CRITIQUE OF THE CODEWORD COMPARTMENT IN THE CIA
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP83B00823R000900180001-6
Release Decision:
RIPPUB
Original Classification:
T
Document Page Count:
57
Document Creation Date:
December 12, 2016
Document Release Date:
November 21, 2001
Sequence Number:
1
Case Number:
Publication Date:
March 1, 1977
Content Type:
REPORT
File:
Attachment | Size |
---|---|
![]() | 1.84 MB |
Body:
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Top Secret
INTELLIGENCE
MONOGRAPH
CRITIQUE OF THE
CODEWORD COMPARTMENT IN THE CIA
i CENTER FOR THE STUDY OF INTELLIGENCE
CENTRAL INTELLIGENCE AGENCY
Top Secret
TR/IM 77-02J
TCS 4530-77
Handle via TALENT-KEYHOLE-
March 1977
COMINT Channels
Approved For Release 2002/01/08: CIA-RDP83B00823R0009001- $of 250
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
THE CENTER FOR THE STUDY OF INTELLIGENCE IN OTR OPERATES
A RESEARCH AND DISCUSSION PROGRAM KEYED TO THE PROCESSES
AND FUNCTIONS OF INTELLIGENCE. THE OBJECTIVE OF THE CENTER
IS TO CONTRIBUTE TO THE PROFESSIONAL UNDERSTANDING AND
TO THE RECORD OF THE ART OF INTELLIGENCE. RESEARCH PROJECTS
ARE UNDERTAKEN BY INTELLIGENCE "FELLOWS"-VOLUNTEER OFFI-
CERS FROM ACROSS THE AGENCY ON FULL-TIME DETAIL TO THE
CENTER. INQUIRIES ABOUT THE CENTER PROGRAM, OR COMMENTS
ON THIS REPORT ARE INVITED BY THE DIRECTOR/CSI, EXTENSION 2193.
Warning Notice
Sensitive Intelligence Sources and Methods Involved
(WNINTEL)
NATIONAL SECURITY INFORMATION
Unauthorized Disclosure Subject to Criminal Sanctions
Classified by 058227
Exempt from general declassification schedule
of E.O. 11652, exemption category:
? 5B(2)
Automatically declassified on:
date impossible to determine
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/Ig8~M4P83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
CRITIQUE OF THE CODEWORD COMPARTMENT
TOP SECRET
HAN, -1Fc~i rA e s dA1dd / -bif4~lR~b ~d6 6 o d t LS
Approved For Release 2002/01/0 SECRET P83B00823R000900180001-6
TOP HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
A CRITIQUE OF THE CODEWORD COMPARTMENT IN THE CIA
Page
FOREWORD i
SUMMARY OF CONCLUSIONS AND RECOMMENDATIONS 1
THE CODEWORD COMPARTMENT 4
Genesis of the Codeword Compartment 4
CIA and the Codeword Compartment Today 9
THE ANARCHIC NATURE OF THE CODEWORD COMPARTMENT 12
Lack of Central Control 12
Psychological Aspects 16
Problems With Other Agencies 18
Structural and Conceptual Problems 21
FRAGMENTATION OF INTELLIGENCE 24
Decompartmentation 29
PROBLEM OF STRUCTURAL REFORM 41
CONCLUSIONS AND RECOMMENDATIONS 43
CIA Management of Compartmentation 45
Community Management of Compartmentation 49
The Problem of Central Control 50
The Problem of Decontrol 51
The Problem of Fragmented Intelligence 52
FOOTNOTES 59
TCS 4530-77
TOP SECRET
HANIAAfoN'i l F~6APMbiEeIO 1Tt"I1 :( ID '3M&MROD9?I 419 415LS
Approved For Release 2002TOP8SECRET 83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
This study, the second in a series of Center papers on
compartmentation in the CIA, focuses on the codeword com-
partment.* This security device is designed to provide
special protection, beyond that provided by the federal
classification system, to a specific category of sensitive
information. Its charter is Section 9 of Executive
Order 11652.1 The Codeword compartment is only one form of
supplementary protection for sensitive materials, although
perhaps the predominant one. Although no exact accounting
is possible, it appears that the use of codeword-protected
materials may be involved in up to one-half of total Agency
man-hours.
What distinguishes the codeword compartment from other
forms, for example those used by the Operations Directorate
to protect sources and methods, is its relatively large
* The first study, dated 6 December 1976, was entitled
Secrecy vs. Disclosure: A Study in Security Classification.
In the series of Center studies, compartmentation has been -
viewed as "formal" as it relates to the codeword systems,
"informal" as it pertains to the generalized need-to-know
found in the classification system or the more rigid need-
to-know practiced in the Clandestine Service, and "bureau-
cratic" as it evolves accidentally from organizational
The author of the final draft of this study was
of the Directorate of Operations. He drew
upon the contributions of other Center fellows.
i TCS 4530-77
TOP SECRET
H A Nfl~i~o ~kid1F a'rff ~e~T~ ~(RJ~~17~4$E -~1~R1~ ~9~~~~~8 0 ~QQ l~ ~ L S
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
interagency membership, its established and enduring character,
its formalized procedures for granting access approvals and
indoctrinating initiates, and its use of distinctive code-
words to control personal access and document distribution.
Usually, it protects information collected by technical
means, such as electromagnetic intercept or overhead pho-
tography.
The codeword compartments--perhaps because of their
size and complexity--have been extensively studied by the
CIA and the Defense Department.2 A recent Defense Depart-
ment announcement described 20 ongoing studies of various
aspects of compartmentation. While taking note of the
earlier studies, many of which are highly technical and have
a Community focus, a point was made in this paper of not
covering the same ground. Instead, this study focuses for
the most part on the operation of the codeword compartment
within the CIA and on the criticisms of it voiced by Agency
officers and bearing on the Agency's intelligence mission.
The paper seeks to assess the validity of these criticisms
and to suggest some remedies.
The methodology of the study involved use of an atti-
tudinal questionnaire, interviews of officers in all direc-
torates, and research in the available literature. For
I I TCS 4,530-77
Aproved ForRele 2pp2 IGxET 0M9_ HANDLE VIA 1 AL~1 1=,EJM ' Sf 8Z?R Wft"9. 1?PPAk?;ELS
Approved For Release 2002fp ,8f 9 83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
25X1A
security reasons, the study was restricted mainly to the
three principal systems of compartmentation--Communications
Intelligence, TALENT-KEYHOLE (see pages 5-7 for 2 1A
definitions); smaller, more sensitive informational compart-
ments were touched on only lightly. Study of the
compartment was limited generally to information available
through the M access approval. Where applicable, the
intelligence product of the compartments was emphasized.
projects designed to protect the collection source,
engineering development, and technology, were not included.
