COORDINATED CODIB CONTRIBUTION ON OBJECTIVE 3

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80B01139A000200130012-5
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 14
• Document Creation Date: December 15, 2016
• Document Release Date: February 13, 2004
• Sequence Number: 12
• Publication Date: August 5, 1963
• Content Type: MF

Approved For qiiease 2004/03/11 : CIA-RDP80601139)11/411160200130012-5
UNITED STATES 1NTELI,IGENC BOARD
COMAITTE.E ON DOCUMENTATION
.5 itagurt 1963
MEMORANDUM FOR: Inspector General of CIA
tIUBJECT: Coordiaated CODIB Contribution la Cbjective 3
REFERENCE: Memorandum froni DDCI, Subject: United States
Foreiga Intelligence Objectives, Top Secret,
dated 8 Joky 1963
There are forwarded hex ewith 13 copies of the coo,!dlnated CODIB
opor on Cbjective 3 as recreated In the 7efereneed memorandum,.
S-E-C-R-E-T
25X1
pone vanoe
'AotirF, Chairman
GROUP I
cluded from .autantatio
downgradirkg atal
declassification
Approved For Release 2004/03/11 : CIA-RDP80601139A000200130012-5
Approved For jease 2004/030.1E?WERIMOB01130,14700200130012-5
*tee' 5 August 1963
PROGRAMS AND PLANS FOR ACTION TO ACCOMPLISH
UNITED STATES FOREIGN INTELLIGENCE OBJECTIVE NUMBER 3
INFORMATION PROCESSING TECHNIQUES
? I. OBJECTIVE NO. 3
That research be intensified to determine the usefulness of data processing
techniques, including mechanized title or summary sentence permutation, to facilitate
review and assessment of the great volume of material that must be dealt with in the
Intelligence community.
THE PROBLEM IN PERSPECTIVE
As of today, only man possesses the intelligence and insight to ask questions,
to exercise judgment, and to recognize a good solution when it turns up; Modern
Information processing equipment and techniques, however, can be of substantial
assistance to an intelligence analyst by drastically reducing the amount of time and
effort required to search for the information he desires, to collate it with other
pertinent information, to update it, to rearrange it in different ways to see if any
patterns are developing, and to perform various types of mathematical computations
and logical structuring. The machine will never replace the intelligence analyst, but,
if skillfully used, it can provide an extremely powerful tool of analysis.
Research and development work in intelligence automation is currently at a high
level and is expected to increase in the years immediately ahead. The rate of progress,
GROUP
SE-C-R-ET Excluded from automatic
Approved For Release 2004/03/11 : CIA-RDP80B01139A0001151M4S and
declassification
Approved For giirease 2004/03/11 : CIA-R P OB011301400200130012-5
-
h,? eeerer, will be governed more by availability of first-rate rivesearch worker:2 thin
y the pply of funds.
oto of the fundamental difficul es in applying automation to intelligence probealas
is
that computer equipsnent in use today is basically designed fee: Uig proceseeng of
nuenerical data. Yet, snost intelligence ADP effort are pa1yconceraed with
re *pt processing, storage, end retrieval of nonenumerical dt
;numerical information processieg deals with t maniendatIon of symbels,
meanings and deaslons in an esuentinlly qualitative and descriiptire manner. As
epAeosed to the processinn af numerical dale which hi characterized by its fundamental
reliance on arithmetic, the processing IA nonenuraerical data includes etr*icrg.
Illsior6ermg, /MUNI
9 Dearching, sorting or rearranging? Dlerglaq, .tandatten: wad tbe
iiPae. Present techniques for performing these tasks are generally not for ilisetl and DO
refiticible to a conaietemanlogtoal notation: in short, the task el defining the problem
so that it can be erogrammed on the conineter is mode TITAore difficult in non-we rice;
than in numeral:al processing.
III. OPERATIONAL AND DEVELOPMENT PROGRAMS
The intelligence community is presently organized an as to be responsive to
changes as a result of the current technological revolution in information handling.
