LETTER TO (SANITIZED) FROM ROBERT W. KLINE
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP86R00893R000100130003-4
Release Decision:
RIPPUB
Original Classification:
K
Document Page Count:
37
Document Creation Date:
January 4, 2017
Document Release Date:
April 15, 2008
Sequence Number:
3
Case Number:
Publication Date:
May 13, 1982
Content Type:
LETTER
File:
Attachment | Size |
---|---|
CIA-RDP86R00893R000100130003-4.pdf | 2.09 MB |
Body:
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
DEPARTMENT OF THE AIR FORCE
AIR COMMAND AND STAFF COLLEGE (ATC)
MAXWELL AIR FORCE BASE, ALABAMA 36112
Assistant National Intelligence Officer
for Strategic Programs
Central Intelligence Agency
Washington, D. C. 20505
I am most pleased to invite you to address selected members of the
Air Command and Staff College (ACSC) Class of. 1982 in the "Intelli-
gence and National Security" Tailored Instructional Program (TIP)
on 1 June 1982 at 0830. A two-hour period has been reserved for
you to discuss Strategic Intelligence Estimates. Please use this
time as you deem appropriate--formal lecture, informal remarks,
question and answer, or any combination thereof.
The TIP is an elective course designed for ACSC course officers
with follow-on assignments to national or Department of Defense
intelligence organizations or to operations, plans, programs, or
R&D staff positions which involve the use of national intelligence
products. It is intended to provide an introduction to the history,
structure, and operation of the US Intelligence Community and the
production and use of intelligence products. Your wealth of
knowledge and experience will provide the course officers a
valuable insight into the intelligence activities of this country.
Lieutenant Colonel Robert E. Setlow, telephone (205) 293-6519 will
coordinate the details of your visit.
I greatly appreciate your support of our educational program and
look forward to having you as our guest next month.
ROBERT W. KLI
Colonel, USAF
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
1. Comprehend the major issues involved in the production of intelligence
estimates on Soviet strategic forces and objectives.
1.1 Summarize the historical record of U.S. Intelligence
estimates on Soviet strategic forces.
1.2 Explain the analytical methodologies employed in
estimating the present and future capabilities of
Soviet strategic forces.
1.3 Give examples of the major problems encountered in
producing estimates on Soviet forces and objectives.
OVERVIEW:
Estimative intelligence on Soviet strategic forces and objectives
constitutes one of the principal production functions of the U.S.
Intelligence Community. As we develop and deploy our own strategic forces,
it is important that military planners and defense policymakers understand
present and future Soviet strategic capabilities and comprehend likely
Soviet responses to our policies and actions. The arms control negotia-
tions of the past decade have given impetus to the efforts of the
Intelligence Community to improve the quality of its estimates.
Planners and policymakers must, however, bear in mind the limitations
of estimative intelligence, as Klaus Knorr points out in one of the later
readings:
Although the future, within some framework of
particulars, can be estimated, it cannot, of course,
be known. To estimate is to guess in order to reduce
uncertainty dictated by lack of knowledge. The
assumptions and preconceptions about reality that
structure the guesswork can be more or less rigorously
deduced from past behavior. But--as the historical
record discloses...--even the most sophisticated
assumptions can lead threat perception astray. To
depend wholly on any one preconception or set of
assumptions is to court surprise. This risk is
magnified by the tendency that the selection of an
assumption about the real world becomes an act of
cognitive closure that easily leads the perceiver
to be close-minded and to ignore or explain away
C
discrepant information. It must therefore be
accepted that although good estimates can reduce
uncertainty about the future, even the best cannot
be depended on to prove it.
Knorr also identifies several kinds of predispositions that intervene to
affect the selection of assumptions and receptivity to incoming information
and which are apt to distort estimates. Among these are emotions, strong
ideological commitments, bureaucratic behavior, and wishful thinking.
Bear these considerations in mind while you review the controversy
over U.S. intelligence estimates of Soviet strategic forces in the
readings. We will have an opportunity to explore them more deeply
during the lecture on weapon system analysis as applied to Soviet
strategic estimates. Two of the historically controversial subject
areas of these estimates--Soviet strategic bombers and ICBM's--will be
highlighted to illustrate the techniques of weapon system analysis.
READINGS:
Assigned:
"Debate Over U.S. Strategic Forecasts: A Mixed Record"
"The National Intelligence Estimates A-B Team Episode
Concerning Soviet Strategic Capability and Objectives"
THIS PUBLICATION IS PREPARED BY THE AIR FORCE AS EXECUTIVE AGENT THE NATURE. MEANING AND IMPACT OF NATIONAL AND INTERNATIONAL
FOR THE DEPARTMENT OF DEFENSE TO BRING TO THE ATTENTION OF REV NEWS DEVELOPMENTS. USE OF THESE ARTICLES HERE. OF COURSE, DOES
DOD PERSONNEL NEWS ITEMS OF INTEREST TO THEM IN THEIR OFFICIAL NOT REFLECT OFFICIAL ENDORSEMENT. FURTHER REPRODUCTION FOR
CAPACITIES: IT IS NOT INTENDED TO SUBSTITUTE FOR NEWSPAPERS. PRIVATE USE OR GAIN IS SUBJECT TO THE ORIGINAL COPYRIGHT
PERIODICALS AND BROADCASTS AS A MEANS OF KEEPING INFORMED ABOUT RESTRICTIONS.
S~ ~l ~1Ll , EGC
REVIEWS
VOL. VIII SUMMER 1980 Pagea 22-59 NO. 3
DEBATE OVER U.S. STRATEGIC FORECASTS:
A MIXED RECORD
LES ASPIN
THE AUTHOR: Congressman Aspin is Chairman of the
Oversight Subcommittee of the House Permanent Select
Committee on Intelligence and serves on the House Armed
'
.:
. Services Committee and the Government Operations Com-
mittee. He was first elected to Congress in 1970. Aspin
served in the U.S. Army from 1966 to 1968 as an economic
adviser in the office of the Secretary of Defense. He is a
graduate of Yale University, received a Master's degree from
Oxford University and a Ph.D. in economics from the Mas-
sachusetts sachusetts Institute of Technology.
IN BRIEF
The charge has resounded in recent times that the United States intelligence community has chron-
ically and woefully underestimated both the pace and magnitude of the Soviet strategic build-up.
Yet, an analysis of the available record of forecasts with respect to eight major Soviet weapons de-
velopments-extending from the first Soviet A-bomb explosion in 1949 to the improvements in So-
viet ICBM accuracy and yields in the 1970s-shows that the performance has been mixed, consist-
ing of overestimates as well as underestimates, and in at least two instances of predictions that
were on or close to the target. Few of the mistakes that have been committed in forecasting can
be attributed to errors in intelligence gathering; most of them have been the function of more-or-
less inevitable human foibles. With the demise of SALT, estimates of future Soviet strategic pro-
grams are apt to be wider off the mark than they would have been under a SALT II Treaty, because
the reference points provided by the Treaty for U.S. intelligence have been removed, and precisely
because tke human element in intelligence evaluation and forecasting is thus again maximized.
+ R B R R x a B t? R R s? R R B R R t R? t R? R R R? R? It B R B B
A %eea.ted arttLcle: DEBATE OVER U.S. STRATEGIC FORECASTS: A POOR RECORD
by WILLIAM T. LEE appeau on Page 14 of .thi.o Laeue. A REBUTTAL by
CONGRESSMAN ASPIN appetvt4 on Page 25.
CURRENT NEWS
SPECIAL EDITION
30 September 1980 No. 618
Approved For Release 2008/04/15 CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
C
"It is ... a matter of record that the growth of
the Soviet ICBM force was underestimated for a
decade after the 'missile gap' by the entire Intelli-
gence community-Including Pentagon 'hawks."'
Lt. Gen. Daniel O. Graham, USA (Ret.)
"But the history of the past twenty years shows
quite the reverse. Few indeed are the instances
be greater than the estimated 'worst case.' Usually,
the government's experts overestimated the danger."
George B. Kistiakowsky
The death of SALT II turns the focus of
U.S. strategic intelligence away from
"verification" and back to the old busi-
ness of "forecasting." SALT provided for some
degrees of restraint and certainty: We knew
how far the Soviets were allowed to go, and the
task was to verify their compliance with these
restrictions. Without SALT, there are no limits
or guidelines. The United States must rely
purely on its skills in strategic forecasting-in
projecting the future, including future Soviet
strategic intentions and capabilities.
The first forecast since the deferral of SALT
II has been completed and leaked to the press.
The new National Intelligence Estimate-NIE
1138-79-reportedly indicates that without
SALT II the Soviets could amass about 14,000
highly accurate ICBM warheads by the late
1980s. By contrast, an extension of SALT II
beyond its 1985 expiration date would allow
the Soviets only about 6,000 such warheads; if
SALT II were in effect, therefore, the presump-
tion would be that the Soviets would build up
only to that limit.' U.S. strategic force plan-
ning would be based on this assumption and
U.S. intelligence agencies would be concentrat-
ing on verifying Soviet compliance. Now, with-
out SALT II,.all we have to go on is this new
',intelligence estimate. Who knows whether it
has validity or not? If U.S. policymakers do
believe it to be valid, however, then they will
have to think about a requisite expansion of
U.S. strategic nuclear forces. Tens of billions of
dollars potentially ride on a decision of whether
or not to trust this intelligence estimate.'
How good is U.S. intelligence at this task of
strategic forecasting? As the passages quoted
above indicate, this question is highly contro-
versial.3 Over the years, many analysts, par-
ticularly those in arms control circles, have
contended that we have consistently overesti-
mated Soviet strategic capabilities. More re-
cently, other analysts, not generally associated
with arms control, have argued that we have
in fact consistently underestimated Soviet
strength.
This controversy can, to some extent, be re-
solved by examining the record. Considering
the salient developments in the history of the
nuclear arms competition; we can ask if the
U.S. intelligence community has been right or
wrong in its forecasts-and if wrong, in which
direction (too high or too low) it has erred and
for what reasons. The key developments have
been:
1. The first Soviet explosion of an atomic
bomb, 1949.
2. The first Soviet explosion of a hydrogen
bomb, 1953.
3. The "bomber gap," 1955-1958.
4. The "missile gap," 1958-1961.
5. Soviet deployment of an anti-ballistic mis-
sile (ABM) system, 1962 onward.
6. Soviet deployment of missiles with mul-
tiple independently targetable reentry ve-
hicles (MIRVs), 1965-1974.
7. Soviet intercontinental ballistic missile
(ICBM) and submarine-launched ballistic
missile (SLBM) deployments, 1962-1969.
8. The rate of improvements in Soviet ICBM
accuracy and yield, 1969 onward.
bSuch an analysis should provide us with
some idea of how well U.S. intelligence will be
able to estimate future Soviet defense capa-
bilities in the absence of SALT. _
The A-Bomb and the H-Bomb
When the Soviets exploded their first atomic
bomb in August 1949, the United States had
very little information about Soviet nuclear re-
search. Before the detonation, General Leslie
Groves, wartime director of the Manhattan
Project, predicted that America's atomic mo-
nopoly would last twenty years. Scientists in-
volved in the project, on the other hand, be-
lieved in 1945 that the Soviets would duplicate
the U.S. achievement within five years. The
scientists' expectation encouraged the Atomic
Energy Commission to establish, through the
Air Force, a program for airborne collection of
radioactive particles in the atmosphere, which
would detect the explosion of any atomic de-
vice anywhere in the world. The program
began operations in 1948 (and continues to
this day).
As the end of the decade approached and no
Soviet A-bomb materialized, the year of antici-
pated danger, from the vantage point of the
U.S. intelligence community, receded progres-
sively. Just before the Soviets actually deto-
nated an atomic device in 1949, they were offi-
cially expected to do so in 1952 at the earliest.'
The hydrogen bomb, set off by the Soviets in
1953, came as less of a surprise: the United
States had predicted that the Soviets would
achieve that milestone by 1954.
Why did General Groves underestimate, the
scientists correctly estimate, and later most
Special Edition -- 30 September 1980
analysts underestimate again how soon the
Soviets would explode an A-bomb? And why
was the H-bomb prediction so close to the
mark?
The problem was not one of optimism about
Soviet intentions. Indeed, in the first five years
after the war, official circles in Washington
generally held to the alarmist image of a Soviet
Union bent on con' sfartt implacable expannsion-
iFrCRath r, the intelligence error on the A-
6omb hinged on habit and personal intuition.
General Groves thought that the Soviets would
take twenty years to build the bomb because,
like President Truman, he simply did not be-
lieve that "those Asiatic" Russians," valiant
though they might be in standing up to the
Germans on the battlefield, had the technologi-
cal talents to duplicate what his scientists at
Los Alamos accomplished in four years. The
scientists' prediction that the Soviets would
have a bomb within four or five years was
modeled on their own experience. That is how
long it had taken them to build the bomb: It
was a fairly straightforward exercise in physics
and engineering, of which they deemed their
Soviet counterparts quite capable. In the end,
intelligence analysts underestimated the devel-
opment pace for the Soviets because of what the
atomic scientist, Isadore Rabi, characterized as
a "peculiar kind of psychology": after the initial
estimate in 1945 that the Soviets could get a
bomb in four or five years, "every year that
went by, you kept on saying 'five years."'
The close prediction of the Soviet Union's
H-bomb detonation in 1953 was purely a matter
of chance-a very good guess and little more.
The principle of radiation pressure, the essence
of the H-bomb, was not even demonstrated in
the United States until 1951. Indeed, some
officials believed the Soviets could get an H-
bomb before 1953. In an attempt to encourage
President Truman to forge ahead with the
American H-bomb project in 1950, General
Loper of the AEC's Military Liaison Committee
argued in a memorandum to the President that
available intelligence (almost nonexistent)
was consistent with the theory that the Soviets
already had the hydrogen bomb.?
The Bomber Gap
In 1955, Air Force Intelligence predicted that
the Soviets would field a force of 6000 -to_200
long-range bo ers by 1959. The National In-
telligence Estimate (NIE) for that year was
slightly more modest, predicting about 500
bombers by mid-1960. As it turned out, by mid-
1961 the Soviets had deployed only 19U1 long-
range bombers.'
Estimates of bombers grew out of a projec-
tion made in 1950-incorporated in a milestone
Cold War document called NSC-68-that the
Soviets would possess a stockpile of 200 atomic
bombs by 1954.8 This projection was based, in
part, on the rate at which the United States had
been able to build bombs. Given this projection
and NSC-68's explicit assumption that the
Kremlin was bent on expansion and that the
United States was the Soviet Union's principal
enemy, intelligence agencies naturally began
thinking about how the Soviet Union would
deliver the bombs to U.S. territory.
In 1954, Western attaches in Moscow ob-
served a new Soviet long-range bomber flying
overhead at the May Day military parade. On
the basis of this report, U.S. intelligence made
some assumptions about_ when the Soviets had
begun-evelolment of this bomber and _how
?uickly they might Ge able to deploy it In sig-
nificant numbers. A study concluded that the
7bmbers design had been completed in 1952
and its first prototype flight made in 1953. In
accordance wi r U- . expeS' n`ence, it was esti-
mated that ma^ Ss production could not begin
pefo~e_1956. and a substantial force could not
be deployed before 1960.-
The next May Day parade, in 1955, rudely
upset these calculations, or at least appeared
to do so. Although the aviation part of the
parade was canceled, Western observers re-
ported seeing as many as twenty of the long-
range bombers in the air during parade re-
hearsals. Intelligence now updated its earliest
estimates. The design of the plane was as-
sumed to have been completed two years earlier
than the original finding, and mass production
to have begun in 1954. `If the Soviets could
Iirodiice twenty aircraft per month over the next
three years, theme force of 700 aircraft by 1959
was plausible.10
Yet, in 1956 and 1957, U-2 flights produced
hard evidence that Soviet production rates fell
far below the pace that had been estimated by
U.S. intelligence two years earlier. Two factors
were involved in this error: an intelligence mis-
take and a misunderstanding of Soviet strategic
intentions.
First, unbeknown to the Western attaches,
the Soviets were flying the same bombers back
and forth in the 1955 parade preparations; the
attaches mistakenly counted each overflight as
a separate bomber.11 Second, the United States,
believing that its own territory was the ultimate
target of the Soviet Union's nuclear ambitions,
nattrally-assumed that the Soviets would pro-
duce intercontinental bombers at the fastest
rate possible. However, the Soviets apparently
decided that the principal threat to the Soviet
Union lay around the periphery of the Soviet
landmass, whence Russia had historically been
threatened and where the United States hap-
pened to be stationing its own nuclear strike
forces. Thus, the Soviets used most of their
production capacity to guild medium-range
bombers rather than a long-range force."
Approved For Release 2008/04/15 CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
The Missile Gap
The Soviet Union launched its first orbital
satellite in October 1957.19 Although the CIA
had foreseen this development years in ad-
vance, the actual launching triggered fears that
the United States would soon be vulnerable to
an ICBM attack. Sputnik: the very word
evoked a nightmare vision of the Soviets out-
pacing the Americans in missile I technology.
Khrushchev exploited this American fear by
publicly making outrageous statements about
the capabilities of Soviet missiles which he
knew at the time-and we know only in retro-
spect-to be false.
Air Force Intelligence warned in a November
1957 NIE that the Soviets could deploy 500
ICBMs by the middle of 1960 and 1,000 by
1961. The CIA-.believed a more reasonable
estimate to be 100 ICBMs by 1960 and 500 by
1961. The wide di erence to tes
hinged on conflicting views of when the Soviets
would be able to begin mass production of their
first ICBM, the SS-6. A halt in the Soviet test
program, in April 1958, was interpreted by the
Air Force as an indication that the missile was
ready for deployment, whereas the CIA saw it
as evidence that technical difficulties were
being experienced in the missile's development.
