting the B Team’s recommendation, we would not agree that net judgments can never be delivered; some judgments in this complex world remain important and susceptible to experienced analysis. Most predictive analyses or interpretations of the policies and expectations of foreign leaders require an analytical model which includes US policies and forces among the influences affecting those leaders. Even estimates of the technical capabilities of a potential adversary’s weapon systems require an “interaction analysis,” one part of which is the US force which the foreign system was designed to engage.

ii. As for comprehensive net assessments, we have all recognized for some time that there is no national level organization responsible for such assessments on a regular basis. I hope my successor will encourage officials of the new administration to identify such an organization—perhaps at the NSC staff level—and pledge the Intelligence Community to cooperate by providing the intelligence data and insights necessary for its operation. For my part, I would object to assigning the responsibility for such full net assessments of the USUSSR strategic balance, or the balance in other situations involving US and foreign forces, to the Intelligence Community. Such an arrangement would give excessive responsibility to the Intelligence Community and would be unlikely to promote the cooperation of policymaking departments whose participation would be essential.

c. An Integrated View of Soviet Weapons and Force Developments. The packaging of national intelligence on Soviet military forces into several operational categories resulted initially from consumers’ requests in the 1960s to organize the presentation of intelligence according to the way the US plans its forces—strategic offensive, strategic defensive, and general purpose forces. The US defense planning process continues to require this type of presentation.

i. In 1974, the separate estimates of Soviet offensive and defense forces for intercontinental conflict were combined, with the concurrence of the Secretary of Defense, into a single estimate in response to consumer requirements for intelligence on the strategic nuclear balance—that is, the balance as the US measures it.

ii. Our present estimative program acknowledges the further requirement, suggested by other groups as well as the B Team, for national intelligence on overall Soviet military and foreign policy objec[Page 832]tives (as in NIE 11–4)3 and on overall trends in Soviet military forces and capabilities (as in an interagency intelligence memorandum issued in October 1976).4 These integrated assessments could not be done without the more detailed assessments of individual aspects of Soviet power and probably need not be done routinely on an annual basis. We will investigate with key users the advisability of further integration of NIEs. We would, however, strongly resist drawing the impractical conclusion from the B Team recommendation that one should never estimate about a part of the Soviet strategic effort unless one appraises the entire Soviet effort.

d. Policy Pressures and Considerations. I certainly would not quarrel with minimizing any possible policy pressure on NIE judgments and preventing the abdication by the intelligence apparatus of its responsibility to provide objective answers. I would note, however, that the estimative process, as carried out by the DCI under the principles cited at the beginning of these remarks, is designed to do just that. The NFIB participants bring to the estimating process differing experiences and professional backgrounds. If some representatives have convictions about US policy that correspond closely to the advocacy of the bureaucracies they represent, the variety of points of view introduces checks and balances into the system. The professional integrity of the participants, moreover, should not be lightly dismissed. In any case, the DCI, the National Intelligence Officers who support him in supervising the estimative process, and the CIA analysts who have a major role in the drafting process, do not represent any department of government involved in the policymaking process. Their independent bureaucratic positions minimize the susceptibility of the DCI, NIOs, and CIA analysts to policy pressures and allow them to serve as an important check on the objectivity of the process.

e. Disciplined Presentation of Conclusions. I have some difficulty grasping what Team B has in mind. I would not prescribe a format for the conclusions and key judgments in NIEs so rigorous that we could not adjust to the nature of the intelligence available and the needs of the policymakers being supported. I would, on the other hand, agree that consistency is desirable; that we should accurately convey uncertainty and alternatives; and that when important changes occur in the judgments of estimates, the fact of and the reasons for these changes should be called to the attention of readers. A periodic track record of key judgments in an NIE has occasionally been useful. Where and how [Page 833] often one is done should be a matter for intelligence managers and consumers to decide on the basis of practical considerations.

f. Procedures. This section contains a curious discussion of institutional bias. Many people imagine they understand the nature and sources of State’s and Defense’s biases; it would have been interesting if the report had discussed the nature and sources of the bias attributed to CIA. In its argument, Team B appears to adopt the following approach: all past errors are the fault of CIA, even when everyone else was in agreement; the reason for this is CIA’s major role in the preparation of estimates; therefore, take the estimates out of CIA’s, and possibly even out of the DCI’s, hands.

