CONSULTANTS REPORT ON THE ESTIMATIVE PROCESS

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP86B00269R000800060001-6
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 28
• Document Creation Date: December 14, 2016
• Document Release Date: June 13, 2003
• Sequence Number: 1
• Publication Date: August 8, 1963
• Content Type: MF

63- 27Y -e-0,
Approved For Release 2003/06/26: CIA-RDP86B00269RO0800060001-6
MEMORANDUM FOR, Acting Director
Consultants Report on the Estimative Process
1. For your and the SCI's eyes onl I would like to record that I
am a bit die ointed in the re ort that 25X1
have submitted on the estimative
prQC945. lWanz 01 Waar, Tiley rla~ suggested has been previously suggested
on several occasions. For example.
(1) Cutting down the number of coordinated papers.
(Z) Keeping a wide variety of views on the Board.
(3) Reflecting diverse views.
(4) Calling in consultants (which we do with Hyland and others).
(7) Making more information about foreign intelligence
organizations available.
(9) Making validity studies in depth.
On the other hand. I do think they have valid points in their technical items
about (5) indicating when earlier estimates are quoted, and (6) being bolder
in estimating the impact of U. S. policy. Their point in (8) on signs that a
mutation in an existing pattern is taking place I think is very important.
2. The annex on "Preconceived Notions" I am afraid is a bit a~4
as perhaps is the entire study, because these fellows are so closely
associated with the Board of Estimates.
Approved For Release 2003/06/26: CIA-RDP86B00
 Approved For Release 2003/06/26 : CIA-RDP86B00269JR000800060001-6
25X1
3. Having read this report a couple of times, I am disturbed that
perhaps there is slippage in getting these done because we have been on
top of Sherman Kent for months to bring in new blood on the Board, which
he has done with d to set up a specific tour of
duty so that there can be ease in rotation. On the other hand, I am still
very concerned that the Board is "too much of the same,'' and I would
dearly like to see a couple of real hair-shirts from Missouri (to mix a
metaphor) to check some of the bureaucratic, governmental, acaderatc
views that I am sure consistently arise in Board discussions.
Lyman B. Kirkpatrick
Executive Director
LBK:drm
Distribution:
Origins - Addressee w/basic
ExDir
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6
 Approved For Release 2003/0S#L:UJRDP86B00269R000800060001-6
22 July 1963
Dear Mr. McCone:
I enclose the report of the committee of consultants which
you appointed to examine the estimative process in the light of
the Cuban experience and to make recommendations for improving
Introduction
Recommendations
Annex A, "Preconceived Notions"
Annex B, Rejected Proposals
Approved For Release 2003/065?W F `( f~ 4O 0800060001-6
25X1
 Approved For Release 2003/06/26 S6IUEJ86BOO269R000800060001-6
22 July 1963
1. On 17 May, 1963, the Director of Central Intelligence asked
to examine the estimative process in the light of the
Cuban experience, and to make recommendations for improving it.
received a form of consultant's contract not compatible
with his other committments and this mistake was not corrected in time
to make it possible for him to serve on the committee. The other three
members of the committee met for three days (8, 9, and 10 July 1963),
in the office of ONE, exchanged draft proposals during the next week,
and met again on 18 July to discuss the final form of their report.
2. It is obvious that no thorough investigation of the estimative
process could be made in this brief period. A number of estimates
dealing with Cuba and the USSR were read, two development files were
studied, and a few members (or former members) of ONE were inter-
viewed. As consultants, the members of the committee have read, over
the past few years, a number of estimates, both in final and in draft form.
was a member of the Board of National Estimates
I evertheless, none of us feels that he
is in any position to make emphatic and final judgments about existing
estimative procedures.
GROUP 1
Excluded tram antamatle
Approved For Release 2003/06/26 : C 00269R0008 001"d"ad
L ! c au
 Approved For Release 2003/0&C iRDP86B00269R000800060001-6
3. Two of the committee believe that a study in depth of the
estimative process by a small group of informed outsiders (not our -
selves, we hasten to add) might yield valuable results. such a study
would require access to all constituents of the intelligence community
and knowledge of personnel as well as procedures. It would take
several weeks, if not months. For these reasons the third member of
the committee believes the project is not feasible, and he doubts that
it would be useful.