The study first looks briefly at the historical devel-
opment of the codeword systems, spotlighting their structural
and hereditary problems. It then examines the principal
criticisms. A final section presents the study's conclusions
and recommendations, a summary of which follows.
TCS 4530-77
TOP SECRET
H A -9 d ~1i-v e I P rI R1 a Ie~I 1~ i 6~ L- 0 6; ~3 DiQ Il ATOO P I' E L S
25X1 A
Approved For Release 2002/1160P :SgIffP831300823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
SUMMARY OF CONLUSIONS AND RECOMMENDATIONS
The difficulties involved in the functioning of the
codeword compartments stem from the following main causes:
--the ability of agencies and departments to uni-
laterally establish compartments and the con-
sequent lack of central control over them;
--the unilateral withholding of compartmented
information by the originator of the information;
--the failure of the compartments adequately to
distinguish between the needs of analysts and
ultimate consumers;
--the fragmentation of intelligence caused by the
operation of the varied compartments; and
--the failure consistently to decompartment non-
sensitive information.
The study's recommendations fall into two broad categories:
those within the competence of internal CIA management to
act upon and those affecting the Intelligence Community as a
whole. The former deal with:
--centralizing record keeping on access approvals of
all sorts;
--the creation of an Agency information center to
assist researchers and others whose jobs require
access to compartmented information;
--CIA leadership in the updating and/or preparation
of guidance manuals for the three main compartments;
--improved security, indoctrination, reindoctrination,
and a new emphasis on substantive indoctrination;
TOP SECRET
HAlY k)vViVFoT1Kk1 'E BN@i(D.E BMWRRg9NQINQ LS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
IIANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
suggestions for the bulk decontrol of imagery
eligible for decontrol; and
---establishment within a CIA Classification Board of
a special section on compartmented information.
To provide central supervision of the national com-
partments, the study recommends the revision of Section 9
Executive Order (E.O.) 11652 to require: (a) the concurrence
of the Director of Central Intelligence (DCI) whenever a
member of the Intelligence Community establishes a compart-
ment that inhibits the flow of information to other members
of the Community; (b) DCI monitoring of the existing
compartments to ensure that they do not impede the production
of national intelligence.
To cope with the problem of decontrol of compartmented
information, the study recommends:
--a mandatory annual review of compartmented infor-
mation to segregate for decontrol information
whose sensitivity has changed during the year;
--appointment by the DCI of a committee or task
force to study the desirability of decompart-
menting to SECRET NOFORN the totality of the
TALENT-KEYHOLE (TK) imagery product; and
--decontrol to SECRET NOFORN or CONFIDENTIAL NOFORN,
ommunications
Intelligence
To counter the fragmentation of intelligence, the study
recommends the establishment within the DCI Committee
2 TCS 4530-77
RR TOP SECRET
HA q~~ A ~Y 1RrFVAK-1?~ Of I81?R O R"9,O1?PPQ1 TELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
structure of a new Committee on Integrated Intelligence. It
also makes several recommendations for structural reform
suggesting that:
--the Agency sponsor legislation for the establish-
ment of a new category of legally protected in-
formation to be called: Sources-and-Methods
Information;
--the DCI and the Intelligence Community abolish all
existing compartments and replace them with one
uniform Sources-and-Methods Compartment. (Access
to this information would require a full back-
ground investigation.); and
--formal criteria for need-to-know be drawn up for
collector, top policy maker, analyst, processor,
and consumer. All Sources-and-Methods Information
would be slugged to reflect these criteria and
routed accordingly.
TOP SECRET
HANDWOV(fid,F'tAR ' '?I? I964. -el3?lKq3MI 08L90coHANNELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
"We should also intensify our
efforts to downgrade, sanitize,
and decontrol where possible
highly compartmented products
so that they may be more widely
disseminated and used. This
will. require greater refinement
in distinguishing categories of
intelligence which must be given
greater, rather than less, pro-
tection in such a new atmosphere."
Director of Central Intelligence
Perspectives for Intelligence
197o-1981.
Genesis of the Codeword Compartment
The Hoover Commission in 1955 remarked that the CIA
might well owe its existence to the surprise attack on Pearl
Harbor. That most disastrous of intelligence failures was
due in no small measure to the mishandling of compartmented
intelligence. The dissemination of decrypted Japanese
communications, codenamed was so restricted 2 1A
that the theater commanders in Hawaii did not regularly
receive them. The decrypted items were not synthesized into
a meaningful whole nor collated with information from other
sources.
4 TC5 4530-77
TOP SECRET
HA ~LoE'e~ ~r"Rl~I?~ ~ ~ ~.c ~Rl$3 @ P k 91 ~4A14JELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
The embryonic compartment that handled the Japanese
communications was managed by the War and Navy Departments.
In 1941 the Navy decryption unit, the Communications Security
Unit, had a total staff of 300; the Army unit, the Signal
Intelligence Service, a staff of 224 in Washington and 150
in the field. A G-2 agreement with the Office of Naval
Information (ONI) of 23 June 1941 on the dissemination of
COMINT designated only nine persons as recipients--essentially
the top civilian and military officials concerned with
national security. By the end of the war, in Washington
alone, the Army unit had a staff of over 10,000 and the Navy
unit one of 6,000.3 Thousands of others were recipients of
the product. The pattern of rapid growth thus established
continued for many years.
COMINT became a formal compartment in the modern sense
of the term-with the signing of a
agreement
in 1946. This agreement established procedures on both
sides for the handling, safekeeping, and exchange of COMINT.
It was an outgrowth of the wartime exchange of COMINT: our
the codename for decrypted German
traffic.
In 1952, the responsibility for COMINT was centralized
in the newly created National Security Agency (NSA) to help
5 TCS 4530-77
TOP S
HAN zo F4rA JqM5~ A%PR. I,KST 001 ANNELS
25X1 A
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
eliminate the duplication in intercept and dissemination
efforts between the various collectors that had been a
problem from the beginning. But even then COMINT was not
considered foreign intelligence (in the sense of being part
of the foreign intelligence collected by the U.S. intelli-
gence effort). This did not change until the establishment
in 1958 of the United States Intelligence Board (USIB--now
the National Foreign Intelligence Board--NFIB) under the
chairmanship of the DCI, when COMM at last became a
recognized part of foreign intelligence.
