371. Central Intelligence Agency Memorandum of Comments on the Dulles Report0

The Committee’s report is an admirable effort to direct the Central Intelligence Agency to fundamentals and is a genuinely constructive effort to improve the production of intelligence on which the policy of the United States should be based. The observations of the Committee are, in general, accurate, and its objectives are sound; its conclusions, [Page 937] however, are, in many respects, faulty, and the recommendations for their attainment are, in many cases, impracticable.

The larger part of the Report is concerned with the amount of coordination exercised by the Central Intelligence Agency in regard to the other intelligence activities of the Government. It is realized that an optimum coordination has not yet been reached; however, not only a start, but considerable progress along the path of complete coordination has been attained. No gaps remain even though all duplication has not yet been abolished. On this subject of coordination, I am sure the members of the National Security Council realize the difficulties encountered; particularly, those members having to do with the unification of the Armed Services. A comparable problem has confronted the Central Intelligence Agency with the added obstacle that one additional department is involved.

The Committee states (page 5): “—coordination can most effectively be achieved by mutual agreement among the various agencies.” There is no question as to the validity of that statement. However, to achieve coordination in such a manner, compromises of wording, emphasis, etc., have to be accepted. For example, three months were required to secure agreement on most of the present NSCID’s, while it required one year to resolve the differences between the Air Force and the Navy on one NSCID.

The Committee begins its discussions on coordination with the above-mentioned declaration, but, in numerous cases thereafter, throughout the Report, the element of “direction” in coordination is intimated and even stressed. This element of “direction” has usually been entirely unacceptable to the interested agencies. Further, the Director of Central Intelligence, per se, has no authority to enforce coordination, and, indeed, the Committee (page 41) recognizes this when it says, “The Act does not give the Central Intelligence Agency independent authority to coordinate intelligence activities.” Yet many of the Committee’s recommendations hinge upon having such authority, and many of the alleged shortcomings are due to this lack of authority. The phrase, “such procedure would violate the chain of command,” was repeated over and over again in the discussions during the drawing-up of the present NSCID’s.

In view of this lack of explicit authority, in view of the progress made in the relatively short time this activity has been in operation, and in view of the progress already made in coordination by mutual agreement, it is considered that in not too long a period the continued augmentation of such a program will more than justify any delays incurred.

A second alleged shortcoming is that the Central Intelligence Agency is “overadministered in the sense that administrative conditions have been allowed to guide and, on occasion, even control intelligence [Page 938] policy, to the detriment of the latter.” The administration of the Central Intelligence Agency has been developed on the following pattern:

Budget, personnel, and other support services are essential to any operation, and are centralized to serve all components of the Agency.
Control procedures are restricted to keeping within the budget requirements, insuring that illegal transactions are not permitted, avoiding waste and duplication in expenditures, and maintaining certain standards required by security, the Civil Service Commission, the Bureau of the Budget, and the General Accounting Office. No agency, regardless of its nature, and most emphatically one that handles confidential government funds, can possibly avoid such controls.

Administration has advised on policy matters and should continue to do so, particularly, in the determination of the ability to support operational proposals and in the achievement of greater efficiency with the means available for operations. Administration has no voice in determining the substance of operational direction, guidance, and production, nor should it ever have.

There is a never-ending argument on the merits of administration versus operations. The operating people would like to be given a lump sum for their operations and complete latitude with regard to numbers and grades of personnel, travel authorities, new projects, etc. The results of such action, without controls, is obvious—chaos—and, again, it simply cannot be done where government funds are involved.

The Committee further declares: “Administrative arrangements which do not at first appear to be efficient or economical may be necessary in the Central Intelligence Agency. Personnel requirements for certain types of work cannot conform to normal civil service standards, and the demands of security often impose special and unusual procedures. This situation must be understood not only by those responsible for the internal organization of the Central Intelligence Agency but also by Congress and the Bureau of the Budget.”—“The centralization of all administration in one office is undesirable since secret operations require their own separate administration.”

The Committee’s report seems to indicate undue emphasis and alarm in connection with complete autonomy for covert components of the Central Intelligence Agency. There has to be an umpire some place, and no one, directly responsible for operations, should be his own final authority and judge in the utilization of funds and personnel.