4. The discussion and suggestions which follow are only opinions
offered by partially informed outsiders. Perhaps we are fairly typical
consumers, and our report might be taken as representing the reactions
of some consumers to estimates. But we should emphasize that we have
not studied the problem in depth, and that therefore many of our sugges-
tions may be superfluous, impractical, or just wrong-headed. We should
also emphasize that our suggestions deal almost entirely with details of
procedure, and that even the best procedure will not guarantee good
estimates. The essential thing is to find and retain men who have wide
knowledge, broad vision and good judgment. As long as the intelligence
community has such men it will do a good job, whatever its procedures.
25X1
Approved For Release 2003/06/26 :16B00269R000800060001-6
 Approved For Release 2003/06/w:"C1'1-dDP86B00269R000800060001-6
RECOMMENDATIONS
The best way to remedy any deficiencies which occur in the
estimative process is to have full and careful discussion of each prob-
lem. In theory the present machinery should ensure such a discussion.
In practice, we believe that there are some obstacles and we offer the
following suggestions as means of reducing or removing these obstacles.
1) Cut down the number of coordinated papers,
Men who have too much to do in too short a time may not
see all possibilities or sense the full impact of changed
conditions. Pressure to meet deadlines forces limitation
of discussion. Effort devoted to trivial papers means less
time and energy for major problems. It would require more
knowledge thah we possess to determine what papers should
be eliminated, but we suspect that many "country papers"
fall in this category. We also suspect that requests for
coordinated papers are made, and accepted, too easily.
But in any case, there is overloading and it has bad results.
(Note that we do not recommend any increase in
personnel. The number of people now involved in the
estimative process is about as large as can be used
efficiently. More people would simply mean more spe-
cialization and more meetings. What is needed is the
- Rl -
Approved For Release 2003/06/2g[c - ( P86B00269R000800060001-6
 Approved For Release 2003
IX- RDP86BOO269R000800060001-6
stun
overall view and time in which too develop it and
discuss it. This can be done only by a small group
of carefully selected men).
2) Make sure that there are always men with a
wide variety of views and experience on the Board of
National Estimates. The present distribution of member-
ship on the Board seems well designed to achieve this end.
As we understand it, it includes a core of men with long
experience in intelligence, other men who have worked
in the foreign or armed services, and a younger group who
will move on to other positions after a few years on the Board.
We suggest that it might be helpful when there are vacancies
to add two or three rotating members who would serve for
only a year or so. These men could be borrowed from other
branches of the government or from universities. If
properly selected, they could contribute new ideas and
encourage re-examination of accepted formulae.
3) Make sure that diverse opinions of other members
of the community are fully represented in contributions and
coordination meetings. Here our knowledge is limited and
our recommendations are only tentative. 'We believe that
Approved For Release 2003/06/26 [ P86BOO269R000800060001-6
 Approved For Release 2003?/i4-RDP86B00269R000800060001-6
that the problem of overloading is even more serious
outside CIA than inside. Some representatives apparently
have little time to inform themselves or to think about the
problems which they are supposed to discuss. We are
also uneasy about the tendency to concentrate military
contributions in the DIA. There are often sharp differences
of opinion among the armed services. Expression of these
differences can be very helpful in forming sound judgments.
4) Call in a small group of consultants when diffi-
cult problems are to be discussed. We advance this idea
with some hesitation and we realize that it might cause
more trouble than it was worth. But it might be a way of
getting new points of view and of avoiding blind spots. Two
or three consultants would probably have a greater impact
than a single outsider who might hesitate to contradict
the insiders.
5) Indicate in some way when earlier estimates
are being quoted or paraphrased. There are many ways
of doing this (footnotes, quotation marks, etc.) and any
one would be satisfactory. But attention should be called
to repetitions, in the hope that this would lead to
re-examination of accepted positions.