The mechanism within the Intelligence Community for effecting improvement in
intelligence handling techniques is exemplified by the DOD. Here each the three
military d artments and the NSA have research and development organizations. iDIA,
. in turn, develops intelligence research and development requirements for the DOD
Nek,
Approved For Release 2004I0Nig :611149,11801301139A000200130012-5
Approved For lease 2004/03/11: CIA-RDP80B0113G0000200130012-5
SEC-RE-.T
Both NSA and DIA are responsible for ensuring that the projects of the military
departments are operationally sound and meet operational requirements in the
oryptoleetic and non-oryptologio intelligent:)e communities, respectively. The Direotor
of Defense Research and Engin ring then coordinates and guides all of these efforts
for the Secretary of Defense. Finally, DDR&E, NSA, DIA, State Dept., FBI, the
Military Departments and CIA have representatives on the Committee on Documentation
(CODD3) of the United States Intelligence Board. CODD3, which is chaired by CIA,
provides ce tral coordination and leadership for intelligence data handling
matters. Problems are brought up by members for review andteimidance and
Information is disseminated about on-going activities, For example,
melte determakied the need for a survsy oZ community falorreakim processing
rlavit-51Ins, CODIB orgsliteed the Stal t 01121/TILVility !derma ion Pr-
%u, I(SalPS) to perform) the survey with p;roup member& supplt ad by COMB
prommtative agencies under CIA direct= Also, COD1B 14wAtifie4 the problem of
00Xtrfp+PltA13/F24 emanations from operaIng ficw;lwriter-type equirment and !a ,7esolttiting
the prq Went on a communliy-wide basis through dtmelcipment ol standmrd operating
apcinfiestkons, design and productiou of eiluVment and co/rating usage agreements
Wring the past 17 years, as the cold Wrr has spread to arditiousi areas and mtv
communist security measures have tightemed, the U. S. Intel lif emote Community has
toes forcsd t re lacmuaingly on mass ,Joilectio; teohniquas: This hat, rearJited in
onorme,Ate vain es of electronic< photographic, acttustic, and bard copy infozzaation
laKiftilch Novo inundated our manual prooss4ing resources.
Approved For Release 2004/.241,3j.,ithetRyP80601139A000200130012-5
Approved For %lease 2004/03/11 : CIA-RDP806011393411100200130012-5
- 4 -
Manual methods of processing have proved to be incapable of explcdting thoroughly
the volume and type of raw data collected in the following fields: ELINT; targeting;
photography; current missile, air, naval and ground forces activity; merchant shipping
activity; and biographic material. Because of this, in the early 1950s efforts began
In the Intelligence Community to apply punch card techniques to certain intelligence
files, primarily in the targeting and document storage and retrieval areas. Each year
sines then, APD has been applied to a growing number of intelligence tasks. At present,
operational and development programs are underway in the following areas:
A. Document Storage and Retrieval
In 1958 the first automated document storage and retrieval system (MINICARD)
?I'began to be implemented in Air Force Intelligence. Today a number of intelligence
organizations possess machine supported document storage and retrieval systems
tailored to their specific requirements.
CIA operates a central information control system in which about 175.000
Incoming items per year are indexed according to a comprehensive subject classification
scheme developed by the Intelligence Community. The index record for each document
is stored on punch cards and the output to the intelligence analyst in response to a
search query is a bibliography of document titles prepared by partially automated means.
The information reports themselves are recorded on microfilm and mounted in aperture
cards from which reproductions are readily obtained in response to requests.
The Air Force has conducted experiments at SAC, NORAD and the Foreign
Technology Division (FTD) in a very promising indexing system for intelligence
Approved For Release 2004/03/11 : CIA-RDP80601139A000200130012-5
S-E-C-R-E-T
Approved For44,glease 2004/03/11: CIA-RDP80B011360W00200130012-5
S-E-C-R-E-T
- 5 -
documents called "Key Word in Context (KWIC) Indexing." For many years, efforts
have been underway to develop automatic indexing techniques based on key words.