Security is adequately served through the physical segregation of covert administrative branches and the establishment of direct channels, both to the covert operational officials and to the responsible administrator who is the third senior official in the Agency. Covert and overt support activities are not intermingled, but a single administrative chief over each phase insures adequate mutual support and minimizes duplication. The Central Intelligence Agency saved over sixty positions [Page 939] by the new organizational approach and so far has had few complaints. From a strictly security point of view, there will be better security than before because of not pointing up autonomous separate support functions in a single agency.

It is all very well for a group with no responsibilities or authority to state that both Congress and the Bureau of the Budget must understand that the Central Intelligence Agency must be given, in effect, a blank check and a free hand. In practice, the Central Intelligence Agency must justify its demands with some reason and logic and must reassure both of those bodies that the Central Intelligence Agency is, at least, somewhat careful with government funds and does its best to guard against waste and fraud.

Another Committee comment is: “Many able persons have left the organization and few qualified ones have been attracted to it. On the higher levels, quality is uneven, and there are few persons who are outstanding in intelligence work.”

The above statement seems to be totally unwarranted, as well as to have no actual basis in fact. Out of 267 employees in the three top grades (P-6 to P-8 and CAF-13 to CAF-15, both inclusive), 20 employees have left, not all of them voluntarily. The following table shows this turnover.

[Here follows a table of losses of CIA personnel in the top three grades for CY 1948.]

For the calendar year 1948, the turnover in the Central Intelligence Agency for all personnel, for all causes—death, sickness, maternity leave, and all other voluntary or involuntary separations—has averaged 1.6% per month. This is a considerably lower rate than for most other government agencies.

The charge that there are few persons outstanding in intelligence work is another general one and is difficult to sustain. It is admittedly difficult to establish an absolute criterion as to ability, but, if scholarship and experience do have some bearing, the following tables will demonstrate that the Central Intelligence Agency is not totally devoid of capable people.

[Here follow two tables showing the qualifications of professionals in the CIA and ORE as of September 1948.]

Another error, states the Report, is “The placing in key positions of a large percentage of military personnel, many of them on relatively short ‘tour of duty’ assignment, tends to discourage competent civilian personnel from looking to employment in the Central Intelligence Agency as a career.”

Omitting the post of Director, which is considered in another part of the Report, there is only one—out of six—Assistant Director who is a military man. Four—out of six—Deputy Assistant Directors are from the military services, but, considering the fact that the military services are [Page 940] both the greatest suppliers and the greatest customers, this does not appear unreasonable. The overall figure is 58 military personnel, or less than 2% of the total number of employees in the Central Intelligence Agency.

The above Committee statement seems somewhat inconsistent with recommendations in other parts of the Report which urge the assignment of military personnel to various branches of the organization.


On page 33 of the Report is the statement: “Both Congress and the Bureau of the Budget have refrained from examining in detail the internal workings of the Central Intelligence Agency in order to determine the justification for the budget.” This statement is only partially correct as, while Congress, so far, has not inquired into detail, the details have been gone over in the Bureau of the Budget by an official who has full security clearance. It was at the insistence of the Bureau of the Budget that a centralized administrative staff was set up.


The Committee recognizes the need for security, “—in the Central Intelligence Agency which has unfortunately become publicized as a secret intelligence organization.” The Central Intelligence Agency concurs completely in this, and the publicity received has neither been sought nor encouraged, but has been actively discouraged. By special plea of the Director, various periodicals and newspapers (Life, Time, Newsweek, U.S. News, New York Herald Tribune, among others) have refrained from publishing articles. Under the existing conditions of press and radio, it would be practically impossible to conceal completely, desirable as it undoubtedly would be, activities of the Central Intelligence Agency. The more practicable manner, which has been adopted, is to conceal the covert sections by allowing the overt sections to serve as a “lightning rod” and draw off attention from the clandestine activities.

Concise comments on the individual conclusions and recommendations of the Committee follow:

[Here follow 11 pages commenting on Chapters II-IV.]

Chapter 7, page 81.1

“(1) In the Central Intelligence Agency there has been confusion between the responsibility of producing coordinated national intelligence estimates and responsibility for miscellaneous research and reporting activities.”