-R3 -
Approved For Release 2003/0 C RETDP86BOO269R000800060001-6
 Approved For Release 2003/0 4 kRDP86B00269R000800060001-6
6) Be somewhat bolder in estimating the impact
of U. S. policy and attitudes on the rest of the world. This
is a difficult and dangerous task because it can lead to implied
criticisms of U. S. policy. But it seems to us that failure
to do this is one reason why preconceived ideas (which are
still correct) sometimes cause faulty estimates. To be
specific: if the Kremlin in the early months of 1962 had
reason to think that the posture and resolution of the U. S.
were weak, then it should obviously have estimated that the
risks of the Cuban adventure were not unduly high. Thus
the accepted doctrine that the USSR would avoid grave risks
was probably correct; the error was caused by the failure
to estimate that the Kremlin estimate of U. S. policy had
changed.
7) Make more information about attitudes and
methods of foreign intelligence organizations available
to estimators. We believe, for example, that the Kremlin
attaches more importance to certain kinds of evidence, and
less to others, than the U. S. intelligence community would
do. We suspect that what information has been acquired
about Soviet intelligence procedures is not very widely
disseminated. But if the Kremlin does use a peculiar scale
- R4 -
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6
SECRET
 Approved For Release 20031:?JA-RDP86B00269R000800060001-6
in weighing evidence and if this is not generally known,
then the result is bound to be a faulty estimate. One
reason for inadequate estimates on Cuba in 1962 was
that the intelligence community, as. a whole, believed
that Soviet leaders must realize how dangerous the intro-
duction of IRBM's and MRBM's would be. Better knowledge
of Soviet intelligence procedures might have made it possi-
ble to suggest that the Kremlin was undervaluing the evidence
which should have led it to this conclusion.
8) K aco?st twatchfour_ nsthatamutes aeon
in an e~xis~t~g pattern is abort to take lp ale. This is pro-
bably a superfluous and useless recommendation. Certainly
everyone is on the look-out for such signs and it is much
easier to identify them in retrospect than at the moment
when they occur. Existing patterns can accommodate them-
selves to a large number of new ingredients and gradual
changes bring about only gradual shifts in policy. But
there are some new ingredients so large or so sharp that
they may break the old pattern and it is not entirely
impossible to identify these. Again to be specific, in
1962 the unprecedented situation in Cuba and the Sino-Soviet
dispute were probably such ingredients.
-R5-
Approved For Release 2003/06/26 : CIA-RD 86B00269R000800060001-6
SECRET
 Approved For Release 2003/06/ E ~3?flDP86B00269R000800060001-6
9) Make a few validity studies in depth. At present,
validity studies usually involve only a comparison between
a new estimate and the immediately preceding estimate on
the same topic. Often there has not been enough change in
the interval between the two papers to reveal errors or biases.
Our experience in studying the Cuban series of estimates
leads us to believe that there would be some value in making,
from time to time, a validity study of a whole group of related
papers. This might expose persistent biases, rigidities, and
inconsistencies. We know that one such study of a group of
Soviet estimates was made some years ago. We suggest
that more exercises of this sort would be useful, but with the
caveat that if they became too frequent (say more than one
or two a year) their value wo.uld_ sharply decrease. We also
suggest that some of these studies be made by consultants,
both to reduce the burden on the ONE Staff and to ensure a
fresh point of view.
Next to the problem of full discussion comes the problem of
communication. It seems to be fairly well agreed that estimates often
save less impact than they should, and that important ideas are missed
by hasty readers. Nothing can be done to change the fact that everyone
Approved For Release 2003/06/2,? CIA- RDP86B00269R000800060001-6
 Approved For Release 2003/0?:URDP86B00269R000800060001-6
in government has too much to read and that hasty readers will always
miss significant points. No one style or form of presentation will be
equally effective with all readers. But perhaps something can be done
to make it easier for most consumers to read estimates with more
understanding.
1) Put key sentences at the beginning of each
section of the estimate, e. g. "We believe that the food
problem in Cuba is (is not) causing unrest." Reasons
for the statement should follow. At present many impor-
tant statements are made only after a long list of pro's
and con's and thus lose their full impact. Conclusions,
as now written, do not always remedy this fault. Perhaps
a string of key sentences at the beginning of a estimate
would be better.
2) Call attention to possible courses of action
which might have serious consequences, even if the estimate
is that they are highly unlikely. Too many readers think
that "probably notf9 equals "never, " or that "the possibility
cannot be excluded" actually means that it is excluded.