Renewed test launches in 1959 proved the CIA
correct.
An entirely separate issue, however, was how
many missiles the Soviets would produce each
year. Apparently the Air Force picked 500 and
the CIA 100 because they were round numbers.
Since no one at that time knew the location of
Soviet missile manufacturing plants, neither
an actual count nor an inference from indus-
trial volume was possible.
Nor did anyone know what a Soviet ICBM
emplacement would look like. The Air Force
anticipated camouflaged sites, whereas the CIA
argued that the deployment sites would re-
semble the missile test launchers at Tyuratam.
Repeated U-2 flights over Soviet railway lines
could not locate any deployed ICBMs, although
Air Force Intelligence suspected various build-
ings to be camouflaged structures hiding mis-
siles. Among these were a Crimean War me-
morial and a medieval tower. A U.S. photo-
reconnaissance satellite took the first clear pic-
tures of a Soviet ICBM site at Plesetsk in August
1960-laid out, as the CIA had predicted, just
like the site at Tyuratam. According to the
early Air Force projection, the Soviets should
have deployed more than 500 ICBMs by this
time, but satellite coverage detected no similar
sites anywhere else.
The identification of an operational SS-6 site
reopened the issue of how quickly the Soviets
could produce the missiles. From studies of the
Soviet economy and the cost of American
ICBMs, the CIA assumed that the Soviets could
start off producing ICBMs either on an "orderly"
schedule of three per month or on a "crash"
program of fifteen per month. Assuming that
the Soviets had been producing missiles since
1959, when their test program ended, the CIA
calculated that under the orderly schedule the
USSR would have 36 operational SS-6s by No-
vember 1960, and that they might accelerate
production to reach 100 by mid-1961 and 450
by mid-1962. The Air Force, meanwhile, stuck
to its original prediction of 500 missiles per
year.
The Army and Navy intelligence organiza-
tions, whose client services carried on weapons
programs that competed with Air Force mis-
siles, pointed out technical deficiencies in the
SS-6 tests and expressed doubt that the USSR
would ever deploy "more than a few" of these
missiles. In August 1961, another successful
recovery of satellite film proved them right.
Indeed, the Soviets had deployed no more than
ten SS-6 missiles, all at Plesetsk. This dis-
covery ended the "missile gap" for good.
The Soviets did have a substantial mile
buil8'up iin process-but-it was in intermediate--
range-and-norlntercontlnedtal -missiles: The
Lts.-intelligence error, again, was bile of mis-
taking Soviet priorities. Between 1958 and
1965, the Soviets deployed about 700 medium-
and intermediate-range ballistic missiles (M/
IRBMs) aimed at Western Europe. This comes
to about 100 missiles per year-a figure be-
tween the CIA's "orderly" and "crash" estimates
of 36 and 180 missiles per year respectively,
but far short of the Air Force estimate of 500
per year.
Anti-Ballistic Missiles
Throughout the 1960s, intelligence analysts
repeatedly predicted that the Soviets would de-
ploy a nationwide anti-ballistic missile (ABM)
system.14 In the early 1960s, the intelligence
community estimated that the Soviets would
-deploy some 2,000-eko-atmospheric and 8,000
endo-atmospheric interceptors.15 In 1963-
1964, the NIE on strategic defensive forces
predicted that before 1975 the Moscow ABM
system, just coming under construction, would
be expanded to cover every major city with 500
to 1,500 interceptors. Furthermore, between
1964 and 1966, Pentagon analysts suspected
that the Tallinn air-defense system would even-
tually serve as a nationwide ABM and man-
aged to insert this speculation into some NIEs.
After 1967, construction of the Moscow ABM
System seemed to halt with only 64 interceptors
fielded. Those Tallinn sites were later proved
to be for defense against high-altitude bombers.
At this point, analysts in the Defense Intelli-
gence Agency (DIA) and John Foster, then the
Director of Defense Research and Engineering,
speculated that the Tallinn sites could quickly
be "upgraded" to a dual purpose SAM/ABM
system. Further analysis, however, revealed
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Approved For Release 2008/04/15 CIA-RDP86ROO893ROO0100130003-4
Special Edition -- 30 September 1980
that many of the Tallinn sites were badly lo-
cated for ICBM interception, and that they
lacked the nuclear warhead storage space es-
sential for a workable ABM system.
Why was U.S. intelligence so eager to detect
a Soviet ABM system that never did materi-
alize? Part of this misjudgment was founded
on an assessment of Soviet strategic doctrine.
The Soviets were greatly concerned about stra-
tegic defense. They had an extensive air de-
fense network to intercept bombers, and they
had something of a civil defense program.
Many intelligence analysts logically concluded
that they would construct a comprehensive
ABM system as well.
The type of Soviet ABM for which these ana-
lysts looked-a combination of exo- and endo-
atmospheric interceptors-reflected American
concepts of ABM design, which eventually were
realized in the Spartan and Sprint missiles.
The Spartan was a comparatively slow missile
intended to intercept approaching missiles at
or near the peak of their trajectories, when
they would be moving at their slowest speed.
The fast Sprint would be launched to home in
on any reentry vehicles the Spartan might miss.
Sprint involved an extremely close radar track-
ing. Perhaps because the endo-atmospheric
approach was so demanding, however, the So-
viets chose a different route altogether: an in-
terceptor that would operate at medium altitude
(200,000-500,000 feet). From this model,
the Soviets developed the Galosh and Griffon
interceptors, which used many of the same
components.
Galosh was, and is, an ABM. Sixty-four of
the interceptors remain deployed around Mos-
cow. However, the Galosh radars use a me-
chanical means of tracking ICBM warheads,
an extremely difficult technique. By 1967, U.S.
intelligence analysts began to raise doubts
whether the Soviets would ever make further
investments in so ineffective a system.
Griffon is the missile deployed in the Tallinn
system, now known as the SA-5 surface-to-air
missile (SAM). NIE judgments with respect to
Griffon's mission wavered from year to year.
The Tallinn sites were successors to a system
which the Soviets began building around Lenin-
grad in the early 1960s and which the 1963
NIE deemed an "apparent" ABM ("apparent"
had been formally established as a codeword
in NIEs to indicate that the analysts had little
confidence in the given judgment). In 1964,
the CIA concluded that Griffon must be an anti-
aircraft missile, primarily because its perform-
ance was so inferior to Galosh.
On the other hand, Soviet public statements
were attributing ABM capabilities to Griffon;
Khrushchev said it could hit "a By in outer
space." The Air Force, Army and the DIA were
convinced that the CIA was grossly underesti-
mating Griffon's capabilities. The 1965 NIE
consequently noted that the intended mission
of the Tallinn sites was uncertain, a judgment
repeated in 1966.
In bureaucratic terms, Pentagon intelligence
analysts had large stakes invested in a Soviet
ABM. The Joint Chiefs of Staff, whom DIA
represented within the intelligence community,
and the Pir Force needed the specter of a So-
viet ABM as a rationale for developing MIRVs
(multiple warheads) for U.S. missiles. The
Army needed a finding that Soviet ABMs were
effective in order to overcome the many doubts
about its own ABM program. The analysts
could (and did) cite Soviet documents-in-
cluding classified ones supplied by that premier
spy, Penkovskii-to show that the Soviets
wanted to build an ABM. They conveniently
ignored Soviet documents written after 1965,
which expressed grave doubts about the feasi-
bility of ABMs. Those documents had not been
obtained clandestinely and as such were dis-
missed as deliberate Soviet misinformation.
Intelligence analysts were also misled by an
assumption about the Soviet military as an
eminently rational, far-sighted institution.
Many thought the Tallinn system must be an
ABM-rather than the high-altitude anti-
bomber system it was-because it could have
no other rational purpose. By that time U.S.-
bombers were simulating penetrations of Soviet
air space at low altitude. If Tallinn were a
high-altitude system, then the Soviets were
building weapons for which there was no mis-
sion-an idea thoroughly unpalatable to those
who viewed Soviet defense programs as undis-
turbed by bureaucratic impulses, quirks or
mistakes.
The CIA eventually adopted the view that as
late as 1967 or 1968 the Soviets still intended
to deploy Galosh nationwide, but that improve-
ments in American strategic forces-particu-
larly the MIRY system-convinced them that
they needed to go back to the drawing board.
This highly doubtful argument salvaged the in-
stitutional self-esteem of the Air Force and DIA
by validating their argument that the Soviets
intended the Galosh and Griffon to be nation-
wide ABMs, while conceding to the CIA the/
accuracy of its contention that the Soviets were
not deploying an effective ABM system.
Soviet MIRVs
The prospect that the Soviets might place
multiple independently targetable warheads
(MIRVs) on their ICBMs was first mentioned
in the 1965 NIE. The NIE stated it would take
four or five years for the Soviets to develop and
begin deploying MIRVs that were sufficiently
accurate for the destruction of "hardened" (i.e.,
blast-resistant) targets such as the newly de-
veloped Minuteman ICBM silo. At the time,
there was no evidence that any Soviet MIRV
program had even begun. Thus the earliest date
for Soviet MIRV deployment, inferred from the
STAT
Approved For Release 2008/04/15 : CIA-RDP86ROO893ROO0100130003-4
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
1965 NIE, was 1969. In 1966 and 1967, Soviet
space shots demonstrated some of the tech-
nology necessary for MIRVing. As a result, the
Air Force insisted that the NIE contain a judg-
ment that the Soviets were in fact developing
a MIRV.'a
In August 1968, the United States observed
the first test of the SS-9 "triplet," the three-war-
head ICBM. The SS-9 was a very large missile.
It was believed that such a missile would be
ideally suited to the task of digging out Minute-
man silos. However, even the highest estimates
of ultimate SS-9 deployments-the Air Force's
projection of 700-envisioned a number that
was insufficient to destroy 1,000 Minuteman
missiles. Thus, analysts who believed that the
Soviets were intent upon capabilities to knock
out Minuteman reasoned that the Soviets must
be planning to place multiple warheads on the
SS-9. The triplet tests seemed to confirm this
suspicion.
The issue then became whether the triplet
was a MIRV or merely an unsophisticated
MRV-i.e., whether each of the three warheads
could be aimed at a separate target, or whether
all three must be aimed at the same general
area. Each warhead of the triplet was placed
on a rail in the nose-cone of the SS-9. The rails
did not rotate to allow repositioning and re-
targeting of the warheads. This feature con-
vinced CIA analysts that the SS-9 was simply
an MRV. Therefore, the 1968 NIE did not ex-
pect a Soviet MIRV until 1978-the end of the
period covered by the estimate.
However, analysts outside the intelligence
community, most notably in the Pentagon's
Directorate of Defense Research and Engineer-
ing, noted that the timing of the warhead re-
leases from the SS-9 could cause the warheads
to fall in various triangular patterns. They con-
cluded from the pattern of releases during test-
flights in the Pacific that the Soviets were
indeed adapting these "triangles" (or "foot-
prints") to match the configuration of U.S. mis-
sile silos. A missile force of 400 to 700 SS-9s,
each with three warheads that could be aimed
at three silos, might be very effective against
Minuteman after all.
The triplet issue took on all the more impor-
tance because the Nixon Administration was
seeking Congressional approval of the Safe-
guard ABM system designed to protect Minute-
man against Soviet attack. If the SS-9 lacked
MIRV capability, then Minuteman needed no
protection; if the triangular pattern of the
triplet coincided with the distance between
three U.S. Minuteman silos, however, then the
case for Minuteman vulnerability might still be
valid. Furthermore, Henry Kissinger wanted
the ABM as a "bargaining chip" in the SALT I
negotiations that were just getting underway.
Consequently, Kissinger summoned the CIA
estimators and the Pentagon DDR&F, analysts
to the White House for a series of special meet-
ings. From these sessions, Kissinger concluded
that the triplet was indeed a primitive MIRV,
and he instructed the CIA to rewrite the 1969
NIE to include more evidence supporting both
sides of the controversy." (During 1969,
therefore, two NIEs on Soviet strategic forces
were dis-eminated: one at the beginning of
1969, which had been prepared the previous
year, and one in the fall of 1969 at the new
Administration's request.)
In an important sense, the whole argument
was artificial. In fact, the Soviets had several
programs in parallel: not just an effort to test
a primitive MIRV for the SS-9, but also a pro-
gram to design more sophisticated MIRVs for
the next generation of ICBMs. The United
States knelt' nothingtbout this next generation.
Judging by the U.S. decision to stop its own
ICBM programs with the third-generation
Minuteman, intelligence estimators may have
believed that the Soviets would not proceed be-
yond the SS-9.
In any event, the first 1969 NIE took a wholly
different approach to the issue of when the So-
viets would be able to deploy a true MIRV. The
estimators postulated two possible Soviet ap-
proaches: low force/low technology and high
force/high technology. The former was based
on the assumption that the Soviets would deploy
the triplet, not attaining a true MIRV until
1974. The latter assumed the Soviets would
skip the triplet and move directly to a MIRV for
the SS-9. It was believed that the Soviets, using
the technology tested in the space launches of
1966-1967, might be able to begin deploying
MIRVs as early as 1971.19
As it happened, the first Soviet MIRV was
deployed on an entirely new, fourth-generation
ICBM in 1975. The Soviets never tried to build
a truly MIRVed SS-9. But the intelligence esti-
mates went through various phases. First they
overestimated (in 1965 the estimate was
1970), then underestimated (in 1968 the pre-
diction was 1978), then overestimated again
(in 1969 the projection was 1971). The vary-
ing estimates reflected the different political'
needs of successive U.S. administrations, as
well as a rather vacuous argument over which
U-.S. terminology (MIRV or MRV) was a more
appropriate description of the SS-9 triplet.
The Soviet ICBM Force Size
In a series of articles in 1974, the prominent
strategic analyst, Albert Wohlstetter, argued
that the NIEs between 1962 and 1969 consist-
ently underestimated future Soviet strategic
offensive capabilities.'s Wohlstetter's ostensible
motive was to challenge the commonly accepted
thesis that military intelligence invariably over-
estimated Soviet capabilities to justify its own
costly defense programs.
Motives aside, Wohlstetter advanced the idea
that these underestimates represented a sys-
Approved For Release 2008/04/15 CIA-RDP86R00893R000100130003-4
IC9M MFq[iIOXMAOF IN 1961
ICBM MEYICiION M/.OF IX 1999
ICeM MEOICIION MAOE IX 1919
T
~
o tr
X
o
/
e
o
t
tematic bias inside the CIA and the intelligence
community as a whole-a bias against recog-
nizing the grand scope of Soviet ambitions for
ICBM procurement. As the charts on page 36
(reproduced from Wohlstetter's text) indicate,
the intelligence agencies did underestimate the
number of Soviet ICBM launchers in making
projections of future Soviet capabilities. More-
over, as the Soviet build-up accelerated, intel-
ligence projections did not improve; in some
cases they even worsened.
Why did this happen? One explanation is
that of "mirror-imaging." After 1965, the CIA
expected that the Soviets would place MIRVs
on their ICBMs just as the United States had
done. This expectation led intelligence analysts
to project that the Soviets would deploy fewer
ICBMs than they finally emplaced. The intel-
ligence community based its estimates on the
finding within the United States Defense De-
partment that qualitative improvements to
ICBMs were far cheaper ways to gain additional
capability than quantitative increases in the
force itself. In response to Wohlstetter's charge
of underestimation, Lt. Gen. Daniel O. Graham,
Director of the DIA, testified in 1975 that "the
continuing evidence of qualitative improvement
was a prime contributor to our underestimation
of ICBM deployment .. . . It seemed logical
at the time that the Soviets would try to use
their advantage in throw-weight by equipping
their ICBMs with MIRVs which could . . . over-
whelm the then-programmed U.S. ABM . . .
and . . . permit multiple targeting [of U.S.
ICBMs]...." 20
The Soviets, however, decided instead to
build larger numbers of ICBMs. Thus, the CIA
did underestimate the number of missile
launchers that the Soviets would construct-
but it did not massively underestimate the of-
fensive capabilities of the USSR as a whole.
Second, the CIA knew that resources in the
USSR were scarce and believed that the major
Soviet military investments were going into
other Soviet military programs. The CIA dur-
ing these years vastly overestimated the number
of ABMs the Soviets would produce-and this,
too, contributed to an underestimation of Soviet
ICBM production. In 1962, when small num-
bers of Soviet ICBMs were predicted, the United
States was also anticipating deployment of
something like 10,000 ABM interceptors. De-
fense Secretary Robert S. McNamara suggested
in his 1964 Posture Statement that ICBM pro-
grams would tend to constrain 'large and very
costly new programs such as an effective anti-
ballistic missile defense system.?21 When the
intelligence community (incorrectly) concluded
that the Soviets were about to deploy a massive
ABM network, it was logically reasoned that the
Soviets would not build a very large ICBM
force. Indeed, the greatest ICBM underesti-
mates, those for 1965 and 1966, coincide with
the greatest ABM overestimates.
Third, the general underestimation of Soviet
ICBMs included a whopping overestimation of
one system in particular, the SS-9. In 1969,
DIA projected 420 SS-9s; the Air Force expected
as many as 700. In fact, the Soviets never de-
ployed more than 280 and devoted most of
their resources to constructing nearly 1,000
smaller SS-11 missiles.22 Had the Soviets gone
ahead with SS-9s, the same resources would
have purchased something closer to the number
of SS-9s predicted by the intelligence commu-
nity (except for Air Force Intelligence). Thus,
in terms of projecting actual offensive capa-
bilities, U.S. intelligence was not so far off the
mark as Wohlstetter suggests. Still, the agen-
cies did err in not foreseeing the new Soviet
emphasis on larger numbers of much smaller
missiles, which greatly enhanced Soviet power
to destroy American industrial and population
targets.