Technical advances in collection capabilities brought
forth new compartments, vaguely patterned after the COMINT
compartment. CIA's use of the U-2 aircraft for high-altitude
photographic coverage led, in the mid-1950's, to the estab-
lishment of the TALENT compartment to protect both the
collection system and its intelligence product. A Presidential
directive of 20 August 1960 broadened the TALENT Control
System to include all information collected by means of
nationally tasked aerial reconnaissance, at the same time
creating a subcompartment, codenamed KEYHOLE, for information
collected by satellite. By 1962, the problem of managing
the development and operation of the growing satellite
systems and of administering access approvals to information
6 TCS 4530-77
Approved VFor IA Release
T-KEYHOLEI BO 8GMgq '8nqX-fKN ELS
HANDLE
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
25X1 A
superintending satellite collection of photographic and
signals intelligence; and TALENT-KEYHOLE containing the
intelligence product of Although not divided func-
tionally between collection system and intelligence product,
the COMINT compartment is stratified into three categories,
based on sensitivity. 's the most sensitive.
Standing by the side of the three main compartments are
an "indeterminate" number of lesser compartments. Members
of the former Intelligence Resource Advisory Committee
(IRAC) were said to require access to some :ompartments
to do their jobs. But the exact number of compartments in
the Intelligence Community at any given time is not easy to
ascertain since it is difficult to find anyone who is cleared
for all of them or even witting of the existence of all of
them. There is no central point in the Agency where one may
go to verify the access approvals held by an individual.
The Compartmented Information Branch of the Special Security
Center maintains records of the three main systems and some
third country ones only. Control officers for the other
compartments are dispersed throughout the Agency. Most, if
not all, of the creation of the smaller compartments appears
to have occurred within the Defense Department.
8 TCS 4,'60-77
TOP SECRET
R01ALS
25X1 A
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
CIA and the Codeword Compartment Today
The codeword compartments today are vastly different
from the wartime compartment out of which they grew. The
main differences lie in the numbers of employees with access
to them, in the complexity of their structures, and in the
vastness of their information holdings. All three major
compartments have grown steadily, and their scope will
further expand with the advent of "real time" collection.
Each advance in collection technology has brought a corre-
sponding increase in the numbers and categories of persons
with a "need-to-know" the information collected. This
increase has even transcended the Intelligence Community.
Witness the contributions of overhead photography to agri-
culture, climatology, exploration for new energy resources,
and drought control.
In 1975, the Office of Central Reference (OCR) dissemi-
nated more than four times as many compartmented as noncom-
partmented intelligence reports (444,870 versus 98,551).
The Intelligence Community's output of compartmented finished
intelligence (monographs and articles) handled through OCR
exceeded that of finished noncompartmented intelligence by
14,359 to 9,742. Although these figures do not represent
the complete output of either the CIA or the Intelligence
9 TCS 4530-77
TOP SECRET
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
'TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
25X1A
Community, they give an accurate picture of the relative
numerical significance of compartmented intelligence reports
and documents. To process the latter, large numbers of CIA
employees have been admitted to the compartments. Thus, 86%
of all employees have access approvals for COMINT, 60% for
TK, and 40% for Large segments of the Agency,
particularly in the Deputy Directorate of Intelligence (DDI)
and the Deputy Directorate of Science and Technology (DDS&T),
work almost entirely on compartmented intelligence. And
even in the Deputy Directorate of Operations (DDO), where
most divisions spend relatively little time on it, one whole
division, works nearly full-time on COMINT.
Probably one-half of total Agency man-hours is spent on some
facet of compartmented intelligence. The size of this
investment underscores the importance of understanding the
compartmented milieu within which such a large volume of
sensitive but valuable intelligence is collected, processed,
evaluated, and disseminated.
Despite the relatively large access of Agency employees
to compartmented information (at least to the main compart-
ments), complaints are still heard from analysts and others
* Since February 1977, part of the DDS&T.
10 TCS 4530-77
Approved For Relea~ 0Q2LqP I,CRET 0 19 ~N.ELS
HANDLE VIA TALI. 43-KL WM"9'
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
of production difficulties caused by denial of codeword
information. At least in the minds of many analysts, the
problem posed by Pearl Harbor is still with us: How do we
reconcile the protection of highly sensitive information
with its optimum use? And on the side of optimum use, an
old question with new ingredients: What is the effect on
the intelligence cycle of the multiplicity of unilateral
compartments with their collection orientation, their
overlap, and their rigidities? A related question is
whether there really is an intelligence gap caused by the
compartments.
For this study, the attitudes of a cross section of
Agency employees toward compartmentation were surveyed.6
Although the 163 persons who answered the questionnaire did
not technically qualify as a "random sample," they constituted
a representative sample in terms of directorates, job cate-
gories, grades, and length of service. One hundred and
thirty-one had access to at least one compartment and 78 to
more than two. What should be stressed is that 73% of the
respondents, a very significant percentage, said they had
not been denied codeword information; on the contrary, they
had received it regularly on a need-to-know basis. Overall
the survey conveyed the impression that the codeword com-
partment was not a serious problem for the CIA.
11 TCS 4530-77
TOP SE -s~ p~~9
HAN E?`'AF?1'`A1~iLE B ~ULOQ~g
CHANNELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
But detailed interviews with persons intimately con-
cerned with the compartments left a somewhat different
picture. Although not contradicting the survey as such, the
picture was of a much more variegated, more complex landscape,
dotted here and there with problems stemming from the operation
of the compartments. For the most part, these problems
involved the operational dynamics of the compartmented
systems, such as:
--the anarchic nature of the codeword compartments;
--the unilateral withholding of compartmented
information;
--the failure to distinguish between the needs of
analysts and ultimate consumers;
--the fragmentation of intelligence; and
--the failure to decompartment nonsensitive infor-
mation.
In the sections that follow these issues are discussed.
THE ANARCHIC NATURE OF THE CODEWORD COMPARTMENT
Lack of Central Control
The unilateral authority of the agencies to establish
compartments permits the originating department to fence off
its information and establish ground rules for access and
distribution. In the name of security, information may be
12 TCS 4530-77
Approved For Release 2002/ P 0
HANDLE VIA TALENT-KEPI IJ #W"1"-cuNW?3R0L8A0R"ELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
25X1 A
withheld in whole or in part, watered down by sanitizing, or
served up as a restricted summary. There are, of course,
DCI committees that attempt to coordinate the use of com-
partmented information and impose order in an area where
autarchy is too often the rule. But the originating depart-
ment remains the final arbiter. No wonder then that the
compartments are dissimilar in structure, conflicting, and
overlapping.