Positive rather than negative formulae would help, but
we must confess that we are not very good at devising them.
Perhaps a phrase such as "there is at least a twenty per
-R7-
Approved For Release 2003/06/j6rC
 Approved For Release 20039 r1i
IA-RDP86B00269R000800060001-6
cent chance that" approaches what we need. And perhaps
a little more argumentation showing the advantages which
might result from the adoption of unlikely courses of action
would make the consumer take them more seriously.
3) Emphasize new ideas or facts. Well informed
readers find much that is familiar in estimates, especially
in the larger estimates. This lulls them into missing
useful new material. * Perhaps some typographical device
(e. g. side-lining) could be used to indicate places where
significant new ideas and facts are introduced. Or perhaps
more of the historical and purely descriptive material could
be put in annexes.
4) Make sure that the consumer is reminded of
key items in other estimates. This is probably an
*For example, there was a very important change in the Cuban
estimate of 19 September, 1962 (SNIE 85-3-62). Up to that time, all
estimates and memoranda had asserted that the USSR had little to
gain by placing MRBM's or IRBM's in Cuba. SNIE 85-3-62, on the
contrary, pointed out that the Kremlin might see decided advantages
in such a deployment. But no emphasis was put on the fact that this
was a changed estimate and we suspect that few readers noticed the
shift. One of our groups reading the Cuban estimates seriatim
(which few consumers ever have time to do) still failed to observe
the fact that there had been a change on this highly significant point.
Approved For Release 2003/06/ IEWT P86B00269R000800060001-6
 Approved For Release 2003/61
f l I RDP86B00269R000800060001-6
unnecessary caution, but there is some danger that drafters
of an estimate will assume that consumers remember related
estimates as well as they do. For example, in SNIE 85-3-62
a little more emphasis might have been placed on the fact
that the estimate of the number of operational Soviet ICBM's
had recently been sharply reduced.
Approved For Release 2003/06/1E VIL
IRD
I P86BOO269R000800060001-6
 Approved For Release 2003/06/Eq&U-1,
In response to one of the requests put before, us, we have
Z
examined the estimates on Cuba from January 8 to September 19, 1963,
It is generally acknowledged that there was a failure of intelligence, in
the sense that USIB did not warn the government that the Soviets might
be planning to place offensive ballistic missiles in Cuba, and indeed
suggested that such a decision was unlikely, though not impossible.
This failure has been widely attributed to the presence of "preconceived
notions" which prevented ONE, and in fact the entire intelligence community
from making a proper judgment. We therefore find it useful to make some
observations on the function, utility and drawbacks of such preconceived
notions. To do so is in fact indispensable to any search for remedies
designed to minimize these drawbacks.
In dealing with a major country, intelligence officers approach
their task with a set of expectations of likely patterns of behavior.
This is especially true in the case of the Soviet Union, the most
frequent target of intelligence. This set of expectations, sometimes
called preconceived notions, is. based on professional knowledge of
political structures, processes and personalities in the country concerned.
It is the distillate of years of experience and has been tested by, and
refined through, repeated intelligence estimates. It must be understood
-Al -
Approved For Release 2003/0c&EDP86B00269R000800060001-6
 Approved For Release 2003/ 6/26 : glA.RDP86B00269R000800060001 -6
that the formulation of such a set of expectations about a country's
likely patterns of behavior is both inevitable and indispensable. It is
inevitable that, in doing their job, professionals will, more or less
deliberately, build up the set. It is indispensable because the set is
a most valuable tool in producing timely, coherent, articulate and, on
a probability basis, accurate intelligence. If no pattern can be estab-
lished, then prediction is almost impossible, and significant variations
can be demonstrated only by comparison with an already accepted norm.
Nevertheless, these sets of expectations have their limitations
and drawbacks. The history of intelligence is full of instances in
which an opponent achieved surprise by adopting a course of action
that seemed unlikely in view of the set of expectations dominant among
intelligenge officers. The Japanese attack in Pearl Harbor is one of
the most conspicious examples. We are not concerned here with what
may be called "technical surprise." A "technical surprise" is not
incompatible with the prevalent set of expectations. The surprise occurs
because the opponent was successful in concealing a particular capability
or in keeping a particular course of action secret. We are concerned
with an opponent's behavior that surprises because it is incompatible,
or seems to be incompatible, with our prevalent set of expectations.