Fourth, about 50 per cent of the intelligence
community's underestimations, for each year in
the lay 1960s, is acc, ..nted for by the Soviet
Union's decision not to retire about 200 obsolete
SS-7 and SS-8 ICBMs, contrary to expectations
of U.S. intelligence. Thus, when Wohlstetter's
chart indicates an underestimate of about 400
ICBMs in 1967, roughly 200 of those were due
to an expectation that the Soviets would retire
older, more inaccurate missiles.
The lesson to be learned from a closer look
at the Wohlstetter study is not, as is now popu-
larly perceived, that the United States has con-
sistently underestimated the offensive capa-
bilities of Soviet missile forces-but rather that,
as Wohlstetter himself avers, we underesti-
mated some aspects of that force, overestimated
other aspects and made some accurate predic-
tions. Perhaps these cases of optimism and
pessimism balanced out when the Defense De-
partment attempted to base its own force
planning on these intelligence projections.
(For example, McNamara testified that the
United States planned forces in the early 1960s
under the assumption that the Soviets would
mount an enormous ABM capability-a belief
that probably more than compensated for the
assumption that they would build a relatively
small ICBM force.)23
The intelligence errors on this score appear
to be interconnected: low ICBM estimates were
directly linked with high ABM estimates. In
short, the error is caused not so much by simple
counting mistakes as by a misjudgment of how
the Soviets planned to allocate their defense
resources.
Soviet ICBM Accuracy and Yield
Estimating improvements in Soviet ICBM ac-
curacy and explosive yield is today's critically
important issue. It is the combination of these
two factors that determines the vulnerability of
the U.S. force of land-based ICBMs in fixed
silos.24
Throughout the 1960s, there was little offi-
cial concern about the vulnerability of Minute-
man. In 1968, Defense Secretary Clark Clifford
wrote a memorandum to President Johnson,
one paragraph of which pointed to the possi-
bility that MIRV deployments of the SS-9 con-
stituted a potential threat to the Minuteman
force, and then suggested various solutions to
the problem. The Joint Chiefs of Staff con-
vinced Clifford to delete the paragraph.25
The Nixon Administration took Minuteman
vulnerability more seriously. If the Soviets
could deploy a force of 700 SS-9s, each with
triplet warheads (as U.S. intelligence was pro-
jetting at the time), they could hypothetically
aim two warheads at each of the 1,000 Minute-
man silos, thereby ensuring the destruction of
nearly all of them. The Office of the Secretary
of Defense believed at the time that the Soviets
could achieve accuracies of .25 nautical miles
CEP (meaning half the warheads would strike
within .25 miles of the target point) with the
SS-9 triplet by 1974-1975. It calculated that
this accuracy, combined with each warhead's
estimated 5-megaton yield, would permit the
Soviets to knock out 95 per cent of the Minute-
man force in a first strike.2"
The CIA disagreed. CIA weapons analysts
did not believe the "triplet" technology could
be improved sufficiently to attain the postulated
.25 nautical-mile CEP. The SS-9 triplet had
failed to demonstrate accuracy better than .5
nautical miles-not nearly enough, even given
the high yields of the Soviet warheads, to de-
stroy missile silos with high probability. In
September 1969 the Board of National Esti-
mates therefore drafted a paragraph to the
effect that the Soviets could not, and would not
try to, achieve a first-strike capability against
Minuteman during the 1970s.
However, Secretary of Defense Melvin R.
Laird had publicly raised, in open Congres-
sional testimony, the SS-9's threat to Minute-
man. Reportedly, Laird's special assistant,
William Baroody, went to Central Intelligence
Director Richard Helms and asked him to de-
lete the contrary paragraph from the ] 969 NIE.
Helms complied. When questioned by Senator
Frank Church's Senate Intelligence Committee
about this matter, Baroody testified that he
could not remember "specifically bring[ing]
pressures to bear on the Director of Central
Intelligence to delete or change any particular
paragraph." However, Abbot Smith, then the
chairman of the CIA's Board of National Esti-
mates, does recall the episode as the only in-
stance of direct political interference with the
NIEs that he could remember in his twenty
years with the agency.''
In April 1971, TRW, Inc. completed a study
sponsored jointly by the CIA and DDR&E. It
demonstrated that Soviet technology for the
SS-9 could not achieve accuracies better than
the .5 nautical-mile CEP estimated previously
by the CIA-an error factor inadequate for an
effective first strike against Minuteman.28 By
this time, however, the deployment of Safe-
guard ABM to defend Minuteman sites had al-
ready been authorized.
In 1973, early Soviet testing for fourth-
generation ICBM programs (the SS-17, SS-18
and SS-19). reopened the controversy over
Minuteman vulnerability. Initial press leaks
suggested that the first tests showed accuracies
for the new missiles to be little better than the
.5 nautical-mile CEP of the older systems. Fur-
thermore, since the new missiles carried more
warheads than the SS-9 and had similar or
lighter throw-weight, the yields of each war-
head would be less than the SS-9's. In short, the
warheads would not threaten the Minuteman
silos. In response to these reports, a Soviet offi-
cer reportedly told U.S. officials during the June
1974 Moscow summit that the United States
was underestimating the accuracies of the new
missiles. He claimed that .27 nautical-mile
CEPs had already been achieved. U.S. analysts
doubted this assertion.2?
Since that time, intelligence analysts have
detected improved performance in Soviet mis-
sile accuracy, which-combined with relatively
high yields-theoretically does pose a threat to
the Minuteman missiles 3? In fact, accuracy
cannot be precisely estimated. In June 1979,
Walter Pincus reported in the Washington Post
that the accuracy of Soviet ICBMs was some-
what better than U.S. intelligence had predicted
for that time." However, he also reported that
the SS-18 and SS-19 warheads were now judged
to have substantially lower yields than had once
been projected for them: the analysts had re-
duced their estimate from 1.5 megatons to
about 600 kilotons. The new pessimism in ac-
curacy estimates and the new optimism in yield
estimates virtually canceled each other out.
This indicates the hazards, and also the impor-
tance, of casting precise estimates. (Had the
CIA reduced the yield estimates without also
accounting for improved accuracy, the per-
ceived vulnerability of Minuteman for the mid-
1980s would have dropped from 90 to 80 per
cent-a perception that might have carried
significant policy implications.)
How Good Is Forecasting?
In sum, the record of U.S. intelligence in
forecasting future Soviet strategic capabilities
is a rather mixed one. Of the eight critical de-
velopments. which we have examined (See
Table 1.), the intelligence community overesti-
mated Soviet capabilities on three occasions,
underestimated them once, and both over- and
underestimated in two cases. The community
was almost exactly on target once, and divided
between accuracy and underestimation once.
The one instance of unmitigated underestima-
tion (in the prediction of the number of Soviet
ICBMs) was linked to overestimates of other
variables (especially deployment and numbers
of Soviet MIRVs and ABMs). The one time
when the prediction was nearly dead right (the
timing of the first Soviet H-bomb) was a case
of fortunate guesswork, based on no hard data,
The record of estimates on Soviet strategic
forces bears out Albert Wohlstetter's conclu-
sion: "Our officials sometimes overestimate,
and sometimes underestimate, and sometimes
even get it right. ..."32 This mixed record is
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
TABLE 1
U.S. Intelligence:
Forecasts vs. Reality
Date
Over: +
Prediction
Under: -
Event
Made
Prediction
Actual
Right: 0
Date of Soviet
1945 (Groves)
1965
1949
-
A-Bomb
1945 (scientists)
1949
. 0
1949 (intelligence)
1952
-
Date of Soviet
1950
1954
1953
0
H-Bomb
Number of Soviet
1955 (Air Force)
600-700
190
+
Long-Range
1955 (NIE)
500
+
Bombers By 1960
("Bomber Gap")
Number of Soviet
1957 (Air Force)
1,000
10
+
ICBMs By 1961
1957 (CIA)
500
+
("Missile Gap")
Number of Soviet
early 1960s
10,000
64
+
ABMs
Date of Soviet
1965
1970
1975
+
MIRV Deployment
1968
1978
-
1969
1971 or 1974
+
Number of
Soviet ICBMs "
By 1967
1964
325-525
570
-
1965
330-395
-
By 1970
1965
410-700
1,299
-
1966
505-795
-
By 1971
1967
805-1,080
1,513
-
By 1972
1968
1,020-1,251
1,527
-
1968
1,158-1,276
-
ICBM Accuracy
and Yield
For SS-9
1969
.25 CEP ?
.5 CEP
+
Accuracy
For New Missile
1973
.5 CEP
.25 CEP
-
Accuracy
For SS-18/-19
1978
1.5 Megatons
600 Kilotons
+
Yield
" Source: Albert Wohlstetter, Legends of the Strategic Arms Race, USSI Report 75-1 (Washington, D.C.:
United States Strategic i
nstitute, 1975), p. 24. All other numbers taken from the body of this
paper.
" Circular Error Probable, the number of nautical miles from target
within which a warhead will land 50
per cent of the time.
obviously due in part to the inherent uncertain-
ties in forecasting. Yet the record suggests cer-
tain patterns for mistaken estimates-some
common sources of error and some lessons to
be learned.
Sources of Error in Strategic Forecasting
As reconnaissance technology has improved
over the decades, U.S. intelligence has become
more proficient in the science of collecting data.
It has more "hard" information about the So-
viet military-industrial establishment-missile
deployments, production facilities, etc.-and,
therefore, a firmer platform from which to
make projections.
Yet, few of the mistakes noted in this retro-
spective have been due to errors in intelligence
Approved For Release 2008/04/15 : CIA-RDP86ROO893ROO0100130003-4
Special Edition -- 30 September 1980
gathering; most are attributable to mishaps in
the far more uncertain art of intelligence
analysis. Here is where judgment comes into
play-and it seems that, in several instances,
misjudgment distorted the view of the future.
There are several principal sources of misjudg-
ment.
Preconceived Notions. It is human to look at
the world with preconceived notions-preju-
dices, excessive attention to some things, in-
sufficient attention to others. These precon-
ceptions shape what we look for and what we
believe we see.
Occasionally, these preconceptions limit our
vision. President Truman, General Groves and
certainly others believed that it would take
many years for the Soviets to build an A-bomb
because they had a preconceived image of the
Russians as technological primitives. With re-
spect to error in forecasts of Soviet ABM, U.S.
intelligence fell victim to a preconceived notion
of what might be called "extended rationality."
The members of the community knew that the
Soviets traditionally emphasized defenses in
their military program-it followed logically
that Moscow would strive for a nationwide
ABM. They recognized that the Tallinn site,
with its SA-5 missile, was worthless for anti-
bomber defenses-therefore, they concluded,
assuming Soviet military planners to be flaw-
lessly logical, that it must be an ABM system.
Mirror-Imaging. In the absence of obvious
facts to the contrary, U.S. intelligence often
strays into the assumption that the Soviets con-
ceive of military problems in roughly the same
way that American analysts do. This, too, is a
natural and understandable human trait. It,
too, can mislead.
U.S. intelligence underestimated the number
of Soviet ICBMs, for example, because Ameri-
can analysts assumed that the Soviets, like the
Americans, would stress quality rather than
quantity in the further development of their
strategic nuclear forces-specifically, opting
for MIRVs on their missiles instead of building
more missiles. It was also assumed that, like
the United States, the USSR would replace old,
obsolescent missiles with new ones. Instead,
the Soviets chose a quantitative build-up of
their missile forces and did not retire older
ICBMs until much later.
Misjudgment of Soviet Strategic Priorities.
The "bomber gap" and the "missile gap" were
not the unqualified intelligence fiascos that they
have been painted to be. The Soviets did pro-
duce and deploy hundreds of bombers in the
late 1950s and hundreds of missiles in the early
1960s. The mistake was in assuming that Mos-
cow would emphasize long-range strategic
weapons aimed at the United States. In fact,
Soviet strategists decided that areas along the
periphery of the USSR-most notably in
Western Europe--were the locus of the greatest
threat to the Soviet Union, and they accordingly
concentrated on the development and produc-
tion of medium- and intermediate-range weap-
ons. The Eurasian peripheries, after all, repre-
sented the historical arena of threats to Russia
-and until the 1960s these were the regions
where most of the U.S. strategic strike forces
were deployed.
In the case of Soviet ICBM forces in the
1960s, U.S. analysts did not underestimate
the magnitude of the Soviet defense effort as
much as they misjudged Soviet priorities. They
believed that the Soviets would go for small
numbers of heavy missiles, put more resources
into quality than quantity, and emphasize de-
fensive missiles. Thus, the intelligence com-
munity projected a large number of SS-9s, low
numbers of ICBMs, early deployment of MIRVs
and thousands of ABMs. Instead, the Soviets
developed only a few hundred SS-9s and about
1,000 smaller SS-11s, took several years longer
to field MIRVs, and halted their ABM program
after only 64 were deployed.
Political and Bureaucratic Pressure. Intelli-
gence is not practiced in a political vacuum.
Direct political interference in National Intelli-
gence Estimates is rare: the reported Baroody
case, alluded to earlier, is the only one on
record. Nevertheless, intelligence estimates are
often highly responsive to the political needs
of the client and to the politics of the moment,
even when the heavy hand of politics is not
visibly applied. The Air Force's need to justify
its MIRV program was one factor in its projec-
tion that the Soviets would build 10,000 ABM
interceptors. Likewise, the Nixon Administra-
tion's desire to deploy the Safeguard ABM sys-
tem was one reason for its initial early estimate
of the Soviet MIRV program.
The CIA's underestimation of Soviet ICBM
deployments coincided with Defense Secretary
Robert McNamara's public testimony that the
Soviets would not try to match the U.S. force
in number. This was his rationale for resisting
pressures to expand U.S. nuclear forces-the
level of which, having been arbitrarily set at
1,000 ICBMs, was difficult to justify convinc-
ingly as opposed to some equally arbitrary
higher (or, for that matter, lower) level. Mc-
Namara did not have to signal CIA analysts
directly in order to have his logic reflected in
their estimates; they read the newspapers as
carefully as the rest of the Washington bu-
reaucracy.
Spurious Learning. Bureaucracy has been
defined as an organization that cannot learn
from its own mistakes." The intelligence com-
munity's record in strategic forecasting bears
this out. When the community reacts to previ-
ous errors, the lessons it "learns" are often
spurious; the community overcompensates for
its errors instead of revising the methods that
produced them. Thus, overestimates tend to be
Approved For Release 2008/04/15 : CIA-RDP86ROO893ROO0100130003-4
Special Edition -- 30 September 1980
followed by underestimates, and vice versa.
The underestimates of ICBM deployments in
the 1960s were, in part, in response to the over-
estimates of the late 1950s. CIA officials were
determined not to repeat the mistakes of the
"missile gap." Similarly, the intelligence com-
munity shifted back and forth in its estimate
of when the Soviets would deploy MIRVs. Airst
they overestimated (in 1965, the expectation
was 1970), then they underestimated (in 1968,
the projection was 1978), then overestimated
again (in 1969, the projection was 1971). The
actual Soviet MIRV deployment came in 1975.
Failure to Use Soviet Sources. The charge
that has been leveled against the CIA is that its
estimators ignore clear statements of Soviet in-
tentions and capabilities that are often to be
found in the open Soviet literature." This is a
difficult issue to deal with. On some occasions,
a heeding of Soviet statements would have
made for more accurate intelligence. For ex-
ample, in a public speech in July 1965, Brezh-
nev contended that the United States was
underestimating the scope of Soviet ICBM pro-
grams-which turned out to be true. Similarly,
if the CIA had given credence to the statement
by the Soviet official that the new Soviet ICBMs
had demonstrated a .27 nautical-mile CEP by
1974, the Agency would not have underesti-
mated the rate of improvement in the accuracy
of Soviet SS-18 and SS-19 missiles.
Still, some of the Soviet statements are
clearly false: an example is Khrushchev's boast
that the Soviets had a missile that could hit a
fly in outer space. Such statements obviously
must be tested against other available intelli-
gence evidence. If the evidence does not match,
however, should one then trust one's own esti-
mates or the statements of a foreign govern-
ment that has not been noted for its addiction
to the truth?
Nobody has proposed a consistent set of
rules for determining which Soviet statements
are true and which are false. Some analysts
follow the rule that any Soviet statement mak-
ing the USSR appear hostile toward the United
States is an accurate representation of Soviet
intentions, while any less hostile statement
represents a planned deception. This rule is
obviously unsatisfactory for intelligence analy-
sis.
Perhaps the CIA has been reluctant to
grapple with the complexities of working with
Soviet documents. There is no guarantee, in
any event, that correction of this deficiency will
make intelligence analysis any more accurate
or unbiased.
Implications of the Demise of SALT. Look-
ing at the intelligence community from the out-
side, the public tends to visualize a machine
spewing wt facts. R. rely does the public re-
alize that the intelligence community is com-
pot d of humans in a bureaucracy that is sub-
ject to the same pressures and pitfalls as any
other.
The problems faced in making accurate fore-
casts are, of course, grounded first of all in the
fact that the Soviet Union is a closed society
and does not supply the world's libraries with
volumes of public testilrlony frqm its generals
about military plans. Given the limits on fac-
tual information that an intelligence system
can draw from any closed society, the intelli-
gence community must rely heavily on its ana-
lytical capabilities. This opens the product of
the intelligence community wide to a host of
human foibles-the preconceived notions, mis-
judgments, spurious "learning" and other short-
comings that have been discussed above. In
fact, given the limited data; base upon which
the intelligence community must build its pro-
jections, it would not have been unreasonable
to expect far more errors than have actually
been committed.