Nor is it surprising that they generate confusion.
Each "subcompartment" in the M ompartment has its own
guardian codeword. Even if an officer possesses the generic
for access to the complex, he does not enter a par-
ticular subcompartment without a specific access approval
for the subcompartment. If one had business with all of
them, it would require a half dozen or so access approvals.
Each codeword, except _ protects a specific photo-
graphic or Signals Intelligence (SIGINT) satellite collec-
tion system. Each collection project is singlemindedly
focused on the development, production, or operation of its
own system. Its preoccupations are those of the designer,
engineer, or system manager, not those of the processor,
analyst, or consumer. The interests of the latter, therefore,
tend to be neglected. Information belonging by definition
13 TCS 4530-77
TOP SECRET pp g
HAND ? P~ KJ3EM
~Q9 N?Rp~~
NN LS
25X1 A
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/OF8$: JA;, T3B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
and an unclassified designator assigned by NSA. 2 1D
This can confuse the user of the intelligence product,
making it difficult for him to remember which designator
applies to which system. Whether the designators bolster
security, as originally intended, seems questionable. These
examples are meant to suggest rather than exhaust the degree
of confusion obtaining in the compartments. As the number
of access approvals held by an individual increases, it
becomes more difficult for him to distinguish one from the
other or even to remember clearly those he holds.
A product of the confusion factor is the compartmenta-
tion expert. The intricacies of compartmentation are such
that few can weave their way through its recesses without a
guide. To publish compartmented information more widely, it
is necessary either to sanitize or to decompartment it. And
it is here that the services of the compartmentation expert
are in greatest demand. The analyst must sometimes consult
with him to verify that finished intelligence has the proper
markings and conforms to the rules of compartmentation.
This can introduce additional delay into the production-
dissemination process. But the experts, located for the
most part in the Requirements and Evaluation Staff in the
Comptroller's Office, are not exempt from confusion. They
15 TCS 4530-77
TOP SECRET
HAA dve~l br IIB ~ ~/1~I1~L~~ ~1E ~O QO KKBO.~t ELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
tend to be highly specialized, and when a compartmentation.
problem involves information moving simultaneously in two or
more compartments, they are sometimes hard put to unravel
its complexities. This is no reflection on them, but rather
a consequence of the structural incompatibility of the
compartments and the inadequacies of existing guidance,
particularly for sanitizing and decompartmenting.
Psychological_Aspects
We know that secrecy by its very nature may affect the
personality of its practitioners. This is true of all forms
of secrecy from the primitive secret society to the codewcrd
compartment. The latter is a heightened form of secrecy
that resembles the former in many ways. It has the aura cf
a secret society. It has its initiation, its oaths, its
esoteric phrases, its sequestered areas, and its secrets
within secrets. And in place of passwords and hand signs,
there are letter designations on badges. There are in-
groups and out-groups. No wonder, then, if the codeword
compartment has unintended psychological effects.
For many, the badge with its distinctive letters has
become a status symbol, and for some of them the symbol has
fused with careerism. Others, equating knowledge with
power, have become collectors of clearances. They have lost
16 TCS 4530-77
'FOP SECRET
Fl A E?TVf
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
sight of need-to-know. The effect on those without access
to a compartment is sometimes adverse. They may wonder if
their loyalty is being measured against those admitted to
the compartment. The aggressive analyst who considers it a
point of honor to know everything going on in his field may
take exclusion as a challenge to outwit the codeword compart-
ment. He feels free to probe and to exploit to the hilt his
personal contacts. In this sense, compartmentation tends to
subvert the formal channels of communication. Whether
aggressive or not, the analyst who knows he is not privy to
compartmented information on a particular subject will often
be more diffident in expressing his views. And if his
interlocutor retorts: "You don't have all the clearances,"
his diffidence may increase.
There are also other more general effects. Some are so
impressed with the trappings of codeword information that
they come to consider it more accurate than other informa-
tion not confirmed by COMINT or photography.8 Others see
codewords as an attention-getting device, pyramiding them on
publications to assure the customer that all sources have
been tapped, thus contributing to overclassification and
overcompartmentation. Some components of the Intelligence
Community--NSA is often accused of this--compartment infor-
17 TCS 4530-77
TOP SECRET
HAN%ordMa
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
mation according to its substantive importance rather than
its intrinsic sensitivity. Those admitted to codeword
compartments frequently believe they have been granted a
certified need-to-know for everything within the compart-
ment. Still, codewords do deter: few would have the
temerity to take codeword material home, few even among
those who take home other types of classified material. On
balance, however, the psychological side effects of the
codeword compartment seem to diminish rather than enhance
security.
Problems With Other Agencies
The interests of the department or agency originating
and administering the compartment and those of the depart-
ments receiving the information contained in the compartment
are not the same. Differences arise most frequently over
the number of persons to be admitted to a compartment and
over the kind and amount of information to be granted to
those admitted. The DCI Committee structure takes the edge
off these disputes, establishes a modus vivendi, but has
been unable to remove their causes. Because of CIA's strong
position with respect to the TK and M compartments, CIA
access to these compartments is usually not a problem.
18 TCS 4530-77
TOP SECRET
HA,ie~rJF4 ~eT2[/~'Llj~J0WJM&W&0fi8 IN-grELS
25X1 D
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
Unfortunately, the same cannot be said of the COMINT com-
partment nor of some of the smaller Defense Department
compartments, particularly those of the Navy.
With respect to the COMINT compartment, CIA's problems
on the substantive side generally involve the difficulty of
obtaining what some analysts believe is a full enough text
of the actual intercepted messages rather than the gists,
summaries, or paraphrases that the National Security Agency
prefers to make available in some cases. These difficulties
extend to disputes over what constitutes "technical data"--
the material associated with the intercept which describes
in essence the "sources and methods" part of its origins.
Some CIA analysts feel that they need more of the latter in
order to make full use of the intercepts, while NSA naturally
wants to protect it, especially in the light of the bad
leaks of intercepted communications which have occurred in
the media in recent years. It is this concern which also
generally formulates NSA's policy in substituting gists and
summaries for full texts of COMINT material.