There are three possible causes of this type of surprise.
First, the opponent's basic pattern of behavior may change as
a result of changes in leadership or various other important conditions,
- AZ -
Approved For Release 2003/06/25 Er
86BOO269R000800060001-6
PbD
 SE
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6
and our set of expectations may not register this change quickly and
correctly. Our set, in other words, is out-of-date. Intelligence
professionals must obviously be on the alert for such changes and,
generally, take care that their working set of expectations does not
become obsolete, stale or unduly rigid.
Second, the opponent may act irrationally. Irrational behavior
is, of course, very difficult, if not impossible to predict. It may take
all sorts of directions that make no sense to the rational mind. Only
if a country's leadership has a record of frequent lapses into irration-
ality will this observed fact become incorporated into our set of
expectations. But even then the estimate of future behavior will be
extremely difficult.
Third, the opponent may adopt a course of action that seems to
us to be in conflict with our set of expectations, but actually is not.
A set of expectations is based on the attitudes and predispositions that,
we have learned, strongly affect the opponent's behavior. The important
point to note is that attitudes, though they shape behavior, do not determine
it. Behavior also depends upon the information which the opponent possesses.
It is this information which brings his attitudes into play and which gives
him a basis for calculating the advantages and disadvantages of a course
of action. Hence, we may go wrong in an estimate, not because our set
of expectations is faulty, but because we assume that the opponent acts
on approximately the same information that we have. Clearly, to minimize
Approved For Release 2003/06/26 S --C1X- -86B00269R000800060001-6
IGh
 Approved For Release 2003/0Ed: RDP86B00269R000800060001-6
this source of possible intelligence failure, we require not only a good
set of expectations but also some idea of the information that the opponent
uses in his cost-gains calculus.
We do not think that the failure of the intelligence community
to predict the Soviet deployment of medium range missiles resulted
essentially from its set of "preconceived notions" about the Soviet Union.
In our opinion, the community's set of expectations was, at that time at
least, sophisticated and realistic. According to the set, the Soviet Union
was considered unlikely to depart from its cautious behavior in the military
field or to undertake actions involving a high risk of war with the United
States. That this expectation about Soviet behavior was basically correct
was, after all, confirmed by the great caution with which Kruchshev acted
when the missile crisis broke in October.
Nor do we believe, as some experts have suggested, that the
Soviet decision was largely irrational. It is true that the Berlin state-
ment, Chinese criticism, and possible differences of opinion within the
Soviet leadership had put Khrushchev under some pressure to score a
success in foreign policy. It is also true that during the last year he has
seemed somewhat less sure of himself than before and has abruptly
reversed himself on several issues. But while he and other members of
the top leadership may have been perplexed by current difficulties there
is no evidence to suggest that they became reckless gamblers. At the
most, they may have been a little more willing to take a chance than
- A4 -
Approved For Release 2003/06/26M86B00269R000800060001-6
 Approved For Release 2003/ 'Ii-RDP86B00269R000800060001-6
they would have been under other circumstances. And a strong case can
be made for arguing that Soviet behavior was entirely compatible with
rational decision-making.
We believe that more important factors in the Soviet decision
were inadequate information and, resulting from this, an erroneous
assessment of the situation. They did not have the fund of information
about the mood and temper of the US government and people which the
intelligence community thought they must possess. They saw the balance
of advantages and disadvantages differently from the way Washington
officials thought they should see it. It is in this area that the intelligence
community failed, although we defer for the moment the question of the
extent to which the community can be held accountable for the failure.
