Once an engagement starts, though, the difference between
deterrence
and compellence, like the difference between de- fense and offense, may disappear.
Schelling - The Art of Commitment
?
?
?
but not certain
? ? ? ? ? ? ? ? 72 ARMS AND INFLUENCE
nonprovocative. The act that is intrusive, hostile, or provoca- tive is usually the one to be deterred; the deterrent threat only changes the consequences ifthe act in question- the one to be deterred- is then taken. Compellence, in contrast, usually involves initiating an action (or an irrevocable commitment to action) that can cease, or become harmless, only if the opponent responds. The overt act, the first step, is up to the side that makes thecompellentthreat. Todeter,onedigsin,orlaysaminefield, and waits- in the interest of inaction. To compel, one gets up enough momentum (figuratively, but sometimes literally) to maketheotheracttoavoidcollision.
Deterrence tends to be indefinite in its timing. "If you cross the line we shoot in self-defense, or the mines explode. " When? Whenever you cross the line- preferably never, but the timing is up to you. If you cross it, then is when the threat is fulfilled, eitherautomatically,ifwe'veriggeditso,orbyobligationthat immediately becomes due. But we can wait- preferably for- ever; that's our purpose.
Compellence has to be definite: We move, and you must get out of the way. By when? There has to be a deadline, otherwise tomorrow never comes. If the action carries no deadline it is only a posture, or a ceremony with no consequences. If the compellent advance is like Zeno's tortoise that takes infinitely longtoreachtheborderbytraversing,withinfinitepatience,the infinitely small remaining distances that separate him from collision, it creates no inducement to vacate the border. Compellence, to be effective, can't wait forever. Still, it has to wait a little; collision can't be instantaneous. The compellent threat has to be put in motion to be credible, and then the victim must yield. Too little time, and compliance becomes impos- sible; too much time, and compliance becomes unnecessary. Thus compellence involves timing in a way that deterrence typically does not.
In addition to the question of "when," compellence usually involves questions of where, what, and how much. "Do noth- ing" is simple, "Do something" ambiguous. "Stop where you are" is simple; "Go back" leads to "How far? " "Leave me
THE ART OF COMMITMENT 73
alone" is simple; "Cooperate" is inexact and open-ended. A deterrent position- a status quo, in territory or in more figura- tive terms- can often be surveyed and noted; a compellent advance has to be projected as to destination, and the destina- tion can be unclear in intent as well as in momentum and braking power. In a deterrent threat, the objective is often communicated by the very preparations that make the threat credible;thetrip-wireoftendemarcatestheforbiddenterritory. There is usually an inherent connection between what is threat- ened and what it is threatenedabout. Compellent threats tend to communicateonlythegeneraldirectionofcompliance,andare less likely to be self-limiting, less likely to communicate in the very design of the threat just what, or how much, is demanded. ThegarrisoninWestBerlincanhardlybemisunderstoodabout what it is committed to resist; if it ever intruded into East Berlin, though, to induce Soviet or German Democratic Republic forces to give way, there would be no such obvious interpreta- tion of where and how much to give way unless the adventure could be invested with some unmistakable goal or limitation-
a possibility not easily realized.
The Quemoy escapade is again a good example: Chiang's
troops, once on the island, especially if evacuation under fire appeared infeasible, had the static clarity that goes with com- mitmenttoanindefinitestatusquo,whilethecommitmentjust to send troops to defend it (or air and naval support) according to whether a Communist attack there was or was not prelude to anattackonFormosalackedthatpersuasivequality,reminding us that though deterrent threats tend to have the advantages mentioned above they do not always achieve them. (The ambiguous case of Quemoy actually displays the compellent ambiguity, seen in reverse: a "compellent" Communist move against Quemoy was to be accommodated, as long as its extent couldbereliablyprojectedtoaterminusshortofFormosa;ifthe Communiststhought we meant it, it was up to them to design an action that visibly embodied that limitation. ) An American or NATO action to relieve Budapest in 1956- without major engagement but in the hope the Soviets would give way rather
? ? ? ? ? ? ? 74 ARMS AND INFLUENCE
than fight- would have had the dynamic quality of "compellence" in contrast to Berlin: the stopping point would have been a variable, not a constant. Even "Budapest" would have needed a definition, and might have become all of Hun- gary- and after Hungary, what? - if the Soviets initially gave way. The enterprise might have been designed to embody its specificintent,butitwouldhavetakenalotofdesigningbacked up by verbal assurances.