Judging the right and wrong of these positions is
almost impossible at present except on a case by case basis.
It is clear, however, that the organizational and juris-
dictional boundaries thrown up around the compartments by
19 TCS 4530-77
TOP SECRET
HAl`~~,~~O1I~~4~fJ$ E~~QQG$QA~P~ELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
the intricate rules and semi-independent ability of the
agencies to construct these rules have contributed in no
small measure to the situation and that any reform must go
in the direction of centralizing the authority for estab-
lishing national compartments and the rules under which they
operate.
Two of the Navy compartments may be taken as typical of
the problems posed by the smaller national compartments.
These compartments (unidentified here for security reasons)
have the trappings of the three main compartments--codewords,
indoctrination oaths, and special transmission channels.
But they also have distinctive features being functionally
tiered into levels for collection, analysis, and production.
The corresponding tiers require separate "tickets" for
access. CIA's principal difficulty with these compartments
lies in the restricted number of "billets" or access ap-
provals it receives. This allotment of "billets" is clearly
not based on a careful assessment of need-to-know within the
Agency. According to one estimate, the Electronics Intel-
ligence (ELINT) activities in the Agency presently require a
tenfold increase in the number of "billets" in one of the
compartments. Since there are not enough "billets" for all
the analysts who need them, the practice of juggling per-
HA9DLo e~FAr'1'~IE '1'-K-EMA
R40e6 ?159 R vELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
sonnel in and out of the compartments, briefing and debrief-
ing them, has developed, offering a good example of compart-
mentation defeating its own purposes.
What NSA and the Navy are doing, according to their
point of view, is unilaterally protecting some of the most
sensitive and fragile information the U.S. possesses. The
culprit is the unilateral nature of the compartments; only a
central arbiter can resolve the differences.
Structural and Conceptual Problems
In the preceding section, some of the ways policies and
procedures imposed by compartment managers can adversely
affect the raw and finished intelligence product was de-
scribed. The compartment in its origin is the creature of
the collector. It reflects his two principal concerns: to
protect the collection system and to maximize collection.
It is only when these concerns conflict with the ultimate
purpose of all collection--the production of national
intelligence--that a problem is created.9 Unfortunately,
the collection bias of the compartments sometimes does just
that.
Compartmental bias expresses itself structurally in the
establishment of separate collection and product compartments
for the same system. The classic instance is the
21 TCS 4530-77
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
25X1 A
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
collection compartment and its TK counterpart. The under-
lying assumption is that the Intelligence Community can be
neatly divided into collectors and consumers. But, suppose
an analyst has received an unconfirmed clandestine intel-
ligence report about a new missile system. He knows over-
head reconnaissance can confirm or deny the report. But to
task the appropriate system, he must have at least a nodding
acquaintance with the collection projects. If he received
photographic coverage and the results are inconclusive
because of the sun's angle, both he and the processor (the
photointerpreter) must know that it is possible to request
additional coverage at a different time of day. Only the
collection system operators need know how this is done, but
the processor and the analyst must know that it can be done
and be able to task the system.
Ignorance of the potential of overhead photography as a
means of solving intelligence problems and of supporting
military and foreign policy decisions is still fairly wide-
spread. For example, relatively few State Department per-
sonnel are cleared for TK. It was only within the last year
that State got around to establishing for the
storage of TK and the use of those cleared for it. There
have been instances when the use of overhead photography in
22 TCS 4530-77
1 T TOP SECRET
HAl~~l~v~ci`~`or F~~1'~2[5~ 2~2t0'T/OS-'kCIA=F~~P83TB0'08Z31~OOOJtlO'I"80DII1-N6ELS
25X1 A
Approved For Release 2002 P88 fifkW83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
analysis would have been useful but, being unfamiliar with
its availability, analysts did not think to ask for it.
Besides analysis and production of finished intelli-
gence, the analyst is sometimes responsible for evaluating
the raw product of the collection system. To do the latter,
the analyst must sometimes know more about the collection
system than is normally provided by access to the product
compartment.
What has been said of overhead photography is equally
true of overhead S:[GINT. The processor (the signals analyst)
and the finished intelligence analyst need to know less than
the collector but more than the consumer of finished intel-
ligence. The latter needs to know that the system exists or
will exist, its intelligence potential and its reliability,
but not the processes employed by the systems operator, the
processor, and the analyst.
Although not divided structurally, the COMINT compart-
ment is nonetheless split along user-collector lines. The
practices of gisting and withholding "technical data"
mentioned in the preceeding section are predicated on the
principle that message externals, the verbatim text, and the
means by which the information was collected and processed
should be kept in the hands of the collector and not released
23 TCS 4530-77
TOP SECRET
HAI Wf&e o T asNeT2M WFcfA bib O'gFW( 0 JELS
Approved For Release 2002 FIP> hl~ft P83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
to the user. The separation of "technical" COMINT from that
released to users takes place by means of need-to-know
rather than by means of formal compartmentation. Regardless
of the means employed, the effect on the analyst is the
same. Whether the bias is structural or conceptual, it has
this in common: it does not fully accommodate the inter-
mediate need-to-know situated between the collector and the
consumer of finished intelligence.
All forms of compartmentation have a fragmenting effect
on information. But the codeword compartment, because of
the volume and importance of the information sequestered,
undoubtedly has the greatest effect. Not only is all infor-
mation on one subject not in one file, but access to each
repository requires a separate access approval. The prin-
ciple that assigns information to a particular compartment
is not a fully logical one; it is based primarily on security
and the convenience of the collector. Intelligence, like
all knowledge, ideally is seamless and indivisible. In its
search for meaning it abhors all compartments, no matter how
necessary they may be.
24 TCS 4530-77
TOP SECRET
I IANA~FoJ~~1FoTrAAp.Ta~~-$~~'/b~~~~i~~3~~0~1~0b~~~LS
Approved For Release 2002( (P8~I%P83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
How serious is the fragmenting effect of the codeword
compartment? a military and warning analyst,
answers the question this way:
The compartmentalization of information--both
intelligence material under special clearances and
policy or operational information by executive action--
is a serious enough problem in normal circumstances.
In time of possible impending hostile action, it could
be disastrous. The gulf which separates the Joint
Chiefs of Staff from the desk-level military analyst
may be immense . . . . From day to day it may not make
much difference whether or not the analyst is privy to
much of this material (in very restricted collection
systems); he may make some minor skips for lack of it,
but the security of the nation will not be at stake.