i. The possibility is raised of a chief estimative officer and staff within the Executive Office of the President. If this chief estimative officer were not the DCI, the arrangement would circumvent the statutory responsibilities of the DCI. If the officer the B Team has in mind is, in fact, the DCI, the question of the location of his estimative function and staff would have to be considered as part of the broader question of the role of CIA in the Community. My judgment is that physical and institutional separation of the DCI from CIA would sharply limit his ability to reach responsible judgments because it would cut him off from his independent analytical base.

ii. This organizational recommendation fails to take into account the checks and balances built into the system. The preparation of the NIE 11–3/8 estimates, for example, involves a program of production by analysts within the military services, CIA, and DIA, their various contractors, DCI Committees, and analytical teams drawn from the several agencies. This specific recommendation reveals naivete about the interactions of policy and intelligence that, in my opinion, tends to undercut the credibility of other observations.

iii. The recommendation is silent on all the big questions—how would the NIEs be drafted; how would the draft contributions be pulled into a single document; how would coordination be achieved; how would the rules of dissent and alternative statement be enforced; and how would final power of approval of the text be exercised? Would the B Team have us reintroduce a monopoly on the drafting of estimative intelligence, one of the weaknesses perceived in the former ONE staff system? And would not the location of the estimative process in the Executive Office of the President in fact subject that process to additional policy pressures without the checks and balances of the current national intelligence production mechanism?

iv. The B Team recommendation concerning the use of a panel of outside specialists to review NIEs is sensible. Such panels have been used at various times in the past. Some months ago, I approved in principle a plan to establish an Estimates Advisory Panel that would in[Page 834]clude a broad range of outside experts with a variety of viewpoints. Because of the impending change of administrations, however, I delayed the formation of this panel, but commend it to my successor.

v. The recommendation that adversarial procedures similar to the B Team experiment be continued, perhaps every other year, is one I oppose. It is not that the experiment was a total failure; to the contrary, the B Team on low altitude air defense made a particular contribution. Rather, it is that, when one sets out to establish an adversarial B Team, one sets in motion a process that lends itself to manipulation for purposes other than estimative accuracy. I am already, incidentally, getting recommendations that, should the process ever be repeated, a C Team of a persuasion opposed to the B Team should be established to review the estimate at the same time. I would prefer to convene panels of experts with a mix of views. Indeed, I would expect that my successor might very well wish to do so. Individual agencies and DCI Committees should also continue the practice of using panels of experts such as those convened by the CIA and the OSD to review technical analysis of Backfire performance and the panel of US experts in the field of directed energy convened by the DCI’s Scientific and Technical Intelligence Committee to review evidence of Soviet research applicable to particle beam weapons.

6. The essence of national intelligence production is that it marshals the full resources of the Intelligence Community to address the most important analytical and estimative problems, that it provides the base which allows the DCI to fulfill his mandate as an independent advisor to the President, and that it displays for policymakers such differing analyses as exist on important issues. The challenge is to produce these results; doing so depends first of all on the quantity and quality of the resources and talent devoted to it. Equally critical at this highest level of need is the willingness of policymakers to help the Intelligence Community concentrate on the issues of most concern and, then, to support the Community when it accomplishes its mission. Both these factors are far more important for the production of national intelligence than the changeable procedures that may be used.

George Bush
[Page 835]


Paper Prepared by Team B5

Team “B” Recommendations

1. Concerning methodology

A. Mirror Imaging. To overcome the bias toward viewing Soviet motives and intentions in U.S. terms, it is urged that:

1) In dealing with Soviet intentions, the NIEs should integrate observed and projected Soviet weapons’ programs and force deployments derived from the “hard” physical data with more thorough analysis of historical, political, institutional, and other “soft” factors shaping Soviet motives and intentions. The search should be for a consistent elucidation of both sets of factors and their interaction. In this connection considerably more attention should be paid to relevant open and clandestinely acquired Soviet pronouncements and writings (especially those directed to internal audiences) than has been the case in the past. In this regard it should be understood that expert analysis of the open material can reveal a great deal, insofar as the Soviet political system often compels the Party to issue to its cadres authoritative guidance on policy matters through unclassified sources;

2) Soviet objectives should be perceived in terms of Soviet concepts: this rule applies especially to the treatment of concepts like “strategy”, “strategic threat” and “strategic objectives”, all of which should be understood in the Soviet context of “grand strategy.” When, for reasons of convenience to U.S. consumers, the NIEs address Soviet military programs in the U.S. rather than the Soviet strategic context, this fact should be made clearly evident to the reader.