Actually, any coercive threat requires corresponding assur- ances; the object of a threat is to give somebody a choice. To say, "One more step and I shoot," can be a deterrent threat only if accompanied by the implicit assurance, "And if you stop I won't. '' Giving notice of unconditional intent to shoot gives him no choice (unless by behaving as we wish him to behave the opponent puts himself out of range, in which case the effective threat is, "Come closer and my fire will kill you, stay back and it won't''). What was said above about deterrent threats being typically less ambiguous in intent can be restated: the corre- sponding assurances- the ones that, together with the threat- ened response, define the opponent's choice- are clearer than those that can usually be embodied in a compellent action. (Ordinary blackmailers, not just nuclear, find the "assurances" troublesome when their threats are compellent. ) l 8
They are, furthermore, confirmed and demonstrated over time; as long as he stays back, and we don't shoot, we fulfill the assurances and confirm them. The assurances that accompany a compellent action- move back a mile and I won't shoot (other- wise I shall) and I won't then try again for a second mile- are
18. The critical role of "assurances" in completing the structure of a threat, in making the threatened consequences persuasively conditional on behavior so that the victim is offered a choice, shows up in the offers of amnesty, safe passage, or forgiveness that must often be made credible in inducing the surrender of rebels or the capitulation of strikers or protesters. Even libraries and internal revenue agencies depend on parallel offers of forgiveness when they embark on campaigns to coerce the return of books or payment of back taxes. In personal life I have sometimes relied, like King Lea, on the vague threat that my wrath will be aroused (with who knows what awful consequences) if good behavior is not forthcoming, making a tentative impression on one child, only to have the threat utterly nullified by another's pointing out that "Daddy's mad already. "
THE ART OF COMMITMENT 7s
harder to demonstrate in advance, unless it be through a long past record of abiding by one's own verbal assurances.
Because in the West we deal mainly in deterrence, not com- pellence, and deterrent threats tend to convey their assurances implicitly, we often forget that both sides of the choice, the threatened penalty and the proffered avoidance or reward, need to be credible. The need for assurances- not just verbal but fully credible- emerges clearly as part of "deterrence" in discussions of surprise attack and "preemptive war. " An enemy belief that we are about to attack anyway, not after he does but possibly before, merely raises his incentive to do what we wanted to deter and to do it even more quickly. When we do engage in compellence, as in the Cuban crisis or in punitive at- tacks on North Vietnam that are intended to make the North Vietnamese government act affirmatively, the assurances are a critical part of the definition of the compellent threat.
One may deliberately choose to be unclear and to keep the enemy guessing either to keep his defenses less prepared or to enhance his anxiety. But if one wants not to leave him in doubt about what will satisfy us, we have to find credible ways of com- municating, and communicating both what we want and what we do not want. There is a tendency to emphasize the commu- nication of what we shall do if he misbehaves and to give too little emphasis to communicating what behavior will satisfy us. Again, this is natural when deterrence is our business, because the prohibited misbehavior is often approximately defined in the threatened response; but when we must start something that then has to be stopped, as in compellent actions, it is both harder and more important to know our aims and to communi- cate. It is particularly hard because the mere initiation of an en- ergetic coercive campaign, designed for compellence, disturbs the situation, leads to surprises, and provides opportunity and temptation to reexamine our aims and change them in mid course. Deterrence,if wholly successful, can often afford to con- centrate on the initiating events- what happens next if he mis- behaves. Compellence, to be successful, involves an action that must be brought to successful closure. The payoff comes at the end, as does the disaster if the project fails.
