IMPROVING THE QUALITY OF INTELLIGENCE

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP89G00720R000100050014-0
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 5
• Document Creation Date: December 27, 2016
• Document Release Date: January 24, 2013
• Sequence Number: 14
• Publication Date: November 19, 1981
• Content Type: MEMO

/
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
19 November 1981
MEMORANDUM FOR: Director of Central Intelligence
Deputy Director of Central Intelligence
FROM : Robert M. Gates
SUBJECT
?
?
Improving the Quality of Intelligence
Now that you have seen NFAC's program (Tab A) for improving the quality of
analysis, you may find my ideas on this subject of interest.
I have written you in some detail on several occasions my perceptions of
the problems of analysis in NFAC. One of these papers you converted into a
memorandum to John when he was first appointed. I would summarize those problems
in three words: stagnation, isolation, and arrogance.
Analysts become stagnant when they are not required to question
their assumptions and make those assumptions explicit; when they
spend too much time behind a desk talking to the same people day
in and day out; and when there is no premium on quality production.
As one of my mentors once said, "Being wrong never prevented any-
one in CIA from being promoted."
Analysts are isolated when their contacts with the outside world
are too limited and they have too little idea of what other
agencies and other authorities are thinking about their subject.
They are too isolated when they cannot read the language of the
country on which they work or when they have not traveled in
that country or have no outside experience.
Analysts are too arrogant when they assume that other intelli-
gence agencies have little if anything to offer; when they reject
outside criticism as being uninformed; and when they and their
supervisors refuse to acknowledge that serious errors have been
made in the past--sufficiently serious as to shake the confidence
of a less smug group of people.
Serious remedies to these problems have been shunned for years perhaps because
many managers had no idea how to correct the problems even if they recognized them
and because there has developed in the analytical directorate a preoccupation
with protecting the sensitivities of the analysts from criticism (or influence
from the policymaker) that might dampen morale.
SE ET
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
25X1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0 25X1
The NFAC reorganization has been beneficial both in terms of reducing bureau-
cratic obstacles to better analysis and in providing the proper foundation for
significantly improved cross-fertilization and multidisciplinary work among
previously isolated political, economic and military analysts. Nevertheless,
the reorganization has left untouched the above-stated three problems and others
importantly affecting quality of analysis. What follows is my notion of specific
steps that can and should be done to get at these problems:
1. Rotations to Policy Agencies: I would require that any intelligence
analyst serve a minimum of a one-year tour in a policy agency or consumer of CIA
analysis before being promoted to division chief (supergrade). These rotations,
at a rate of perhaps six to eight a year initially, would be to policy desks at
State (not INR), Defense/ISA or elsewhere in OSD, the NSC Staff, Commerce,
Treasury, the Congress and perhaps other intelligence organizations as well. By
way of analogy, we have a large number of people in NFAC charged with designing
and building cars who have never seen one, ridden in one, or driven one. The
results show it. A better understanding on the part of senior managers in NFAC
of how the policy process works and how CIA intelligence analysis is used is
essential to improving the quality and responsiveness of our product. For divi-
sion chiefs already in place without this experience, further promotion would
depend upon their acquiring it.
2. Research Programs: It is evident that there is still too much "bottom-up"
preparation of NFAC's research program. Too many marginal or simply "nice-to-have"
reports are being prepared and too much analysis is being prepared for other
analysts, thereby creating a shortage of resources to address the really important
questions. I would require that the research programs of each of the offices be
vetted by the regional office chief with the appropriate Assistant Secretary of
State or Deputy Assistant/ISA, as well as appropriate NSC Staff officers and others
in the government (including the DDO division chief) at the same level to get their
reading on the value of the proposed work. In each office's proposed research
program, I would require a brief review of research on the same subject area being
undertaken in the other intelligence analysis units of the Community. I believe
there are a significant number of areas where competitive analysis is not necessary
and where work done by DIA or INR or others could usefully stand on its own without
CIA devoting the resources to duplicating it. That one would go down hard but I
believe it has merit. Finally, I believe you both should have an opportunity to
review the main elements of the research program. No one brings a better perspective
or understanding of the limited resources available to the analytical problem than
the DCI and DDCI. I would still make considerable allowance for the self-initiated
paper that was relevant or which addressed a prospective problem or crisis that has
not yet come up on the policymakers' horizon.