The problem has been to combine key words into meaningful key terms, since an
effective indexing system is most frequently based on terms rather than on single
words. Cee of the most promising approaches now appears to be the KWIC system by
means of which it is possible to retrieve a listing of all appearances of a given word in
Intelligence documents, together with the words appearing immediately before and after
that word in each appearance. Thus, by seeing the word in context one can more
readily find the document he is searching for. Such a system will soon become opera-
tional on a limited scale in the DIA Current Intelligence and Indications Center.
%of NSA has been conducting experiments in capturing titles of reports automatically
and preparing therefrom permuted word indexes. The first results of this effort have
become available recently and appear to be very encouraging. (NSA's applications of
EDP techniques are otherwise in areas not pertinent to this paper.)
The Department of State is currently developing a pilot system in the area of
Cuban affairs based on a computer and designed to produce permuted subject indexes
for the analyst. At present, sentences describing the contents of approximately 5,000
documents of an 88,000 document library on Cuba have been prepared and machine
input is expected to begin shortly.
In CIA a major systems study was initiated in 1962 to investigate the
application of computers to the Agency's central document storage and retrieval
system. The study is now completing an investigation into analyst needs and is
Approved For Release 2004/03/1g_ic6VER1301139A000200130012-5
Approved For Wease 2004/03/VaVr491301139,4e00200130012-5
- 6 -
expected to proceed with detailed system design, and implementation of initial segments
over the next three years. The proposed system would cover all document sources of
significance to the analyst, all categories of information (persons, organizations, places.
things, subjects), and all geographic areas, with minimum handling during processing.
It is expected to incorporate input-output devices to speed data transcription and file
querying, large capacity memories for index data storage, and a large-capacity document
storage system in micro-image form.
Both the Army and the Air Force have sponsored research in automatic
abstracting from Intelligence documents. Under this concept. the ADP equipment
automatically selects and prints significant sentences from a report. Attached as the
Nereovering sheet of a report, the Auto Abstract provides a good summary of the full
contents.
B. Biographic Intelligent", Support
The WALNUT System, now at an advanced stage of development in CIA, is
a large file on foreign personalities of counterintelligence interest. Automated document
storage and retrieval equipment has been installed, a sophisticated name-searching
technique has been designed, and development of very large random access computer
memory equipment to store the entire biographic index is at an advanced stage. A
proposal is under study to develop a name tracing communications network that would
provide quick access to information on foreign personalities stored anywhere in the
Intelligence Community.
Approved For Release 2004/03/U1-EC4 '41)B01139A000200130012-5
Approved For Rejease 2004/03/11 .? CIA-RDP80B01139,001)0200130012-5
S-E-C-R-E-T
-1 -
C. Military Intelligence Support
Targeting, as we know it today, could not exist without computers to store,
collate, and retrieve the detailed data on 110,000 targets world-vide. Computers have
also proved their worth in assisting in the production of radar order-of-battle from
ELINT; in producing air, naval and ground forces order-of-battle; in following current
air and merchant shipping activity; in compiling damage assessment information; in
Intelligence support to war gaming; and in performing various mathematical computations.
Efforts are currently underway to extend ADP to such functional areas as Current
Intelligence and Indications, Collection Assets Inventories, Ports and Harbors, Coasts
and Landing Beaches ? Foreign Railroad and Highway Transportation Systems, and
14101 Photographic Storage and Retireval.
In the late 1950's the Army began to develop a computer system for collating
Incoming fragmentary reports on foreign ground forces orderof-battle. Under this
system, detailed items of information on foreign military units, e.g., names of unit
commanders, unit designations, locations, will be organized and collated to establish
full identification and movement of military units. In 1963, this system was transferred
to DIA and is currently undergoing testing.
D. Overhead Reconnaissance Support
Computer support to overhead reconnaissance has increased significantly
during the past two years in CIA and DIA. Computers are being used in the following
areas: (1) determining what to collect; (2) technical support of operations; (3) information
and technical support of photographic interpretation. The computer-supported system
for infornfrgeerAndelleggetbiht, ei'Metlfgat?