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 76 ARMS AND INFLUENCE
THE ART OF COMMITMENT 77
? The compellent action will have a time schedule of its own, and unless it is carefully chosen it may not be reconcilable with the demands that are attached to it. W e cannot usefully threaten to bomb Cuba next Thursday unless the Russians are out by next month, or conduct a six weeks' bombing campaign in North Vietnam and stop it when the Vietcong have been quiescent for six months. There will be limits, probably, to how long the compellent action can be sustained without costing or risking too much, or exhausting itself or the opponent so that he has nothing left to lose. If it cannot induce compliance within that time- and this depends on whether compliance is physi- cally or administratively feasible within that time- it cannot accomplish anything (unless the objective was only an excuse for some act of conquest or punishment). The compellent ac- tion has to be one that can be stopped or reversed when the enemy complies, or else there is no inducement.
surrender statement or acknowledgement of submission, some symbolic knuckling under, will itself achieve the object, verbal compliance may be enough. It is inherent in an intense crisis that the conditions for bringing it to a close have to be of a kind that can be met quickly; that is what we mean by an "intense crisis," one that compresses risk, pain, or cost into a short span of time or that involves actions that cannot be sustained indefi- nitely. If we change our compellent threat from slow pressure to intense, we have to change our demands to make them fit the urgent timing of a crisis.
Notice that to deter continuance of something the opponent is already doing- harassment, overflight, blockade, occupation of some island or territory, electronic disturbance, subversive activity, holding prisoners, or whatever it may be- has some of the character of a compellent threat. This is especially so of the timing, of who has to take the initiative. In the more static case we want him to go on not doing something; in this more dy- namic case we want him to change his behavior. The "when" problem arises in compelling him to stop, and the compellent action may have to be initiated, not held in waiting like the de- terrent threat. The problems of "how much" may not arise if it is some discrete, well defined activity. "At all" may be the obvi- ous answer. For U-2flights or fishing within a twelve-mile limit, that may be the case; for subversive activity or support to in- surgents,"atall"mayitselfbeambiguousbecausetheactivityis complex, ill defined, and hard to observe or attribute.
Blockade, harassment, and "salami tactics" can be interpreted as ways of evading the dangers and difficulties of compel- lence. Blockade in a cold war sets up a tactical "status quo" that is damaging in the long run but momentarily safe for both sides unless the victim tries to run the blockade. President Ken- nedy's overt act of sending the fleet to sea, in "quarantine" of Cuba in October 1962, had some of the quality of deterrent "stage setting"; the Soviet government then had about forty- eight hours to instruct its steamers whether or not to seek collision. Low-levelintrusion,as discussed earlier, can be a way of letting the opponent turn his head and yield a little, or it can be a way of starting a compellent action in low gear, without the
? ? ? If the opponent's compliance necessarily takes time- if it is sustained good behavior, cessation of an activity that he must not resume, evacuation of a place he must not reenter, payment of tribute over an extended period, or someconstructiveactivity that takes time to accomplish- the compellent threat requires some commitment, pledge, or guarantee, or some hostage, or else must be susceptible of being resumed or repeated itself. Particularly in a crisis, a Cuban crisis or a Vietnamese crisis, there is strong incentive to get compliance quickly to limit the risk or damage. Just finding conditions that can be met on the demanding time schedule of a dangerous crisis is not easy. The ultimate demands, the objectives that the compellent threat is really aimed at, may have to be achieved indirectly, by taking pledges or hostages that can be used to coerce compliance after the pressure has been re1ie~ed. IO~f course, if some kind of
? ? ? ? ? 19. Lord Portal's account of the coercive bombing of the villages of recalcitrant Arab tribesmen (after warning to permit evacuation) includes the terms that were
? demanded. Among them were hostages- literal hostages, people -
as well as a fine; otherwise the demand was essentially cessation of the raids or other misbehavior that had brought on the bombing. The hostages were apparently partly to permit subsequent enforcement without repeated bombing, partly to symbolize, together with the fine, the tribe's intent to comply. See Portal, "Air Force Cooperation in
? ? Policing the Empire," pp. 343-58.