[Page 836]

B. Net assessing.6 Whatever their intentions, the drafters of the NIEs do engage in implicit net assessments of sorts, particularly when advancing major judgments in the executive summaries. These assessments are usually so rough, so poorly documented, and essentially so speculative that they invite—indeed, cannot be immune from—the injection of the authors’ general biases. Where NIE judgments demand net assessment, the netting should be done explicitly, analytically, and thoroughly, not implicitly or perfunctorily. The interface between NIE judgment and net assessment should be identifiable.

C. An integrated view of Soviet weapons and force developments. The NIEs tend to an excessive extent to analyze each Soviet weapon system in isolation from the totality of the Soviet military effort (and indeed from other relevant non-military factors as well), with the result that the overall Soviet military effort appears as less significant than it actually is. Team “B” urges that in the future weapons systems and force developments be examined in a more integrated manner to yield “combined evaluations” more indicative of Soviet total military capabilities and overall intentions.

D. Policy pressures and considerations. In the opinion of Team “B”, total avoidance of policy pressure on the intelligence estimating process is an impossible goal. The normal and proper function of policy makers in raising questions which are to be addressed by the intelligence estimators in and of itself influences the answers the latter provide. Some awareness on the part of the estimator of the impact of intelligence judgments in support of or in opposition to policy is unavoidable. Nonetheless, improved methods and procedures adopted for the preparation of the NIEs should be able to minimize the policy pressure on judgments and prevent the abdication by the intelligence apparatus of its responsibility to provide objective answers.

E. Disciplined presentation of conclusions. Key judgments of NIEs are presented in various styles and formats. This on the one hand permits statements to be made with a certainty that is not warranted by the available evidence, and on the other hand permits statements, better supported by the evidence, to be degraded in the reader’s mind through the insertion of a clause or sentence that have the effect of dismissing their impact. A more disciplined (though not necessarily rigid) format for NIE key judgments, summaries, and conclusions should be constructed. The format and style should ensure that the various rea[Page 837]sonable interpretations of the available evidence are laid out without semantic embellishment; that the pros and cons of evidence supporting each are discussed briefly; that the likelihood of occurrence of each is assessed; and that the requirements for additional data to resolve remaining uncertainties are identified. Further, each major intelligence estimate should contain as an annex a review of the past 2 to 10 years’ “track record” of U.S. estimates on the major aspects of the relevant subject matter. Such an annex would be best prepared by a body of analysts not responsible for the estimates critiqued. The purpose of such an annex would not be to criticize or chastise but rather to throw light on possible trends of misanalysis or mistaken judgments so that a compounding of error by continuation into future intelligence estimates can be avoided.

2. Concerning procedures.

A. Some, though undoubtedly not all of the methodological shortcomings which Team “B” found in the National Intelligence Estimates can be overcome by improving the process of their preparation and review. The authors of the NIEs will always remain in some measure prone to perceive the USSR in U.S. terms and to allow political considerations to affect their judgments. Nevertheless by minimizing inherent institutional biases and broadening the range of judgments brought to bear on the NIEs it should be possible to weaken considerably the impact of factors which have accounted in the past for NIE misperceptions.

B. Team “B” considers the organizational position of the NIE function within the national defense-security-foreign policy complex less than optimal for guarding against both policy and institutional biases. Current and previous organizational entities charged with preparation and processing of NIEs have been subordinate to the Director, CIA, and staffed almost exclusively with CIA officials. This arrangement was intended to compensate for the real or alleged biases of the Departments of Defense and State, but it can over-compensate by encouraging the institutional biases of the Central Intelligence Agency itself.