S-E-C-R-E-T
'61 II
? ASS
earlotli pg information
Approved For Rapase 2004/03/11 : CIA-RDP80B01139M00200130012-5
- 8 -
handling field. It brings pertinent data on each target (what's known and what's needed)
to the elbow of the photo interpreter as he exantines new photography. New information
gained from the readout is promptly incorporated in the record used to establish new
target priorities and to support new studies of the given target.
Both the Navy and the Air Force have under development semi-automated
tactical photo interpretation systems. These are designed to provide the photo
interpreter at a console with the neceasary reference material without the attendant
bulk that is usually associated with reference materials. The consoles also perform
certain standard mathematical computations for the photo interpreter. One of these
systems is currently installed and undergoing testing on two of the Naves aircraft
Iftsiarriers.
E. Machine Translation
All three Military Services and CIA have been sponsoring research in
machine translation of foreign language documents. The present level of Ciovermnent
support to machine translation is about $3 million annually. Both general and special
purpose equipment is being used for this purpose.
Three current developments are especially significant. First, CIA plans
to contract for equipment which should both augment translation capabilities and yield
valuable data on an operating machine translation system. Second, the Air Force
program has reached the stage where the special equipment involved will be transferred
from the contractor's facility to an operational environment at PTD in Dayton, Ohio.
N4sl'his equipment will soon permit machine-assisted translation of 100,000 words of
Approved For Release 2004/0/J1 : 9IA-RDP80B01139A000200130012-5
Approved For Release 2004/03/11 : CIA-RDP80601139A000200130012-5
%WI
S-E-C-R-E-T
-? -
material per day. Third, a Joint Advisory Group on Automatic Language Processing
has been formed, with membership from the National Solemn Foundation, Department
of Defense elements, and CIA. This group has agreed to support jointly a central staff
to coordinate Government-supported machine translation research, evaluate its
results, and guide its course in order to accelerate progress in this field.
F. Intelligence Support of Command and Control Systems
A considerable amount of research and development work is currently
underway to provide intelligence in machinable form to the various command and
control systems. The latter include the entire range of facilities from the National
Military Command Center and its Alternates, which will serve National Command
kw/Authorities, all the way down to such tactical systems as the Naval Tactical Data
System (NT
0 .4i
In individual ships. The intelligence provided must be tailored to
the decision-making process of the command served over a wide spectrum of
emergency conditions.
IL CRITICAL AREAS IN FUTURE RESEARCH
Three broad areas of research of critical importance to intelligence are being
considered: (A) conversion of hard copy material into machinable form; (B) development
of massive random access memories; and (C) development of concepts and systems
for non-numerical intelligence data structuring and processing. Each of these is
discussed briefly below:
A. conversion at hard cep material Into rnathInaja form. At present, the bulk
of the intelligence material processed on ADP equipment is first formatted manually
sari
from free text, then key punched onto punch cards, verified, converted to magnet*: tape,
Approved For Release 2004/0ilkiElifIVP,F0B01139A000200130012-5
Approved For Rejease 2004/9514144%80601139711,60200130012-5
- 10 -
and finally entered into the computer. This is a costly procedure in terms of money,
personnel and time.
For an intelligence ADP applioations which entail huge data bases of
historical material or which are characterized by voluminous daily inputs, the coat
of data conversion is so great as to be a major problem. This is especially true of the
processing of data on foreign missile, air, ground, naval, and merchant shipping
activity. It is also true of projects pertaining to foreign transportation systems, ports,
and harbors, coasts and landing beaches and biographic information. It is apparent that
ADP will never be applied to the bulk al the historical or current intelligence material
until the data conversion problem is solved. The solution may lie, in part, in
Notieseareh efforts currently underway to develop an optical scanner which can recognize
lower and upper case letters and digits, and automatically punch the text on cards or
paper tapes. As for the formatting problem, the solution may lie in programming
efforts designed to permit the processing in computers of material which is virtually
In full text.