? ? ? ? 78 ARMS AND INFLUENCE
THE ART OF COMMITMENT 79
or costly,wecancallita"coercive"or"deterrent"defense. The language is clumsy but the distinction is valid. Resistance that might otherwise seem futile can be worthwhile if, though in- capable of blocking aggression, it can nevertheless threaten to make the cost too high. This is "active" or "dynamic" deter- rence, deterrence in which the threat is communicated by pro- gressive fulfillment. At the other extreme is forcible defense with good prospect of blocking the opponent but little promise of hurting; this would be purely defensive.
Defensive action may even be undertaken with no serious hope of repelling or deterring enemy action but with a view to making a "successful" conquest costly enough to deter repeti- tion by the same opponent or anyone else. This is of course the rationale for reprisals after the fact; they cannot undo the deed but can make the books show a net loss and reduce the incen- tive next time. Defense can sometimes get the same point across, as the Swiss demonstrated in the fifteenth century by the manner in which they lost battles as well as by the way they sometimes won them. "The [Swiss] Confederates were able to reckon their reputation for obstinate and invincible courage as one of the chief causes which gave them political importance. . . . It was no light matter to engage with an enemy who would not retire before any superiority in numbers, who was always ready for a fight, who would neither give nor take quarter. " 2o The Finns demonstrated five hundred years later that the princi- ple still works. The value of local resistance is not measured solely by local success. This idea of what we might call "punitive resistance" could have been part of the rationale for the American commitment of forces in Vietnam. *'
"Compellence" is more like "offense. " Forcible offense is taking something,occupyingaplace,ordisarminganenemyor a territory, by some direct action that the enemy is unable to block. "Compellence" is inducing his withdrawal, or his ac-
20. C. W. C. Oman, The Art of War in the Middle Ages (Ithaca, Cornell University Press, 1953), p. 96.
21. An alternative, but not inconsistent, treatment of some of these distinctions is in Glenn H. Snyder, Deterrence and Defense (Princeton, Princeton University Press, 1961), pp. 5-7, 9-16, 24-40.
conviction that goes with greater momentum but also without the greater risk. Instead of speeding out of control toward our car that blocks his way, risking our inability to see him and get our engines started in time to clear his path, he approaches slowly and nudges fenders, crushing a few lights and cracking some paint. If we yield he can keep it up, if not he can cut his losses. And if he makes it look accidental, or can blame it on an impetuous chauffeur, he may not even lose countenance in the unsuccessful try.
Defense and Deterrence, Offense and Compellence
The observation that deterrent threats are often passive, while compellent threats often have to be active, should not be pressed too far. Sometimes a deterrent threat cannot be made credible in advance, and the threat has to be made lively when the prohibited action is undertaken. This is where defense and deterrence may merge, forcible defense being undertaken in the hope, perhaps with the main purpose, of demonstrating by resistance that the conquest will be costly, even if successful, too costly to be worthwhile. The idea of "graduated deterrence" and much of the argument for a conventional warfare capability in Europe are based on the notion that if passive deterrence initially fails, the more active kind may yet work. If the enemy act to be deterred is a once-for-all action, incapable of with- drawal, rather than progressive over time, any failure of deter- rence is complete and final; there is no second chance. But if the aggressive move takes time, if the adversary did not believe he would meet resistance or did not appreciate how costly it would be, one can still hope to demonstrate that the threat is in force, after he begins. If he expected no opposition, encountering
somemaycausehimtochangehis mind.
