a 4
SPECIAL ESTIMATE
IINTTEIXIGENCE IMPUCATIONSENSUS AND VERIFICATION OF ARMED FORCES AND ARMAMENTS
ay/
CENTRAL INTELLIGENCE AGENCY
T0f> SECRET
: /7
a 4
DISSEMINATION NOTICE
This copy of this publication is for the information and use of the recipient designated on the front cover and ofunder the jurisdiction of the recipient's office who require the information for the performance of their official duties.
This copy may be either retained or destroyed byin accordance with applicable security regulations, or returned to the Central Intelligence Agency by arrangement with the Office of Collection and Dissemination, CIA.
DISTRIBUTION:
The President
National Security Council
Intelligence Advisory Committee
SPECIAL ESTIMATE
INTELLIGENCE IMPLICATIONSENSUS AND VERIFICATION OF ARMED FORCES AND ARMAMENTS
Number 4
Published1
The Intelligence organizations of the Departments of State, the Army, the Navy, the Air Force, and the Joint Staffln the preparation of this estimate. All members of the Intelligence Advisory Committee concurred in this estimate onay.
CENTRAL INTELLIGENCE AGENCY
SE-4: INTELLIGENCE IMPLICATIONSENSUS AND
VERIFICATION OF ARMED FORCES AND ARMAMENTS
THE PROBLEM
To analyze, from the Intelligence point of view, the Implicationsensus and verification of armed forces and armaments (including atomic).
ASSUMPTION
* Including para-military, internal security, and police forces.
ontinuing census and verification of all armed forces* and armaments, including atomic weapons, of the US and the USSR is to be undertakenhased basis as the first step of an agreedsystem for control, regulation, and limitation of armed forces and armaments, including atomic energy activities.
CONCLUSIONS
We are convinced that the USSR would enter into any agreement of census and verification in bad faith and would carry it out in bad faith.
Since the USSR has far more information on the US than vice-versa, it Is probableensus and verification could be so devised that in the initial phases the US would secure more valuable information than would the Soviet Union. This would require that the USSR be prevented from securing in those initial phases information Intended to be withheldater phase, or not to be disclosed at all.
Beyond the Initial stages, phasing ofensus andwill become increasingly difficult and there will be agreater danger that the USSR would secure Information of relatively greater value than would the US.
Adoption of the following principles in the phasing of census and verification would aid in protecting US interests:
the earlier phases, only the least sensitivebe released.
aspects of research and development inmanufacturing processes and details of new weaponsof US technological superiority) should be
freedom of movement and access of inspectionIn general be carefully limited to quantitative verificationtypes, sizes,lthough within these limitsbe as full and free as possible. Detailed inspection of
technical specifications, performance data,hould be avoided. Although this limitation on the freedom of Inspection would reduce the amount of Intelligence the us would acquire, it would bein order to guard against exposure of the highly sensitiveexcluded from the census and against premature disclosure in an earlier phase of information reservedater phase.
d. Eath phase, and within it each step In the inspectionshould be carried out simultaneously in the US and USSR.
t Is certain that the USSR would attempt to exploit any system of census and verification to Its advantage and there is grave danger that the USSR might succeed in so exploiting it. ystem based on the principles enumeratedbove would in the initial phases theoretically secure for the US more Information on the USSR than vice-versa, and aid in protecting the technological superiority of the US, we believe that these results would be extremely difficult to achieve in practice. It would be particularly difficult to implement the system in detail Inay as to prevent the USSR from securing In an earlier phase information which was to be withheldater phase, or not disclosed at all. An analysis of these difficulties, as well as the possibilities of surmounting them, will require careful and extensive study by operating as well as intelligence agencies of this government.
discussion
PART L RELATIVE ADVANTAGES AND DISADVANTAGES TO THE US
Relative Superiority of Soviet Intelligence on the US
We believe that in almost every field the USSR has far better intelligence on the US than the US does on the USSR. The intelligence available to the Soviet Union from overt US sources alone is ofvalue. The postwar atomic espionage trials in the USf UK, and Canada, among other things, are ample evidence of the extent of Soviet covert penetration. While the effectiveness of Soviet intelligence penetration may well be decliningesult of growing Western security consciousness, it is only prudent to assume that the USSR will continue to secure relatively superior intelligence on the Western Powers in many fields. Under these circumstances, the USSR might consider that the value of the additional intelligence it could gain through census and verification would be relatively less than that which might accrue to the US.
On the other hand,esult of its superior intelligence on the US, the USSR would be far better able to "pinpoint" critical activities which it wished to inspect without being diverted to less important targets or following false leads. Moreover, because much of its intelligence Is gained overtly or semi-overtly, the Soviet Union would find it less necessary to risk compromising sensitive sources. Owing to the paucity of overt data on the USSR, the US would be far more hampered by fear of compromising such sources.