It is fairly clear in retrospect that the intelligence community
both underestimated the gains that Moscow believed it could derive from
introducing offensive missiles in Cuba and overestimated the risks which
the Kremlin thought it ran in adopting this course of action. In reading
the various estimates seriatim, we were struck by the fact that, although
the question of the deployment in Cuba of Soviet medium-range missiles
was raised in a Memorandum to the Director of January 8, 1962, it was
concluded repeatedly until early September that the Soviet Union could
not thereby add substantially to its strategic capabilities. The reason
given was that targets in the United States capable of being reached by
medium-range missiles from Cuba were already covered by ICBM's
Approved For Release 2003/0652 DP86B00269R000800060001-6
 p EMU
Approved For Release 2003/06/A . P86B00269R000800060001-6
deployed on Soviet territory. Only on September 6 were doubts raised
about this conclusion. The special NIE of September 19, conceded that
Soviet planners might see some advantage in deploying IRBM's and MRBM's
in Cuba in order to supplement the limited number of ICBM's believed to
be operational in the Soviet Union. The delay in coming to this conclusion
is surprising in view of the fact that it had been estimated several months
earlier that the USSR had only a small number of operational ICBM's. In
this situation one obvious way for the Soviet Union to supplement its small
force of ICBM's was to deploy a proportion of its large supply of medium-
range missiles within appropriate striking distance of the United States.
We are not, in fact, sure that such a deployment constituted, objectively,
a substantial improvement of their strategic position, though Soviet planners,
of course, may well have thought so. We are sure, however, that the Soviet
Union would have gained enormously in prestige and, indirectly, greatly in
the military balance of power and deterrence, if it had succeeded in
installing "offensive" missiles in Cuba and maintaining them there. Soviet
leaders, like all rational leaders, must be expected to accept a higher
level of risk for great gains than for small gains. We note that this was
not acknowledged in the estimates we examined.
Even when it was admitted that the Soviet Union might secure
possible strategic advantages by employing IRBM's and MRBM's in
Cuba (NIE 85-3-62, September 19, 1962), the conclusion remained that
the Soviets were unlikely to do so because the risks would be too great
- A6 -
Approved For Release 2003/06/26 A-RQR86B00269R000800060001-6
 Approved For Release 2003/0 2 091-RDP86BOO269ROO0800060001-6
and acceptance of so great a risk would represent a sharp departure from
Soviet practice. This was the crucial view that prevented assigning any
high degree of probability to the course of action that the USSR pursued.
However, this view was justified only if the Kremlin assessed the risk
factor approximately as it was assessed in Washington. The intelligence
community thought that the risk was very high. It knew it to be so. Yet
it is fairly clear in retrospect that the Soviet leaders did not share this
knowledge, that they assessed the risk at a lower level and that, given
this assessment, their deployment of "offensive" missiles in Cuba, did
not seem unduly risky to them. In that case, their action did not, as
they saw it, constitute a sharp departure from their normally cautious
military behavior.
The extent to which the intelligence community can be held
accountable for neglecting the possible gap between their information
and our information is not clear to us. Such gaps are a possible source
of intelligence failure to which the professional should be alert. In
retrospect it would certainly have been better if the possibility of this
gap, and its possible implications, had been acknowledged in the estimates.
If it had been, it might have qualified the impact of an erroneous
conclusion.
On the other land, it is patently very difficult for our intelligence
services to have sufficient knowledge of Soviet intelligence to ascertain
or predict intelligence failures in the Soviet Union. Moreover, there
- A7 -
Approved For Release 2003/06/26uECRR86B00269R000800060001-6
 Approved For Release 2003/06/:Lqt-9DP86B00269R000800060001-6
is reason to believe that much of the intelligence community, including
ONE, is not provided with such knowledge of Soviet intelligence activities,
and their organization as is available to a part of the community. Yet
this is the kind of knowledge which would have been required for a correct
prediction of the Soviet move in Cuba.
Moreover, the intelligence officer is also handicapped in speculat-
ing about national differences in estimating the risk of particular courses
of action as long as he may not take relevant United States policies fully
into account, and have adequate information about these policies and their
implementation to the extent that they are known, or must be assumed to
be known, to the Soviet Union.
However, even if these handicaps were removed, or at least
appreciably diminished, the consumer of intelligence would be wrong
to expect the intelligence community to be able in all cases to obtain
sufficient information on the opponent's estimates of the advantages and
disadvantages of contemplated courses of action. For this reason alone,
and not necessarily as a result of "preconceived notions, " the best
intelligence service cannot help failing from time to time. It would
be unfair and utterly unrealistic, in our opinion, to expect otherwise.