3. Training: Training is needed at two levels. For the analyst as described
above, there is a need not only to improve analytical skills and knowledge of the
subject country or area, but also a better understanding of how policymaking is
done, how research programs are planned, and how these things are done in other
agencies. Analysts' skills must be refreshed. There is a tendency in NFAC to do
this with extremely expensive year-long sabbaticals to distant schools of high
reputation. I place a greater premium on an analyst being required every year or
two to take one or two courses in their field of specialization at Agency expense
2
SECRET 25X1
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0 25X1
at metropolitan D. C. universities. Many such courses are available and many
are of high quality. The Agency could support this with relatively little cost.
There also is a need to train managers, not in the sense of traditional management
skills, but as they rise to positions of responsibility more detailed knowledge
of how the other elements of the Agency and Intelligence Community work, how the
policymaking system works, how we relate to Congress, and so forth is required.
A considerable revamping of NFAC's training program probably is needed.
4. Promotions: One way to improve the quality of intelligence is to make
it clear to every analyst that his promotion depends on the quality of the
analytical work he does. His interpersonal skills, his ability to coordinate
smoothly, how he gets along with his branch chief, and so forth would not have
the high priority they have now. Instead, the primary requirement for promotion
would be to demonstrate a consistent record of quality analysis and accurate analysis.
A list of an analyst's production and copies of that production should accompany
any recommendation for promotion beyond the grade of GS-ll. All members of a pro-
motion panel would be required to review that production and greater weight to
production would be given than to any other factor, perhaps 50 to 60 percent.
(No NFAC office keeps a file of its analysts' work so that a manager could review
it to assess in toto how accurate an analyst is, whether he is improving or getting
worse, or otherwise make a judgment on the skill for which he is paid--analysis.
That should be remedied immediately.)
5. Evaluation: NFAC has no internal evaluation process worthy of the name.
The Senior Review Panel does not bring the urgency or the energy to the task of
evaluation, even if they had the charter to undertake it. I believe a small in-
house evaluation office should be established, perhaps headed by an energetic
outsider with a reputation as a critic of CIA analysis, and with a fairly broad
charter to examine and evaluate virtually any product of the Directorate and
provide its views directly to the D/NFAC. Its evaluations would go into an
analyst's production file.
The above five areas are the most important and should have top priority.
There are, however, a number of additional steps that would help improve analysis
and/or help ensure that resources were better managed:
-- Getting Up and Getting Out: An analyst too long stuck behind a desk
is a serious enemy of quality intelligence. It is essential that our analysts
spend more time developing contacts on the outside and benefitting from outsiders'
experience and views. NFAC should develop both on an office basis and on a
directorate basis a much more aggressive program of contacts, conferences and
seminars on important subjects. By the same token, analysts need to take greater
part in academic and business conferences on relevant topics. Incentives should
be created to promote this. Finally, of course, I would continue and try to
-- Language: Too many analysts do not speak the language of the country
or countries on which they work. For example, I am told NFAC has one analyst
3
SECRET
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
25X1
25X1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
who has Polish; it has no analysts who read Bulgarian. This lack of a language
capability is important in its own right but also is symbolic of a broader
problem--the fact that in a variety of areas people with the requisite knowledge
of the culture and society of the country on which they work is very thin. A
long-range program needs to be undertaken to remedy this. A large part of it
is the proper targeting of recruitment efforts to get people with those specific
skills.
-- Provocation: Changes need to be built into the system to ensure the
questioning of assumptions. NFAC papers need to be shared more widely with the
DDO--particularly at higher levels of the DDO--as well as with other elements of
the Intelligence Community (even though they will remain NFAC products). NFAC
also needs to take steps to hire a smattering of good, provocative analysts who
can be brought in as lateral entries to spice the intellectual stew. Finally,
certain important papers might usefully be shared with contractors for review or
serve as the subject of seminars with a group of outside experts who have been
asked in to review the paper and offer their thoughts.
-- IC Work: If the Director of NFAC is to remain the Deputy Director for
National Foreign Assessment, he needs to shoulder additional responsibilities for
keeping track of analytical work underway in the Community. The DDNFA (and through
him the DCI) cannot shirk responsibility for having information on where the
Community's analytical assets are targeted. He ought to know not only how many
analysts are addressing what subjects in CIA, but also in other agencies. Thus,
if the Director were asked what recources in the Community were devoted to the
balance of forces in Korea, this information could be provided without having to
query each agency individually. The circulation of such a compendium also should
help all agencies target their own resources better and avoid duplication.