With increasingly more comprehensive SALT
agreements, the intelligence community was
finding its tasks made easier. The SALT agree-
ments set concrete numerical ceilings for many
categories of measurement of military power.
The intelligence community did not have to rely
on a murky crystal ball in examining every
realm of Soviet activity. The SALT agreements
narrowed the analysts' task: in those areas
covered by SALT, they needed only to focus
their capabilities and efforts on ascertaining
whether the Soviets were adhering to their
treaty pledges. Resources heretofore devoted to
predicting future missile numbers could be de-
voted to other areas not covered by SALT.
With the death of SALT II, analysts must
dust off once again the murky crystal ball.
Estimates of future Soviet activity are likely to
be wider off the mark than they would be under
a SALT II Treaty, simply because the reference
points provided by the Treaty have been re-
moved. The human element is maximized, and
with it the likelihood of human foibles in-
creases.
Approved For Release 2008/04/15 : CIA-RDP86ROO893ROO0100130003-4
Special Edition -- 30 September 1980
1. Michael Geller and Robert G. Kaiser, "Intelligence
Estimate Said to Show Need for SALT," Washington
Post, January 31, 1980.
2. This cuts both ways. Acceptance of the estimate
may lead to greater spending in some strategic arms,
but it may also lead to rejection of the land-based
multiple-shelter basing scheme for the MX missile on
grounds that too many shelters would have to be con-
structed to "absorb" so many Soviet ICBM warheads.
3. The quotes are from Lt. Gen. Daniel O. Graham,
former Director of the Defense Intelligence Agency,
"Intelligence: Realities and Myth," Wall Street Journal,
March 11, 1977, p. 16; George B. Kistiakowsky, "False
Alarm: The Story Behind SALT II," New York Review
of Books, March 22, 1979. Kistiakowsky was a member
of the President's Science Advisory Committee, 1957-
1963.
4. See Lawrence Freedman, U.S. Intelligence and
the Soviet Strategic Threat (Boulder, Colo.: Westview
Press, 1977), p. 64; Richard C. Hewlett and Oscar E.
Anderson, History of the AEC, Vol. 1: The New World,
1939-1946 (University Park, Penn.: Pennsylvania
State University Press, 1962), pp. 358-360; Herbert
York, The Advisors: Oppenheimer, Teller and the
Superbomb (San Francisco: W.H. Freeman & Co.,
1976), pp. 34-36.
6. Cited in Freedman, op. cit., p. 64, and informa-
tion from D.A. Rosenberg, a researcher studying the
U.S. decision to develop the H-bomb.
7. U.S. Senate Select Committee to Study Govern.
mental Operations with Respect to Intelligence Activi-
ties (the "Church Committee"), Final Report, Book IV,
Supplementary Detailed Staff Reports on Foreign and
Military Intelligence (Washington, D.C.: Government
Printing Office, 1976), p. 56; Freedman, op. cit., p. 67;
The Military Balance, 1975-1976 (London: Interna-
tional Institute for Strategic Studies, 1975), p. 73.
8. For the text of NSC.68, see U.S. Department of
State, Foreign Relations of the United States, 1950,
Vol. I: National Security (Washington, D.C.: Govern-
ment Printing Office, 1978), pp. 235-292; for the 200-
bomb estimate, see p. 251. For a thorough analysis of
NSC-68, see Paul Y. Hammond, "NSC-68: Prologue to
Rearmament," in Warner R. Schilling, Paul Y. Ham-
mond and Glenn H. Snyder, Strategy, Politics and De-
fense Budgets (New York: Columbia University Press,
1962).
9. Freedman, op. cit., pp. 65-66.
10. Ibid.
11. Ibid.
12. Even now, the Soviets emphasize medium-range
far more than long-range bombers. They have 156
Bear and Bison long-range bombers, compared with
613 Badger medium-range bombers and 80 Backfires
that have limited intercontinental capability but are
deployed and exercised mainly for "theater" missions.
See The Military Balance, 1979-1980 (London: Inter-
national Institute for Strategic Studies, 1979), p. 89.
13. This section is based on material in Edgar M.
Bottome, The Missile Cap: A Study of the Formulation
of Military and Political Policy (Cranbury, N.J.: Fair-
leigh Dickinson Univt rsity Press, 1971); Freedman,
op. cit.; Arnold Horelick and Myron Rush, Strategic
Power and Soviet Foreign Policy (Chicago: University
of Chicago Press, 1966).
14. Material in this section is based on Freedman,
op. cit., Chapter 5; Edward R. Jayne II, The ABM De-
bate: Strategic Defense and National Security (Cam-
bridge, Mass.: MIT Center for International Studies,
1969); Ronald Tammen, MIRV and the Arms Race
(New York: Praeger, 1973).
15. Paul H. Nitze, commenting on articles by Albert
Wohlstetter, Foreign Policy, Fall 1974, p. 82.
16. See Freedman, op. cit., p. 116.
17. Ibid., pp. 137 ff.; Senate Foreign Relations Com-
mittee, Hearings, Intelligence and the ABM (1969),
p. 24; John Newhouse, Cold Dawn: The Story of SALT
(New York: Holt, Rinehart & Winston, 1973), p. 161.
18. Melvin R. Laird, Department of Defense, Fiscal
Year 1971 Defense Program and Budget (February 20,
Race?" and "Rivals But No Race," Foreign Policy,
Summer and Fall 1974. These articles were reprinted
in Strategic Review, Fall 1974 and Winter 1975, and
then published together as Legends of the Strategic
Arms Race, USSI Report 75-1 (Washington, D.C.:
United States Strategic Institute, 1975). The Wohl-
stetter articles sparked a debate involving articles and
replies by Paul Nitze, Joseph Alsop, Morton Halperin,
Jeremy Stone, Michael Nacht and Johan Hoist, in
Foreign Policy, Fall 1974 and Summer 1975.
20. U.S. Congress, Joint Economic Committee, Hear-
ings, Allocations of Resources in the Soviet Union and
China-1975, Part 1, pp. 97-98.
21. Robert S. McNamara, Department of Defense
Posture Statement for FY 1964 (February 1963), p. 22,
.
22. Freedman, op, cit., p. 146.
23. Robert S. McNamara, Department of Defense,
Posture Statement for FY 1967 (1966).
24. An index of vulnerability has been calculated as
YN divided by CEP2, where Y = weapon yield in mega-
tons, and CEP = Circular Error Probable, or the dis-
tance from the target within which a warhead will land
50 per cent of the time. The gist is that increasing
yield or accuracy will boost a warhead's "kill proba-
bility" against a target, but that doubling accuracy will
have the same effect as boosting yield by eight times.
See Kosta Tsipis, Offensive Missiles (Stockholm: Stock-
holm International Peace Research Institute, 1974).
25. Newhouse, op. cit., p. 129.
26. Freedman, op. cit., p. 141.
27. See U.S. Senate Select Committee to Study Gov-
ernmental Operations with Respect to Intelligence Ac-
tivities ("Church Committee"), Final Report, Book 1,
Foreign and Military Intelligence (Washington, D.C.:
Government Printing Office, 1976), pp. 77-79; and
Laurence Stem, "Agency Forced to Alter Own Data,"
Washington Post, April 27, 1976.
28. Michael Geller, "Russian Missile Faulted," Wash.
ington Post, June 17, 1971.
29. Freedman, op. cit., p. 173; U.S. Senate Foreign
Relations Committee, Hearings, Briefings on Counter-
force Attacks (September 1974).
30. According to data released by Paul Nitze, the
newest version of the SS-18 (the 10-warhead Mod-4
variant) has a CEP of .17 nautical miles, as does the
latest SS-19 with six warheads. See Nitze's testimony,
Senate Foreign Relations Committee, Hearings, The
SALT II Treaty (July 1979), P. 459.
31. Walter Pincus, "U.S. Downgrades Soviet ICBM
Yield," Washington Post, May 31, 1979.
32. Albert Wohlstetter, Legends of the Strategic
Arms Race, op. cit., p. 14.
33. Michael Crozier, The Bureaucratic Phenomenon
(Chicago: University of Chicago Press, 1964).
34. See for example the views of Senator Malcolm
Wallop, in U.S. Senate Select Committee on Intelli.
gence, The National Intelligence Estimates A-B Team
Episode Concerning Soviet Strategic Capability and
Objectives (Washington, D.C.: Government Printing
Office, February 1978), p. 13.
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
DEBATE OVER U.S. STRATEGIC FORECASTS:
A POOR RECORD
WILLIAM T. LEE
THE AUTHOR: Mr. Lee is a consultant to several govern-
ment agencies and private research organizations, and he
has written widely on Soviet military strategy and economic
matters. Mr. Lee served with the Central Intelligence
Agency from 1951 to 1964. He is the author of Soviet De-
fense Expenditures in an Era of SALT (USSI Report 79-1).
IN BRIEF
Congressman Aspin's assessment, while heralding a welcome Congressional attention to the prob-
lems of U.S. strategic forecasting of Soviet weapons developments, does not portray accurately the
U.S. intelligence community's past performance in this crucially important arena. His scoring of the
eight cases of forecasting selected not only is too generous to the CIA and other U.S. intelligence
agencies, but it also neglects the relative weight of the mistakes committed-particularly in the
failure to forecast the formidable build-up of Soviet strategic capabilities in the 1970s. The record
of intelligence estimates becomes even more grievous when looked at in the larger compass of the
CIA's responsibilities, notably its estimates of Soviet defense expenditures. A ratified SALT II Treaty
could not ease the problem; the solution, rather, lies in badly needed improvements in the intelli-
gence interpretation of the ample evidence available.
Congressman Les Aspin has offered an
assessment of U.S. intelligence fore-
casts of Soviet weapons systems devel-
opment and deployment that says, in effect:
We have won some, lost some and tied some.
Moreover, he implies that this is about the best
one can expect from intelligence forecasting of
Soviet weapons technology and deployments.
Congressman Aspin's assessment is welcome
on three counts. First, it needs pointing out, as
he does, that the U.S. intelligence services have
a mixed record in forecasting Soviet weapons
development: there have been some overesti-
mates as well as many underestimates. Second,
public recognition of some of the intelligence
underestimates by a member of the Congres-
sional Select Committees on Intelligence is long
overdue. Third, at a time when the Congress
and the Executive Branch are negotiating a
charter to govern the activities of covert intelli-
gence collection and action, our attention needs
to be focused also on the neglected question of
how to accomplish improvements in the intelli-
gence analysis and projection of Soviet forces
that threaten our survival. We are about a
decade late in realizing that the repeated and
systematic misestimates of future Soviet forces
by the analytical components of the intelligence
community have had far more serious negative
consequences for U.S. national security than
whatever excesses the covert intelligence com-
ponents may have engaged in. Moreover, un-
like a recent assessment by Messrs. Robert
Ellsworth and Kenneth Adelman; Congress-
man Aspin does not blame the covert operators
of the CIA for the failures of the Agency's in-
telligence analysts. It is hoped, therefore, that
Congressman Aspin's assessment represents
the beginning of a serious bipartisan effort by
the Congress, in cooperation with the Executive
Branch, to improve the quality of U.S. intelli-
gence analysis, particularly long-range fore-
casting.
On the negative side, Mr. Aspin's assessment
is less than comprehensive, representing a very
small sample of intelligence community fore-
casts of Soviet military programs. A number
of instances where the CIA has unambiguously
or inferentially underestimated Soviet programs
are conspicuous by their absence from Con-
gressman Aspin's review. He also fails to note
that the most grievous underestimates of So-
viet programs have occurred since the revolu-
tion in intelligence collection capabilities that
he dates to August 1960. Finally, certain
"facts" cited by Congressman Aspin require
further (probably official U.S. Government)
documentation before they can be accepted as
such.
Quantitative Measures of Performance
Congressman Aspin evaluates U.S. intelli-
gence estimates of Soviet weaponry on eight
counts-four forecasts of Soviet strategic mis-
sile and aircraft deployment, and four forecasts
of critical nuclear and missile technologies.
Table 1 attempts to summarize Congressman
Aspin's assessment of the U.S. intelligence com-
munity's performance. The "box-score" that is
reflected seems consistent with one of Con-
gressman Aspin's summary statements: "... the
intelligence community overestimated Soviet
capabilities on three occasions, underestimated
them once, and both over- and underestimated
in two cases. The community was almost ex-
actly on target once and divided between ac-
curacy and underestimation once."
Two general observations are in order. First,
this is far too small a list of cases on which to
base an evaluation of the effectiveness of the
CIA and other members of the U.S. intelligence
community in forecasting Soviet forces and
technologies five or ten years into the future.
Second, one must take issue with Congressman
Aspin's account, as presented, of the record of
initial intelligence forecasts made during the
1960s of Soviet MIRV deployment, and of the
Table 1
Congressman Aspin's Box-Score of U.S. Intelli-
gence Forecasts of Soviet Military Technologies
and Weapons Systems Deployments
Technology/ Under- Over- Near the
System estimate estimate Mark
1st A-Bomb Test X
1st H-Bomb Test
"Bomber Gap" X
"Missile Gap" X
ABM
Deployments X
ICBM
Deployments X
MIRY
Deployments X X
ICBM Accuracy
and Yields
(since 1969) X X
Definitions:
Underestimate: The Soviets achieved a caps.
bility earlier, or deployed a larger force, than
the U.S. intelligence community forecast.
Overestimate: The opposite of underestimate.
Near the Mark: A relatively small discrep-
ancy between forecasts and observed events.
accuracy and yield of Soviet MIRVed ICBMs
when these systems were deployed in the mid-
1970s.
In each decade of the 1960s and the 1970s
the Soviets procured some 100 major weapons
systems-offensive and defensive, of strategic
reach and shorter ranges. A number of tech-
nologies, some standard and others that were
more or less advanced, were incorporated into
each of these weapons systems. In each of the
two decades, about half of the weapons systems
procured had been developed in the previous
decade.
This pattern is likely to continue through the
1980s.2 Consequently, four Soviet weapons de-
velopment programs and four technologies
represent a rather small sample-even if one
wants to restrict the evaluation of the intelli-
gence community's prowess to "strategic" weap-
ons systems and technologies. Such a restric-
tion of the field of inquiry is much too narrow.
In terms of overall military capabilities, "tac-
tical" systems are as important as, if less glam-
orous than, "strategic" systems. Both are
equal parts of the intelligence community's
charter. Both weigh on the U.S.-USSR military
balance and, most importantly, in the percep-
tions of that balance by the U.S., the USSR,
their respective allies, and the "third world."
Nor does Mr. Aspin mention the CIA's abysmal
record in underestimating Soviet defense ex-
penditures and long-term trends in national
economic priorities.
Before turning to the question of how the
intelligence community's record would look if a
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
larger sample of forecasts of Soviet weapons
programs were used, let us examine Congress-
man Aspin's eight cases in a little more detail.
Congressman Aspin's count of intelligence fore-
casting successes and failures is sensitive to his
definition of terms. He recognizes that both the
bomber gap" and the "missile gap" embrace'
two distinct families of weapons systems in
each case: medium and heavy bombers, and
IR/MRBMs and ICBMs, respectively. Both ap-
pear as overestimates in Table 1.
Yet, if we split the "bomber gap" and the
"missile gap" into their two components, heavy
and medium bombers and ICBMs and IR/
MRBMs respectively, then the count for under-
estimates increases from five to seven because
U.S. intelligence services underestimated Soviet
medium bomber and IR/MRBM programs by
roughly the same degree that they overesti-
mated heavy bombers and first generation
ICBMs. Mr. Aspin excuses U.S. intelligence
services on the grounds that they did not grasp
either Soviet strategic priorities or the Soviet
perception that the immediate threat to the
USSR emanated from NATO and U.S. overseas
bases rather than from the aircraft bind Mis-
siles based in the continental United States?
This is true, indeed. U.S. intelligence has
conspicuously failed to understand Soviet stra-
tegic concepts, perceptions and priorities. So-
viet IR/MRBMs and medium bombers, which
are considered by the Soviets just as "strategic"
as ICBMs and heavy bombers, continue to be
referred to in the United States as "gray area"
and "peripheral" systems. For more than a
decade the United States has put all of its stra-
tegic systems into the hamper of SALT nego-
tiations while allowing the Soviets to withhold
about one-third of their strategic systems from
the bargaining process. The persistent use of
the terms "peripheral" and "gray area" to de-
scribe Soviet strategic missiles and bombers, the
exclusion of about one-third of Soviet strategic
forces from the strategic arms control agree-
ments negotiated with the Soviets over the past
decade, all indicate that both the CIA and most
of the intelligence analysts in other agencies of
the U.S. Government did not learn much from
the so-called "bomber" and "missile" gaps.
If the CIA had absorbed the lesson of the un-
expectedly large Soviet medium bomber and
IR/MRBM deployments in the late 1950s and
early 1960s, the Agency might not have under-
estimated Soviet ICBM deployments from the
mid-1960s onward. Soviet IR/MRBM force
levels-some 700 launchers-were designed to
meet the requirements of Soviet nuclear target-
ing strategy. Widely publ.cized after 1962, So-
viet targeting strategy required not only the
1,600-odd ICBMs the Soviets eventually de-
ployed, but also MIRV technology because the
Soviets could not afford to deploy sufficient
numbers of ICBMs with one warhead each in
order to satisfy their required capabilities
against either hard or soft targets. As has been
pointed out previously, taking into account So-
viet requirements is a prerequisite for effective
long-term forecasting of Soviet strategic force
deploym,ni s.-
Congressman Aspin's history of the intelli-
gence community's estimates of Soviet MIRV
programs is questionable in part and less than
reassuring in toto. According to Mr. Aspin, U.S.
intelligence services made the following fore-
casts of Soviet MIRV technology:
? The 1965 National Intelligence Estimate
(NIE) predicted MIRVs would be opera-
tional on Soviet strategic missiles in
1969-1970.
