Outline of a Study for the Plans Analysis Section
The purpose of this paper is to present a rationale for a study to be undertaken by the Plans Analysis Section. A very tentative design for the study was presented in D(L)-921. The present effort is largely an attempt to recast this tentative design into a language which will be more generally understandable to the members of the mixed team which will engage in the study.
The section proposes as a first study—with some additional social science and military aid—to examine the major strategic and logistic variables affecting weapons choice, and to do this in the context of various types of war likely at different times and in the different phases of such wars. This will provide an opportunity, among other things, to define and quantize concept of salvage value and flexibility which depend on the performance required by the successive phases likely in a given kind of war and on the relative likelihoods of various kinds of wars at various times. It is proposed to develop the study of the basic factors affecting weapons choice in terms of one specific extended application—a particular weapon job or set of weapon jobs, such as long-range bombing by missiles vs. aircraft—which will illustrate and test the basic method and hypotheses.
The object of study may be divided into four parts:
- The types of (hot and cold) war situations likely to occur in the fairly near future.
- The broad strategies open to the U.S. in such situations as they might be illustrated in successive campaigns.
- The range and relative importance of tasks indicated for the Air Force by (1) and (2)—by the kinds of conflicts and campaigns likely—and the indicated performance requirements for Air Force weapons systems.
- The implications of the possible Air Force tasks on the construction and evaluation of a weapons system analysis. By way of example, the implications of (3) for efficient allocation of resources as in the choice of "efficient" and "flexible" ways to destroy economic and military targets at long range.
Seven Things This Study Is Not
This study is not intended to yield an account of:
(a) All possible wars with all possible variants, minor and major.
(b) Fractions—exact to five significant digits—measuring the probability of the occurrence of each of the preceding.
(c, d) Ditto for campaigns and their probabilities.
(e, f) Ditto for Air Force tasks generated by each campaign.
(g) The complete set of systems analyses for all alternative Air Force weapons systems covering all possible tasks.
How It Is Limited
This study, while general in its applicability, will be limited in application. The types of wars and campaigns considered will be limited sharply to the major, decisively different alternatives which seem likely to make an important difference in weapons choice. No exact attachment of numerical probabilities to these alternatives is projected or, in fact, regarded as possible. However, while such measurements are not feasible, rough empirical judgments of the likelihood of alternatives are. (They are implicit in all preparations for future action. When we allocate our resources in time and money we judge the relative importance of various tasks. This depends, in the case of warding off damage, on
- how damaging we surmise a given event will be if it should occur; and
- how likely we think the event if to occur.)
Extended time will be devoted to the analysis of the range and relative importance of the assumptions relevant to RAND analyses—and comparatively little effort, except by way of illustration, will be spent on mathematical model-building and the computational elaboration of such analyses themselves. This first study then will perform primarily a service function within RAND, useful precisely because of its generality. Our focus is on the factors affecting choice, not the extended application thereof. Finally, to narrow the problem still further, for the time being we are going to limit our attention, as has been explained, to a particular kind of weapon task. (Which one has not been firmly decided.)
Relation to Basic RAND Research
It is evident that even broad judgments of the likelihood of various kinds of conflicts and of the sequence of their development are hazardous. They involve very grubby and poorly defined economic, military, and political considerations. However, some crude probability judgments on the subject are implicit in any analysis relevant to the choice of weapons. Making them explicit makes it possible to reduce their crudity by using related empirical work going on at RAND and at other agencies. And such work is, of course, going on.
For example, we are interested in the likelihood of war at various times. That war with the Soviet Union is most likely at times when we are most vulnerable is a simple-minded and probably sound judgment on which a good deal of U.S. foreign policy is based. The continuing work of RAND economists on relative military capabilities should help fix these time periods. But the date of outbreak of war which, in accordance with U.S. foreign policy, is up to the Soviet Union to choose depends not only on the actual military capabilities and the objective likelihood of success, but on the Politburo's view of the matter and on its view of U.S. intentions. This is the subject of continuing work in the Social Science Division, and therefore their aid is of critical importance to this study.