[Page 941]

This is a broad statement that needs qualification. Both types of activities may have been performed, but there has been no confusion between them, and both were found to be necessary. It is believed that what the Committee describes as “confusion” has resulted from the Central Intelligence Agency’s recognition of its threefold responsibilities as the producer of national intelligence estimates, as the intelligence facility of the National Security Council, and as the agency responsible for performing services of common concern. The Committee has failed to recognize the gap that exists between the Central Intelligence Agency’s explicit responsibilities, as set forth in NSCID #3, and the Central Intelligence Agency’s implicit responsibilities as the intelligence facility of the National Security Council. Moreover, NSCID #3 is so rigid with respect to the Central Intelligence Agency’s production responsibilities and so flexible with respect to those of the departmental intelligence agencies that its usefulness as a working document is seriously impaired. The Committee’s conclusion is made on the basis of a dangerously limited view of the Central Intelligence Agency’s responsibilities for intelligence production. Such responsibilities should be considered not solely in terms of the production of “coordinated national intelligence estimates” and “intelligence in fields of common interest” but also in terms of the Central Intelligence Agency’s role as the intelligence facility of the President, the National Security Council, and such agencies as may be designated by them.

The Central Intelligence Agency’s intelligence production responsibilities should include the following, and NSCID #1 and NSCID #3 should be modified accordingly to include a revised and more appropriate definition of national intelligence:

To produce intelligence required to reduce to a minimum the element of surprise in foreign situations and developments which affect United States national security.
To produce coordinated intelligence estimates required for national planning, policy, and operational decisions.
To produce intelligence required to support those agencies designated by the National Security Council to receive it.
To produce intelligence in fields of common interest to the Central Intelligence Agency and the departmental intelligence agencies, including economic and scientific intelligence.
To produce intelligence required for Central Intelligence Agency estimates not normal to any departmental responsibilities.

(It should be noted that paragraph (2) above is the only category of intelligence which conforms to the definition of national intelligence in NSCID #3.)

“(2) The provisions of the National Security Act for the production of national intelligence estimates, as interpreted by the [Page 942] National Security Council Intelligence Directives, are sound but have not been effectively carried out.”

Concur in the assertion that the provisions of the National Security Act for the production of national intelligence estimates are sound, but, after almost a year’s experience, believe that NSCID #1 and NSCID #3 should be revised for the reasons cited immediately above and for the following reasons:

The directives do not provide for adequate treatment of the matter of priority within the departmental intelligence agencies with respect to their support of the Central Intelligence Agency. Consequently, the Central Intelligence Agency is unable to make optimum use of departmental intelligence and cannot depend upon timely departmental action in meeting Central Intelligence Agency commitments and deadlines.

The directives do not adequately allocate production responsibilities and do not define intelligence fields of common concern.

“(3) There should be created in the Central Intelligence Agency a small Estimates Division which would draw upon and review the specialized intelligence product of the departmental agencies in order to prepare coordinated national intelligence estimates.”

The Report is vague concerning the details of the proposed reorganization, and, therefore, opinion is divided on the necessity or desirability of a drastic internal reorganization now. However, if the Committee’s recommendation means that the Estimates Division is to be large enough to provide its own relatively high level research support, the recommendation is workable and might result in simplification of the process of producing estimates. The problem of a greater duplication of effort, than now exists, immediately arises. It is believed, therefore, that this recommendation is premature and should be considered only after a revision of NSCID #1 and NSCID #3.

  • “(4) Under the leadership of the Director of Central Intelligence, these estimates should be submitted for discussion and approval by the reconstituted Intelligence Advisory Committee whose members should assume collective responsibility for them.
  • “(5) Provision should be made in these arrangements for the handling of crisis situations when coordinated estimates are required without delay.
  • “(6) Coordinated intelligence estimates produced in this way must, in order to be effective, be recognized as the most authoritative estimates available to the policy makers.”