? Three years later, the 1968 NIE moved
the operational date for Soviet MIRVs to
1978.
? The 1969 NIE (prepared in 1968) esti-
mated Soviet MIRVs could be operational
as early as 1971?
? In September 1969 the CIA's Board of Na-
tional Estimates concluded that "the So-
viets could not, and would not try to,
achieve a first-strike capability against
Minuteman during the 1970s."
Congressman Aspin provides no documenta-
tion for this chronology. Even if one accepted
it as an accurate record, the CIA's 'vacillation
on the MIRV issue is staggering. Consider once
again the alleged sequence as outlined above.
In 1965, Soviet MIRVs were predicted for 1969
or 1970. In 1968, they were not projected a
threat until 1978. In the 1969 NIE suddenly
the conclusion was reached that the Soviets
would have MIRVs in 1971. If this is a valid
history of the sequence of CIA and NIE fore-
casts of Soviet MIRVs, then one has to wonder
about the sobriety of the estimators.
Yet some who were involved in the 1965 NIE
do not remember it this way. According to their
account, the Air Force argued that the Soviets
would develop MIRVs by the end of the 1960s.
As usual, the Air Force arguments were de-
rided by CIA analysts. As usual, the Air Force
deposited its footnote to the NIE, this time con-
tending an early development of Soviet MIRV
capabilities. According to a senior Air Force
analyst who participated in the 1965 NIE on
Soviet strategic offensive weapons systems, the
NIE did not forecast Soviet MIRVs for 1069-
1970.
In retrospect, the Soviet MIRV program prob-
ably was approved in about 1965. Feasibility
tests evidently were carried out in two 1964
space shots." Hardware for the first flight test
-in August 1968-of the triplet version of the
SS-9 ICBM probably was being fabricated by
the end of 1965 at the latest.' The development
of the SS-16, SS-17 SS-18 and SS-19 ICBMs
must have been approved no later than 1966,
probably by the end of 1965 as part of the
1966-1970 Five Year Plan, in order to allow
for the flight test programs that began in
1972.
The FY 1971 Department of Defense Pos-
ture Statement confirms that by 1969 the NIEs
expected Soviet hard-target MIRVs as early as
mid-1971-a date which turned out to be about
three years premature.8 In this case, however,
the intelligence community evidently had a
good excuse, which Congressman Aspin does
not mention: by early 1970 the community
had detected construction of new types of silos
in Soviet ICBM complexes.? Presumably this
was taken as an indicator that flight tests of
new model ICBMs were imminent. This, in
turn, could have been interpreted to mean that
one or more new-model ICBMs would become
operational by the end of 1971. In any case, the
1969 strategic force NIE evidently did forecast
a Soviet MIRV capability by as early as mid-
1971. The new silo construction detected by
early 1970 quite understandably could have in-
duced such a prediction-although, given the
complexity of MIRV technology, mid-1972
would have been a more prudent dpte.
Yet, Congressman Aspin reports that in Sep-
tember 1969 the Board of National Estimates
wrote that the Soviets "could not, and would
not try to" achieve the capability to attack
Minuteman silos (effectively) in the decade of
the 1970s. If true, this represents an extraordi-
nary confusion. Within a period of less than
twelve months in 1968-1969 the CIA's Board
of National Estimates took this position while
the NIE predicted, without making a choice,
either Soviet counter-silo MIRVs by mid-1971,
or counter-silo multiple RVs in 1974.10
If true, all this also constitutes a most damn-
ing indictment of the CIA. An overestimate of
three years for initial Soviet MIRV capabilities
in the earlier 1969 NIE was not a grievous
error: it was at least on the right track. But to
argue in September 1969 that the Soviets could
not and would not pose a threat to U.S. ICBMs
in the 1970s was a gross underestimate that
(1) flew in the face of all evidence then avail-
able, and (2) was fraught with serious policy
implications that continue to haunt us today.
If it was advanced, as Congressman Aspire re-
ports, the Board of National Estimates' argu-
ment of September 1969 exemplifies the disre-
gard for evidence and the proclivity for "mirror
imaging" that most concerns those critics who
have concluded that the CIA could and should
have done better-and who believe that the
strategic predicament in which the U.S. finds
_tselt'-to`day i9--d we inT no smalLpart tos`uch badl y
mistaken estimates, in which the CIA play-4 ominant ureb aucratic and intellectual
th--t~-_.m
roles within e intelligence comunitq.
If in September 1989 tEe CIA Boaz- of'Na-
tional Estimates so grossly misperceived the
thrust of Soviet strategic missile programs into
the 1970s, it was not from lack of evidence. In
the context of that evidence, the arguments at
that time over the accuracy of the "triplet" ver-
sion of the SS-9 have little relevance. The CIA
probably was right in its estimate of the ac-
curacy of that missile, as tested in 1969. De-
partment of Defense officials probably also were
correct in believing that the accuracy could be
improved sufficiently with a few more years of
testing to make the SS-9 a silo-killer. The evi-
dence available in the public domain marks this
argument as one of those tempests in a teapot
which so often has muddied long-range esti-
mates in the intelligence community. Soviet
strategic objectives and the long-term trends on
this issue were quite clear by 1969.
As has been noted, Soviet targeting strategy
rated the enemy's nuclear delivery systems as
the first priority. The Soviet IR/MRBM force
of some 700 launchers was sized for counter-
force operations against nuclear delivery sys-
tems and other military targets in NATO terri-
tory and elsewhere along the Eurasian
periphery. One or two hundred IR/MRBMs
would have been more than adequate if the So-
viets had been interested only in targeting cities
in a minimum deterrent strategy. The SS-9 had
been predicted as a missile designed to attack
hardened Minuteman launch-control centers.
Both the design and the deployments validated
that prediction. The triplet version of the SS-9
made sense only as an attempt to acquire a
system which could attack individual Minute-
man silos even if the necessary accuracy was
not achieved. If the CIA's Board of National
Estimates failed to understand Soviet targeting
priorities by 1969, it was not because of lack
of evidence.
Congressman Aspin's observation that the
Soviet counterforce threat that emerged in the
1970s had not been taken seriously in the
1960s, at least not in the NIEs, is supported by
Secretary Robert McNamara's 1967 Posture
Statement." Because the NIEs were not fore-
casting a threat to the U.S. Minuteman ICBMs
in the 1970s, Defense Department analysts in-
vented the "greater than expected threat" in
order to help justify U.S. MIRV programs. This
threat projection deliberately went far beyond
that of the NIEs. Had the DoD planners not
exercised such prudence, the United States
probably would have faced a "MIRV gap" in the
early 1970s.
Yet,. the Defense Department planners went
one step further: in an effort to discredit vari-
ous options then (in the mid- to late-1960s)
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
being considered for deploying American anti-
ballistic missile defenses (ABM), they invented
the "greater than the greater than expected
threat." They contrived this threat by feeding
into the computer the assumptions needed to
arrive at the predetermined answer-namely,
that U.S. ABM deployments not only would fa.l
to reduce U.S. fatalities in the event of a Soviet
attack (which, despite all the evidence to the
contrary, was assumed to be designed to maxi-
mize civilian fatalities) but that such deploy-
ments would even result in increases in the
fatalities. This contrived Soviet threat served
as the basis for a table in Mr. McNamara's
1967 Posture Statement, which painted the
horrendous picture of 70 to 120 million casual-
to improve the accuracy of its MIRVed ICBMs
only because the United States sold to the So-
viets the precision machine tools needed to pro-
duce the requisite small, very precise ball bear-
ings for Soviet missile guidance systems."
As regards U.S. estimates of Soviet MIRY
warhead yields, Mr. % -,pin probably is both
more mistaken and more correct than he re-
alizes. He refers to a lowering of U.S. intelli-
gence estimates of the yield on the SS-18 MIRVs
from 1.5 to 0.6 MT. While it appears that there
has been some scaling down in yield estimates,
has it been of this magnitude? Caution is in
order because the SS-18 is reported to have been
tested with various MIRV payloads involving 8
and 10 reentry vehicles (RVs). Before Mr.
to the available evidence, the CIA in 1967 gave
its blessings to the "greater than the greate
forecasts of Soviet MIRV technology reads quite
differently from that presented by Congress-
man Aspin. The 1965 NIE probably did not
overestimate the time that it would take the
Soviets to achieve a MIRV technology. The
early 1969 NIE probably did overestimate it,
by about three years, but for understandable
reasons. Nevertheless, both that NIE and later
versions failed to forecast the full scope of So-
viet counterforce capabilities that emerged in
the late 1970s. Indeed, if Congressman Aspin
is correct in his description of the late 1969
NIE and its optimistic predictions of both Soviet
strategic intentions and ICBM capabilities
through the 1970s, then this must be rated as
one of the most dismal chapters in the history
of U.S. intelligence. It bears out the obvious
fact that errors in forecasting cannot be rated
on the same scale: some are more serious than
others. Consequently, instead of charging the
U.S. intelligence services simply with one over-
estimate and me underestimate, it seems more
realistic to :.core one "near the mark" for U.S.
forecasts of MIRV technology and one grievous
underestimate with respect to the potency of
'Soviet ICBM capabilities in the 1970s.
Finally, Congressman Aspin counts one un-
derestimate for the U.S. intelligence commu-
nity's judgment of the accuracy of the MIRVed
ICBMs introduced by the Soviets in the 1970s
and one overestimate with respect to the yields
of those weapons. These cases concern current
ine.lligcu' : estimates of Soviet weapons c'.ar-
acteristics, not forecasts of future develop-
ments. Over the years the estimated CEPS
(circular error probable) of Soviet MIRVed
ICBMs have shrunken from about 1,800 feet to
between 600 to 900 feet as the result of a vig-
orous Soviet development effort." It is worth
noting that the Soviet Union probably was able
field of the SS-18 RVs by a factor of two and
ne-half, he needs to demonstrate that the
figures of 1.5 and 0.6 megatons refer to one
nd the same RV model. This he has not done.
On the other hand, it is probably true that
intelligence community estimates of the yields
of Soviet strategic missile RVs tend to be over-
stated on the average. When aiming at hard
targets, the Soviets need all the yield they can
pack into the RV. Consequently, the single RV
version of the SS-9 probably did carry a war-
head on the order of 20 MT, and the three-RV
version probably delivered warheads with yields
of approximately 5 MT. Yet, megaton and
multimegaton warheads make no sense against
soft targets-and it is important to keep in
mind that Soviet targeting strategy has clearly
deemphasized strikes against the general U.S.
population and heavy industry.15 As regards
such "soft" (urban-industrial) targets, the So-
viet literature specifies the use of minimum
yields.'" Consequently when aimed at soft
targets, even the older and relatively inaccurate
Soviet ICBMs and IR/MRBMs seldom required
yields of even one megaton: depending on the
degree of damage required, yields in the range
from 0.05 to 0.5 MT were adequate. Given the
accuracy of current Soviet MIRVs, 0.15 MT is
about the most yield that is required for soft
targets, and the lower limit drops to about
0.025 MT.'7
Apparently the intelligence community has
been estimating Soviet strategic missile yields
at the maximum levels considered to be tech-
nically feasible. This is probably a reasonable
approach for estimating the explosive power of
missiles designed to attack hard targets, such
as the SS-9, S' :. hl?..`, 1P. But
it probably overestimates, by as much as an
order of magnitude or more, the yields the So-
viets are likely to employ against soft targets.
Before turning to a more comprehensive
survey of intelligence performance, it may be
appropriate to deal with one of the alibis offered
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4 -
for the CIA's quantitative underestimates of So-
viet ICBM deployments: namely, that the as-
sumption was made in each succeeding NIE
that the Soviets would retire some 220 older
SS-7 and SS-8 ICBMs. If these are added to the
annual estimates, Congressman Aspin argues,
the NIEs were not too far off base.
No documentation is offered for this alibi.
Moreover, the notion that every succeeding NIE
expected all 220 of these ICBMs to be retired
seems too neat an excuse. What was the basis
for such an expectation-if indeed it was enter-
tained and so consistently? Throughout the
mid-1960s and until the late 1970s the Soviets
did not have nearly enough ICBM warheads to
cover the whole array of both hard and soft
targets in the United States. Their reserve also
was inadequate." Until several years ago, the
Soviets needgd every one of those old ICBMs,
and there should have been no question that
the missiles could be kept in service that long.
The United States still has 54 Titan II ICBMs
in its inventory and evidently intends to keep
them until the mid-1980s at least.
Indeed, the Soviets had no difficulty in main-
taining their SS-4 and SS-5 missiles-many of
which antedated the SS-7s and SS-8s-in ser-
vice during the 1960s and 1970s. True, these
missiles were soft and vulnerable, but what did
that matter in the context of a preemptive strike
strategy-to say nothing of the launch-on-warn-
ing concept which the Soviets adopted in the
mid-1960s? 19 The whole excuse is threadbare.
And if the CIA really clung so stubbornly to the
expectation that the Soviets would retire those
SS-7s and SS-8s, then this represents no excuse
at all, but simply another mistaken estimate
to be added to all the others.
A Wider Appraisal of Strategic Estimates
? Up to this point we have addressed only Con-
gressman Aspin's "box-score" of U.S. intelli-
gence forecasts of Soviet weapon technologies
and systems deployments. Closer examination
already has changed the roughly "even" score
adduced by Mr. Aspin to one that gives the edge
to underestimates by the intelligence commu-
nity. But to gain a more accurate appraisal,
we need to consider the CIA's record on several
other strategic issues and programs which Mr.
Aspin omitted from his analysis.
Thus, while he concedes that the CIA under-
estimated Soviet ICBM deployments from the
mid-1960s through the early 1970s, Congress-
man Aspin overlooks parallel intelligence un-
derestimates of Soviet SLBM programs. Such
underestimates emerge even from Department
of Defense Posture Statements, and additional
documentation of what has long been an open
secret has been provided by a former CIA
analyst, Ray Robinson.'" In the mid-1960s the
CIA officers adamantly rejected evidence that
the Soviets had a large SLBM program under-
way-one that eventually resulted in the de-
ployment of 34 Y-Class SSBNs carrying 544
SS-N-6 SLBMs. At one point Robinson was
asked to resign for the sin of sticking by the
evidence." Thus, one more major underesti-
mate must be added to the record.
Later, in the mid-1970s, MIRVs for Soviet
SLBMs evidently were not forecast. In his FY
1977 Posture Statement, Secretary Rumsfeld
noted: "There is no evidence that any missile
other than the 4,200 nm single-RV SS-N-8 will
be carried in the near term" by the 16-tube ver-
sion of the Delta Class SSBN." Yet, in his FY
1978 Posture Statement, Secretary Rumsfeld
reported the testing of the SS-N-18 MIRV?a
which probably became operational in late 1977
or early 1978. The Soviets admitted to having
144 of the MIRVed missiles deployed by June
1979, when they signed the SALT 11 Treaty .24
So another underestimate must go up on the
scoreboard.
At this point, the score reads three overesti-
mates (heavy bombers, and first-generation
ICBMs and ABMs in the 1960s) and six under-
estimates (medium bombers, IR/MRBMs,
third-generation ICBMs, SLBMs, the counter-
force threat from Soviet ICBM MIRVs and
MIRVs for SLBMs). The gist of this record is
that not a single strategic missile or aircraft de-
ployment program appears to have been esti-
mated correctly in the course of the past twenty-
five years. The batting average reads no hits in
nine times up-not exactly an all-star perform-
ance.
CIA estimates of Soviet defense expenditures
represent another area of significant errors-
one to which, curiously, Congressman Aspin
does not allude. Perhaps the reasoning is that
the CIA's defense expenditure estimates were
not forecasts. If this is the explanation, it is
not convincing for two reasons. First, it is in-
consistent: the estimates of Soviet MIRV ac-
curacy and yields that Congressman Aspin
included in his assessment were "current esti-
mates" of existing Soviet systems, not forecasts.
Second, the CIA does make long-term forecasts
of Soviet defense expenditures.
From 1960 to 1975 the CIA estimated Soviet
defense spending, excluding RDT&E and space,
at essentially the same levels as the published
Soviet "defense" budgets-notwithstanding the
evidence that since at least 1951 the latter
figure had not included weapons procurement.
Moreover, between 1960 and 1975 the CIA
probably spent between $0.5 and $1.0 billion
of taxpayers' dollars in producing an estimate
of Soviet defense expenditures that was readily
available on the shelves of the Library of Con-
gress and many university libraries. Time and
time again the CIA argued that its estimates
were correct and denounced both critics and
criticism of its methods and its estimates.
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
In 1975 the CIA discovered that its estimates
of Soviet defense spending in 1970 had been
off by a factor of two-that the USSR had spent
at least 50 billion rubles rather than the 24
billion rubles which the CIA had estimated pre-
viously. About the same time, the CIA dis-
cerned that its estimates for 1972 were off by
somewhat more than a factor of two-amount-
ing to some 58 billion rubles rather than the
27 billion rubles estimated. In 1976 the CIA
published its revised estimates for Soviet de-
fense expenditures in 1970. Within this revi-
sion of the total, the CIA tripled its previous
estimates of Soviet military procurement in
1970: from between 5 and 6 billion rubles to
about 17 billion rubles.