But it is entirely different when the enemy is getting
ready for some surprise action, for then the analyst's
"need to know" may be as great, if not greater, than
that of the policy maker.
Indeed, it is arguable that all intelligence failures,
starting with that of Pearl Harbor, are attributable in some
degree to abuses of compartmentation, to a failure to
integrate information. According t "The 2 1A
intelligence analyst who did not know that the Pueblo was
off North Korea prior to its seizure, and most did not, was
not likely to have alerted his superior to some anomaly in
North Korean naval activity. The list of such examples is
endless."11
Conspiring against the integration of information in
the Agency is not merely the codeword compartment but also
the information explosion itself, the increasing special-
25 TCS 4530-77
TOP SECRET
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
ization among analysts, and the functional organization of
the DDI and DDS&T.12 One consequence is the increasing
difficulty of assembling all needed information on a given
subject. Although there is a plethora of indices, few of
them are all-source. When intelligence data are not entered
into compatible computer programs, the computer is unable to
perform its unique function of bringing together related but
dispersed fragments of information; there is a loss of
"convergence." The basic information on lasers lay dormant
in the Intelligence Community, no one knowing it was there,
until a concerted effort was made to find it. Recently, a
team
had difficulty getting the information it needed out of some
of the more hermetic compartments. Some Agency officers
believe that the latter not only reduce the flow of infor-
mation to a trickle, but channel it in directions contrary
to need-to-know. Sensing this situation, the DCI in his
Perspectives for Intelligence: 1976-1981 called for a more
integrated approach to intelligence: "The growing inter-
dependence nationally and among disciplines will require a
greater integration of many activities which in prior years
could be handled in separate compartments."
The mischievous effects of fragmented data also affect
requirements, collection, and dissemination. For example,
25X1 D
26 TCS 4530-77
TOP SECRET
IIA1`~IllrL? e7~~r'1R~~, r ~~ LE~~' OQ 4~Q@~I91 I 1 ELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
the Fiscal Year 1975 performance report on Key Intelligence
Question (KIQ) 14
con-
cluded that: "Where data were disseminated widely enough to
permit coordinated tasking, collection tasks were substan-
tially or almost completely satisfied." The report explained
that compartmentation of "certain relevant data" (pertaining
to had hampered Community efforts
to collect and disseminate information concerning the
requirement; exploitation and tasking feedback had suffered
for the same reason. The report added that although the
problem "has been alleviated in part by wider distribution
of certain of the relevant data . . . there still remains
the difficulty of disseminating to the interested customer
meaningful finished intelligence on these data." In order
to exploit and analyze available information on one aspect
of KIQ 14, the report suggested this could probably be done
"if all-source data--some of which is highly compartmented--
were exploited by an analytical group." But here again,
because of the restrictions governing the particular com-
partment, the analytical group would have had difficulty
getting the results of its work to appropriate customers.13
Since the compartments are organized around a technique
of collection, the information they contain is there because
27 TCS 4530-77
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
it derives from that technique, not because it constitutes a
discrete body of knowledge that can stand on its own feet.
The information in the various compartments is therefore
incomplete and only becomes fully meaningful when synthe-
sized and meshed with collateral and other compartmented
information. This is obvious to all analysts. But for a
variety of reasons, some of which have already been men-
tioned, the full integration of intelligence data still
eludes the intelligence Community. This is particularly so
in the area known as "fusion," the coordinated use of
different collection techniques to obtain a rounded view of
a single "event."
Perhaps in no other area are the barriers of compart-
mentation so formidable. For fusion demands close inter-
compartmental--and therefore interdepartmental--coordination.
Multitechnique collection and subsequent integrated analysis
of the products is not one of the strong suits of the
Intelligence Community. The physical and administrative
dispersion of the processing elements may compound the
difficulty, although the simple amalgamation of organiza-
tions is by no means an easy answer to what may be more a
matter of attitude and perception than of structure. NSA
does the readout for COMINT, FLINT, and telemetry; the FLINT
28
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 D
25X1A
unit of CIA does the readout for ELINT and telemetry col-
lected by CIA or received from third parties; the National
Photographic Interpretation Center (NPIC) does the readout
for overhead photography,
M and the Office of Computer Services assists both the
ELINT unit of CIA and NPIC in their readout.14
Advances in technology are thus putting strains on the
traditional compartment. To remedy this, some have proposed
the establishment of a multisensor compartment. In the
interim, the National Intelligence Officer (NIO) system, the
KIQs (which ignore compartments), and the pioneering concept
of the are integrative forces
acting to counter the fragmentation of intelligence.
Decompartmentation
Decompartmentation or decontrol is the removal of
information that has undergone a change in sensitivity from
the compartmented state to the simple protection of the
federal classification system. As such, it is an important
measure of the health and legitimacy of the codeword compart-
ment. In terms of the intelligence process, if properly
conducted, it fosters the integration of intelligence. On
the other hand, it is often controversial, triggering dis-
29 TCS 4530-77
TOP SECRET
HAM bvvdWoTReJ "-E-2O 11G1 D$EC R BQQ 2R QJ4WPA LS
25X1 D
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
putes over the residual sensitivity of a piece of information,
the extent of foreign knowledge of our technical collection
systems, and the risk of compromise if the information is
decontrolled. In arguing these points, opponents and pro-
ponents often reason as if decontrol were tantamount to
declassification.
In assessing how decompartmentation is functioning, it
is necessary to note at the outset that there is no man-
datory review of compartmented information for purposes of
decontrol. The latter usually takes place only upon request,
when someone has noticed a nonsensitive item that would be
useful to analysts not enjoying access to the compartment.