Similarly we want to know how these wars might develop. The economists' work on relative war potentials and the dynamics of war economies is relevant here again. So is the work of the military planners one of whose major concerns is the way wars, starting at a given time and place, develop and the various kinds of weapon performance called for.
The fact that the RAND studies mentioned have not been completed does not mean that the sort of analysis we project must await their completion and final result. Such studies probably never are or should be finished. However, the scientists doing the continuing work are in the best position to aid in the selection of fruitful hypotheses as to the major alternatives we face.
Forming explicit and reasoned judgments on such basic empirical alternatives before the ongoing thorough investigations of various parts is carried much further is particularly dirty work. It is however, we feel, essential. We expect such work to occupy most of our time.
The Slow and Dirty Method
If there is any gap in RAND's work so far—and from time to time this possibility is suggested by various loyal though troublesome elements—it seems to us that it is in the frequently professed quick and dirty analyses. These have been too quick and not dirty enough. And there haven't been enough of them. The method of this study therefore will be slow and dirty.
Likelihood of War and Weapons Choice
In order to select one or more desirable weapons from a set of possible ones, we must compare the performance of these weapons under the campaign conditions in which we expect them to be used. Thus the time, the place, the effectiveness of enemy's counter weapons all influence the choice of the desirable weapon. Since we clearly cannot procure a special best-suited weapon for each campaign we may expect to fight, and since a weapon which is best for one of the conceivable campaigns may perform very poorly indeed in some of the others, it becomes necessary to investigate the probability or (to use a weaker and less offensive word) the likelihood of occurrence of each campaign. Thus, if we wish to make constructive comments on the problem of weapons choice, we are faced with the formidable problem of predicting the likelihood of war.
Viewed in its entirety, the problem is indeed formidable. Let us arrange the various approaches to its solution on a decreasing scale of complexity. The most complex, and to the best of our current knowledge, the "correct" approach is to treat the complicated interplay of economic, political, military, and social factors as an n-person (non-zero sum) game. At present we lack the conceptual framework, the mathematical tools as well as the measuring techniques, required even for a bare start on the solution of the complete game.
At the extreme opposite end on the scale of complexity lies the so-called most-probable-course-of-events approach. Here one possible date and class of war is selected (though in general not on the basis of a lengthy and explicit analysis) and then weapons are examined within the campaigns of that war. The particular war is selected either because it is judged to be most likely or, even if the likelihood is low, because the harm done by our failing to prepare for it is very large (that is, the expectancy of gain or loss is very high).
In a gamelike situation, in which the enemy is well informed of our moves, the danger of such "all eggs in one basket" approach is apparent. The danger is even greater in view of our current determent (no preventive war) policy which in effect lets the enemy choose the time and the place of the conflict.
It is important to note that in using the most probable course approach we in no sense avoid the problem of likelihood of war; we in fact assume a solution on the basis of subjective judgment, and we assume a very poor one indeed, since we ignore all other alternatives—the sum of whose likelihoods may be very sizable, perhaps greater than that of the single most likely alternative on which we are concentrating. In any case, it is clear that this procedure invites the enemy to choose one of the less likely alternatives.
Can we do any better in the face of the highly uncertain future? A step up our scale of complexity which appears consistent with our determent policy would be to examine the set of possible wars which we can anticipate in the future and to select weapons which would perform well, if not brilliantly, in each of the anticipated wars. Regardless of what our feelings are with respect to explicit estimate of the likelihood of war, it is very probable that we will immediately reject the above proposal for it prepares equally for all conceivable eventualities. Instead, we will tend by a combination of reasoning and judgment to discard certain alternatives as wholly improbable ("the USSR will not willingly start a major war until it has attained atomic parity"). Among the remaining alternatives, again a combination of reasoning and judgment, we will invariably single out certain alternatives as "important," certain as less important, and certain others as of comparatively little importance. If we are careful, within a well-designated economic, military, and political framework, to secure the judgments of a group of men who are expert in international politics, we shall have the best available although fairly crude set of estimates of the relative likelihood of armed conflict at different times in the fairly near future.