Development_of_Massive Random Access Memel.% Because cf the limited
Internal memory capacity of oomputers, data bases today are stored on punched cards,
magnetic tape, disc files or drums. None of these are entirely satisfactory. Storage
capacity on cards or magnetic tape is theoretically unlimited, but processing speeds
are relatively slow. Processing speeds on disc files or drums are much faster but
storage capacity is limited.
S-E-C-R-E-T
Approved For Release 2004/03/11 : CIA-RDP80601139A000200130012-5
Approved For Rahlease 2004/03/11 : CIA-RDP80B01139/41$10200130012-5
S-E-C-R-E-T
Due to the enormous volumes of material encountered in intelligence
processing, there is a strong requirement for massive random access memories.
The largest typical internal random amass computer memory today contains about
300,000 words. Disc memories containing three million wards are just becoming
commercially available and even larger ones - up to 15 million words - are in the
development stage. The DOD is currently sponsoring several projects to develop
Increased capacity memories.
In order to attain a real breakthrough in ibis area and achieve, say a multi-
billion bit magnetic memory, great strides in batch fabrication techniques for both
the magnetic elements and the semi-conductor circuitry are necessary. Superconductive
thin film technology offers the possibility of a very large capacity memory because
of its ideal storage and switching characteristics combined with its integrated
miniature batch fabrication possibilities.
(C) Non-numerical intellisence data structuricessi
Three of the most promising non-numerical techniques being pursued at
present are adaptive procedures, self-organizing procedures, and content-
addressing procedures. All of these are dependent in varying degree on the
development of what has come to be known as "associative memory techniques."
Under this concept, the computer selects information on the basis of content,
rather than on the basis of location of information or of indexing. All of memory
can thereby be interrogated at one time. Such content addressable, or associative,
+ieliemories could make searching for information very simple and could either simplify
Approved For Release 2004/0gji.;6442180B01139A000200130012-5
Approved ForWease 2004/03/11: CIA-RDP80B0113011400200130012-5
S-E-C-R-E-T
or make unnecessary such tasks as ordering, merging, sorting, and collating of
information which require so much time in today's serial processing.
Very little has been written either on associative techniques or on the
organization of a data processor which includes an associative memory. At the
present time a bibliography of such papers includes less than 20 authors. In turn, very
little of what has been written has been directly funded by the Government. Most has
been company- or university-sponsored. Current efforts are highly scattered and are
not focused on any definable goal. In particular, there are as yet no theories of machine
organization or of information processing adequate to guide hardware design efforts.
As a result, associative memories are being designed and built with minimum
IlliPteellance on logical design concepts. Within DOD no projects aimed at development of
data processing organization concepts, of techniques, or of applications for associative
memories have been funded (although there are several hardware programs for
construction of small associative memories). Approximately $1 million is available
in FY44 for research in this area. however.
V RECOMMENDATIONS
A. Sher t age of Research Personnel
Intensification of basic and applied research in ADP techniques of intelligenoe
interest will be dependent primarily on increasing the number of first-rate research
workers in universities, laboratories, and private research institutes who are engaged
in tasks such as those mentioned in Section IV above. MatATRairgEWAL.
..Zational Science Foundation its concern over the acute shortage of skilled revarA.
Permunal4MarKAWItu12085/Ar?8%-ffirlisibil4Y--
Approved Forilaplease 2004/N1 aCikW8060113111600200130012-5
of working with thea/rki oundation to determine what remedial actions might be taken.
B. Interagency Mechanisms for Cooperative Effort
Present-day problems in handling information are not unique to the Intelligence
Community. Major investigations are underway in many quarters, particularly in the
scientific community. It is likely that vigorous action by the Federal Council on
Menne and Technology's Conunittee on Scientific Information could and should contribute
In areas of importance to intelligence. The UM should maintain *donate ties with
inter-agencyzeohanismi engaged in research in information handling Vohnktues
makcji...ble to intelligence.
S-E-C-R-E-T
Approved For Release 2004/03/11 : CIA-RDP80601139A000200130012-5