The relative advantages and disadvantagesystem of census and verification must also be assessed against theof present and potential US intelligence on the Soviet sphere. Largely because of the exceedingly effective and all-inclusive Soviet security system, US Intelligence on many aspects of the
T^JP SECRET
3
Soviet bloc is gravely inadequate and based on fragmentary and scattered evidence. US estimates of Soviet plutonium and atomic weapons production, for example, are probably accurate only to withinercent. Because of the Iron Curtain on the one hand and the large reservoir of Communist informants and easier Soviet access to the US on the other, the US is far less able to collect overt and also covert intelligence than the USSR.
the relative superiority of Soviet intelligence onmight be turned to US advantageensus andIt might be possible toood deal to thematerially adding to Its knowledge. For example,of US and NATO order of battle is probablyas is the USSR's knowledge of plant locations,military installations, etc. Soviet intelligence onof our forces and equipment stockpiles is probably also Therefore, any system which merely allowed the USSRthese details, while assuring the US opportunity ofintelligence, might be relatively advantageous toIf such categories could be included in the initial phases ofand verification, they might serveeans ofintelligence at minimum cost and testing the efficacysystem.
Relative Advantages of Disclosure in Various Categories
the following fields even partial Soviet disclosureand verification might be relatively advantageous to thean intelligence point of view;
a. Size, composition, and equipment of existing forces; size of conventional arms stocks; and location of these forces and stocks. Soviet intelligence in this category probably is relatively complete, while US intelligence is largely based on fragmentary and scattered evidence.
/
location, capacity, and production rates ofmilitary installations and munitions production facilities.field, too, existing and potential Soviet knowledge isand far superior to that of the US, at least in thesphere.
information. In view of Soviet secrecyto all types of economic information, as comparedfull US disclosure, Inclusion of as much as possibledata in any census would be of distinct relative advantage to Knowledge of armed forces and armaments, and even ofplants, is incomplete without knowledge of overallIndustries supporting military end-item production,labor force, raw materials, etc.
aspects of military research andatomic). Even in the research and developmentare certain aspects which it might be relativelythe US to disclose if there were reasonable prospect forSoviet disclosure. From an intelligence point of view,probable greatest needor detailed drawingsto enable it to bridge the gap between aquantity production;or insurance against theof decisive new weapons. With respectain might accrue to the USSR from mere physical surveyand development facilities and prototypes withoutWith respect, however,urvey ofalready known to the USSR might to some extent removeelementew weapon which had reached the
Because of the probable continuation of Its currentlead for the next few years, the US need is not so much for detailed technical data to permit exploitation of Soviet developments as for indications of those areas in which the USSR is giving priority
- 5SECRET
' ^
to development and procurement of new weapons. This requirement could largely be satisfied byurvey of research andfacilities and prototypes, with comparable access to production facilities. In this phase the US might stand to gain relatively more, even considering probable Soviet bad faith.
e. Certain aspects of atomicimilar situationln this field. As long as inspection included only uranium and thorium mining, fissionable material production facilities, and even the current stockpile (excluding new prototypehe US should gain considerably or at least break even from reciprocal inspection.
n the other hand, in the three categories below there would be major risks in US disclosure which might well outweigh the advantages to the US. These categories are doubtless among the highest priority Soviet intelligence targets, as it is in these fields that the existing or potential US lead over the Soviet Union largely counterbalances superior Soviet conventional strength in being.
Production processes and manufacturingn this field the relative advantage to the USSR would rise sharply-asbecame more detailed. So long as plant inspections were brief and general, the US might gain more ln general intelligence, though It would have little to gain in knowledge of production techniques. On the other hand, the USSR would gain substantially from detailed and frequent inspections of US production processes, designs, machinery, etc.
of military research and development (exceptthe USSR's probable greatest need in this field is (orand specifications of research and developmentthe US has no comparable need, disclosure in this categorymajor risks for the US.
energy. If inspection should become detailedatomic research and development, the USSR would gainit has made wholly unexpected progress or Its espionagemuch more successful than believed.
In any census and verification, each side would be bound toertain amount of "collateral" information (data on general conditions in the country, economic conditions, quality of weapons, standards of training, level of maintenance of equipment,n fields not directly covered by any census phase. While the USSR,olice state, could more effectively control the collateral intelligence obtained in the course of inspection and verification, we believe on balance that the US might gain more in this field. Even assuming Soviet bad faith, the paucity of US intelligence on most aspects of the USSR gives ahigh value to almost any collateral information that might be gleaned, while much of the collateral information available to the USSR could be easily secured in any case.