It is proposed to limit the scope of the estimates of two ways:
- The problem is to be treated as one of current and probable future strategies (as determined by constraints such as NATO, U.S. policy of determent, political leadership, etc.) rather than a problem of selection of the best strategies which the U.S. might use under a modified set of constraints.
- The precision of the required estimates will be low. Primary attention will be devoted to orders of magnitude and directions of movement. Thus, for example, instead of requiring a numerical probability of war in 1956, the problem may be to determine the criticality of the period 1954-1957 relative to other such periods in the proximate future; further, the relative criticality is to be measured on a fairly rough scale, say, 0, 1, 2, 3.
It is proposed to construct a framework and a set of assumptions within authoritative judgments on the likelihood of various types of wars can be correlated. A sample framework and set of assumptions for such analysis is indicated in Appendix A, and a sample development of such analysis is given in Appendix B.
These estimates of the likelihood of war next can be applied to the problem of selecting flexible weapons which will perform efficiently in a number of contingent campaigns (rather than in a single, most probable situation). This application is illustrated later in this paper.
Consider the possible broad types of war which the USSR might choose to initiate at any time and the possible broad types of response open to the U.S. during Phase I of the war, the kinds of strategic alternatives available to the opponents might be described in terms of the broad alternatives of action fixed by the scale and type of weapons employment. For each such strategic alternative open to the enemy we can estimate the likelihood of each possible U.S. response. The following chart will serve as a simple illustration of such a set of strategic choices for the year 1956. (See Fig. 1.)
As an example of the sort of considerations that determined the filling of the blanks in this matrix, take the "doubtful" mark in the second column, first row. We say that in the year 1956 when, we have assumed, the S.U. has the critical number or more of A-bombs and carriers to destroy the greater part of the U.S. industrial centers, it is doubtful that the U.S. will use the A-bomb in answer to a hot war opened by the S.U. without an A-bomb. We think it is doubtful because the great destructiveness of the bomb deters other than retaliatory use by the U.S. in the hope that, like poison gas in the last war, it may be used by neither side. By assumption, the A-bomb would give us no distinctive advantage, and we might expect to be strong enough then to win by other devices. In the year 1951 when it might be assumed Russia has less than the critical number of bombs and NATO is hardly of any effect, our response would be likely to include the A-bomb.
A sample application of this kind of analysis to weapons choice say, to SAC, might be pointed out. A glance at the chart of strategic choices will indicate that, in the example given, a class A war with A-bombing is "likely" if the S.U. attacks with A-bombs, and a class B war (no A-bombing) is likely if the S.U. makes a class B move. In the former war obviously a long-range heavy bomber will be required, at least in the early stages, while in the latter war only a light or medium bomber is needed for such things as interdiction bombing. If the Air Force prepares exclusively for a class A war it may be ill-prepared for a class B war, thereby making it more attractive than ever for the enemy to start a class B war. If we are interested either in deterring wars or in fighting them successfully once they start, attention should be given to designing a SAC which will guarantee preparedness for a number of types of possible conflict. (This doesn't say that we shouldn't have a strong intercontinental capability, but it does say that we shouldn't have a strong intercontinental capability, but it does say that this capability must not be built up to the exclusion of everything else.) a development of a given strategic choice into a sequence of campaigns (discussed below) may further indicate that within a single war there are SAC missions both of long-range strategic bombing and interdiction bombing in support of a land campaign; some consideration must therefore be given to the selection of weapons which will be efficient in the performance of several different jobs during a war.
The systematic display of the alternative tasks that a weapons system may be required to perform together with the likelihood of each task provides a basis for a precise description of the erstwhile nebulous military requirement for flexibility of weapons.
The unlikely boxes in the chart can be eliminated from further consideration thus reducing the total number of alternatives which require further examination. We next examine, in turn, each of the "likely" and "doubtful" alternative courses of action and explore the consequences of each strategic choice.
It is very doubtful that a nation could ever formulate strategies in the sense of von Neumann prescribing an appropriate response for every possible move of its enemies. That is to say, its response to every single move is not completely determinate and predictable on the basis of its fundamental interests. However, a partial strategy might be formulated (it appears in military plans under the label "overall strategic concept") which permits sorting the countermoves open to a country at any given time into those that will be clearly unacceptable, those that will be advantageous, and those countermoves that are not clearly one or the other.
Furthermore, as the war progresses each move made under the partial strategy narrows the range of strategic moves acceptable to the nation and represents an accumulating commitment. A given strategic choice is to a considerable extent "hereditary" with respect to its acceptable successors in later phases of the war.
Take as an example an important strategic choice in World War II: The decision to concentrate the major allied effort in the European Theatre. Though there was no completely determinate strategy for World War II, the partial strategy which was employed included this choice. Given the presumption that the forces available were limited enough to require a choice between Europe and the Pacific, the selection of Europe was to be expected on the basis of the critical interests of our principal allies, which were European powers. It was obviously to their interest, and therefore in the interest of our alliance with them, to meet and destroy at the earliest date the forces closest to their own population and industrial centers—namely, the German forces.
The hereditary property of the decisions made in the course of the war is very clear. Having once decided to strike at the supposed "soft underbelly of Europe," Italy, an all-out campaign in the Balkans, for example, was excluded from the next choices.
In general the course of the war can be divided into phases. At the beginning of each phase a selection must be made from a set of mutually exclusive strategic alternatives subject to the constraints indicated above. To be admissible the selected alternative must be feasible (not call for more than the available resources) and it must be consistent with the overall strategic concept. A strategic alternative may consist of a single campaign or a combination of campaigns fought simultaneously. A campaign is defined here by specification of military tasks in terms of the geographic location, the type of armed action, and the scale of weapons employment.
The interplay of enemy and U.S. strategic alternatives at the beginning of each phase may be represented in chart form as was done earlier for the first phase. As before, this procedure eliminates a number of unlikely alternatives and will attach measures of likelihood to the remaining alternatives. (It is of course clear that after the initial strategic choice the initiative will no longer belong exclusively to the S.U.) For each pair of alternatives (enemy and U.S.) we may evaluate one or more possible outcomes of the phase, expressed in broad terms (e.g., "U.S. and S.U. industrial potential virtually destroyed," "Europe taken by S.U.; U.S. holds England," etc.).
By tracing chronologically such possible choices and outcomes we may attempt to delineate all likely (not all conceivable) courses of action. This may be represented in the form of a tree (see Fig. 2). Such a tree may be constructed starting with any of the individual boxes in Fig. 1.
A sequence of strategic choices from the beginning of the war to the end of the last phase represents one possible war. Given the major type of war, and the D-day, each choice at each phase will have associated with it some measure of likelihood of that choice being made. Considering the likelihood of each of the links in a sequence forming each admissible war, we can arrive at a measure of the relative likelihood of the entire sequence beginning on a given D-day.
It is expected that the large number of alternatives available at each phase can be drastically reduced. The elimination of the low likelihood alternatives, the restrictive influence of the overall strategic concept, and the hereditary character of each prior choice have already been mentioned.
Several equally important selective procedures can be mentioned: The campaigns and the strategic choices will be so defined as to give the highest level of aggregation compatible with concrete implications with respect to weapons choice. The study will be oriented toward specific weapon systems and the aggregation determined with respect to these systems. Thus, for example, the class of wars will be restricted to a few major types, as shown on the chart in Fig. 1; the consideration of possible D-days will be limited to some manageable number, (say, every two years, or whenever there is a drastic change in the likelihood of war), the class of campaigns will be limited by considerations of the types of weapons employed in these campaigns.
In summary, a given type of war starting at a particular date generates alternative sequences of strategic choices or campaigns. Each campaign may be described by a series of military tasks and from these tasks the associated weapons requirements can be derived as will be discussed in the next section.
Air Force Employment
Each campaign may be described by a group of military tasks a part of which will be assigned to the Air Force. Thus, the Air Force role in one of the possible wars can be defined by the series of tasks assigned to it from the several campaigns comprising the war. Since the campaigns are defined by geographical location, type of armed action, and scale of weapons employment, multiple duplication of campaigns will occur among the wars considered for a given D-day. Consequently, the total number of Air Force tasks abstracted from the campaigns is expected to be manageable.
The characteristics of each task (such as targets, location, time, enemy strength, etc.) will generate a set of required weapons performance characteristics such as range, amount and type of firepower, maneuverability, etc.).
It can thus be seen that broad strategic considerations introduced at the beginning of this paper can be reduced to a set of Air Force tasks with associated performance characteristics. Further, each task will have attached to it a likelihood of occurrence. If this analysis were to be carried no further, it would provide data which should be quite useful in design of systems analyses. The analyses could either be designed to study new weapons in a number of highly likely situations, or conventional analyses of a most probable (single theatre) situation could still be made. In the latter case, however, the available data should permit improved salvage value estimates through a display of possible subsequent uses.
It is proposed, however, to expand the study beyond this point and to introduce considerations which will provide a new approach to the solution of some conceptual problems encountered in systems analyses. The following discussion illustrates the proposed approach to three such problems: planning over time, salvage value, and alternate uses.
For purposes of this discussion the Air Force can be described as a product mix of combat aircraft together with supporting logistics.
Ideally we would of course like the mix to consist of a single type of aircraft which would perform brilliantly in all conceivable tasks. However, since the performance characteristics compete against one another (we exchange maneuverability for weight, payload for range, etc.), the real Air Force consists of a number of classes of aircraft designated according to their primary function such as fighter, heavy bomber, etc.
The composition of each class varies with time. Thus, for example, in 1951 the heavy bomber class consists exclusively of B-36's. In 1954, under the conditions of no major war since 1951, the class will very likely consist of B-36's supplemented by heavy jet bombers. If no war occurs until, say, 1960 a part of the long-range (heavy bombing) job will be taken over by surface-to-surface missiles. Thus a class of airplanes is a continuously changing mixture of types with differing performance characteristics. If we now wish to examine the effectiveness of a new type, we must phase it in at various times and in various proportions into the currently existing class; and we must correspondingly assume certain phasing-out schedules for the current airplanes. Thus any particular phasing-in-and-out schedule will give a continuously changing class containing the proposed airplane. In the numerical illustrations proposed for this study the actual numerical composition of the class will be made compatible with the monetary appropriations projected under the current war plan and NATO.
The performance of a given class can next be studied in a consecutive series of tasks which constitute a probable war (in the sense defined in an earlier section). Since the initial composition of the class is here given, the performance can be measured by the accumulating damage done as the force remaining after each consecutive task is sent on to the next. If successful, this procedure eliminates the salvage value problem encountered in current systems analyses by evaluating the actual uses in later war phases of the weapons left over from earlier campaigns.
If we next consider the results of all probable wars together with their associated likelihoods, it will be possible to measure the flexibility of a class of weapons (and, consequently, of the new weapon within the class) in terms of the expected damage. If several possible classes of weapons are considered, the most flexible (or the preferred) class may be one which will cause the maximum expected damage. It may also be possible to consider a primitive analogue of additional moments of the probability distribution beyond the mean by examining the range between different levels of damage and introducing risk premium estimates.
In this study it is proposed to develop the preceding concepts in precise form and to demonstrate their applicability by a number of numerical examples.
A Sample Framework for Political Analysis
To a very good approximation the determination of the date of transition from cold to hot war can be reduced from a two-person, non-zero sum game between the U.S. and the S.U. to an optimization process. This holds under the following assumptions:
- The U.S. national policy will continue to exclude preventive war (or an intentional provocation of a preventive war on the part of the U.S.) as a means for attaining national objectives.
- The U.S. national strategy will continue to be one of rapid buildup of the NME by a pre-assigned date. (Thus the counter-deterrent effect of a too rapid rate of buildup is either neglected or taken as a calculated risk by U.S. policy makers.)
- From the point of view of the Politburo the conflict with the West is a struggle for survival. It is the intention of the capitalist democracies to destroy the political systems within the Soviet bloc. Lacking means and know-how for subversive operations, the capitalist democracies plan to force an ultimate showdown by means of a large-scale armed conflict.
- The USSR has complete information on Allied military strength.