The new information, however, only told the
CIA its aggregates had been too low by factors
of two to three: it apparently did not tell the
CIA how to correct its ponderous direct-costing
methodology. This writer's pleas to the Agency
to correct its methodology before publishing its
revised estimates were rejected. In 1976 the
CIA published its revised estimates which-to
oversimplify only slightly-represented merely
the prior estimates moved up the scale. Predict-
ably,-the revised CIA estimates showed only a
slightly higher rate of growth in Soviet defense
spending than its prior estimates. (Any'Soviet
defense posture falling between unilateral dis-
armament and full mobilization registers a 3
to 4 per cent per annum growth rate on the
CIA's direct-costing model.) Soviet defense ex-
penditures for the years prior to 1970 now were
overestimated; after 1970 they were underesti-
mated as before. By 1972 the revised CIA esti-
mates were already about 20 per cent short of
the 58 billion rubles which Brezhnev reportedly
had indicated the Soviets spent for defense that
year." In 1980 the revised CIA estimates are
once again off by a factor of two, or nearly so.
As before, the CIA estimates of Soviet de-
fense expenditures are produced at considerable
cost to the taxpayer, but can be replicated (in
the aggregate) by almost any analyst with
minimal effort. Total CIA personnel costs alone
for producing these estimates probably are on
the order of $5 million annually. To replicate
the revised CIA estimates, simply plot its prior
estimates on a piece of graph paper. Move the
prior 1970 estimate of 24 billion rubles up to
the revised estimate (mid-point) of 45 billion
rubles. Through this point draw a line parallel
to the prior estimate. This line represents the
announced Soviet defense budget. It is a very
cost-effective method, indeed.
The score of the CIA's underestimates is now
increased by two: one for the prior estimates
of Soviet defense expenditures and one for the
revised estimates. To be sure, the CIA has not
yet admitted to the second underestimate, but
the evidence for it is as good as or better than
before. It may take another decade or two for
the CIA to discover its new error-by which
time its estimates may be awry by a factor of
three or four or more.
Before leaving the arena of strategic esti-
mates, a few words may be in order on the sub-
ject of Soviet anti-submarine warfare (ASW)
programs. With on.y one exception, every
major surface combatant vessel that has been
converted or constructed for the Soviet Navy
since the early 1960s has been assigned ASW
as a primary mission. (Soviet designators are
"BPLK" and "PLK" for "large anti-submarine
ship" and "anti-submarine cruiser.")26 It re-
mains to be seen whether the same will apply
to the four new classes of Soviet surface com-
batants currently reported to be fitting out or
under construction, but it is likely that most
and perhaps all will be equipped and trained
for the ASW mission." Similarly all Soviet
SSNs not equipped with cruise missiles that
have been built since the late 1950s have been
assigned the strategic ASW mission.
Did the CIA forecast such a Soviet emphasis
on ASW even in the late 1960s and on into the
1970s? The answer is: apparently not. The
annual Department of Defense Posture State-
ments have reported the construction of each
class of new Soviet surface combatant and SSN
as it appeared, but nothing can be inferred from
the Statements that suggests intelligence fore-
casts of the vast ASW effort that has dominated
Soviet ship construction. Little or no heed was
paid to the characteristics of the Alpha-Class
SSN as they were described by two Soviet ad-
mirals in the General Staff journal, Military
Thought, in the late 1960s.28 Series production
of this SSN in the late 1970s evidently came
as a surprise: even Secretary Harold Brown's
FY 1981 Report indicates such production only
by implication, noting that the Soviets are now
producing 10 "general purpose" SSNs per
year.2" In fact, in the Soviet view all of these
SSNs are "strategic" systems. Unless it can be
demonstrated that the DoD Posture Statements
somehow have ignored intelligence forecasts of
the Soviet emphasis on strategic ASW over the
past two decades, we must chalk this up as an-
other CIA underestimate.
Forecasts of Soviet General Purpose Force
Programs
This brings us to CIA forecasts of major
weapons programs for Soviet Ground Forces,
Frontal Aviation (FA) and Military Transport
Aviation (VTA). Did the NIEs between 1968
and 1971 prepare U.S. planners for the contin-
gency observed by Secretary Rumsfeld in the
FY 1977 Annual Report that "This total force
appears to be larger than would be required
for even the most stalwart of defenses ..."10
-or for the similar observations made by
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
Secretary Brown in the FY 1981 Annual. Re-
port? Sl
It seems clear from the various Posture State
ments by the U.S. Secretaries of Defense that
the NIEs written; in the late 1960s . and early
1970s forecast neither the increase in the num-
ber of 'weapon, systems introduced into the So- '
viet'Ground Forces, Frontal Aviation and Mili-
tary Transport. Aviation during the 1970s, nor
the quantities in which they were produced,
nor the advances in technological sophistication
which went into the Soviet weaponry. Each -
new weapon system was reportedas it appeared
in the 1970s, but the whole thrust of the Soviet
Union's effort in increasing and modernizing its
general purpose: forces evidentlywas not dis-
cerned.by?the CIA until after its implementa-
tion-was well advanced. Even Secretary. Rums-
-feld's FY 1977 Report did not anticipate, the
new family of Soviet tactical surface-,to-surface
missiles which have been or are currently re-
placing the older models.32
;,Learning From Mistakes This brief survey is about as far as one can
proceed within" the bounds of what has been
reported and what can be inferred from the
material availablein- the public domaiazn with..
-respect to.ClAand intelligence community fore-
casts of Soviet weapons deployments and tech-
nologies. Table 2 attempts to summarize the
historical: record, of intelligence' forecasts and'
estimates as.bestlas can be deduced from the
available. evidence in:the public domain. The
record is flattering neither to, the CIA nor to the
intelligence coiutfiunityas a ,whole, nor can-it.
be reassuring [o' the policymaker or. to the tax-
payer.- The "best and brightest" should have
done better.
Predicting, the future always'will-be a'diffi'---
,r_cult enterprise 'Itis. anoccupation in which,
ui'oider to be right, oe has to have the courage
to be wrong.: Under the,best of efforts' and cir-
cumstances;no one can realistically aspire to
a' flawless record in forecasting. Nevertheless, .
the record ought to be substantially better. than
is suggested in Table 2.
One obstacle tIa improvement seems,to be the, ''
reluctance, I particularly' ink the CIA, to admit'
-
errors Intelligence is a business iniwhich one
learns more,, tror'h ones mistakes than. from
s";.
one's}successe5 But-in'orderto learn from past
!'errors oemust first recognize them and then
analyze them. Some time.agb a former Deputy
Director of 'the- CIA's Office of Strategic Re-
search told this writer: "What you. do not on
derstand is that while we may not always be,,,.
right; we are never wrong:-',; In a recent book
on the CIA one reads .''. -... On the 'whole; the
CIA seems, to have ;dorte,its job well; especially
where Soviet military capabilities have been
concerned' ' There' have. been: no major sur-
prises, no sudden discoveries of developments
Under- Over- Near the.
estimate estimate. Mark
Weapon System
Deployments
Heavy Bombers X
Medium Bombers X
-lit Generation ICBMs - X
IRIMRBMs X
3rd Generation
ICBMs ' . . X
'2nd Generation
'.SLBMs , X
ABM Deployments. X
Advanced Nuclear-
Powered ASW
Submarine ' X
Ground and Air Force
Modernization in
1970s
Tanks and Armored
Personnel Carriers X
'Self-Propelled Guns' - X
Operational-Tactical ,
Missiles ' X
Fighters and - '
Fighter-Bombers ' X
SAMs - X
Selected Technologies
1st A-Bomb Test ? X
1st H-Bomb Test ' X
ICBM MIRVs' X
4th Generation ICBMs
(MIRVs) .
Accuracy
Yields
SLBMs MIRVs , X
Operational-Tactical
Missile and Aircraft '
Avionics &
Electronics- --'r 'X'
Soviet Defense
Expenditures
1960-1976 Period. ' X
:1976-1980 Period . X ,
Source: The underestimates of the scale and
pace of Soviet modernization and deployment
programs for the Ground Forces, Frontal Avia-
tion and Military Transport Aviation in the
1970s have been inferred from the pertinent sec-
tions of the annual Posture Statements by the
various Secretaries' of Defense.
in Soviet military technology that might leave:
the United States, in a dramatic disadvan-
It is difficult to reconcile this statement with
the record as shown in Table'2. One can'grant'
that perhaps n&single one of the series of sur-
ipnseslput;the United States at a "sudden"'mili-
tart' disadvantage.' The cumulative effect of all
the underestimates on U.S: arms-policies, how-
Some mild surprises are, probably `unavoid-
able. 1 A case in point is the recent appearance
Approved For Release 2008/04/15 : CIA-RDP86ROO893ROO0100130003-4
Special Edition -- 30 September 1980
of a large new Soviet nuclear submarine said
not to he the expected "Typhoon" SSBN.'4 (In-
cidentally, the rate at which the Soviets are
building large, expensive new naval combatants
leads one to believe that the only constraint on
Admiral Gorshkov's budget may be the cham-
pagne that it takes to 'aunch all his new v(,.
sels.) In any event, it seems that U.S. intelli-
gence analysts knew that a large Soviet subma-
rine was under development: an SSBN referred
to by the Soviets as "Typhoon," which may be
larger than Trident and the first boat of which
probably will be launched soon. But the new
submarine already sighted is not the "Typhoon."
Thus the surprise is that there are two new
classes of very large nuclear-powered subma-
rines rather than just one, both apparently
being built in the same Soviet shipyard, That
sort of thing is difficult to detect. Yet, the inci-
dent should serve to impress upon us the limits
of U.S. "national means of verification." The
Soviets would not be able to build a fleet of new
submarines in secret; nevertheless, the fact that
such a large Soviet weapons project could reach
this stage before being detected should give
everyone pause for reflection on the lihtitations
of external collection sensors.
Given the magnitude and continuity of the
Soviet arms build-up since 1958, more surprises
probably lie in store for us. There is some in-
telligence which only human sources (report-
edly much downgraded in the past four years)
can provide.
This brings us to the related issues of infor-
mation sources, forecasting methods and the
quality of the forecasts. Mr. Aspin dates what
may be called the revolution in U.S. "national
means of verification" to August 1960: "A U.S.
photo-reconnaissance satellite took the first clear
pictures of a Soviet ICBM site at Plesetsk in
August 1960...... The 1960s inaugurated the
era when the U.S. could, by non-intrusive tech-
nical means, count with some confidence such
phenomena as ICBM launchers, aircraft on air-
fields, naval vessels and eventually a great deal
more. Prior to the early 1960s the U.S. did not
know how many missiles the Soviets had or
had not deployed; from then on we knew. In-
deed, it was this revolution in "national means
of verification" that made SALT possible in the
1970s, because until at least 1978 U.S. intelli-
gence provided all the data on Soviet weapons
systems-the numbers and characteristics-re-
quired to negotiate the various SALT agree-
ments.
Also in the early 1960s the Soviets began to
publish at great length the rationale for their
decisions on weapons system acquisition and
their objectives in the amts competition with
the United States. As we now know from
declassified documents, from 1960 onward the
United States had reasonably continuous access
to the journal of the Soviet General Staff, Miii-
tan/ Thought. Under such circumstances one
might have expected the CIA to have improved
the quality of its forecasts of Soviet weapons
programs.
The record, however, does not show any im-
provements. In the .S..os when the CIA had
very little "hard" intelligence on Soviet deploy-
ments and not much intelligence on the Soviet
strategic rationale, the national estimates both
over- and underestimated Soviet strategic force
deployments and technologies to a more or less
equal degree. After the revolution in U.S. "na-
tional means of verification" and after the So-
yiet strategic rationale could be gleaned from
bt tp unclassified and classified Soviet sources,
the` CIA might reasonably have been expected
to have learned from past mistaken estimates.
If anything, however, the record since that time
is much worse. Only one Soviet strategic de-
ployment program (ABMs) was overestimated
after 1960; all the rest were underestimated.
The forecasting of Soviet weapons technologies
did improve a bit. Since 1960 four major So-
viet weapons technologies have been underesti-
mated and two overestimated. ,
The CIA's dssessments and forecasts of Soviet
defense expenditures and national economic
priorities have been the most unambiguously
dismal failures of all. We have described two
gross underestimates, which have come back-
to-back and for essentially the same reasons.
As was pointed out earlier,, even the ex-
panded list in Table 2 is still a modest sample
of the hundred-odd major procurement pro-
grams in which the Soviets have engaged in
during the past three decades. Noteworthy
programs missing from the list are Soviet air
defense radars, SAMs and interceptors. The
record of the CIA and the other intelligence ser-
vices in forecasting Soviet programs in these
areas may well have been better. Unfortu-
nately, the information available in the public
domain is not adequate to assess the forecast-
ing record with respect to these Soviet defensive
weapons systems.
Even if the CIA's forecasting record would
improve a bit by further expansion of the
sample, the general conclusion would remain
the same: U.S. intelligence has underestimated
Soviet weapons programs far more often than
it has overestimated them; and the forecasting
record since 1960-1961 seems worse than be-
fore, despite the revolution in technical intelli-
gence collection capabilities over the past two
decades and the relative abundance of evidence
available from classified and unclassified
sources on the Soviet rationale for decisions on
weapons acquisition.
The Question of Methodology
What methodology (or methodologies) has
the CIA developed to forecast Soviet weapons
production and deployment programs and So-
viet weapons technologies? Given the record,
the methodologies have not worked very well.
On the basis of the record, it also is difficult to
believe that forecasting methodologies have
been (1) consistent from one class of NIE to
another-that is, from strategic offensive to
strategic defensive to general purpose forces-
or (2) consistent among various editions of the
same NIE-for example, successive editions
of the NIE for Soviet strategic offensive
forces.
One glaring deficiency in CIA forecasting
methodology for Soviet ICBMs has-been pro-
vided in the public domain by former CIA
analyst David Sullivan, who points out that dur-
ing the 1960s the CIA ignored the role of the
Soviet Five Year Plans in determining future
deployment programs 85 In effect, CIA analysts
assumed that all the ICBM silos known to be
under construction each year would be com-
pleted, but that no new silos would be started.
Hence, every year the NIEs forecast that Soviet
ICBM forces would peak when all the silos then
under construction were completed. This ap-
proach ignored both the five-year planning cycle
as well as any type of requirements analysis to
determine how many ICBMs, or warheads, the
Soviets might ultimately need to carry out their
well-publicized targeting strategy.
In the final section of his article, Congress-
man Aspin offers six reasons for "misjudg-
ments" that "have distorted our vision.- Only
some short observations are possible here be-
cause a full commentary would require at least
as much space as already has been used. Suf-
fice it to say that most of these factors listed by
Mr. Aspin are not exactly novel ones. None of
them is "fatally inevitable," to borrow a Soviet
phrase. If intelligence analysts would con-
sciously seek to avoid these error-prone factors,
they could significantly upgrade their intelli-
gence forecasts. For example, simply paying
more attention to Soviet sources-the last
factor on Congressman Aspin's list-could have
avoided successive underestimates of Soviet de-
fense expenditures and national priorities.
Better forecasts were possible. Indeed, better
forecasts were made-outside the CIA and the
intelligence services and even without benefit
of access to classified Soviet literature on the
Soviet strategic rationale.
SALT II and U.S. Strategic Forecasts
Some concluding comments are in order on
Congressman Aspin's linkage of the SALT II
Treaty to U.S. strategic forecasts. His main
point is that, under a ratified SALT II Treaty,
the U.S. intelligence community could concen-
trate its efforts on the task of verifying the com-
pliance of the Soviet Union with the agreed-
upon limitations on strategic weapons systems.
Conversely, he concludes, "Estimates of future
Soviet activity are likely to be wider off the mark
than they would be under a SALT II Treaty,
simply because the reference points provided by
the Treaty have been removed."
This linkage of SALT and forecasting does
not stand up on three grounds. First, the SALT
II Treaty applies to only a small portion of the
Soviet military establishment. About 30 per
cent of Soviet strategic offensive forces are not
accountable under the Treaty. Except for ABM
systems, strategic defenses are not covered at
all, and there is substantial cause to fear that
the Soviet Union will abrogate the ABM Treaty
in the mid-1980s. U.S. intelligence agencies
are charged with forecasting the disposition of
some 40 to 50 major Soviet weapons systems
that are not covered by SALT, as well as such
related issues as how and to what ends the So-
viets would employ their armed forces in the
event of war, Soviet defense expenditures, the
military burden on the Soviet economy, etc.
The SALT II Treaty would not help the intelli-
gence community in forecasting the weapons it
does not cover, nor do the Treaty's "reference
points" promise to contribute much, if anything,
to forecasts of Soviet technological develop-
ments even for those relatively few strategic
systems that are subject to the Treaty's provi-
sions.
Second, there is no assurance that the SALT
fI Treaty would be of any greater help to the
U.S. intelligence community in its strategic
forecasts than was provided ostensibly by SALT
1. When SALT began, the United States pos-
sessed unquestioned superiority in strategic nu-
clear systems vis-a-vis the Soviet Union. Today,
the debate centers not on the fact of U.S. stra-
tegic inferiority, but rather on its degree. None
of these changes in the military balance was
forecast by the CIA a decade ago. What will the
SALT II Treaty do for U.S. intelligence fore-
casts of Soviet military capabilities that differ
from the past SALT experience?
To note one example: Already there are
press reports that the Soviets have been testing
a new reentry vehicle on the SS-18 missile 96
U.S. analysts evidently are concerned that this
RV may have some sort of terminal guidance
system which could give the Soviets much more
confidence in their ability to destroy U.S. ICBMs
on the ground. Unfortunately, but not surpris-
ingly, analysis of the new RV is being hindered
by Soviet telemetry encryptment, which is per-
mitted under the SALT II Treaty. It remains
to be seen whether this new RV being tested on
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
the SS-18, the prim, threat to U.S. ICBMs, is
compatible with either the spirit or the letter
of the SALT II Treaty.
Third, the implied notion that the SALT II
Treaty will somehow help the CIA to do better
what it has done badly is the brittlest of argu-
ments in favor of ratifying th. Treaty. As ha:
been contended above, the poor record of the
CIA in divining Soviet strategic objectives and
forecasting Soviet strategic capabilities has not
been for lack of evidence or "reference points";
rather, the mistakes have issued from the hu-
man failure to interpret accurately the ample
evidence that has been available. To be sure,
strategic forecasting can never become an exact
science, and some surprises in Soviet weapons
developments and deployments are probably in-
NOTES
1. Robert F. Ellsworth and Kenneth L. Adelman,
"Foolish Intelligence," Foreign Policy, Fail 1979, pp'
147-159.
2. For a summary of the historical pattern, see
William T. Lee, "The Soviet Defense Establishment in
the '80s," Air Force Magazine, March 1980, pp. 100-
108, most specifically, Table I on p. 103.
3. 1 shared this faulty perception of Soviet strategic
priorities in the early 1960s and previously have dis-
cussed it at some length. William T. Lee, "Understand-
ing the Soviet Military Threat" (New York: National
Strategy Information Center, Agenda Paper No. 6,
1977), pp. 27-30.
4. Ibid., pp. 33-39, 42-45.
5. This was the "high force/high technology" esti-
mate which assumed the Soviets would drop the three
RV version of the SS-9 and move on to MIRVs, which,
as it turned out, the Soviets did. However, it took them
until 1974 to do it, the date Congressman Aspin says
the 1969 NIE forecast Soviet MIRVs via the "low
force/low technology" route.
6. Charles S. Sheldon II, "Soviet Space Programs,
1966-1970," Staff Report prepared for the Senate
Committee on Aeronautical and Space Sciences (Wash-
ington, D.C.; Government Printing Office, 1971), p.
168. These two shots were the only launches in the
"Elektron" series. For an independent evaluation of
Soviet programs leading to the same conclusion, see
David S. Sullivan, "The Legacy of SALT I: Soviet De-
ception and U.S. Retreat," Strategic Review, Winter
1979, p. 33.
7. The first flight test of the three RV, "triplet"
version of the SS-9 ICBM occurred in August 1968,
Statement of Secretary of Defense Melvin R. Laird on
the Fiscal Year 1972-76 Defense Program and the
1972 Defense Budget, (March 9, 1971), p. 46.
8. Secretary of Defense Melvin R. Laird, Fiscal Year
1971 Defense Program & Budget, (February 20, 1970),
p. 104. This forecast followed the "high force/high
technology" route. According to Secretary Laird, the
NIE did not offer a "most likely" Soviet course between
the "high force/high technology" and "low force/low
technology" routes. Furthermore, the confusion is
compounded because Secretary Laird's Report is dated
February 20, 1970. Therefore the NIE he referenced
must have been written in 1969 when Congressman
Aspin says the Board of National Estimates ruled out
Soviet counterforce capabilities through the 1970s.
Secretary Laird said the MIRV forecast would give the
Soviets a "formidable hard-target capability by the mid-
1970s' while the multiple RV forecast would provide
a "considerable" capability in the same period.
evitable. Saying this, however, does not exon-
erate the consistently poor record of the CIA
and the other intelligence services in forecast-
ing Soviet forces, at least not since the revolu-
tion in the "national means of verification" that
took place two decades ago.
The solution to tht problem is not a treaty
with the Soviet Union that might somehow ease
the burden of responsibilities on the CIA and
the other intelligence services. The solution,
rather, lies in such improvements as are neces-
sary to ensure that forecasting is done more
competently. Meanwhile, let the fate of the
SALT II Treaty ride or fall on the putative claim
that it will control strategic arms in ways that
contribute to the security of the United States
and its allies.
9. Admiral Thomas H. Moorer, United States Mili-
tary Posture for FY 1972, March 9, 1971, p. 7; Senator
Gordon J. Humphrey, "Analysis and Compliance En-
forcement in SALT Verification," International Secur-
ity Review, Spring 1980.
10. Laird, FY 1971 Defense Program & Budget, p. S.
11. Statement of Secretary of Defense Robert S.
McNamara, Before a Joint Session of the Senate Armed
Services Committee and the Senate Subcommittee on
Department of Defense Appropriations in the Fiscal
Year 1968-72 Defense Program & 1968 Defense Budget,
(Washington, D.C.: Government Printing Office, Jan-
uary 23, 1969), pp. 43-45.
12. Ibid., p. 53.
13. Congressman Aspin cites a CIA estimate of 0.5
can CEP for the SS-9 triplet in 1971 and "initial press
leaks" (undocumented) in 1973 giving the same CEP
for "first tests" of the new Soviet MIRVs. The first
documented report available to this author puts the
accuracy of Soviet MIRVs at 0.3 nm in the mid-1970s,
Air Force Magazine, March 1977. Secretary of De-
fense James R. Schlesinger apparently told Congress in
1975 that Soviet MIRVs were accurate to 0.33 to 0.25
nm. Clarence Robinson, "Cabinet Shifts May Speed
SALT," Aviation Week and Space Technology, Novem-
ber 10, 1975, p. 12. Later reports, reflecting subse-
quent testing, put the CEP of the SS-17, SS-18 and
SS-19 at 0.1 to 0.06 nm, Aviation Week and Space
Technology, February 11, 1980, p. 12.
14. For the link between these precision tools and
improvements in Soviet MIRV accuracy, see Senator
Harry F. Byrd, Jr., commenting on the testimony of
Dr. Jack Vorona of the Defense Intelligence Agency
before the Subcommittee on General Procurement of
the Senate Committee on Armed Services, Congres-
sional Record, Senate, November 13, 1979, p. S. 16506.
15. For a survey of much of the pertinent Soviet
literature, see William T. Lee "Soviet Nuclear Targeting
and SALT" in Steven Rosefielde, editor, World Com-
munism at the Crossroads (Boston: Martinus Nighoff,
1980), pp. 55-88.
16. General Major V. Kruchinin, "Contemporary
Strategic Theory on the Goals and Missions of Armed
Conflict," Military Thought, No. 10, 1963.
17. For a first approximation of such calculations,
see W.T. Lee "Soviet Nuclear Targeting and SALT,"
op. cit. Refinement using Soviet instead of U.S. esti-
mates of soft targeting vulnerabilities, damage criteria
and objectives now in process reinforce these pre-
liminary calculations.
18. For a first approximation of Soviet strategic
missile allocations and reserves against the Conus
target array, see W.T. Lee "The Soviet Defense Estab-
Approved For Release 2008/04/15 CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
lishment in the 80's," op. cit., p. 105. (Note that the
ordinate in Figure 1 should be in thousands.)
19. Marshal N. Krylov, "The Nuclei r Missile Shield
of the Soviet State," Military Thought, No. 11, 1967,
p. 18.
20. R.A. Robinson, "The Evolution of Soviet Naval
Policies and Programs (Part 3), Cuba Missile Crisis to
the Late 1960s," American Intelligence Journal, Spring
1980.
21. Ibid. This happened in mid-July 1966, probably
just a few months before the first Y-class SSBN was
launched.
22. Secretary of Defense Donald H. Rumsfeld, An-
nual Defense Department Report FY 1977, p. 53.
23. Secretary of Defense Donald H. Rumsfeld, An-
nual Defense Department Report FY 1978, p. 63.
24. "Statement of Data on the Numbers of Strategic
Offensive Arms as of the Date of Signature of the
Treaty." signed by V. Karpov, Chief of the USSR Dele.
gation to the Strategic Arms Limitation Talks, printed
in U.S. Department of State, Bureau of Public Affairs,
Selected Documents No. 12A, SALT 11 Agreement
(Vienna: June 18. 1979).
25. Jack Anderson, Washington Post, December 12,
1976.
26. Captain John H. Moore, ed., Jane's Fighting
Ships 1975-76 & 1979-80 (London: Jane's Yearbooks).
27. Secretary of Defense Harold Brown, Annual De-
fense Department Report FY 1981, p. 103.
28. Admiral A. Chabenko, "Combat Rocket Carrying
Atomic Submarines," Military Thought, No. 12, 1967,
pp. 45-47. Admiral N. Kharlamov, "Some Trends in
the Development of Navies," Military Thought, No. 10,
1967, pp. 66-67. That such a submarine would have
a titanium hull was indicated by Captain N. Shatrov,
"Trends in the Development and Employment of Naval
Fleets," Military Thought, No. 1, 1972, p. 55.
29. Secretary of Defense Brown, FY 1981 Report,
p. 103.
30. Secretary of Defense Donald H. Rumsfeld, An-
nual Defense Department Report FY 1977, p. 97.
31. Secretary of Defense Brown, FY 1981 Report,
p. 38.
32. Secretary Rumsfeld, FY 1977 Report, p. 81.
33. Thomas Powers, The Man Who Kept the Secrets:
Richard Helms and the CIA (New York': Alfred A.
Knopf, 1979), p. x.
34. "Giant Soviet Submarine Reported," New York
Times, May 8, 1980. Given the pattern of Soviet nu-
clear powered submarine developments, both a new
model cruise missile SSN and a new model ASW SSN
should appear at any time. Given the reported size of
this new SSN, it probably is the former rather than the
latter. It will be particularly interesting to see if this
new submarine also is a titanium hull model designed
to realize the tactical advantages of greater depths and
high speed as described by Admirals Kharlamov and
Chabenko and Captain Shatrov in the Military Thought
articles previously cited.
35. David S. Sullivan, "Evaluating (l.s. Intelligence
Estimates," in Roy Godson, ed., Intelligence Require.
ments for the 1980s: Analysis and Estimates (New
York: National Strategy Information Center, 1980),
p. 54.
36: Henry S. Bradsher, "New Soviet Missile Tests
Watched Closely by U.S., Washington Star, June 3,
1980, p. 7.
A REBUTTAL BY CONGRESSMAN ASPIN
William Lee's criticisms of my article ap-
pear to fall into four rough categories:
(1) that I did not analyze every intelligence
estimate concerning Soviet military matters of
the past two decades; (2) that I did not clearly
state where I obtained some of my facts; (3)
that some things I classify as overestimates
were actually underestimates and that I pro-
vide misleading alibis for CIA errors; and (4)
that SALT does not really provide strategic pre-
dictability. In addition, Mr. Lee raises the im-
portant point that some misestimates are much
more significant than others, a point he claims
I neglect.
Mr. Lee's first two criticisms can be dealt
with quickly. Basically, he chides me for not
including in my article such intelligence esti-
mates as those pertaining to Soviet theater
forces and defense budgets. Let me just say
that Mr. Lee should have addressed himself to
the article I wrote, not to the one he would have
written. I state very clearly, up front, that the
article is about strategic forecasting; I do so
chiefly to put the issues of a world without
SALT into some perspective. While conven-
tional force estimates and the size of the Soviet
defense budget are interesting issues-and they
are certainly on the menu of topics I will be
investigating through the House Intelligence
Committee-they fall outside the realm of the
topic I was addressing in my article. Inciden-
tally, I certainly regard Mr. Lee's comments as
valuable indicators of what problems may be
worth examining. I should also remark that I
find his estimates of the Soviet defense budget
-which proved more accurate than those of the
CIA prior to 1975-to have been an important
contribution.
As to the sources of my facts, this is more
difficult terrain. I have access to classified in-
telligence data as a member of the House Per-
manent Select Committee on Intelligence.
Committee rules forbid me from disclosing
classified information received through the
Committee. Thus, I used only public state-
ments and other open literature. This con-
straint shaped the choice of estimates and ex-
amples I included in the article. Many of the
topics raised by Mr. Lee in his commentary
were excluded by me because they are treated
inadequately or not at all in public sources.
This applies notably to the 1965 National Intel-
ligence Estimate (NIE) and Mr. Lee's com-
ments about what he had learned about its con-
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Special Edition -- 30 September 1980
tents and the internal controversy surrounding
it.
Third, there is Mr. Lee's criticism of my clas-
sification of certain CIA errors as under- or
overestimates. His reworking of my table seri-
ously distorts the matter. First, he misunder-
stands my point about the role of peripheral sys-
tems (IR/MRBMs and medium-range bombers)
in the "bomber gap" and "missile gap." He criti-
cizes me for not recognizing that the CIA's un-
derestimate of Soviet medium-range forces was
caused by, or at least proportional to, CIA over-
estimates of intercontinental forces.
Yet, this is precisely what I did say, except
in reverse order-i.e., that the overestimate of
long-range bombers and ICBMs was caused by
lack of appreciation that the Soviets planned to
build more forces targeted in the theater. Mr.
Lee says, "Mr. Aspin excuses U.S. intelligence
services on the grounds that they did not grasp
either Soviet strategic priorities or the Soviet
perception that the immediate threat to the
USSR emanated from NATO and U.S. overseas
bases...." (Emphasis added.) Yet, I did not
excuse them for this error; I criticized them for
it. Indeed, a main point of my article was that
CIA errors have often been caused by misjudg-
ment of Soviet priorities and intentions and
that this constitutes a highly serious-perhaps
the critical-problem. If Mr. Lee does not un-
derstand that I was making this point, then I
am afraid he has badly misread my article.
He also misread my article when he claims
that I misinterpreted estimates of the date when
the Soviets would have MIRVed ICBMs. He
writes as if I were talking about when the So-
viets would develop MIRVs, when my article
clearly refers to estimates of MIRV deployment.
Mr. Lee argues, furthermore, that the 1969
NIE's forecast of Soviet MIRV development
( 1971) was only three years off the mark,
which is quite good compared with the 1968
NIE (1978). Actually, since I was talking
about estimates of MIRV deployment, and since
the Soviets deployed MIRVs in 1975, both esti-
mates are equally off the mark-by four years.
Mr. Lee is also misleading on the matter of
the Bureau of National Estimates' statement in
1969 that the Soviets "could not, and would not
try to" attack Minuteman silos effectively in the
1970s. Mr. Lee considers this a dramatic mis-
judgment, but it needs to be considered in con-
text. While the Soviets did, indeed, develop
some capability of endangering Minuteman
missiles in the 1970s, the BNE was talking
about "first-strike capability"-i.e., the potential
to knock out Minuteman without suffering sub-
stantial retaliation. On this matter, the BNE
was right in principle, as the Soviets are only
now beginning to acquire a capability to knock
out almost the whole Minuteman force.
Moreover, Mr. Lee calls this BNE statement
"a most damning indictment of the CIA." This
is polemical: surely Mr. Lee knows that the
BNE and the CIA are not the same institution.
In fact, Mr. Lee's accusation against the CIA for
underappreciating the Soviet threat to Minute-
man should instead he leveled against the Joint
Chiefs of Staff, who convinced Defense Secre-
tary Clark Clifford to delete from a memo to
President Johnson a warning based on (as it
turned out, false) intelligence that the SS-9
might pose a threat to Minuteman. (Inciden-
tally, Mr. Lee repeats the old error of referring
to the SS-9 "triplet as a MIRV-a tendentious
and misleading epithet, as my article points
out. )
Mr. Lee is also misleading, or perhaps simply
confusing, when he challenges my claim that
the intelligence community overestimated the
yield of SS-18 RVs on the grounds that I need
"to demonstrate that the figures of 1.5 megatons
[the overestimate] and .6 megatons [the revised
and now-believed correct estimate] refer to one
and the same RV model." This is illogical. If
intelligence community evaluations of explosive
yield refer to deployed RVs, and if these RVs
are in fact .6 megatons, then a forecast that
RVs carrying 1.5 megatons would be deployed
is much too high, whether or not the two RVs
are the same model.
A basic problem in Mr. Lee's rebuttal is that
he simply misses the point of my article. I at-
tempted to address the perception that the intel-
ligence community is biased toward overesti-
mating or underestimating Soviet strategic
programs. My conclusion is that the commu-
nity does both. This conclusion is insensitive
to the coincidence that the cases I examined
were divided about equally between overesti-
mates and underestimates. The point would be
equally valid if the division were not so equal.
(If, as Mr. Lee claims, the division were lop-
sided in one direction, we would have a serious
problem, but for strategic estimates it is not
lopsided.) The important conclusion for policy.
makers is that intelligence forecasts cannot be
as reliable as we would like them to be, and
therefore a major contribution of arms control
treaties such as SALT is that they introduce a
measure of certainty (or at least bounded ex-
pectations) into an otherwise uncertain envi-
ronment.
Mr. Lee's criticisms of this conclusion con-
cerning SALT distort the argument. Mr. Lee
first observes that SALT II does not cover all
Soviet systems. But SALT does cover the sys-
tems of most importance to the United States.
The ability to know the limits of Soviet options
for the systems covered by SALT is a plus, even
if the other systems remain difficult to predict.
Second, Mr. Lee claims that SALT I did not
enable the intelligence community to forecast
the onset of U.S. strategic inferiority. This of
course is not the intelligence community's job.
Any forecast of the strategic balance requires
prediction of American as well as Soviet forces.
Prediction of American forces is improper for
intelligence. In recent years the intelligence
community has begun comparing its predic-
tions of Soviet forces to plans for U.S. forces.
These so-called "net assessments" are a valid
intelligence function, but they do not permit
forecasts of the strategic balance except in the
conditional sense: if the United States carries
out its plans.