Under these circumstances; overcompartmentation and over-
classification are inevitable. The COMINT compartment is a
case in point. An earlier Agency study of compartmentation
concluded: "We are compartmenting an unnecessarily large
volume of intelligence and intelligence information, notably
COMINT products, the bulk of which has in effect been
sanitized when published by the Cryptologic Community.ttl5
The study explained that most "COMINT information is passed
to consumers in the form of combined, multiCOMINT source
reporting." This, the study stated:
30 TCS 4530-77
TOP SECRET
HAl RMv C1,Ko'rTR4lMcT216I A. P1938 IkWoMoC MiXELS
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
Since it is difficult to determine the probability of
unauthorized disclosure of any kind of information, compart-
mented or not compartmented, most discussions of the risks
of decompartmenting end up as judgments concerning the
relative sensitivity of a piece of information and the
probable effects of its unauthorized disclosure. The only
official measure of sensitivity is the three levels of the
federal classification system corresponding to degrees of
damage that unauthorized disclosure would cause to national
security: simple "damage" for Confidential, "serious
damage" for Secret, and "exceptionally grave damage" for Top
Secret. These three levels equate to the "probable effects
of disclosure." Through a study of the latter, we arrive at
32 TCS 4530-77
TOP SECRET
HAN04iovV"oTAk1?ii-' UNdilDLE IbI918R 000RBMNWLS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
the proper classification of a piece of information. We
can, as a rule, decompartment to the Top Secret collateral
level without any loss of protection. In fact, the
study concluded that "there is less control of documents
classified Top Secret when handled in COMINT control channels
(and to a lesser extent in TK channels) than would be the
case if given the same classification and not compartmented."19
Normally, decompartmentation is to the Secret or lower
level, reflecting a corresponding decrease in the sensi-
tivity of the information decompartmented. The argument
sometimes heard that decontrol "tends to increase the flow
of information toward the public domain" is a specious one
that ignores the classification system. Almost equally
specious is the contention that we should not decompartment
to the Secret level because many holders of a Secret clear-
ance have not had a full background investigation. To the
extent this contention has merit, it argues for changing the
clearance requirements for Secret or limiting certain
categories of Secret information to those who have had a
full background investigation. In the meantime, E.O. 11652
is the only operative guidance.
The notion that some compartmented information, such as
COMINT, is eternally sensitive and therefore
33 TCS 4530-77
TOP SECRET
HAA iibt veVIt [sE*'I QQZV L: EJINe flWUOQ M01MPPARft1ELS
25X1 A
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1 A
25X1 A
automatically exempt from downgrading or decompartmentation
finds no basis in E.O. 11652. Compartmented information is
just as subject to mandatory review after ten years upon
request and to compulsory 30-year review as any other type
of classified information. Yet, COMINT regulations do not
envisage the possibility of decompartmenting
information except in the military context in wartime.20
Nor is there any provision for decompartmenting M
information. The decontrol of TK information is in the
hands of a subcommittee of the DCI Committee on Imagery
Requirements and Exploitation (COMIREX) which is more
attuned to the _ collectors than to the TK users.
On the other hand, there are secrets connected with
overhead reconnaissance which must be protected. They
concern the advanced technology of the systems and their
collection capabilities. These are the secrets that the
-and TK compartments protect. But do all these
secrets require supplementary protection or would simple
classification suffice for some or all of them? The answer
hinges to a great extent on how much the Soviets already
know about our capabilities and what, if anything, they
could do and would do, if they knew more. There is no doubt
that the Soviets are acutely aware of our overhead recon-
34 TCS 4530-77
TOP SECRET
I TAN?`'AFA' ~ILE -~I~10~4ELS
25X1 A
25X1 A
25X1D Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Next 2 Page(s) In Document Exempt
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
thousands of analysts, researchers, and contractors who are
cleared for Secret only. A senior CIA analyst has observed
that the Army's Training and Doctrine Command, which draws
up training doctrine for the future, is largely ignorant of
Soviet capabilities and needs the relevant compartmented
intelligence. Large numbers of Defense Department technicians
are striving to counter Soviet weapons systems without being
privy to compartmented information on the subject. Stepped-
up decompartmentation would improve the quality of analysis
among Defense and other contractors. Counting the tactical
commander, there is therefore a large, untapped market for
decompartmented or sanitized intelligence, one that will
probably expand greatly with the advent of real-time systems.
Closely allied to decompartmentation is sanitization,
an alternative means of removing information from the code-
word compartment. It also involves a calculation of the
degree of sensitivity, but, more particularly, it requires
explicit knowledge of what constitutes the sensitivity of a
given piece of information. This may be the source or the
method of collection or exploitation, the source-revealing
nature of the content, or some combination of these. The
job of sanitization is to delete or obscure the sensitive
elements adhering to the information. "Sanitization is
38 TCS 4530-77
TOP SECRET
fir" Ig 12 - 1 ~I~PI~SBO ~bC9~d0 K~ A ~E,LS
IA fyeff
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
accomplished by deliberately changing the original material
through a process of synthesis with other source materials
or by additions or deletions of substance. A prerequisite
of sanitization is the existence, or a reasonable presump-
tion of the existence, of separate and logical but less
sensitive sources or methods."28 It has been estimated that
from one-fourth to one-third of the original information may
be excised. The result is often a significant loss.29
But sanitization, which is inherently difficult and
time-consuming, is made even more so by the lack of precise,
up-to-date guidance. The TK control manual dealing with
sanitization was issued in 1956; the M ontrol manual
issued in 1966 makes no provision for decontrol; the Communi-
cations Intelligence Security Regulations (CISR) contains no
guidance on sanitization, although it provides for decom-
partmentation of military information in time of war, and
25X1 A
authorizes the publication of information in a 2 1A
separate NOFORN series. There are CIA compartmentation
experts who provide invaluable guidance on knotty questions,
but the bulk of sanitizing must necessarily be done by
analysts. Typical perhaps of the approach to sanitizing of
most officers is the technique of two experienced analysts
interviewed for this study. Because of the lack of precise
39 TCS 4530-77
TOP SECRET
H pp 9Ee4'ft4r`FAl? ' ' qlMLI;WW Q?gM@%M1 Ik1ELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
FOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
guidance and the pressure of deadlines, they "play it safe:,"
and without regard to any specific criteria, quickly cut out
everything that looks sensitive. A more considered approach
might reduce significantly the percentage of information
excised and the corresponding loss of substance.
Because of the cost and time factor, sanitization is
clearly not a substitute for the bulk decompartmentation of
eligible material. The latter approach is the only econom-
ical, timely means of getting large quantities of informa-
tion to those who need it but lack access to it in the
compartment. Despite a failure to follow through on the
full intent of the Presidential directive, great progress
25X1 A
40 TCS 45,30-77
TOP SECRET
HA ?IPVTA ~I - ~I LE ~3 ~? LS
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
25X1A
25X1 A
intelligence compartment." Once admitted to the latter,
need-to-know would determine further access. One senior
officer interviewed called for the abolition of all codeword
compartments, the imposition of stiff penalties for dis-
closure, and the use of only these five designators:
CONFIDENTIAL, SECRET, TOP SECRET, NOFORN, and TOP SECRET
M Others saw no problem in combining COMINT and TK
into a single compartment. A leitmotif of military studies
of compartmentation has been emphasis on the "gap between
the sensor community and the weapons applications community"
which has been preventing combat commanders from receiving
needed information.