Under these assumptions the choice of the first move leading to a large-scale employment of military weapons is left to the USSR. From the point of view of this study it is important to classify the levels of weapons employment involved in any such move. It will probably suffice to distinguish between Class A wars (global conflict), Class B wars (localized armed conflict), and Class C wars (cold war status—military weapons not employed).
It is proposed that the probability of such a move be estimated by, say, two-year periods by examining (from the Soviet point of view) the set of choices available to the USSR at a given time and by comparing the estimated success of these choices with the success which could be attained if similar choices were made at a later time. The relative likelihood of making the choice at timeis then determined by the ordering sign which holds in the relation:
A suggested methodological procedure is as follows: First, design a strategic framework similar or equivalent to the assumptions presented above. Second, within this framework design a set of economic, political, and military criteria by which the choices of the success function, 'si,' are to be made. Third, design and employ a polling technique on a group of prominent political and military scientists to obtain values of the function.
Political Considerations in Armament Planning
1. Peacetime policies of arms procurement and development are obviously sensitive to political considerations. Some of these considerations are relatively unproblematic. There will be, for instance, little quarrel about such generalities as that (a) there is, behind such policies, some conception of which powers may be considered as prospective enemies, prospective allies, and probably neutrals, and (b) that the armaments of each power plus its prospective allies must somehow "match" those of the prospective enemies. Thus, a considerable armament effort undertaken by any one power will call forth a comparable effort in powers which consider themselves to be its potential enemies. The application of these general principles is not entirely unproblematic, since an individual power may overlook a potential threat and neglect matching it in time; it is also possible to make wrong estimates about the magnitude of the matching effort required. Examples of this abound in recent history. To mention only one: the U.S. certainly failed to recognize Soviet power as a potential enemy for some time after World War II and therefore neglected to match Soviet armaments.
2. When the general problem of who the prospective enemies and allies are is well understood, and the magnitude of the necessary matching effort is determined, finer and more problematic political questions emerge which must also be taken into account in procurement and development. These concern the probable political and military decisions the prospective enemy may take, as well as our own probable political and military decisions.
3. The first question of this kind of effective determent. In the present world situation, the U.S. is interested in stopping expansion of the Soviet power. It is generally understood that superior military strength in being on our side is essential to this. But what kind of "strength in being" achieves the best results? And can we expect a "strength in being" that is effectively deterrent today to remain effective as time goes on?
4. Further: a force in being which has the best deterrent effect up to a certain time may not be the best in actual use in war. As long as we expect our strength in being to have a reliable deterrent effect, we may plan for developing a force which has the best deterrent potentialities. But if we look upon the cycle of determent as finite, we also have to consider our strength in being from the point of view of actual military use; and this means that we have to work with two different concepts of "optimum." We may, for instance, plan for maximum determent for the longest possible time, with the proviso that when determent no longer works, our strength in being will be optimal for use. Clearly, such planning depends on some estimate of the probable minimum length of the "determent" cycle.
5. For instance, our aim may be to discourage Soviet expansion in the Middle East or in Europe. Presumably, a strong SAC, deployed close enough to the Soviet Union, is the optimum instrumentality on which we can rely to deter such expansion. Strong anti-aircraft defenses in the continental United States will contribute relatively little to inhibiting Soviet decisions to move only in regions far remote from the U.S. We cannot, however, neglect air defenses, for if determent should cease to work and the Soviets moved somewhere regardless of the possibility of U.S. atomic reprisals, a good defense system may be essential. It will be vital, of course, in case the S.U. decides to strike at the U.S. itself. The same applies to air defenses for the allies of the U.S.
6. Another question, raised in acute form by the Korean conflict, is the following: Supposing that the Soviet union is "deterred" effectively enough from attacking the U.S. or the major European powers directly, will it not use existing potentialities elsewhere for moving communist forces into troubled vacuum areas? The Soviet leadership may estimate that in such a case, the U.S. will not be able or willing to threaten all-out war; hence, the deterrent strength in being will not necessarily influence such decisions.