On the other hand, the inevitability of occasional failure does
not justify either complacency or resignation. No organization stays
at the same level of performance over a long time -- if only because
- A8 -
Approved For Release 2003/0 t k DP86B00269R000800060001-6
 Approved For Release 2003/06/2 E IRDP86BOO269R000800060001-6
of changes in personnel -- and there may always be room for improve-
went. There are good reasons for reviewing institutional practices
from time to time with a view of raising the batting average by a few
per cent.
Approved For Release 2003/Ol ffDP86BOO269R000800060001-6
 SECRET
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6
Rejected Proposals
We have considered, and have rejected as impractical, the
following proposals.
1) "Devil's advocates", professional dissenters, etc.
We hope that the intelligence community will always include
tough-minded, skeptical individuals who will advance unpopular
ideas and who will query accepted doctrine. But any attempt
to institutionalize this function would be self-defeating. No
one can be skeptical all the time and the views of a man who
was being paid to act as a professional dissenter would not
carry great weight.
Z) Dual or multiple estimates, For example, one
group might make an estimate from the Soviet point of view
while another estimated the same problem from the US point
of view. The difficulty here is what to do in case of disagree-
ment. Either some higher group would reach a final decision,
in which case we would be back where we started, or the
consumer would make a final estimate, in which case the
community would have abdicated its responsibility. Moreover,
there is certain artificiality about this procedure which would
probably erode its value very quickly.
-B1 -
Approved For Release 2003/06/260 Erp6B00269R000800060001-6
 Approved For Release 2003/0DP86B00269R000800060001-6
3) Make less effort to reach agreed language in
estimates, and encourage dissents. It is possible that under
pressure of time some dissents which might have been helpful
are never made. But while a few footnotes are useful in alerting
the consumer to the difficulties of a problem a steady diet of
footnotes would blunt his attention. If there were too many
points on which agreement could not be reached the consumer
would have to make his own estimate and the community would
again have abdicated its responsibility. In short, dissents on
minor points should not be encouraged, and it seems unlikely
that dissents on major points are, or could be suppressed.
We note that in the Cuban estimates of 1962 ( and in others
which we can remember) the worst: mistakes were not caused
by suppressing disagreements or by watering down correct
judgments in order to obtain an agreed text. They were caused
by mistaken judgments which were held almost unanimously
throughout the intelligence community.
4) Make more effort to keep estimates up-to-date by
issuing frequent revisions. Carried to an extreme, this would
obviously lead to a confusion between estimates and current
intelligence. Even a moderate increase in the number of
revisions would probably do more harm than good by forcing
estimators to concentrate on minor fluctuations in policy rather
- B2 -
Approved For Release 2003/06/28[ClR 14IP86B00269R000800060001-6
 SECRET
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6
than on longer-range problems. When an estimate has been
overtaken by events nothing can be gained by issuing a hasty
revision which simply says that the preceding estimate was
wrx,ng. When a situation has gone operational revision of
earlier estimates should wait until there are some firm indica-
tions of new policies and courses of action. At this point current
intelligence will give policy-makers the flow of information which
they require. To refer once more to our case study, we doubt
that more frequent revisions of the basic estimates on Cuba
between 1 August and 19 September, 1963 would have changed the
opinions held by the intelligence community in any important
way.
Approved For Release 2003/06/iE CWUMP86B00269R000800060001-6
 CENTRAL INTELLIGENCE AGENCY
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6
OFFICE OF THE DEPUTY DIRECTOR
7c"~-4, ref oc*- t
25X1
Approved For Release 2003/0616 :C RbP86B00269R000800060001-6
 App 4ed I's e 2003/06/26 : CIA-RDP86B0 R00080006000_1-6
A J P- 91
.Jr?,touc,~.eLda.~r~u4. h /aa/f -I
!?'+n' v. ~~L ~a ,C A.a~ttt ? ~c " ,hE.COuc a,.~u c/a h QZ t S
74x?yc #?9J?
74.
15 Cac X Mf. ~~ oueciacoo~ ~oTtou... ~~ (1 f
I A GGU a U-wx --c. .J'co /~ c c/ ~b ~e av~..a . ~ 13-3
25X1
Approved For Release 2003/06/26 : CIA-RDP86B00269R000800060001-6