These three recommendations have been considered together as parts of the same problem. The Central Intelligence Agency cannot agree with the view that members of the Intelligence Advisory Committee should assume collective responsibility for Central Intelligence Agency estimates. It is considered that the main question is what sort of estimates should emerge from the national intelligence structure. Those who [Page 943] developed the original concept of a central intelligence group had clearly in mind the ever-present factor of departmental bias as well as the shortcomings of joint intelligence, which often produces estimates so watered down in the process of reaching agreement as to be valueless. The Central Intelligence Agency estimates can be free from such faults only if they are produced with full cognizance of departmental views but without subservience to departmental prejudices. Intelligence Advisory Committee review of the Central Intelligence Agency drafts should undertake to eliminate minor or apparent differences but not to gloss over divergences of informed opinion. The resultant estimate must be essentially that of the Central Intelligence Agency and not that of the Intelligence Advisory Committee collectively. Any statements of dissent should be limited to substantial and well-defined issues, as the directives now provide. Moreover, these directives should be further modified to restrict dissent to the field of responsibility of the dissenting agency. (For example, the Office of Naval Intelligence has an interest in political matters but is not responsible for estimates in that field, and the Central Intelligence Agency should not be required to publish a dissent by the Office of Naval Intelligence on purely political matters if the State Department concurs in the Central Intelligence Agency position.) The logic of this contention is simply that, just as a dissenting member of the Intelligence Advisory Committee cannot be expected to share in the responsibility for a Central Intelligence Agency estimate in which he dissents, so, by extension, the Central Intelligence Agency has sole responsibility for its estimates, even when Intelligence Advisory Committee concurrence is complete, and does not share this responsibility collectively with the members of the Intelligence Advisory Committee. Concurrence should fortify the estimate; it should not create an artificial responsibility. The Report does not reflect a clear understanding of this vital aspect of the coordination theory. To the Committee, “national” intelligence is merely “coordinated” intelligence, and coordinated intelligence is joint intelligence, at the mercy of departmental bias.

Although the Report favors the method of preparing drafts in an ad hoc committee, experience shows that this procedure is time-consuming and inefficient. The Central Intelligence Agency believes that coordination procedures will be effective only if its own estimates personnel produce the initial draft as a basis for discussion with a permanent, full-time substantively competent Intelligence Advisory Committee sub-committee, prior to the submittal of the estimates to the Director for Intelligence Advisory Committee action.

There must be effective means for handling crisis situations. In the past, an informal procedure, which worked, was in effect. This procedure is now in process of being formalized by agreed, written understanding.

[Page 944]

The Central Intelligence Agency’s estimates must earn their distinction as the most authoritative intelligence available to the policy makers. It is, however, recommended that the National Security Council enjoin the policy makers to give the Central Intelligence Agency such guidance in matters pertaining to national policy that Central Intelligence Agency support will be more directly responsive to their needs.

[Here follow 16 pages commenting on Chapters VI-IX.]

Chapter X, page 138.

“(1) The directing staff of the Central Intelligence Agency has not demonstrated an adequate understanding of the mandate of the organization or the ability to discharge that mandate effectively.”

Do not concur. Had the Central Intelligence Agency been constituted in a vacuum, where relations with other established agencies need not have been taken into account, progress in coordination, organization, and production might have been much more rapid. However, the relationships and impacts between not only the other intelligence agencies, but also the administrative agencies of the Government—the Bureau of the Budget, the General Accounting Office, the Civil Service Commission, etc.—had to be adjusted and had to be lived with. Dismissing the theoretical conditions which one might have liked to enjoy, and considering the realistic conditions which have confronted the Central Intelligence Agency, it is believed that much progress along sound lines has been made. And, what is more gratifying, this tendency is increasing in both scope and in the right direction.

“(2) Administrative organization and policies tend to impede the carrying out of the essential intelligence functions of the Central Intelligence Agency under the Act.”

Do not concur. The development of our organization and functions has been based on a policy of centralization of specific responsibilities in single offices and the coordinated work of the offices as a team, each concentrating in its own field and looking to the others for support from other fields. Only in this manner can we avoid the confusion, duplication of effort, inefficiency, and major frictions which always attend a situation wherein every component tries to be self-sufficient and attempts to do the same things, at the same time, in the same sphere, and in constant competition. Our organizational policy has resulted in the centralized groupings shown below:

Research, evaluation, and production, and coordination in these fields:
  • Office of Reports and Estimates (except science)
  • Office of Scientific Intelligence
Collection and liaison control, dissemination, liaison service, and information service, and coordination in these fields: [Page 945]
  • Office of Collection and Dissemination
Overt field collection:
  • Office of Operations
Covert field collection and counterespionage:
  • Office of Special Operations
Covert field operations:
  • Office of Policy Coordination.