Now it is clearly essential to procurement and development policies to have some idea about the political reaction of the U.S. government to such an emergency. Two methods of resisting such moves are possible: the U.S. may decide to use total threat to make the Soviets desist, or it may decide to correct the situation by local action. Which decision is taken makes of course an extraordinary difference to the mobilization and armament effort to be undertaken. Heavy bombers and atomic arms are of little use in Korea.
A political question of the type under consideration is: Can we "extrapolate" the Korean experience and prepare for possible new Koreas? Or shall we assume that the next similar aggression will be handled differently—i.e., that the U.S. will then address an ultimatum to the Soviets to withdraw the forces used (either its own, or those of its satellites) or face immediate atomic attack? And if the second, will the Soviets back down or accept the challenge?
7. A parenthetical remark: We are not concerned here with the question of what posture we need in order to win an over-all conflict with the Soviet Union—e.g., whether we can "knock out" the Soviet Union by air attacks alone, or need also sufficient ground forces to defeat it. This is a strategic problem to be decided by the top military leadership. Our problem is narrower; it concerns Soviet political behavior in certain foreseeable situations, and the kind of preparation best calculated to discourage or to meet the moves the Soviet leaders may be expected to contemplate.
8. A strong SAC alone may be effective in deterring aggressive Soviet moves for some time to come, regardless of whether it would also "win the war," for the Soviets may be reluctant to invite an atomic attack even if they were convinced (and rightly so) that they would come out on top in case of such an attack. They may decide, in individual cases, that a local success would not be worth risking a general war,—even a victorious one; and their calculations will point to such conclusions in every case in which they estimate either that they could reach their major objective without an all-out war, or that they could win such a war more easily at some future time. On the other hand, they may reach different conclusions if they estimate that a total attack upon them will be forthcoming in any case, or that winning the war will be more doubtful, costly, and difficult if they wait longer. A third possibility is that the Soviet leadership turns defeatist and concludes that it cannot win in any case. In that case, of course, we could make it accept any terms we might wish to impose, except for the possibility that they will decide upon a desperate last stand, not because they hope to survive, but to exact some toll from the enemy for their inevitable downfall.
9. It may be assumed with a high probability that the Soviet leaders at present are anxious to avoid all-out war with the U.S. The main reason for this is not their overall estimate of the war-making strength on the Western side, but the fact that while they cannot now reciprocate atomic strikes, they hope to acquire a "critical" atomic capacity in not to long a time. Under these circumstances, they can only gain by postponing an all-out conflict.
10. During this interval, the Western powers will presumably build up a more balanced fighting force; while the balance of atomic strength will develop favorably to the Russian side, the balance of ground and tactical strength will change in favor of the West. How is this likely to influence Soviet political decisions? Will it make "war"—that is, Soviet or U.S. decisions to make "irreparable" moves leading inevitably to total conflict—more or less likely?
11. It is very likely that the Soviet leadership will interpret the building up of a "matching" ground and tactical strength as a prelude to aggression. In that case, they may decide to choose an "optimum time" for all-out conflict. The "optimum time" is one that is neither too early nor too late. From the Soviet point of view, starting a conflict before they have critical atomic strength (say, t1) would be too early; starting after the West has critical ground capabilities (say, t2) would be too late.
Now it may be that in the Soviet estimate t1 < t2, or t1 > t2. If the former, the interval between t1 and t2 will be highly critical: at any time during this interval, the S.U. may issue an ultimatum to the West, summoning it to stop its buildup or face all-out atomic war. This pressure may be concentrated upon the European NATO powers. If, on the other hand, t1 > t2, the S.U. is likely to adopt an appeasing attitude during the interval and try to "contain" the West after t1. The calculation upon which such "containment" policies would be based may be analogous to those underlying their present reaction to containment. Even if the West is superior in terms of its chances to win an all-out conflict, it may prefer to avoid the conflict in case the enemy has sufficient atomic strength to hurt it very badly: victory would be too costly to be worthwhile.
12. The period characterized by the highest "likelihood" of war is the interval (t1, t2). One might argue that if there is such an interval, i.e., if t1 < t2, we would be "better off" without building up ground forces in Europe. For our efforts at strengthening Europe can have only one result—namely, that of "provoking Russians into starting a war sometime before our ground forces reach adequate strength. If, on the other hand, we concentrate upon strategic air power, it is true that we cannot hope to "defend" Europe with arms, but we may achieve continued containment; Europe will be "defended" by pure atomic containment plus the avoidance of defensive arming which the Soviets would consider aggressive, and which would "provoke" them into a move we could not stop.