C. Team “B” recommends that some combination of the following three steps be considered:

1) The first involves building as much immunity to institutional pressures as possible into that entity which is charged with preparing NIEs on Soviet strategic objectives. There are various ways to accomplish this end. One attractive possibility is to identify an official in the Executive Office of the President who would be charged with assuring such immunity and who would report directly to the President. His staff would be small and guarded against acquiring an institutional life of its own. Members of the staff would be drawn from the various intel[Page 838]ligence organizations and serve relatively short tenures (3–4 years). The official charged with this function would be genuinely removed from and independent of the operating membership of the NFIB by the devices of a separate budget, a separate staff, and a separate physical location. He should have the authority to subpoena substantive intelligence officers from any agency and to require of pertinent departments and agencies such net assessments as may be necessary to the NIE process;

2) The second step involves the marshalling of expertise in and out of government to offset the temptation to mirror-image. The official charged with assuring the objectivity of the NIEs (as specified above) should enlist the part-time services of a panel of prominent outside specialists for the purpose of reviewing estimates so as to identify judgments that are based on questionable assumptions concerning Soviet strategic doctrine and behavior. Such reviews should be carried out immediately post facto, but they should not form a part of the NIE preparation process itself.

3) The third step involves periodic independent checks on both the process and the substance of the NIEs by employing procedures similar to the PFIAB-conceived Team “B” approach. Intermittently, perhaps initially every second year, a team of outside experts who owe no formal responsibility to the existing governmental intelligence agencies would be assembled to play the adversary role. The composition of the Team would vary every time. Team members would have available all the pertinent information from all the sources. The effort would be reasonably time constrained. The report of the Team would be subject neither to review nor to revision but would be made available directly to the President, Secretary of State, and Secretary of Defense. (After the Team had made its report, it would become available to other governmental agencies for criticism but not for revision). While this step would not eliminate the particular views and biases which the non-governmental experts would bring to their study, it would be free of the bureaucratic pressures or biases of the existing governmental intelligence—or indeed policy—agencies.

D. Team “B” has not addressed itself to substantive national intelligence issues other than Soviet strategic objectives. Should similarly critical issues arise—e.g., with regard to China or the Middle East—the above recommended processes could help to ensure objective intelligence support to top policy makers.

  1. Source: Central Intelligence Agency, National Intelligence Council, Job 91M00696R: Subject Policy Files, Box 7, Competitive Analysis, 1977. Secret. Team A also later commented on Team B’s reports. Stoertz forwarded Team A’s comments to Acting DCI Knoche under a February 23 covering memorandum. (Ibid., Job 85B00134R, Box 2, Competitive Analysis, Part V (2), Commentary on the A Team B Team Experiment, Dec. 1976 through Completion)
  2. NIE 11–3/8–76 is the attachment to Document 170.
  3. A series of NIEs treating broad trends and issues in Soviet strategic policy. The series was published annually until 1968 and then irregularly until 1977, when it was again published annually. See the attachment to Document 173.
  4. IIM 76–039J, Trends in Soviet Military Programs, October 1976, was not found.
  5. Secret. Team B members—Richard Pipes (Team leader), William R. Van Cleave, General Daniel O. Graham, Paul Nitze, Seymour Weiss, and Paul Wolfowitz—forwarded the paper through Bush to Cherne under an undated covering memorandum, which reads as follows: “In our critique [see Document 171] of current and previous National Intelligence Estimates, we made a concerted effort to identify those aspects of methodology, procedure, and institutional structure which we believe have contributed to unsound estimative judgments. In the attached paper we proffer our recommendations to PFIAB concerning improvements in methodology, procedure and structure aimed at correcting the perceived deficiencies. Evidence for our conclusion that the cited shortcomings, do, in fact, exist in the NIEs is to be found in the main body of our report.” (Ibid.)
  6. What we mean by net assessment in this context is a judgment on the balance between U.S. and Soviet military capabilities based on the relevant static indicators extant or projected, or based on a dynamic analysis of the balance assuming that those capabilities actually are to be called into use. The latter type of net assessment assumes a scenario, but may or may not assume actual warfare. [Footnote in the original.]