There is still a distinction here between forcible defense and
defensive action intended to deter. If the object, and the only hope, is to resist successfully, so that the enemy cannot succeed even if he tries, we can call it pure defense. If the object is to induce him not to proceed, by making his encroachment painful
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ARMS AND INFLUENCE
quiescence, or his collaboration by an action that threatens to hurt, often one that could not forcibly accomplish its aim but that, nevertheless, can hurt enough to induce compliance. The forcible and the coercive are both present in a campaign that could reach its goal against resistance, and would be worth the cost, but whose cost is nevertheless high enough so that one hopes to induce compliance, or to deter resistance, by making evident the intent to proceed. Forcible action, as mentioned in Chapter 1, is limited to what can be accomplished without enemy collaboration; compellent threats can try to induce more affirmative action, including the exercise of authority by an enemy to bring about the desired results.
War itself, then, can have deterrent or compellent intent, just as it can have defensive or offensive aims. A war in which both sides can hurt each other but neither can forcibly accomplish its purpose could be compellent on one side, deterrent on the other.
Once an engagement starts, though, the difference between deterrence and compellence, like the difference between de- fense and offense, may disappear. There can be legal and moral reasons, as well as historical reasons, for recalling the status quo ante; but if territory is in dispute, the strategies for taking it, holding it, or recovering it may not much differ as between the sidethatoriginallypossesseditandthesidethatcovetedit,once the situation has become fluid. (In a local tactical sense, American forces were often on the "defensive" in North Korea and on the "offensive" in South Korea. ) The coercive aspect of warfare may be equally compellent on both sides, the only difference perhaps being that the demands of the defender, the one who originally possessed what is in dispute, may be clearly defined by the original boundaries, whereas the aggressor's demands may have no such obvious definition.
The Cuban crisis is a good illustration of the fluidity that sets in once passive deterrence has failed. The United States made verbal threats against the installation of weapons in Cuba but apparently some part of the threat was unclear or lacked credibility and it was transgressed. The threat lacked the autom- aticity that would make it fully credible, and without some
THE ART OF COMMITMENT 81
80
automaticity it may not be clear to either side just where the threshold is. Nor was it physically easy to begin moderate resistance after the Russians had crossed the line, and to increase the resistance progressively to show that the United States meant it. By the time the President determined to resist, he was no longer in a deterrent position and had to embark on the more complicated business of compellence. The Russian missiles could sit waiting, and so could Cuban defense forces; the next overt act was up to the President. The problem was to prove to the Russians that a potentially dangerous action was forthcoming, without any confidence that verbal threats would be persuasive and without any desire to initiate some irrevers- ible process just to prove, to everybody's grief, that the United States meant what it said.
The problem was to find some action that would communi- cate the threat, an action that would promise damage if the Russians did not comply but minimum damage if they com- plied quickly enough, and an action that involved enough momentum or commitment to put the next move clearly up to the Russians. Any overt act against a well-defended island would be abrupt and dramatic; various alternatives were appar- ently considered, and in the end an action was devised that had many of the virtues of static deterrence. A blockade was thrown around the island, a blockade that by itself could not make the missiles go away. The blockade did, however, threaten aminor military confrontation with major diplomatic stakes- an en- counter between American naval vessels and Soviet merchant ships bound for Cuba. Once in place, the Navy was in a position to wait; it was up to the Russians to decide whether to continue. If Soviet ships had been beyondrecall, the blockade would have been a preparation for inevitable engagement; with modem communications the ships were not beyond recall, and the Russians were given the last clear chance to turn aside. Physi- cally the Navy could have avoided an encounter; diplomati- cally, the declaration of quarantine and the dispatch of the Navy meant that American evasion of the encounter was virtually out of the question. For theRussians, the diplomatic cost of turning
? ? ? ? ? ? ? ? 82 ARMS AND INFLUENCE
THE ART OF COMMITMENT 83
threats of more violent action, had no effect. 23If the North Vietnamese dramatically issue a call to the Vietcong to cease activity and to evacuate South Vietnam, it is a conspicuous act of submission. If the Americans had evacuated Guantanamo when Castro turned off the water, it would have been a con- spicuous act of submission. If an earthquake or change in the weather had caused the water supply to dry up at Guantanamo, and if the Americans had found it wholly uneconomical to supply the base by tanker, they might have quit the place without seeming to submit to Castro's cleverness or seeming afraid to take reprisals against their ungracious host. Similarly, the mere act of bombing North Vietnam changed the status of any steps that the North Vietnamese might take to comply with American wishes. It can increase their desire, if the tactic is successful, to reduce support for the Vietcong; but it also increases the cost of doing so. Secretary Dulles used to say that while we had no vital interest in Quemoy we could not afford to evacuate under duress; intensified Chinese pressure always led to intensified determination to resist it. "
If the object is actually to impose humiliation, to force a showdown and to get an acknowledgement of submission, then the "challenge" that is often embodied in an active compellent threat is something to be exploited. President Kennedy un- doubtedly wanted some conspicuous compliance by the Soviet Union during the Cuban missile crisis, if only to make clear to the Russians themselves that there were risks in testing how much the American government would absorb such ventures. In Vietnam the problem appeared the opposite; what was most
freighters around, or even letting one be examined, proved not to be prohibitive.