As Mr. Lee well knows, furthermore, doom-
sayers in the intelligence community went well
beyond their mandate by forecasting U.S. in-
feriority in the early 1980s. Along with ele-
ments of the intelligence community, I have
challenged the claim that these forebodings
have proven correct.
Third, Mr. Lee asserts that the "implied
claim" that SALT will help the intelligence com-
munity forecast Soviet strategic forces is the
"brittlest" of all arguments in favor of ratifica-
tion. While SALT II does ease the intelligence
agencies' forecasting task, this is a side issue.
The benefit of SALT is that it offers to policy-
makers a more predictable strategic outlook.
Judged by the record, the intelligence commu-
nity cannot supply this predictability, which is
useful for rational choices about future forces
and diplomatic policy.
Mr. Lee notes that some misestimates by the
U.S. intelligence community are more serious
than others. I agree with the conclusion, but
differ somewhat on the criterion. Misestimates
are serious when they cause us to make choices
different from those we would have made if
faced with a correct estimate, and when these
poorly based choices place us in a disadvanta-
geous position. Few intelligence estimates meet
this criterion of significance. Many factors
totally unrelated to intelligence enter into
policymaking, and it is rare that these other
factors are so evenly balanced that intelligence
can be shown to have made the decisive differ-
ence. However, on occasion it has made a
difference, and that difference has been caused
by overestimates as well as by underestimates.
For example, the intelligence community's
overestimates of Soviet MIRV and ABM capa-
bilities almost certainly played a role in the
Senate's 1969 approval, by one vote, of deploy-
ing an American ABM system. Had the intelli-
gence community realized then, as it did two
years later, that the SS-9 "triplet" could not
threaten Minuteman and that the Soviets would
not try to deploy a nationwide ABM, one Sena-
tor might very well have voted "no" instead of
"yes," and we would have been spared the waste
of deploying an inadequate, unworkable ABM
that we later dismantled. Mr. Lee seems to
think that overestimating Soviet capabilities is
more acceptable than underestimating them.
However, this example indicates that whether
one kind of error is more harmful than the
other depends on the specific details of the
case.
The main point of my article is that intelli-
gence errors are caused more by misunder-
standing of Soviet priorities and requirements
and by imperfect analysis than by inadequate
technical collection. I infer that Mr. Lee gen-
erally agrees with this. I concur with his notion
that we must have an intelligence community
able to admit its own errors. I can also com-
ment that whatever the failings of the past, the
House Intelligence Committee has not found
intelligence officials unwilling to admit where
they have gone wrong once it can be demon-
strated that something is wrong.
I should add that, contrary to Mr. Lee's sug-
gestions in his final remarks, simply because
individuals can make predictions that occasion-
ally turn out correct, this does not mean that
all equally objective and knowledgeable ob-
servers would reach the same views ahead of
time. In intelligence there are always differing
.opinions, even when all the analysts are un-
biased. There is also the problem of political
interference, some cases of which are discussed
in my article. It is against improper political
interference-and not all political actions
which influence intelligence are improper-
that we must take aim. We need to encourage
vigorous, technically competent research and
analysis within the intelligence community;
and we need to make policy in recognition that
crystal-ball gazing has inherent limitations
when it comes to forecasting Soviet strategic
forces.
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
95th Congress l
2d Session f
THE NATIONAL INTELLIGENCE ESTIMATES A-B
TEAM EPISODE CONCERNING SOVIET STRATE-
GIC CAPABILITY AND OBJECTIVES
SENATE SELECT COMMITTEE
ON INTELLIGENCE
SUBCOMMITTEE ON
COLLECTION, PRODUCTION, AND QUALITY
UNITED STATES SENATE
TOGETHER WITH
SEPARATE VIEWS
Approved For Release 2008/04/15: CIA-RDP86R00893R000100130003-4 i
THE NATIONAL INTELLIGENCE ESTIMATES-A-B TEAM
EPISODE CONCERNING SOVIET STRATEGIC CAPA-
BILITY AND OBJECTIVES
The Senate. Select Committee oil Intelligence. as part of its over-
y of the
19i fi ".A 'Ie~uu-li 'I'ruu"
sight function, has conducted a stud
experiment in uuuparative. assessnienis of Soviet strategic strength
which was initialed by the President's Iorvigu liltelligence Advisory
Board (PPIAB). The conmuittoo conducted this inquiry under its
nuunlate to evaluate the collection, production, and quality of [-.S. in-
telligence. in this case assssiug whether the, A-B experiment. had
proved to be a, useful procedure in improving Nulional Intelligence
Estimates (N I K's) 01 1:1 centrally important question.
The pt?rtinei t facts of the A-B case are (a) that PFI.Ali ce.nniis-
sioned three ail hoc outside. groups (composing the "B Team') to
examine the chin acailablo to the.IT.S. intelligence connnnnity:s ana-
lysts (the "A 'leant"). to determine whet lid' such data would sup-
port conclusions on Soviet strategio capabilities and objectives dif-
ferent from those presented in the conummrity's NIE's: and (b) that
during the exercise details of these sensitive questions leaked of sev-
era.l occasions to the press.
The committee has prepared a classified report. on the subject,
sent copies of that report to the. executive branch, made, copies avail-
able to certain members of the, B Team for review and comment, and
subsequently reelecked the record thoroughly and accommodated
some of the B Train members' comments. A summary of the. commit-
tee's report follows.
The committee sought, to determine the facts -and issues central to
the A Team-B Team case, and to give it critique of the, procedures
which underlay the principal judgments and conduct of both the. A
and B Tennis. The committee's report makes no attempt to judge
which group's estimates concerning the U.S.S.R. are correct. The
report focuses on the processes followed; its finding, and recom-
mendations for improving the quality and utility of future NIE's on
Soviet strategic capabilities and objectives are primarily directed at
procedural issues.
In the broadest sense, the NIE-I3 Train episode derived from, a
growing concern over the U.S.S.R.'s steady increase. in strategic
weapons strength over the course of the past decade and disagree-
Approved For Release 2008/04/15 : CIA-RDP86ROO893ROO0100130003-4
C
meat within the U.S. intelligence comnlnuit.v r on the meaning of
this growth.
The B Team experiment in competitive analysis stemmed from
the PFIAB's opinion that the NIE's had been underestimating the
progress of Soviet strategic weapons: In an August 1975 letter to
President Ford, PFIAB Chairman George AV. Anderson, Jr.,
proposed that the President authorize the NSC to implement a "com-
petitive analysis." The then Director of Central Intelligence (DCI)
William E. Colby, speaking with the unanimous agreement of the
U.S. Intelligence Board (the chiefs of the intelligence community
components), responded with a proposal that the PFIAB first ex-
amine an applicable NIE then underway and thereafter determine
what specific course of action to take.
The PFIAB found weaknesses in that NIE and, after having made
further investigations of its own, again proposed' (in April 197(1) an
experiment in "competitive analysis." The PFIA13 recommended that
the exercise be placed under the DCI's jurisdiction and that it address
certain critical estimative issues.
i'ltINCiPAL JUDGMENTS AND IECOMM1N DATIONS
The committee's report includes these central judgments:
That the concept of a review of the NIE's by outside experts
was it legitimate one.
That the 13 Team made some valid criticisms of the NIE's,
especially concerning certain technical intelligence questions, and
some useful recommendations concerning the estimative process,
but those contributions were less valuable than they might have
been because (1) the exercise had been so structured by the
PFIAB and the Director of Central Intelligence (DCI) that
the 13 Team on Soviet objectives reflected the views of oily one
segment of the spectrum of opinion; and (2) that. Team spent
much of its effort on criticizing much earlier Nil?'s rather than,
as had been earlier agreed upon by the PFIAB and the DCTproducin_, alternative estimates from certain of (hose of the 1()7(:
NIE.
That the value, of the A-13 experiment. was further lessened lv
the fact that details concerning these highly classified questions
leaked to the press, where these appeared in garbled and one-
sided form. It has not been determined who was responsible for
the leaks.
That, most importantly, NIE's on Soviet strategic capabilities
and objectives still need improvement in a number of important.
respects.
r Tn the past. the 11.0. Intelligence commanity inch tied the Central Intelligence Agency,
the Defense Tntelligenre Agency. the National Security .Agency, and the intelligence com-
ponents of the State Department. Army, Navy, Air Force, Fill. Energy Resmirers Develop-
ment Ad,nlnistratlon. and Treasury.
As of Aucust 1075, the PFTAB's member,. In ndddtinn to Chairman George W. Ander.
son. ,Jr. (Aim. IJSN, Ret.). were William O. Raker. Len Chrrne, John S. Foster, .Jr..
Robert W. Calvin. Gordon Crny, Edwin IT. Land, Clare Booth T,uce. George 1'. Shultz. and
Ed Wa rd Teller. As of mid-1070, Mr. Cherne had heroine rim o Irmo n, anti these additional
nernhers lind joined the PFTAB: John R. Connally, Gen. Lyman L. Tmmnltzer. Robert D.
Murph, and Edward Bennett Williams. The I'FTAB function has since been abolished by
President Carter.
Thronch its Committee on NIE Evaluation (Messrs. Robert Calvin. Edward Teller, and
John Fester).
Approved For Release 2008/04/15 CIA-RDP86R00893R000100130003-4
J
The report's principal recommendations include:
That a collegial estimative group be formed in place of individ-
ual National lntelli%ence Officers.
That outside critiques of NIE's should continue to be con-
ducted, but should, in each instance, be made by expert groups
which are broadly representative in character, and whose pro-
cedures are thereafter more strictly monitored by the couuuis-
sioning authorities than obtained in the A-B case.
The committee's investigation was based upon study of prinuuy
documents; examination of the NIE record since 1959 on Soviet
strategic weapons developments; and interviews with principals front
the. A and B Teams, the intelligence continuuity. and the PFIAB.
191, committee has enjoyed the full cooperation of all the above
parties. The continents of DCT Stansfield Turner on the report and the
present. statement and the views of curtain members of the B Team
on Soviet Objectives have been given consideration by the roumtittee.
Responding to the. PFIAB initiatives, the new DCI, Mr. George
Bush, consented to the experiment. and by Time 1976. the PFL\B
and tie DCT had worked out ground rules for it competitive assess-
ment experiment. The DCI, through his represeniat.ives. trade ar-
rnngenients for. and monitored the experiment in accordance with,
those ground miles. Members of the PFTAB were called upon to assist
in the formation of the three B Tennis and tool; an active role in the.
selection of team mcnihers.
The exercise did not, simply pit an A (or NTE) team against a B
Team. There were three. B Tennis: two on technical questions and one
on Soviet objectives. As for the N side, an NIE on Soviet strategic
weapons had already been regularly scheduled earlier in the year, and
work on it by the. intelligence community had already begun before
the B Teams came into being. This NIF, was much broader in scope
than the particular estimative questions the. B Teams had been com-
missioned to address, and the individual civilian and military analysts
involved in producing that NIE represented a wide range of views
held within the departments and agencies of the intelligence com-
munity on the NTE's many questions.
The NTF participants and the B Teams proceeded to product, their
two sets of studies independently. with only occasional direct contact
during the drafting phase. After the initial drafts of the three B teams
were completed, the two sides confronted one another formally on three
occasions. Once the decision to proceed with the exercise had been
made, procedural cooperation was good between the intelligence com-
munity and the three respective B teams. The specific results differed,
however, in the three cases. Those concerning technical questions were
the most rewarding: there was a, mutual give-and-take. and these B
Teams clearly made a constructive contribution. By contrast, the dis-
ciissions concerning Soviet objectives were more controversial and less
conclusive. The. B Team on Soviet. Objectives contributed some useful
critiques concerning certain technical intelligence questions, but there
was not much give-and-take on broader issues. The view cited in a.
December-1976 press article4 that the 13 Team challenge turned the
NTE "around 180 degrees" is incorrect.
New Turk Time,,, nee. 26. 1976.
23-542-T6-2
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
C
Ti. is the view of the committee that past NTE's could have profited
from drawing on experts on Soviet strategic questions from outside
the intelligence community, both in and out of Government, and from
subjecting NIE analyses and judgments on this and other areas to
competing assessments from such sources. The PFIAB's 1975-76
proposition that outside expertise should be used to criticize and evalu-
ate the NIF.'s was a legitimate one. The exercise in practice, however,
fell Short of the initial conception.
The composition of the B Teani donning with Soviet. objectives was
so structured that the outcome of the exercise. was predetermined and
the experiment's contribution lessened. The procedures followed by the
intelligence community in the, A-B episode also weakened the overall
effort to some degree. The, intelligence agencies were cast inaccurately
in the role of "doves," when they in fact represented a broad spectrum
of views. They needlessly allowed analytic mismatches, by sending
relatively junior specialists into the dehating arena against prestigious
and articulate B Team authorities. And the monitoring of the pro-
cedures of the B Team on Soviet Objectives was subsequently fairly
loose.
The B Team contributions and the 1976 NIE can also he faulted on
various substanti ve grounds. Because of its narrowly specified purpose
and scope. influenced strongly in recent years by the preferences of
senior policymaking readers regarding format, the NTE did not ad-
dress the question of how Soviet strategic weapons development fits
into important larger concerns [the entire panoply] of Soviet domestic,
military, diplomatic, economic. and cultural efforts. As a consequence,
the NIE's discussion of Soviet objectives was too brief to he useful. In
the view of some readers, its discussion of 'Soviet military hardware in
certain respects was inadequate to be helpful to high-level officials.
_1 weakness in both I lie NTE and the B Teton report is their lack of
expressed sensitivity to the fact and the significance of world develop-
ments other than those directly related to the U.S.-Soviet arms race.
The strategie weapons balance is the chief subject of both documents,
but both documents nonetheless are dominated by military hardware
questions and define "strategic power" quite narrowly. By design. in
neither the NIE nor the B report are U.S.-Soviet strategic matters
set within the wider framework of other dynamic world forces, many
of which are essentially the creatures of neither U.S. nor Soviet
initiative or control.
n/1MMI71TEE FINDINGS
Estimates should, of course, be written in an accurate and dispas-
sionate, manner. They should reflect. the best and most. broadly rep-
resentative expert knowledge possible, from both inside. and outside
the Government. The sensitive estimative questions at hand should not
be, argued in the press. These, requirements did not obtain in the case
of the NTE-B Team exercise.
The field of strategic weaponry is complex, and there is much valu-
able expertise on the subject outside of the intelligence community.
The quality of NIE's on these subjects would benefit from more. exten-
sive use of this outside knowledge than is now the case In this respect,
the PFIAB initiative was justifiable and desirable.
Approved For Release 2008/04/15 CIA-RDP86R00893R000100130003-4
To be of maximum value, however, such efforts most employ the best
and most competent expertise available. Panels representing only one
perspective, whether "hard" or "soft," are not desirable. In this
respect, the B Team "experiment" was not as constructive as it could
have been concerning Soviet objectives.
The exercise in competitive analysis was devalued by the fact that,
contrary to the expressed directions of both DCI George Bush and
PFIAB Chairman Leo Cherne, word of these sensitive matters leaked
to the press, where it appeared in garbled form.
The A-B Team experience sharply demonstrated the intense pre-
occupation of the CIA, the rest of the, intelligence community, the
PFIAI3, and policyunakers with Soviet strategic weapons and their
consequences. This subject is of enormous significance to U.S. policy-
mnaking, but, there are also other significant questions. The greatest
intelligence attention often is given to the least likely Soviet actions,
nuclear attack, rather than to Soviet intentions and assertive world
activity short of those extremes.
Of most significance, the A-B Team case has demonstrated : (a) that
the key question of Soviet strategic intentions and conduct is one
which demands the best possible marshalling of U.S. intelligence re-
sources and American brainpower; and (b) that the estimative process
needs improvement in this area of concern.
The committee's recommendations for improving National Intelli-
gence Estimates concerning Soviet strategic weapons capabilities and
objectives included these judgments :
The intelligence community must more effectively meet the
particular needs of particular policymakers. Creative use should
be made of other estimative formats, in addition to the current
categories of NIE's, tailored to the particular needs, but not the
views, of different policymaking entities and levels.
There is need for competitive and alternative analyses. Both
within the estimative body and with respect to outside expertise,
competing and on occasion alternative estimates should be en-
couraged. To be fully useful, such initiatives must avoid panels
with narrow preconceptions, of whatever kind, to assure the. bal-
ance necessary for the competitors to evaluate evidence which is
often both conflicting and ambiguous.
Estimates must openly express differences of judgment, and
clearly indicate, the assumptions, the evidence, and the reasoning
which produce alternative readings.
Estimates should highlight significant changes from related
past estimates, including changing probabilities, the emergence
of new important alternatives, and findings that make past esti-
mates false or less relevant.
NIE's should define "strategic matters" more comprehensively.
than has obtained in recent years, so that. Soviet military develop-
ments can be better seen within the context of Soviet interests
and policies, and in interaction with U.S. and world developments.
Enchantment with the details of military hardware must not.
permit either the producers or the policymaking consumers of
intelligence to become, deflected from pursuit of the most impor-
tant estimative questions at hand, those of intentions.
Approved For Release 2008/04/15 : CIA-RDP86R00893R000100130003-4
Reliable net assessments are needed to complete an effective
estimative process, so that policymakers can better appreciate
Soviet strengths and v, aknesses by hev' g systec..atically com-
pared them with those of the United States-a function which
the NIE's are not designed to perform. The NSC should commis-
sion such net assessments, to be prepared by experts at the national
level, including some from the intelligence community.
Policvmakers must define the questions, not the answers. The
DCI and the intelligence community's estimative hodv must re-
main independent in judgment. Judgments must not be bent or
suppressed by outside pressures or fear of an uncongenial reaction.