Inevitably, any phased inspection system will leave, as anincreased internal security problems for the US. The Soviet inspection teams will note the places and facilities they are not allowed to visit, and some of these facilities will then become targets for intensive covert Intelligence exploitation. Since Soviet intelligence facilities in the US are extensive, the internal security problems will become more serious through such targeting.
PART IX CENSUS AND VERIFICATION PROCEDURES
satisfactory census and verification procedure shouldcriteria. It should provide for: (a) comprehensive andfor disclosure within agreed phases, thus reducingscope of Soviet bad faith; (b) clear documentation offaith at any stage, thus providing the best possible case(c) prompt termination at any stage in event of suchand (d) protection of the US, insofar as possible, fromof highly sensitive information, including prematurein earlier phases of information reserved for later phases.
Scope of Census and Verification
physical and geographical scope of census andobviouslyreat bearing on the overall advantagesversus West. The USSR has called for Big Four disarmament,
and presumably any census would Include at least Britain and France, if not the other US allies. The state of existing "open" disclosure in other key Western countries appears almost as great as that in the US, and Soviet intelligence penetration, largely through localis in many cases probably substantially greater. Since so much of the research and development, new and unconventional weapons, arms production, and technical know-how of the Western Powers (which are probably the key Soviet intelligence targets) is concentrated in the US, census and verification in other US allies (except the UK) might have even fewer disadvantages arising from premature or unintended disclosure than in the US. The potential disadvantages for the UK might be roughly comparable to those for the US, however, because of extensive UK research and development, arms production, and technicals new and secret US or UK weapons are transferred to other allies and as US "know-how" and production processes are used to stimulate production overseas, the potential disadvantages arising from premature or unintended disclosure by these allies would tend to grow.
lthough presumably the USSR would strongly favor limiting any census and verification to the Big Four powers, we consider itfrom the intelligence point of view, that any system cover the European Satellites and Communist China as well. Inclusion of East Germany and Czechoslovakia would be particularly important from the scientific standpoint, because they contribute substantially to Soviet research and development. If the Satellites and China were omitted, the USSR could conceivably conceal forces, stockpiles, and research and development activities by transferring them there.the Iron Curtain around other Soviet bloc countries, while by no means as impenetrable, is roughly comparable to that around the
USSR.
ecause of the serious risks to the US of disclosure of sensitive information in such fields as research and development, manufacturing processes, technicalnd details of new weapons (the measures of US technologicalhe US should exclude such sensitive information from any census. Protection of such
'P SECRET
sensitive Information would be facilitated If census and verification could be construed as meaningumerical count andin the case of armed forces and armaments; this would avoid detailed Inspection which would reveal qualitative data. Similarly, census and verification of manufacturing facilities, research and development installations, and technical manpower might be limited to quantitative inspection of facilities and records to establish size, production rates, etc. Certain aspects of research and development, including atomic, might be Included in this manner without revealing sensitive data. The groat "difficulty would be ln devising Inspection procedures which would permit quantitative verification withoutqualitative details.
Phasing of Census and Verification
Because of probable Soviet badareful phasing of any census and verification would be Indispensable to minimize the risks to the US of disclosures without adequate compensatory disclosure from the USSR. Any phasing should be so designed as to allow breaking off with minimum risk at any phase In event of demonstrable Soviet bad faith. In view of the propaganda disadvantage to the party which withdraws, the census and verification system should be so constructed as to permit the strongest possible proof of Soviet bad faith.
The USSR clearly has much greater opportunities than the US to conceal data effectively: (a) the US cannot "pinpoint" targets as effectively as the USSR because of the relative paucity of US intelligence on the USSR; (b) the USSR Is much more skilled at large-scaleand evasion; (c) the Soviet Union has great physical capabilities for concealment owing to the vastness and inaccessibility of much of the USSR; and [d) the police state techniques of the USSR and their absence in the US would hamper US inspection and facilitate that of
the USSR. Moreover, the USSR would probably take advantage of these opportunities for concealment while the US presumably must act ln good faith. Therefore, any phasing must be designed to minimize these Soviet advantages by first stressing those categories where the US has the least to lose.
SECRET
From the intelligence point of view, any phasing system should take account of the relative superiority of Soviet intelligence on the US. The implicit concept upon whichystem should be based should be that of seeking initially to put the UScomparable* intelligence footing with the USSR. In other words, the initial phases of the system should be designed to make available to the USSR data which it is already known or believed the USSR has. By this means it might be possible to guard against Soviet bad faith, and to reveal relatively less of significance to the Soviet Union than it would be compelled to reveal to the US. However, since the USSR will obviously seek the same advantages in its proposals, the US proposals would have to appear reasonable to assure the maximum propagandaif the negotiations failed.