13. Is this reasoning valid? First of all, we do not know whether t1 < t2. But even supposing that this is the case: it is extremely doubtful whether we would actually be better off by responding to the Soviet's acquisition of full atomic strength by the decision not to strengthen European ground and tactical defenses. It seems true to say that a particular constellation of circumstances (full Soviet atomic strength, plus the prospect of stronger European forces in the future) would have an influence upon the timing of the Soviet move against Europe. It may be said, in this sense, that rearmament would "provoke" a Russian ultimatum or attack which otherwise would not occur at that time. The question is, however, whether avoidance of such an attack at the "critical" time in the interval (t1, t2) would ensure peace and European liberty in the long run. In other words: Would the greater atomic and strategic air superiority we would buy at the cost of leaving ground defenses in Europe weak enable us to contain Soviet expansion over the vital areas of the West? It seems certain that it would not. Under the circumstances outlined, containment would become inoperative.
14. The reason for this is the following: "Containment," that is, the threat of strategic air attack in response to a Soviet move, can be effective only if there is "aggression" characterized as such; that is, the Soviets may be "deterred" from moving into an area if there is, in that area, a government which would resist their penetration at least verbally and invoke outside help. If a government simply moves into the Soviet orbit by its own volition, e.g. by concluding first a non-aggression treaty and then gradually deepening its collaboration, the threat of reprisals cannot be operative. This, however, is the inevitable course of events if we hollow out our European alliances by stopping land rearmament. Then, the Soviets need not launch an armed attack against Western Europe to extend their control over it; they can achieve this by political pressure alone, using both diplomacy and military demonstrations and the subversive forces that exist within Western Europe. Supposing that the conclusion of a non-aggression pact between, say, France and the Soviet Union is announced, there is nothing we can do about it, no matter what our atomic capacities are; it would be senseless to address an ultimatum to France to cancel the treaty, or else we would atom-bomb Paris. A weapon which is good for determent purposes is not necessarily an effective instrument in controlling the political behavior of everybody else. If, however, we let disintegrate the Western system of alliances in this way, we shall be isolated and vulnerable to Soviet political pressure, in spite of our atomic capacity in existence.
15. The best political objective for the U.S. seems to be to accelerate Western rearmament as much as possible, so as to prevent the interval (t1, t2) to come into being at all—or to make it very short if it does emerge.
16. If the interval (t1, t2) is safely cleared, a long equilibrium is possible. Neither side may then resort to a "preventive" attack, provided that such a preventive attack is not expected to be so crippling as to prevent any effective counterblow.
17. The preceding analysis seems to point to the necessity of a procurement and development effort that is not concentrated on strategic air offense and defense alone but includes also adequate ground and tactical forces to match Soviet land strength. The main reason for this is that while an atomic force might best deter unequivocal military aggression, ground forces alone seem to be effective in checking political aggression.
18. One question facing armament planning is, then, the following: How much land armament is needed to counteract Soviet political pressure upon our alliance system? In formulating an answer, we have to consider various factors, including not only the probable trend of foreign (S.U. and Western European) attitudes, but also U.S. domestic considerations. Any budgetary proposal we put forward must be politically feasible.
19. Another political question to be taken into account in armament planning has already been mentioned: Do we have to plan for future "Korean-type" wars? No clear-cut answer to this question is possible, for there is one other conflict of the same type already going on (in Indo-China), and it is also uncertain how the Korean conflict itself will develop. It may be assumed, however, that a completely new armed move by communist forces, particularly if it occurs in a sensitive area (e.g. Yugoslavia) is not likely to remain localized. Also, the U.S. people and government seem to have become averse to handling any fresh crisis on a purely local basis. This does not mean to say that the first reaction to any fresh aggression will necessarily be global; however, we have to expect that fresh local conflicts will show a stronger tendency to spread.