-- Future Probabilities: As Bill Colby wrote in his recent Studies in
Intelligence article, we need to develop a capability to step beyond ordinary
analysis through new techniques to project future probabilities rather than
explain the past. There have been experiments with a variety of techniques in
this new discipline, but greater attention needs to be devoted to refining and
improving them in the search for a better system. (See the excerpt from Colby's
article at Tab B.)
-- Defining Likelihood: Over the years there have been a number of
debates over the meaning of terms like "probably," "likely," "almost certainly,"
"on balance," and so forth. Our estimates in 1962 said that "on balance, the
Soviets will not put missiles into Cuba." As Dick Betts has pointed out, the
policymaker would have a very different view of a prognosis if it stated that
the odds were 60:40 that the Soviets would not undertake that step. DIA has
worked with this kind of an approach off and on over the years. I think there
is some merit to it, in the sense that it requires an analyst to define a little
bit more closely just what odds he really has in mind when he blithely applies
one of the above terms. I would be inclined to move in this direction, at least
on an experimental basis.
-- Bibliographies: The life on an intelligence research paper or assess-
ment is a short one. In part this is because we publish so many reports that
4
SECRET 25X1
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
 ._J 25X1
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0
25X1
analysts and policymakers alike lose track. The fact is that there is a wealth
of intelligence research on the shelves, most of which has not been touched in
years, but which still has great value. I believe the Office of Current Reference
in NFAC should begin preparing bibliographies of intelligence work prepared over
a five-year period on a range of countries and subjects. Most of this is on
computer; I cannot imagine it would require any enormous application of resources.
Once the bibliographies for NFAC were done, we might take a look at doing some-
thing on a Community basis as an overall service. In any event, these bibliographies
would be circulated fairly widely among senior staffs and other analysts, including
our own, because in many instances new papers are requested on subjects where
analysis done a year or two ago is still perfectly valid. This could help protect
scarce resources for addressing new subjects.
-- Global Issues: Steps should be taken to reconstitute in this office work
that had been started on instability, and which now has faltered with dispersal
of the analysts to various regional offices. As we look at the past six years
and the years ahead, the lack of anyone working on insurgency--its causes, character,
weaknesses, and support--is a scandal. Two or three good people need to be assigned
to this immediately. In cases of both instability and insurgency, the Office of
Global Issues should offer a core group to track the problem generally and work
with the regional specialists in other offices.
These comprise my first cut at a specific program for improving the quality
of intelligence. While most of these ideas are ones that I have developed over
the past ten years or so, I freely acknowledge that I also have drawn on the
Heritage Foundation Report; Leo Cherne's December 1976 PFIAB report; articles
in Studies in Intelligence by Bill Colby, Tom Lattimer, Dick Betts and others;
and a variety of other evaluations of CIA's analysis. You probably will have
additional ideas as well, which would be welcome. None of my proposals require
reorganization. None require substantial additional resources. Few will be
popular. Any effort to implement them will need to be preceded by an educational
program designed to try to break through the shell of arrogance and smugness
that NFAC is highly esteemed and its work widely respected throughout the policy-
making arena and on the Hill. I envision a series of officials, beginning with
the two of you, who would come and speak frankly for 15 to 30 minutes with the
office chiefs outlining your overall dissatisfaction and problem with NFAC research
and analysis. I would probably try to enlist to do likewise former officials like
Brzezinski, perhaps Dick Holbrook so that the office chiefs would not think I was
stacking the deck with "hardliners," and others who submitted thoughtful responses
to NFAC's survey of consumers in the last Administration. You might also reach
back to the Ford Administration for some officials and perhaps some of your colleagues
from the earlier incarnation of the PFIAB. I do believe, however, that a series of
such presentations is required to at least get the attention of the office chiefs
and lay the basis of credibility for any serious, broad program to address the quality
of analysis.
A final thought. Many of these present NFAC managers would probably tell you
that they are implemented in some manner or another. What is required, however,
is not exhortation or "emphasis" but concrete programs and specific requirements
in areas such as promotion, rotation, evaluation and so forth. Without such require-
ments, such a program will have little real effect.
Robert/p. Gates
5
SECRETJ 25X1
Declassified in Part - Sanitized Copy Approved for Release 2013/01/24: CIA-RDP89G00720R000100050014-0