(The last two functions which had been integrated into one covert office were separated as a result of a directive issued by the National Security Council.)

“(3) Continuity of service is essential for the successful carrying out of the duties of Director of Central Intelligence.”

Concur. This principle is obvious, and a lack of the assurance of continuity works a hardship on both the personnel of the Central Intelligence Agency and the quality of work performed. Changes of Directors with the following changes of organization tend to make the Central Intelligence Agency employee feel uncertain and insecure in his job with consequent harm to the quality and quantity of his output. This was particularly noticeable last Fall just before election when literally dozens of rumors were extant in Washington that one of the Committee members was to become Director of Central Intelligence as soon as the election was over.

“(4) As the best hope for continuity of service and the greatest assurance of independence of action a civilian should be Director of Central Intelligence. If a Service man is selected for the post he should resign from active military duty.”

This is a question that has arisen time and time again. Perhaps the best discussion of this problem is that stated in the report of the Eberstadt Committee which is quoted, as follows:

“A moot question is whether the Director should be a civilian or a professional military man. The argument in favor of a service man is that he will command more confidence from the armed services who talk his language and will respect his position and security. With a military man, the present pay scale will not prove a deterrent. The job could be developed into one of the top staff assignments available to members of the three services.

“Against this, it is said that the position requires a broader background and greater versatility and diplomatic experience than is usually found in service personnel; that the best qualified and most competent officers would not accept the position if to do so meant permanent retirement and an end of the road to important command or operational responsibility. If a military man is assigned to the position as a tour of duty, he will, it is said, inevitably be influenced to some degree, in the [Page 946] execution of his duties, by his rank and status as compared with that of other officers with whom he deals. He may also be influenced by concern for his next billet.

“The principal argument against a civilian is the difficulty of getting a good one. It will be difficult to attract a man of force, reputation, integrity, and proven administrative ability who has an adequate knowledge of foreign history and politics and is familiar with intelligence technique and the working machinery of the Government and the military establishment. Not only is the pay low in comparison to industry and the professions, but the reward of success is anonymity. The wisdom of putting an individual who lacks intelligence experience in charge simply because he is a competent administrator is dubious. A civilian would have the advantage of being free from taint of service ambitions or rivalries. On the other hand a civilian may be more subject to political pressure than a military man. In certain foreign countries this has occurred. In any event a civilian would have to be a man of commanding reputation and personality in order to secure the respect and cooperation of the services. CIA’s relations with the State Department would undoubtedly benefit from the presence of a civilian director, known and respected by the Secretary of State and his assistants.

“The intrinsic interest of the work, its potential influence on policy, and recognition of public service to be performed might combine to persuade a competent civilian to accept the position. If so, his appointment would seem desirable. A change in the statute that would disqualify a military man is not, however, recommended.

“Moreover it would not be wise, at this time at least, to amend the statute to include a mandatory requirement that a military man, appointed as Director, must retire from the service. A competent officer could be persuaded to retire from his service and abandon his career to become Director of CIA only if he felt some assurance of a reasonable tenure of office. That no such assurance exists today appears from the fact that three different Directors have been appointed since January of 1946. A provision requiring the retirement from service of any commissioned officer appointed Director might appropriately be included in the statute—if coupled with the provision for adequate retirement pay in case he is removed as Director.”

[Here follow 9 pages commenting on Chapters XI and XII and charts showing the organization of the CIA.]

  1. Source: Central Intelligence Agency Records, Job 86-B00269R, Box 5. Secret. The full text, including the tables not printed here, is in the Supplement. Sent to the Executive Secretary of the National Security Council who transmitted copies to the members of the NSC. (Memorandum from Souers to the NSC, March 2; Central Intelligence Agency Records, Job 86-B00269R, Box 5) See the Supplement.
  2. This and similar chapter, page, and paragraph references are to the text of the Dulles Report; see Document 358.