A recent study by the former USIB Security Committee
recommended replacing the present system with two compart-
ments, one for technical data (collection data) and one for
the product of collection. "Efficiency and utility," the
study said, "would seem to indicate the establishment of a
single system of access based strictly upon intelligence
needs once personal eligibility is determined. The local
senior intelligence officer is best qualified to determine
intelligence needs. Therefore a single system of product
compartmentation should be established using the current
42 TCS 4530-77
FOP SECRET
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
personnel security standards with decentralized access
granting authority." There is thus something approaching
consensus on the need for a unified system in which need-to-
know would play a greater role than now. The nemesis of
proposals for reform of so complex a thing as the codeword
compartments is that what looks simple in theory may be
confusing in practice. Thus, the scheme replaces
old codewords with new ones and substitutes internal com-
partmentation for external. And the single unadorned
compartment fails to come to grips with the complexities of
routing sensitive information throughout the Community. A
synthesis of the best features of these various proposals
may nevertheless be possible.
CONCLUSIONS AND RECOMMENDATIONS
A high percentage of CIA employees enjoys access to at
least one of the three main compartments. Generally,
employees feel they are receiving compartmented information
in accordance with the need-to-know principle.31 The
complex comes closest to being a problem. Some 40% of
Agency personnel have access to one or more of its "compart-
ments," but this is a differential access that does not
25X1 A
25X1 A
43 TCS 4530-77
TOP SECRET
HAXO veVIF teRO72QO J f $L? 1 1AOPflMO $WMVVPP,Qkfk1ELS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
I [ANDLE VIA TALENT- KEYHOLE-CO MINT CONTROL CHANNELS
always meet the needs of analysts. Even when it does not
exclude, this partitioned compartment confuses, acting as a
drag on the intelligence process. One may rightly ask
whether its structure is not more responsive to the require-
ments of managerial convenience than to those of security.
Overall, the relatively broad access to compartmented intel-
ligence enjoyed by Agency employees has apparently not
caused significant security problems.
There is a problem of access to the smaller Navy com-
partments and a problem of access to adequate substance in
the materials in the compartments. By fragmenting intel-
ligence data, by overcompartmenting, by hindering computer
retrieval and convergence, by complicating the fusion of
information collected by different techniques, the compart-
ments have made more difficult the integration of intelli-
gence.32 It has indirectly led the Agency to misprize a
potential consumer market consisting of personnel not
cleared for compartmented information.
The present costs of compartmentation in confusion,
wasted manpower, less meaningful intelligence, and enhanced
risk of intelligence failure could be ameliorated, if not
eliminated. As steps in this direction, a series of recom-
mendations are set forth below. They are discussed under
44 TCS 4530-77
TOP SECRET
HANM~v 1Aorl'Wef I2- ~~ ~L:E~ U 1 0~ 1 E LS
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Next 7 Page(s) In Document Exempt
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
the other problems of compartmentation: the establishment
of centralized control; vigorous decontrol of information;
and rationalization of the structure of the compartments.
Still, some steps could be taken to improve the present
system and obtain a more holistic intelligence product.
Extension of Recommendation 1 (a central repository for
information on access approvals) and Recommendation 2 (A
Center for Compartmented Information) to the Intelligence
Community would be helpful.
The DCI Committee structure is largely a reflection of
the divisions attendant upon compartmentation. The NIO
system was established to fill the gaps left by the com-
mittees. The geographic makeup of the DDO and the predom-
inantly functional one in the DDI and DDS&T hamper coordi-
nation among these directorates. And within the latter
directorates, there has been a "fragmentation of analytic
responsibility." "Because of the many academic disciplines
involved, the analysis of weapons systems has been broken
into pieces and distributed among the analytic branches in
several offices."40 The recent DDI/DDSeT reorganizations
were certainly designed in part to overcome these fragmenting
tendencies, but they persist and are reinforced by the
operation of the compartments.
53 TCS 4530-77
TOP SECRET
HAW]idbvV EdFRkJ T-20MM(DLE 3 BM RRQDSDGUMQ LS
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Next 4 Page(s) In Document Exempt
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
TOP SECRET
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
1. Section 9 covers "Special Departmental Arrangements."
It states that "the originating department or other appro-
priate authority may impose, in conformity with the provi-
sions of this order, special requirements with respect to
access, distribution and protection of classified information
and material, including those which presently relate to
communications intelligence, intelligence sources and
methods, and cryptography."
25X1 A
3. Roberta Wohlstetter. Pearl Harbor, Warning and
Decision. Stanford University Press, 1962, p. 174.
4. An article by Peter Szanton and Graham Allison
entitled "Intelligence: Seizing the Opportunity" gave this
description of the NRO: "National Reconnaissance Office.
The largest agency in terms of budget is the National
Reconnaissance Office (NRO) also lodged in Defense. The NRO
operates the numerous 'overhead' (principally satellite)
reconnaissance programs for the community, working largely
through the U.S. Air Force. Its products are medium-resolution
photographs of wide areas and high resolution pictures of
selected points; these are useful to economic analysts and
essential to those concerned with military and arms control
issues. The NRO is subordinate to the DCI and a deputy
secretary of defense." Foreign Policy. First quarter 1976,
p. 187. Note: The NRO considers this information "classified
and compartmented."
59 TCS 4530-77
TOP SECRET
HA1AWrEvOARo 1 ~L:I~~E iIPI Q~ ~IQ~LO M a1ELS
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Next 4 Page(s) In Document Exempt
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Approved For Release 2002/01/08 SECRRDP83B00823R000900180001-6
HANDLE VIA TALENT-KEYHOLE-COMINT CONTROL CHANNELS
48, op. cit., p. 4. In an annex to this study
serves that "the idea of a single compartmentation
is already pretty far along. We have a USIB policy
on physical security requirements to protect any compart-
mented information, a DCID 1/16 on common standards to
protect any compartmented material stored or processed on
computers, a courier system to handle any compartmented
material, and a pretty common understanding of how to log
and mark compartmented materials."
6S TCS 4530-77
TOP SECRET
HAN oYkFc(TMeQ b)ild8E-&W8T3W0 j 90 LS
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Top Secret
Top Secret
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
25X1A Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6
Next 1 Page(s) In Document Exempt
Approved For Release 2002/01/08 : CIA-RDP83B00823R000900180001-6