We believe that any phasing should start with the general and non-detailed quantitative aspects of conventional armed forces, armaments, and over-all economic potential, with inspection designed to verify numbers, types, and locations. Only after these phases were completed to the satisfaction of the US should the US even consider implementing further phases with access to detailed qualitative data. Phased disclosure in the field of atomic and
other unconventional weapons and in research and development might begin concurrently but there should be excluded from any census and verification the following items in these fields:
the atomic field: details of plants producingmaterials and all research and development work.
the non-atomic field: detailed drawingsof research and development projects.
Types of Census and Verification Procedures
from phasing, the types of census and verificationemployed in each phase would be of vital importancethe risks to the US and compelling maximumby the USSR. It is apparent that these two objectives are
-
TOP SECRET
largely contradictory. The more frequent, thorough, and unrestricted the Inspection the less would be the possibility of Soviet concealment, but the greater would be the risks of sensitive disclosures by the US. This contradiction is further apparent when we examine two basic problems implicit in any verification system. While it is probably feasible to verify such quantitative data as had previously been disclosedensus, it would be difficult, If not impossible, to determine whether all relevant data had actually been disclosed. It would be impossible to determine whether there had been full disclosure unless the inspectors were allowed to travel anywhere to look for anything included in the particular phase. On the other hand, only by careful restrictions on the movement of Inspectors could the USSR be prevented from acquiring ln the earlier phases Information reserved for later phases, or not to be disclosed ln any phase.
Although full Soviet disclosure could only be assured byfree and unrestricted inspection in each phase,rocedure would entail serious risks for the US of prematurely revealing at earlier phases information which the US planned to disclose only in later phases or not at all. Consequently it is doubtful whether in any phase the US should allow Sovietfreedom to travel anywhere and inspect everything. The problem, therefore, is to devise an inspection system which, while less than complete, would still increase the relative gains to the US while minimizing the risks of premature US disclosure.
If inspection were restricted, the US woulderious problem in determining whether the USSR had made full It would be exceedingly difficult to determine whether Soviet data already revealed in the census was either exaggerated or understated. The US would have nothing more than its pre-agreement intelligence to useest of possible Sovietor understatement. In some Instances this intelligence Is shadowy and incomplete; in others it is based on highlysources and is reasonably accurate. Thus, in the first
-
case, pre-agreement intelligence might be considered insufficiently authoritative to warrant challenging Soviet good faith; and in the second case, where the challenge was Justifiable, the US might be unable to do so without compromising its sources.
initial agreement on census and verification should be
as comprehensive and clear as possible, spelling out in great detail each phase and step of the procedure. Moreover, the census itself should call for as complete and detailed answers as possible,exact disclosure of numbers, designations, sizes, locations,o that subsequent verification will be facilitated.
There should be provision for Immediate termination of the census in event of an act of bad faith, and the agreement andshould be framed so as to facilitate the determination and proof of bad faith. At best an accusation of bad faith would be difficult to sustain. In some cases the US could prove its charge of bad faith only by revealing sensitive intelligence.
The first over-all step in any census might be aaerial survey of all cities, towns, and installations served by air, rail, or highway links. The intelligence value of this atone to the US would be extensive, while the gains to the USSR would be minor by comparison. These air photos might then be usedeans of identifying plants andto be included in various phases of the census and
There should be simultaneous census and verification in each phase. For example, the USSR would not be permitted to inspect the first US facility until the US team had arrived at the first Soviet facility and was ready to Inspect. Each nation should be
allowed to set its own priorities for inspection of facilities which have already been revealed in the census, so that the US couldhighest priority targets first. The Soviet team would not be permitted to travel to the second US facility until the US team had
-
satisfactorily completed inspection of the first Soviet facility. Each new phase should begin only after both countries had agreed that the previous phase had been satisfactorily concluded.
Some means, short of unrestricted freedom to travel anywhere and inspect anything, would have to be devised to forestall probable evasion of full disclosure in the census on the part of the Soviet Union. Some means of formal challenge of census figures at any stage of the verification process should also be provided.
Inspection should be limited to quantitative rather than qualitative verification, and the freedom of action of inspection teams should be limited accordingly. Inspectors should be allowed to check visually the numbers, size, and type of forces andand pertinent quantitative records, but not to make detailed inspections of weapons, equipment, or facilities to discover details of their design, construction, composition, etc. Special precautions might be required in some cases to detect dummies.
On the other hand, the fewer the restrictions on the number and frequency of inspections, the size and composition ofteams, the timing and duration of inspections,he more the US has to gain. Whenever possible, Inspections should be conducted by qualified US technical personnel. Other nationals might be less thorough or well-briefed, and perhaps subject to Soviet intimidation.
Verification would yield greater Intelligence value if itinspection of records at the ministry and chieflevels, partly toross-check on plant-by-plant inspections.
Original document.
Comment about this article, ask questions, or add new information about this topic: