J'OP SEBB • 0 0 S'UBILifl' • 1 OlUTllllON • FOR IVWJ A TDIG lNSPBCTIOJJ 1BD NTROL SYSTDlS 0 0 · ' '- 0 I I Vincent V McRae Pbilip ID c n a 1ow17 0 Joseph O Sanison Jr DECLASSIFIED E O 1a0aa seo 3 e ctober 29 19 8 f •1 I Thi• dOCQIIBnt consists oth2 page• Bo - ot 40 Copies Serie• l· I _ I TABLE OF CONTENrS STABILITY A CJUTERION FCR EVAWATINO INSPECTION AID CONTROL SlSTEMS Page 11 SUMMARY PART I - A COl'EEP CF STABILlTI l i l-1 f THE FROBLEM l DISCUSSION l A M DEL F R S'l'RAT IDIC STABILITY 5 Criteria A s Criteria B 15 P ART II STABILITY 0 PURE MJSSIIE SlSTEMS 21 I Nl'RODUCTION 21 CONDITIONS AID ASSUMPTIONS 23 STABILITY FCR VERY FAST RESPONSE SJ5TE 5 26 STABILITY F lt VERY SUM RESPONSE SISTmfS 30 EFFECT OF FACTCRS OMITTED PR JII THE ANALISIS 3S 1 c qiep 5ECRE'r SUMMARY The purpose of this paper is to present and explain criteria tor evaluating the stability of various strategic postures ot the US and USSR with respect to surprise attack Part I explains the concept upon which the criteria are based defines the criteria and relates them to stability against surprise attacko Part II applies the criteria numeri- cally to pure ICBM systems In Part I a situation is defined to be unstable when either the US or USSR finds it in his national interest to attack the other by surprise attack A rational aggressor wi -attack only when he can inflict at least a specified level of damage on the enemy while limiting his own damage from retaliation to a second specified level which he can toler- ate The stability of any situation can be determined by relating strategic postures to these levels of potential damage In Part II the stability of ICBM systems with various technical characteristics is determined numerically under specific assumptions concerning the vulnerability of population and industry and the magnitudes of the levels of damage postulated in Part I The effects of the response times of the missile systems the dispersion of the missile systems active defenses and passive defenses are discussed Although the substance ot the paper is largely expository certain general conclusions concerning the stability of possible present and future postures of the US and the USSR can be drawn 0 ii TOP SECRET - 9 410 8 9 f J Conclusions lo Damage to cities is the proper measure of stability for strategic systemso 2 The value of agreements for inspection and limitation can only be determined by a detailed examination of the relative damage to US and USSR cities comparing the case when the USSR makes a surprise attack to the case when the US makes a surprise attack 3 o The degree of st bility deteriorates as ICBM technology improves unless this improvement is balanced by a correspondµig improvement in base hardness base configuration populati on shelter and active defenses 4 It appears to us that since less than 100 uccessfully d livered second-generation Soviet missiles could destroy the United States beyond hope of recovery the best stability we could get is a tability based on fear of mutual annihilation in the absence of almost perfect controls or effective anti- ICBM systems iii TOP OOCRm' STABILITY - A alITERION F R EVALUATING INSPFL'TION AND CONl'ROL S S'l'Elf 1 Applications to Pure ICBM Systems PART I - A CONCEPl' OF STABILITY THE PROBLEM This paper is the first in a series aimed at developing the criteria for judging the merits of various proposals for inspection systems as safeguards against surprise attack Only the problem of strategic surprise attack is considered in a mathematical model which permits the gross quantification of the results of surprise am retaliatory attacks Thie model is applied to various probleJllS relating to securing missile forces against surprise attacks by other missile forces However the concepts on which the model is based are equalq applicable to surprise attack by more complex strategic and tactical systems DISCUSSION The work of the Summer Stuey group reported in the Interagency memoramwn led to the qualitative conclusion that ins_pection 'am report- ing procedur es by themselves were or little value in pre- enting surprise attack It was carefully pointed out that such procedures would yield a large amount of information which would or course have great intelligence value but that such information is unlikely to remove incentives for surprise attack which might be possessed by East or Westo In recent staff discussions it has been suggested that the information revealed 1 SECRE'f by an inspection system oould increase the curr- nt instability or the two strategic postures so far as surprise attack is concerned and by others that 1 this information were combined with actions which could be taken unilaterally the situation would become more stable and the probability or surprise attack reduced -In the first case tbe result could be disastrous to either East or West or to both but in the latter case the results would be quite the contrary lt is therefore most important that the concept of stable strategic postures be fully explored and defined am that as tar as possible quantitative aids be developed to evaluate the relative stability ot postures with inspection and without inspection This study is aimed at developing such aids Now just when are the strategic postures of East and West unstable with respect to surpise attack A complete answer to this question is beyom the scope or this discussion but it is clear that if either East or West judges that it is within his total national interest to attack the other then since the benefits of surprise are so great a ' BUrprise attack is likely and the situation is unstable An understanding of the basis on which such judgements might be made 1s very dU'fi cult to acquire since East and West tend to attribute di ferent values to various elements of their national substance Whereas the West considers its moat valuable national substance to be population the East tends to place a much lower value on human lifq The East on the other ham places a very high priority on economic capability industrial production and real estate in relation to that placed on human lives than does the West 2 9rOP SECR 11 r ____ In addition to these asymetries in values assigned to national assets there are also differences in the strengths or desires to increase one's influence in international affairs and to bring other areas of the world umer direct or indirect controlo These political factors as well as macy economic factors must be considered in assessing the strength or 8 J Y' desire to inflict a level or damage upon the other party which is either critical or perhaps lethal It is unfortunate but perhaps true that these politico-economic factors are not very susceptible to nwnerical evaluatiO'no However since such desires must be viewed against a back- groum or capability to inflict damage on one's opponent while not sustaining a critical or perhaps lethal level of damage to one's self and since such capabilities are more susceptible to mathematical treatment it appears appropriate to examine mathematical models of strategic stability We propose to examine a mathematical model which requires for its input estimates of enenor force levels and operational procedures as well as estimates of our own force levels and or course of the operational capabilities of our weapons systems We expect to obtain as outputs from this model gross estimates of the stability of the postures involved The use of such a model should therefore aid in evaluating the extent to which the probability of surprise attack would be reduced if various inspection procedures were adopted We propose that the mod el be extremely comprehensive in scope since it is our opinion that one cannot adequately evaluate the stability of a strategic posture even with respect to central war without looking at a very larg number of TOP SECRIW 9 4 10 8 9 -f weapons systems simultaneously at their interactions and at the levels of damage that might arise if employed using their best tactics Since the inputs required are estimates of force levels warning times response times and the like it is clear that this model will be useful in examining the effects or errors in estimating en81117 capabilities at future time periodso It also is clear that the effect of arms limitations and restrictions could be explored using the same model It is recognized that within the scope of the conference as outlined the USSR might not be willing to enter into technical discussions relative to limitations -It is also recognized that m I Y' of our allies might not be prepared to enter into these discussions However it has been suggested b Y lllaniY that the attempt to arrive at a reasonably secure system of inspections which would in fact severely- reduce the probability of surprise attack might lead to the concl sion that such a system cou1d not exist unless some limitations on arms were imposed Consequently it has been felt that our delegates ought to be aware of the implications of such limitations on our own strategic posture Moreover it has been suggested that if it were possible to present a concept of stable strategic postures which involved such limitations at this point fruitful discussion might ensue on a conceptual basis and a wa y might be opened towards the exploration of controls ani inspection procedures which would be useful in arms limitations As a result it is also one of the Plll POSes of this docwnent to present such a concept and to illustrate it by considering limitations imposed on ICBM forces onlyo 4 9 4 1 Q 8 9 --ftY A MCDEL F E STRATEGIC STABILITY The model for strategic stability is a two-sided gaming model in which either the US or the USSR may initate a surprise attack The aggressor has the option or allocating his strategic striking force in any proportion batween cities and population targets and offensive forces He does so with the knowledge that the attacked country may devote all of its remaining strategic power to the destruction of the aggressor's cities We assllllle that the decision to attack or not to attack is based upon a combined estimate of population and economic damage to be received by both sides and their relation to each other Such a decision must be based upon estimates of outcomes when the Red forces strike first and when the Blue forces strike first We shall es ablish two sets of criteria on which such decisions could be based Criteria A We postulate that a potential aggressor rational sets some level of damage to his cities and population that he is willing to accept but that he is deterred from making such greater damage than the postulated level an attack if he expects We also postulate that there is a level of damage to the enemy• s cities and population which the aggressor sets as being a critical level of damage and one which he would like to exceed if an attack is to be successful We illustrate these levels in Figure l fl'his level is based on the assumption that without clear evidence that he is about to be attacked no nation knowingly attacks an ene which he believes he cannot defeat dec L'sively 5 • TOP SiCRJ l' I q I 1o89--YIJ c TOP SEc M r FIGURE I THE OUTCOMES OF CEN l'RALWAR Criteria I G C tlO Qi B m A Q s r-i al0 ·n 4 •n ·M 0 ' Critical Red Damue A D 0 0 Blue Damage I 6 TOP SECM'f 9 410 8 9-ftf The levels of critical damage to each nation are presented as fractions of national substance which for simplicity_will be referred to in terms of industrial population deaths in this study We have however insisted in our calculations that all attacks aimed at destro7ing population include attacks on cities which will give at least a comparable level or economic damage Consequently our value criteria are based upon both population and economic losses In terms of Figure 1 a potential Red aggressor would assess the outcome ot his aggression as falling in one of the four areas indicated It this outcome tell in area A then the attack would be considered indecisive though tolerable b7 the Red If the outcome tell in region D from capability considerations Red might be motivated to attack since he would inflict critical damage on Blue but would sustain less than critical damage himself It the outcome falls in region B or in region C then the potential Red aggressor would be deterred it rational by considerations of his own damage A potential Red aggressor must also consider the possible results of a first strike or surprise attack by Blue If he assesses the results of such a blow to be 1n areas A or D then be has little tear of a surprise attack by Blue This is certainly the case it he estimates the outcome to be in area Ai and even more the case i f be estimates the outcome of a Blue surprise attack to be in area D In this region he may be confi- dent that Blue will not initiate a surprise attack for here Blue's damage is critical whereas Red's damage is not However i f the outcome 1 'fOP SEJCRF r - 9 41 0 8 9--11 were estimated to be in area c then Blue has strong motivation tf1l' an initial first strike of his own and Red must reconsider and reevaluate the desirability of a first strike by hiaself Estimating that he will sustain more than a critical level of damage if Blue strikes first Red probably increases the dam age tbat he is willing to sustain as a result of a surprise attack of' his own At the very least he is motivated to increase his counterforce capability so that in the future he can eliminate this very real threat to his existence and also his city defenses so that he can absorb a more severe blow We regard this situation as unstable even though inspection agreements am ar agreements on arms control might bave been reachedJ tor Red would be motivated strongly to abrogate these agre8Jllenta and to increaae bis military effort covert' q the connr tuation applies e q to Blue and hence it must be coRclutted that the probability that a surprise attack will be initiated depends not only upon the capability or Red to intlict intolerable damage upon Blue while sustaining less than critiQal damage himself but also very much upon Blue's capability to intlict intolerable damage on Red llhile sustaini less than critical damage to himself The relationship to stabi ity of the differences in outcomes of first strikes gy Red and Blue can be better seen if' we order the outcomes in terms of preference Assumipg that Red say accords preYenting damage to himself a higher priority than inflicting damag• OD Blue but that for equal levels of damage to himself he prefers higher levels of damage on Blue am conversely for Blue's preferences we obtain the ordering in Table lo 8 TOJ' SEC RP r 9 4 1·o8 9 -ii TABLE l Olltcome Preferences Order or Preference Blue Red Outcome Description 1 D C Attacks Motivated 2 A A Indecision 3 B B Deterrence 4 C D Positive Deterrence Red prefers the outcome D most or all because it represents the least damage to himself and at the same time the greatest damage to Blue He pref'ers A over Band C because A represents less than critical damage to himself while B and C represent critical damage to hilaselt He prefers B to C even though the damage to him self is the same because in B Blue sustains greater damage than in Ca By combining sind lar Blue preferences with the Red preferences we are able to order the joint outcomes and hence the strategic postures which lead to these outcomes with respect to stabili t7 a Referring to Table 1 it is obvious that the most stable posture is one in which a surprise attack by Red results in outcome C and a surprise attaclc by Blue results in outcome Do Neither side is motivated to attack the other side because to do so would not only result in an into1-rable level or damage to himself 9 but also would not damage his opponent criticallyo An attack by either side would accomplish nothing but would 9 • 'fOP SEGRHP 9 41 O8 9 --1t7 be very paintul At the other extreme we find that an attack b7 Red would result in a critical level or damage to Blne but the reta111ator strike would not damage Red beyond his tolerable levelJ coJ1Versel 7 a surprise attack by Blue would be critical to Red and the retalliatory strike bJ' Red vould be tolerable b7 Blue Thia pair ot outcome represent extremely unstable postures because either side is motivated to attack not only because he can do so successfully without critical damage to himself while inflicting critical damage to his opponent but also because if he doesn't attack and his opponent does his opponent is successful both in limiting his the opponent's own damage and damaging the potential aggressor criticall - Representing the most stable case b7 C D the most unstable case b7 D C and the intermediate cases in a similar mamer am applying the relative preferences in Table 1 we arrive at the varying degrees or stabilit7 p sented in Table 2 TABLE 2 Degrees or Stability Stability Level I Criteria A outcomes of Surprise Attacks USSR Attacks US US Attacks USSR Deterrence - deterrence C 4 3 II 2 1 III IV -1 v -2 3 VI -4 C B B D B B Deterrence - Indecision C A A D B A YB Indecision - Tndecision A A _ indecision - Attacks Motivated A C 2 D A Deterrence - Attacks Motivated c c D D B C D B Attac s Motivated - Attacks Motivated D C 16 ToP SECRET Now how do these results relate to inspection systems It is clear that it an inspection 7stem is proposed which results in an unstable posture ioiicated in Table 2 b7 Stabilit7 Levels IV V or VI and by stabilit7 indices -1 through -4 then either one side or the other will reject the system it the7 assess it a potentialities correctq The strengths of the objections raised ought ·to be ordered according to the magnitude of the negative indexD In contrast both sides probabl 7 would accept Level III postures if convinced of their realit7 and stabillt7 ov8l tine and it the · strategic capabilities permitted them to deal adequately with third countriesD Restrictions and or merel 1' inspection systems aimed at achieving level III or limiting the outc0lll9a of surprise attacks and retaliator y blows to a Na A are shown to be so unlikeq as to require little considerationo It might be worthwhile to point out that with nuclear weapons with megaton yj elds the critical level or population loss or population and industrial loss combined generate extremel low operational requiremem s Hence if' we consider inspection agreement a which restrict the outcomes to area A we are speaking ot severe limitations in operating procedures and of almost impossible inspection and control procedureao Aa will be shown later in this report even if' limitations are considered$ they involve numbers of ICBMs alone which are r t the order or 1 $-100$1 depending on technological progress and ignoring the effects of manned bomber and submarine attacks o Agree- ments to limit just the number of ICBMs and certainly the total level of strategic forces to this level of' capability would therefoN require ll OP SEGRE' ' 9 4 1 O8 9 --tf1 the destruction or nwnerous forces in being Accordingly restriction to area A will not be considered further in this report therefore o U the vel I and Level n There remains postures which are strategicq stable and which may _possibly- be attained by inspection systems accompanied by unilateral actions or even perhaps by a combination of arms controls and inspection procedu reso Inspection systems assessed as representing Level II stability denot e postures in which neither side has motivation to attacko At stability Level 1 one side could launch an ineffective but non-painful attack 9 while the other side cO lld launch an attack which inflicted critical damage to his enemy but only at the cost of critical damage to himself o At stability Level 2 one side is still able to launch an ineffective but non-painful attack here one side the second is more •deterred11 than at Level lo Note that at Level 2 the side with the ineffective attack could provoke the side with the ineffective but also painful attack But at Level 3 neither side can provoke the other for he fears that the other might react 1rrational ly in which case be will be injured criticallyJ at this level both sides seek assiduously' to avoid accidents and to prevent irrational behavior by the other side At Level 4 we find that if' one side attempts an attack which limits his own level of damage to less than critical in the process of doing so he will fail to inf'lict a critical level of damage upon his enemy Thus he is less motivated to attack than beforeo The opposite side is still able to launch a critical but painful attack but he has less fear of lt BECRi 'r 9 4 10 8 9 --fi ti irTational behavior or of quick response to accifients than before he is or course more able to bluff or provoke than befor but has no greater motivation to launch a surprise attack Since one side is less capable ot launching a surprise attack at Level 4 than he was at Level 3 while the other side has equal capability- but leas tear we regard Level as the more stable oneo As indicated previously Level S is 4 obviously the most stable ot all The la k or symmetry in offense effectiveness and other secondary constraints such as greater ability to provoke ons•s enemy through actions less than but includi •accidental11 single rm clear explosions will make Levels 1 11 2 am 4 ditf'icu lt to attain by inspection or control agreement but the1 are preferable to the unstable postures Level 3 is regarded as a more likely goal certainly tor the early sixties when missile forces will be not too weU protected however Level 5 remains the ultimate goal sot as stability- with respect to surprise attack is concerned This level is so stable in fact that it it were achieved there would be very little fear that strategic forces would ever be used by rational opponents Agreements aimed at limiting the outcomes of strategic thermonuclear exchange to those in area B require first of all that in the face ot a surprise attacki each side have the capability of doing a critical level of damage to the aggressoro This implies that these levels of damage set absolute Minimum force requirements but in fact they also specify the quantity or strategic power which must remain after a surprise 9 4 1 0 8 9 -- f z attack b the enemyo Inspection systems accompaiµ ed by llllilateral actions might in tact change the forces involved in a surprise attack ard or modity timing so that the strength of a surprise attack would be reduced and the required quantities of strategic power would remain after a surprise attack by either East or West It is also clearly possible that without appropriate unilateral actions the inf'ormation produced by inspection systems might reveal wlnerabillties or improve target information to the point where such quantities of strategic power would definitely not be left after an attack These are points which obviwsly must be investigated in great detail and the eftectiveness ot an inspection system or of limitations superimposed on inspection systems in restricting outcomes to area B cannot be assessed without such detailed study in a broad strategic framework Even within Area B there are some asymmetries which might lead to I unstable strategic postureso critical levels It should be recognized at once that the or damage which we have postulated are not very clearly- detined and that a more accurate representation would show these critical levels as bars rather than as lines o The width of the bars would indicate a region of uncertainty or a region in which ·covary was insured but in which the length of time required or such recover would be limited but might not be definitely determinable We might consider the case in which a Red surprise attack would result in Red damage which was only slightly more than critical but would result in Blue damage which was clearly beyond any level from which Blue might recovero In terms of 14 9 410 8 9-y2 2- times required tor recuperation Red might be able to rebuild his economy and become a viable nation say in a period ot S years hand it might take Blue something of the order of 20 to On 25 the other years to do the same thing or in tact he might be so severely damaged that he could never rebuild his country and would always be a second-rate power If the converse situation were also true when Blue attacked first then one would have to consider the str gic postures relatively' unstable Both sides would be motivated to launch a surprise attack in order to prevent his own country from being devastated agreeing to accept a fairly' critical level ot damage but a level at wh lch his recovery would be assured in a reasonably' short period of tiqt According to the criteria already established an attack by either party would be considered irrational since he would be subjecting himself' to a fairly' critical level ot damage However if the international situation became very tense or it he had aey reason to believe that his opponent was irrational and or if he distrusted his opponent's intentions as appears to be the case between East and West today' then such an attack might assume a much greater degree ot rationality · In order to take care ot circumstances such as these we have attempted to consider a secom level of damage which we call lethal level ot damage Criteria B We postulate a second level or damage which we consider to be on the threshold or national destructiono We shall refer to this level as the lethal level in subsequent discussions This threshold is taken 15 ' 'OP SECRfi 'l' 9 410 8 9 -- 7 ' COP SBCRE fl in terms of the nation's ability to recuperate over a very long period ot timei say 15 to 20 or 25 years Damage at the lethal level would not merely be critical but would tend to be catastrophic am would subject the nation to the clear possibility that a now secom-rate power could in tact dominate it completely We suggest that a nation would be absolutely deterred from initiating a surprise attack if the retalia-- tory capability were such that this level of damage would be intlicted with a reasonable probabilityo This level provides some degree of pro- tection against irrationality and against premature retaliatory strikes as a result or misinformation or accidents In either inspection or control systems we suggest that this level ought to provide a limit to the damage that could be intllcted by either side no matter what their intentions might be am believe that at the very least unless such systems assured lethal damage to both sides when lethal damage was inflicted on one sidea agreement to accept these systems could not be reached The combination or criteria A and criteria B requires that the relative levels ot strategic power aimed at an inspection or control agr eement must be sufficiently high to permit ef'tective retaliation 1n the face of surprise attack but not large enough o that covert produc- tion by either side or evasion would not permit a greater than lethal damage to either s ae asymmet rically regardless or intent in a surprise attacko 16 TOP SECRET The combined result Qt the two sets ot criteria is illustrated in Figure 2 Here the shaded area and the two heavy lines represent the region of outcomes resulting from total war initiated by surprise attack in the ear sixtiee The upper' right hand corner or this region repre- sents a situation in wnich neither side pays aey attention at all to eneJI T retaliatory- capability but merely fires away at cities attempting to destroy the nation The lines leading from the lower left-hand corner to the axes on the other hand represent the outcome when both powers employ their strategic forces in an attempt to destroy each otber 1 s retaliatory capability in a surprise attack It should be observed that no actual outcomea would arise along these heavy lines in practical cases because ot the coupling which exists between population and otfens1 Te forces For suppose that Red attacks Blue by surprise and concentrates the bulk or his strike on retaliatoryforces Because of this coupling between offensive forces and tional substance the damage done to Blue cannot be zttro The damage done to Red in the return strike would depend entirely on Blue's tactics If he chose to use all of his retaliatory capability against Red's cities then Red would sustain the critical level of damage On the other ham it Blue's retaliatory strike attempted to eliminate the strategic power which had not been launched in the initial surprise attack then the damage done to both sides would be limited to that involved in the coupling between strategic striking forces and cities In some cases this cQUld be quite high but for the pure missile exchange with ·17 P'IGURE 2 THE OJTCQMm CF TCJrAL WAR Criteria I II I 8 Q 2 1 3 9 0 Blue Damage 0 c TOP $ECBWC 1 • currently stockpiled weapons it appears that this coupling is somewhat less than the critical level of damage even in the absence or fallout shelters but under future technology a strategic posture which would result in outcomes confined essentially to area 4 without arms impossible may be controlo Stability with respect to surprise attack um er the combination of criteria now takes on a somewhat di ferent meaning Not only are there postures in which one side may produce critical damage without sustaining critical damage but also postures in which one may produce -lethal damage without sustaining CE itical or lethal damageo order both Red's and It is quite simple to Blue's preferences in terms of outcome However the number of combinations is much larger in this case tbaQ in Figure 1 For example Red obviously prefers areas 9 3 1 in that ordero Blue prefers 8 6 7 2 4 2 7 6 and 8 4 5 1 3 and 9 in that ordero The possible pairs or outcomes have not been analyzed in the same detail as in the case or criteria A alone However Table 3 is suggestive of an ordering of a number of these outcomes We see that the most stable cases are those in which the attacker regardless of which side he happens to be receives a lethal level or damage The most unstable cases are obviously those in which either side receives a lethal level of damage if he does not attack but inflicts a lethal level or damage on his opponent it he does and at the same time reduces his own damage to less than criticalo It appears that the pairs or outcomes for the two criteria can be ordered with respect to the stability of the postures which they denoteo However not as simply as in the case 19 c 'f'OF S' e CM r or Crit ria A alone In summary in this Part we have related the probability of a surprise attack being initiated to the combined outcomes or surprise attacks by Blue am by Red We have ordered the pairs of outcomes with respect to stability when only criteria A are considered and suggested a partial ordering or the outcomes for the combined criteria More important ¥ we have presented a concept of' a two-sided war game and of a strategic framework within which we believe inspection am or control systems must be evaluated if' the total effect of' such systems on the strategic posture of the United States vis-a-vis that of' Russia is to be determined 'fffll SECRE 1' 9 4 10 8 9- tz PART II - STABILITY OF PURE MISSILE SYSTEMS ItITRODUCTI ON As an application of the concept developed in Part I the stability or postures resulting from pure missile systems alone will be determined numerical ly' for a range of conditions These conditions pertain to the force levels technological eff'iciency of missiles configuration ·'of missile bases and hardness - ot missile bases SpecificallJ' the strategic posture under each combination of these conditions will be equated to one of the pos ures 1 1 2 - 9 shown in Figure 2 of Part I The calculation of the strategic posture will be carried out accord- ing to the following model Defender Attacker anti ssile - attack '- - retaliatory attack ·on cities a - coupling damage ' direct a t t a c k on cities The attacker launche a surprise anti-missile attack upon the defender's missile sites b In conjunction with hi_s attack on the defenderts missile bases the attacker damages the defender's cities If the coupling damage i e 21 'POP HBCRET 9 4 1 O8 9-f 2 1 bonus damage to cities by an attack on missile bases reaches the requisite level critical or lethal both cases are considered no 1 direct attack on cities is made It the coupling damage does not reach the requisit level a direct attack on cities is ma de so as to bring the defender's city damage up toitbe requisite level Co With his remaining missiles the defender launches a retalia- tory attack upon the attacker's cities Whether the retaliatory attack does less than critical damage between critical and lethal damage or greater thlµl lethal damage is determined In order to carry out calculations according to this model it was necessary to select numerical values tor the critical level o damage and the lethal level of damage Obviously specification of these levels must be quite arbitrary at best The critical level of damage was taken to be 20% urban population deaths the lethal level of damage was taken to be 50% urban population deaths fhe rationale tor the selection of the 50% urban deaths as the lethal level was as follows If' the 50% urban deaths was incurred in an attack against cities something over would be destroyed This combined with 50% of industrial floor 50% urban deaths apace and 1 addi- tional urban casualties should sufficiently impair the economy ot the country and disrupt and social organization that neither the USSR or the US could continue to exist as a nation One might question that 50% urban deaths to the USSR when incurred by fall-out coupling in conjunction with an anti-missile attack would constitute a lethal level 22 But in this case 501 of the entire population is killed and ma are injured The 20% urban deaths for the critical level was sele t@d as being rougb4 equivalent to 30% industrial floor space - certainl7 critical level o i damage to both sides but probably not lethal COlIDITIONS -AND ASSUMPl'IONS The calculations were carried out under conditions of symmetry between the two sides so far as force levels tecbnolog l cal etticiencyof missiles configuration ot missile bases and hardness of missile bases were concerned The calculations are also based on conditions ot symmetry so tar as the egree ot shelter afforded the populations is concerned a Force Levels The two sides were g l ven the same numbers of ready- missiles namely 100 200 400 or 800 The term ready missiles is interpreted here to mean missiles which are physically located at the launching sites and could be tired during the initial attacker's blow or the defender's counter blow Duds and non-operational missiles at the launch sites were not specifically considered However they are taken care of implicitly by- considering them to be non-ready according_ to the above detinition b· Missile Base Configuration On both sides all bases were assumed to possess three ready missiles The bases were considered to be sufficiently dispersed so that no more than one base could be knocked out with one bomb This is not a strictly accurate picture of 23 the planned US base structure either on the near term or the long term but is a rough average over bases which will in tact vary from lxl configuration to a lx9 configuration In the absence of information on i the planned USSR base structure it was assumed_to be a mirror image of that of the us Both sides were assumed to have perfect intelligence on locations of their opponent ' s missile bases c 0 Technological Efficiency Yields and CEP•s were selected so as to bracket those which might be expected to occur in practice Yields ot ½ 2 and 10 MT and CEP•s ot i l 2 and 5 nautical miles were used The yields and CEP wen varied independentl - giving 12 combinations Bach of the 12 combinations was given symmetrically- to the two sides d 0 • • fatdness of Missile Bases The hardness of the missile bases was varied from 2 psi taken to be representative of bases presently' under construction b - the US to a degree of hardness such that the target missile must be within the area of the crater and lip ot a surface burst abbreviated CPL Numerical values of 2 10 2§ and 100 psi ' and CPL were used in the calculations The same degees ot hardness were ' attributed to both sides e Vulnerability of Urban Population to Blast Damage Vulnerability of the USSR urban population to blast damage was taken from Urban Blast Weapon Yields and Delivery Accuracies U b7 Norman Hanunian RM-1671 This document gives the traction ot urban population killed by blast in each USSR city ot ove 1001 000 population as a function of yield CEP and number ot weapons A doctrine of targeting is used which 24 _ 9 4 1 Q8 9 -yJV destroys the largest fraction of industrial fioor space A nal agous figures on the VUfD8rability of the US urban population were taken f'rolll Thermonuclear Weapon Effects on United States Industry and Population Concentrations U ORO-R-16 Vol I This document gives number of deaths in each of the principal US cities for one weapon burst at specified und zero as a function of yield and ground zero B7 a graphi'ea l method the effect 0 1 a ·number of bombs with specified CEP•s targeted against specific points may be obtained r Vulnerability of Urban Population to Fallout The vulnera- bility of urban populations to fallout both for the US and the USSR were taken from Simple Formulas for Calculating the Distribution and Effects of Fallout u--u rge Weapon Campaigns with Applications b7 Everett d Pugh Although total J pul ation deaths rather than urban pop tlation deaths were given tor purposes of this paper the urban '- population is considered to receive its pro rata share or the total deaths fallout deaths from two· types of attacks were used direct attacks on cities and attacks on misstle sites In the document curves are given for attacks to max1m e population deaths attacks on airbases and attacks -uniforma distributed over the country Fallout deaths from direct attacks on cities were taken trom the tirst case Fallout deaths from -attacks on missile sites were assumed to be balf'wa7 between •-·· · ' the airbase and uniform attack curves since it is reasonable to assume that future missile sites would be correlated with pppul ationt tenters only to half the degree ot present airbases 25 -39P SECBfiU'- --- 94108 9-fJJ STABILITY FCR TIRY FAST RESPONSE SYSTEM The effects of warning and response time is a critical factor in estimating the success of an attack or a defense In this paper two cases will be considered to provide upper and lower limit in warning and response time At one extreme we will assume that the defender can tire all his missiles before the first attacking missile lands This is a case of perfect warning and perfect response At the other extreme we will assume that the defender can fire no missiles until after the attacker's last missile lands All real eases lie somewhere in between these two ext remes We will deal first with the fast response system Our purpose is to provide estimates of the number of missiles re- quired by the US and the USSR to damage the population and eeonom y of the other These estimates will indicate how few missiles in some cases are needed to annihilate a country and how the Soviet requirements to damage the United States can differ markedly rrom the US requirements to inflict the same damage on the USSR We will use the concepts of critical and lethal damage introduced earlier in this paper to calculate these requirements Finally we will draw some rather obvious con- clusions regarding the effects of these missile requirements on jnapection systems and the possibilities of achieving stability by inspection and by inspection plus limitationso The basic data is given in Table II-1 Here we show the number of weapons required to inflict 20 and 50 percent blast deaths to the population in cities over 100 000 in the US and USSR for the yields and accuracies likely to be achieved in the period from 1960 to 19700 26 TABLE ' II-1 WEAPON REQUIREMENTS CEP inn mi 1 20% us 70 20% USSR 35 500 KT so 85 50 85 240 450 250 26o 290 50% USSR 200 380 56o 1600 50 55 6o 75 20 20 lJ 100 50% us 210 220 250 370 50% ussa 100 llO 210 620 mus 20 20 20 35 20% 10 15 15 5 50% us 90 90 95 1 J 50% 70 85 85 220 us USSR USSR 27 - 75 • 20% USSR 10 MT 5 ' 50% us 20 2 HT 2 -To-F - S b 'Cl 'l '1RE tt' l 'l v- · Two facts stand out when this table is examined the very low number of missiles required to inflict decisive damage on a nation and the asymmetric situation of the US vis-a-vis the USSR in requirements Between 10 and 240 missiles On Soviet cities will kill 20 percent of his industrial population and destroy 30 percent of his industry 20 and 100 missiles on US cities will do the same Between More weapons are re- quired to kill a fraction of the US industrial popuJ ation than to kill the same fraction of Soviet industrial population except for very large yields or very inaccurate missiles This is because US cities are larger in area and less densely populated than Soviet cities To raise the damage from 20 percent to 50 percent requires about four times as missiles on US cities about six times as many Table Il-1 considers only blast deaths o many missiles on Soviet cities If we include fallout deaths the requirements for fast response systems are not reduced as tar as damage to industr y and industrial populations are concerned Fallout deaths in blasted cities are negligible Damage to farms and deaths to farmers cannot be treated in a straightforward manner for a variety of reasons They can be considered purely as a bonus effect for the attacker Fallout deaths on cities resuJ ting from attacks on missile bases w lll be treated only for slow response systems since there is no reason for tast response systems to attack missile bases Two among many reasons for the complexity of the farm damage problem are the relationship between reduced crop production resulting from fallout and fewer mouths to f ed resulting from deaths to city populations on the one hand and the 'callous Soviet attitude t'oward farm populations on the other 28 1'0P SECBET -The conclusions from this brief analysis are simple and direct First parity in· DllJllbers of missiles rarely means parity in damage because US cities present a different target system from that of the USSR Second the number of missiles needed to inflict critical damage on industry and population is less than 100 today and with progress in technology may be no more than 20 by 1970 The implication of the conclusion t6 the question of stability by inspection relates to the Soviet pepchant for parity The imbalances in the concentrations ot population and industry of the US compared to the USSR are such that numerical - and qualitative par ty in missiles practically guarantees that the results of an att·ack will favor one side or the other Parity and stability therefore are not synonymous but antithetical The implications of the second conclusion tothe problem of creating a stable situation by inspection are also simple and direct Both sides have the capability of producing covertly enough missiles to inflict critical damage to the other If 10 megaton missiles with i-mile CEPs are built neither side could ever hope by inspection to preven the other from inflicting critical damage by these missiles if they w re deployed covertly and were consequently invulnerable to attack The stabiiity which is achieved under these conditions therefore is the stability of mutual deterrence or mutual aqµihilatiort until and I The fact that Table III-1 indicates that parit in missi e favors the US should not be taken too seriously The table does no represent the results of a war game but an extreme case to provid a lower bound to the numbers of ICBMs required to infi ct a cert in amount ot damage ' 29 --Pet' SECRE' L' unless an active anti-missile missile defense can be created Without such a defense both sides will live with the certainty that the other could if it wished deploy enough covert force to insure etfective retaliation f both sides create simultaneously an anti-missile missile capability the effectiveness of such covert deployment drops rapidly As the AICBM improves the problem ot covert deployment disappears On the other hand it one side creates even a marginally effective anti--missile missile capability stability itself disappears The side possessing the J CBM can attack with considerable assurance that the covert and hence unknown missile sitee of the other will pose no threat Our general conclusion therefore is that advancing technology in ICBMs-will prevent inspection from achieving more than a margina1 kind of stability and that the introduction of the AICBM by one side can destroy even that margin or stability STABILITY F R VERY SLOW RESPONSE SYSTEMS We now examine the other end of the spectrum - the case when the attacker can fire all of his missiles before the defender can fire back This single-attack model was described in Section n The attacker fires enough missiles at his opponent•s bases to reduce his n damage to 20 percent If he has any missiles lert1 he fires them at cities The defender fires all or his missiles at cities This extreme case should be the most favorable possible for the attJlcker so far as ONDITIOt6 AND ASSUMPTIOlf 30 9 410 8 9-1J response times are concerned Even then as we shall see the attacker sometimes infiicts more damage than he receives The analysis considers both critical 20 percent and lethal 50 percent damage to the industrial plant and or population Both blast and fal_lout were considered as lethal agents Fallout sometimes can cause critical or lethal damage to the nation solely as a by product of attacks on missile bases The model assumes identical numbers of missiles identical characteristics and identical base deployments and vulnerabilities The widelyrvarying outcomes are another demonstration of the dangers of assuming that parity in any of these things make for parity in the outcome of a war or tor stability in peace $tability has been discussed at length in Part I For this section we use the 12 steps between the most stable to the least stable situations to illustrate the effect of varying yi Glds 1 CEPB base hardening and numbers on the stability The definition of stability requires an analysis from the Soviet point of view ot the outcome it he makes a surprise attack on the US as compared with the outcome if the US makes a surprise attack on him The most stable situations are those in which the initiative leads to disaster for both The least stable situations are those in which the initiative on both sides leads to annihiliation of the other without damage to the initiator• The in-between cases include mutual deterrence mutual ineffectiveness and attack without risk and hence without l l OP 8 Prr ET - • fear of retaliation The most stable case we will designate f 6 the most unstable case -5i and O the case where neither side by surprise attack either produces or receives critical damage The results are shown in Table ·n z We have shown the degree of stability for each of three missile yields and four missile stockpiles ranging from 100 to 800 ICBMs for varioµs CEPs and base hardnesses The patterns in these charts are straightforward Soft bases accurate delivery and larger yields make for instability seem much less i Jnportant Numbers The chief effect seems to be that larger numbers make unstable cases more unstable and stable cases more stable with a very slight trend toward increasing stability as the number of weapons increases Base hardening will compensate tor yield to some extent but it cannot compensate for increaseiyield and lower CEPs in combination 10-MT missiles with a ½-mile CEP are unstable for all stockpiles even against bases which can be destroyed only if they lie within the crater-plus-lip of a surface burst weapon The conclusions drawn from this table are not unexpected Techno- logical advances will make stability more difficult rather than less difficult to achieve effect on stability Parity in numbers and in quality has little The asymmetries in the geogr phy and population concentration or the US and USSR make it unlikely that parity in numbers of identical missiles and missile bases would create stable There are a few apparent anomalies for example the 2 MT 5 n m CEP Crater plus lip case which is unstable This results from the asymmetry between US and USSR cities The USSR is incapable or inflicting 20 percent deaths on the US with 100 2 MT weapons so that the entire US attack in both attack and retaliation can be directed at Russian citieso The same thing nearly occurs 500 KT 32 9 410 8 9-tr0 TOP SECRE'J ' • TABLE II--2 srow RESPONSE • I STABILITY FOR ATTACKS 500 KT CEP n mi 2 1 5 l CEP n mi 1 2 5 CEP n mi CEP n mi 2 1 1 2 -4 -4 -4 0 -5 -5 -5 -2 -5 -5 -5 -2 0 4 -4 -1 2 -5 -4 -5 5 -5 -5 l-5 µ 0 0 -4 5 5 -6 2 2 0 -1 J 5 6 µ µ 6 -5 5 t2 5 15 5 -2 -5 6 2 -4 -4 -4 -4 -4 -1 25 -4 -1 00 -1 PL -5 0 400 missiles 200 missiles 100 missiles -5 j 6 800 missiles 2 MT 1 2 CEP n mi CEP n mi CEP n mi 5 _ CEP n mi 1 2 1 2 5 l 1 2 •' -t -L _i -5 _ _1_ _ _ --5 -5 11 6 2 -4 -4 -1 -1 10 -z -I ·-0 -0 -5 -5 -4 l 5 -5 -5 -4 -5 -5 -5 -5 25 -4 -4 -2 2 -5 -5 -5 -5 5 -5 5 6 -5 -5 00 -4 1 2 2 5 -5 5 - 5 -6 µ -6 6 -5 µ I' - µ -5 5 16 -6 5 5 µ -6 µ -6 µ 6 -6 - µ -2 ½ CEP n mi 1 2 5 CEP n mi l 2 5 i 800 missiles 400 missiles 200 missiles 100 missiles lOMT CEP n mi 1 µ -6 CEP n mi 2 5 l 1 2 5 2 -5 I -4 -4 -4 -5 -5 -5 -5 ' -5 -5 -5 -5 -5 -5 -5 -5 10 -5 -4 -4 -4 -5 -5 -5 -5 -5 -5 -5 A -5 -5 -5 -6 25 -5 -4 -4 -5 -5 -5 j 2 -5 -5 -5 -4 6 -5 -5 -3 6 00 -l-5 $ -4 p 5 6 -5 1 6 -6 -5 -6 -5 -6 16 -2 µ -2 -3 -6 -6 6 µ -6 1 6 -6 µ -5 4 100 missiles -2 400 missiles 200 missiles 33 -6 800 missiles 9 410 8 9 --if situations The areas of Soviet cities differ radically from those of US cities The vulnerability of the popul tions as a whqle to fallout is almost precisely in the ratiOi l or the average population density in the two countries Not shown in the table are the excess missiles remaining on both sides for the unstable cases -The excess on the Soviet side would then be available for attacking IR BM bases SAC bases surface-to-air missile defenses and radar installations If we coMider only unstable cases for 400 and 800 missiles with ½-mile CEPs we find approximately one-half of the missiles remain unused for attacks against either ICBM or population targets analysis This is the reason for neglecting reliability in the For reliabilities of 80 percent or better tew unstable cases would become stable The slow response case indicates that inspection even coupled to limitations of ICBMs alone can have little chance of achieving ICBM stability in a world of advancing technology It appears that AICBM and perhaps progressive disarmament keeping pace with advancing tech t i nology are needed Even if both sides could be reduced to no more than 100 missiles which were placed in extremely hardened sitea vulnerable 1 only if within the crater plus lip we will have instability when 10 MT l I L 1° 11 r warheads on ½-mile CEP missiles become available Since the tast response case indicated that 10-20 such missiles cou1d inflict catastrophic I I I damage there is little hope that inspection and ICBM limitation alone would provide stability against strategic attack with thermonuclear weapons 34 -srop SECRET 9 4 1 O8 9 --11 4'0f' 5ECRE r- EFFECT OF FACTCRS OMITTED FROM THE ANALYSIS The model for the analysis has been grossly ove simplified The purpose or this section is to estimate the effect of some of the things omitted from the analysis on the stability or the strategic postures of the USSR vs the us It should be recalled that we are 2 concerned with the outcome of realistic campaigns but oncy with the degree ot stability resulting from the views or each side regarding the ease or difficulty or the other's making a successful surprise attack Five factors will be considered response times intermediate between the two extremes covered in the preceding sections population warning and shelters active defenses and warning systems more dispersed ICBM deliloyments and other strategic attack systems In each case we will discuss briefly the effects when the US and USSR doe not have identical capabilities l Intermediate Response Times It is or course unrealistic to assUDle that the attacker can complete his attack before the defense can retaliate or that the defender can complete his retaliation during the time of flight ot the attacker's missiles To gain some quantitative insight into the effect ot an intermediat case a two-attack model was set up and rough1y analyzed Here the attacker fires part of his capability in the initial wave his remaining missiles are vulnerable to the defender's initial retaliation the defender's remaining missiles are vulnerable to the second attack which to avoid overcomplexity we will ass e completes his attack The defender's terminal capability is tired to complete the exchange 35 For initial attacks of one-half to two-thirds of the attacker's capability- and initial retaliations of one-third to one-half of the defender's capability the attacker's ability to reduce his own damage from the defenderls retaliatory missiles is lower by 5 to 20 percent than for the single-attack model A more realistic analysis of response times therefore tend to increase stability but the increase is not large For accurate larg• 'ieldweapons it is negligibleo Our overall con- clusion is that the single attack model is an adeuate representation of the situation 2 Population Warning and Shelters Deaths to the urban population and damage to industry from both blast and fallout were calculates on the assumption of no warning and no shelter The _number of missiles required to kill by blast 20 or 50 percent of the industrial population in shelters is from four to ten times as many missiles as are required when the people are not in shelters On the other hand it is impossible for an attacker to warn bis lation own population without also warning the en8JD1' popu- It is reasonable thereforej to assume that both the attacker's and defender's populations are· initially without shelter The fallout deaths on the other band can be radically altered by- shelter Since it takes times for fallout to reach areas beyond immediate blast damage it shoul d not be difficult to reduce sharply the fallout deaths in cities There is one case which might radically alter this conclusion if the one side made a surprise attack while a nation-wide air-raid drill was in progresso It might be well for both countries to hold simultaneous air-raid drills if stability is desired An agreement to do this might be on the agenda for the conference 36 't P SEC fler from attacks on missile sites If fall-out deaths by attacks on mis- sile sites are eliminated examination of the pattern of stability shows that there is a very slight increase in stability for large numbers ot large yield weapons Since these cases are stable anyway this factor does not appear to be important Let us now consider the asymmetric case where one power bas protected its population from fall-out more than the other Two facts should be recalled the Soviet urban population is more wlnerable to blast than the US urban population but the Soviet population is much less wlnerable to fall-out than the US population Leaving aside any I consideration of the relative values given to human life by the two governments it is apparent that fall-out shelters in the US would affect i - stability to a greater extent than shelters in the USSR If the US under- took an extensive fall-out shelter program many unstable cases would become stable This is because Soviet attacks on US missile sites would no I longer produce critical or lethal damage in terms of urban deaths The comparable casQ of an attack on Soviet missile sites affects stability to only a minor degree since fall out deaths in cities are low even without shelters We may conclude therefore that hardening the population tends to increase st J ilit70 The tendenc7 is more pronounced it the US provides shelter than it the Soviets do 3o Active De tenses and Warning Syatems In $ection IIIf we indicated that tqe development of AICBM defenses see11 13to be the only way to achieve stability in the tast response case when both sides isTABILITI F R vmY FAST RESPOR3E SISTEMS 31 TOP SECREfr can infiict decisive damage by only 10-ro covert missiles In the slow response case although there are also instances when a very few missiles may make the difference between stability and instability there are many more_situations in which UCBMs may contribute to stability Table ll-2 showa clearly the effect or base hardening and CEP on stability Suppose each ICBM base were given the capability of three ICBM kills on the average Three is also the average number of missiles fired at a missile site for those cases which lie very close to the boundary between stability and instability The ICBM capability doubles the attacker's requirements to keep his own damage from retaliation below the critical level There has been talk of the capability of AICBMs to increase the CEP of missiles by near misses If this capability exists and it is possible to d9uble the CEP of ICBMs from ½-mile to 1-mile the attacker would require 3 to 4 times as many missiles on the enemy's capability in order to be safe from retaliation Even a modest AICBH program might make it possible to achieve stability in the face of very accurate high-yield ICBMs I The assumption has been in the preceeding discussion that both sides have identical AICBM capabilities It has been pointed out howeverJ that geography favors the US over the USSR so tar as early warning is concerned It is doubtful whether a Soviet ground-based ballistic missile early warning system could achieve warning times 38 9 41O89 -' ti under 10 minutes of an ICBM attack The US on the other hand appears to possess an average capability of more than 20 minutes This asym- metric position could place the USSR at a substantial disadvantage in the BMEWS-AICBM race The advent of the reconnaissance satellite however may store the balance from the Soviet point of view One furtner point dese v s mention It may ·become more and more difficult to distinguish between ICBMs and AICBMs as missile technology improveso Arguments such as these demonstrate that the effectiveness of AICBMs in producing stable strategic postures cannot be treated in a simple manner The problem requires far more study before reasonably valid conclusions can be drawn 4 Missile Site Configuration The slow response system assumed a single missile base configuration lx3 This means that a single perfectly delivered ICBM can destroy three ICBMs by surprise attack with slow response Let us now examine the effect on stability of • other types of missile bases One extreme is the lxl configuration In ·t bis·case one perfectly delivered missile can never destroy more than one missile ¥ the number of bases is the same tor both sides this configuration would be the most stable one other things being equ l by making it most dif' ficult for either side to eliminate by surprise attack the other's retaliation on himself Despite the asymmetries in missile requirements to inflict equivalent damage on the US and SSR possession of 1 x 1 bases by- both 39 TOP SECRET 941009 -ttl sides increases stability tor b9th small and large yield missiles The important asymmetry is asymmetry in missile base configuration It Soviet missiles by surprise attack could kill 3 US missiles while US missiles in a urprise attack could only kill one Soviet missile the situation would be rather unstable direct and obvious advantage or an --- Here perhaps is one inspect on system in improving ·--- It would allow both sides to keep ICBM base configuration - - - ·---- ----- --- - -- -·- from being a source of instability• ----p - --- _ _ 5 Ot - Strategic Systems We have examined only a small part stability L _ _ _ ______ -·-· of the problem of strategic stability in our discussion of pure ICBM systems The present analysis will become more relevant to the overall problem as missile forces replace bombers as the principal means of strategic attack and retaliation This will not occur in the US for at least 5 years although by the late 60 1 s it will be true extent that ICBMs can be regarded as the hard core ' and retaliatory capability the stability a nation's attack ot these systems will indicate the stability of the general strategic posture USSR or To the or the US vis-a-vis the It then becomes necessary to determine the extent to which such a stability if it could be achieved could be upset b r the existance ot other strategic systems Manned bombers submarine-launched missiles merchant ships and other surface ships capable of launching cruise of ballistic missiles and IRBMs are all capable of attacking imBM basea With reasonable waming systems and a reasonably large number of bases manned bombers ____ 'rOP SECIU a' ____ 9 41O8 9 1 TOP S RET do not appear to have the capability of making a surprise attack on ICBM sites with a reasonable chance of success particularly if the ICBM bases are deep in_the interior of continents Manned bombers appear to be Ir more effective in attacking a nation's population and economy anti-aircraft and interceptor defenses are weak even a small bomber force could achieve decisive damage with multiple loads of thermonuclear weapons such a situation would upset any established stability with respect to ICBMs only The bomber force itself would require a por- tion of the missile force for its destruction which would not then be available tor the destruction of missile bases or cities Sul marine-launched missiles could be a force tor stability as well as a force against stability In the immediate future the relative invulnerability of the submarine is balanced by the small warhead carried by the POLARIS restricted also because Their part in a counterforce action may be or communication difficulties Overall it does not appear that the submarine-launched missile will play a major role in creating or destroying stability until and unless very large numbers ot very large yield warheads are available IRBMs could have an asymmetric effect on stability They con- tribute directly to the US capability of making a surprise attack on the USSR but they do not contribute substantially to the US retaliatory force because of their exposed position and soft condition The mere existance of IRlJ s in overseas bases may change the timing of USSR surprise attacks particularly if inspection systems are agreed ' FOP SECREI' 'TOP smRET to and may- contribute indirectly to the US retaliatory capability by providing a means of early waming There are many other strategic attack systems which could be considered but little would be gained by further listing of possible developnents during the next few y-ears Most of these systems increase stability when the size of the ICBM force is low Most of them also have little effect when the ICBM stockpile is large with high-yield accurate warheads because the requirements in such missiles to destroythe retaliatory capability and inflict catastrophic damage are so small that additional systems cannot have a significant effect TOP SEC Wr