Thus an initial deterrent threat failed, acompellent threat was called for, and by good fortune one could be found that had some of the static qualities of a deterrent threat. 22
There is another characteristic of compellent threats, arising in the need for affirmative action, that often distinguishes them from deterrent threats. It is that the very act of compliance- of doing what is demanded- is more conspicuously compliant, more recognizable as submission under duress, than when an act is merely withheld in the face of a deterrent threat. Compli- anceislikelytobelesscasual,lesscapableofbeingrationalized as something that one was going to do anyhow. The Chinese did not need to acknowledge that they shied away from Quemoy or Formosa because of American threats, and the Russians need not have agreed that it was NATO that deterred them from conquering Western Europe, and no one can be sure. Indeed, if a deterrent threat is created before the proscribed act is even contemplated, there need never be an explicit decision not to transgress, just an absence of any temptation to do the thing prohibited. The Chinese still say they will take Quemoy in their own good time; and the Russians go on saying that their intentions against Western Europe were never aggressive.
The Russians cannot, though, claim that they were on the point of removing their missiles from Cuba anyway, and that the President's television broadcast, the naval quarantine and
22. Arnold Horelick agrees with this description. "As an initial response the quarantine was considerably less than a direct application of violence, but considerably more than a mere protest or verbal threat. The U. S. Navy placed itself physically between Cuba and Soviet ships bound for Cuban ports. Technically, it might still have been necessary for the United States to fire the first shot had Khrushchev chosen to defy the quarantine, though other means of preventing Soviet penetration might have been employed. But once the quarantine was effectively established- which was done with great speed- it was Khrushchev who had to make the next key decision: whether or not to risk releasing the trip-wire. '' "The Cuban Missile Crisis," World Politics, 16 (1964), 385. This article and the Adelphi Paper of Albert and Roberta Wohlstetter mentioned in an earlier note are the best strategic evaluations of the Cuban affair that I have discovered.
23. The tendency for affirmative action to appear compliant is vividly illustrated
? ? by the widespread suspicion -
one that could not he effectively dispelled- that the U. S. missiles removed from Turkey in the wake of the Cuban crisis were
? part of a bargain, tacit if not explicit.
24. Almost everyone in America, surely including the President and the Secretary
of State, would have been relieved in the late 1950s if an earthquake or volcanic action had caused Quemoy to sink slowly beneath the surface of the sea. Evacuation would then not have been retreat, and an unsought commitment that had proved peculiarly susceptible to Communist China's manipulation would have been disposed of. Such is the intrinsic value of some territories that have to he defended!
? ? ? ? ? ? ? ? ? Skill is required to devise a compellent action that does not have this self-defeating quality. There is an argument here for sometimes not being too explicit or too open about precisely what is demanded, if the demands can be communicated more privately and noncommittally. President Johnson was widely criticized in the press, shortly after the bombing attacks began in early 1965,for not having made his objectives entirely clear. How could the North Vietnamese comply if they did not know exactly what was wanted? Whatever the reason for the Ameri- can Administration's being somewhat inexplicit- whether it chose to be inexplicit, did not know how to be explicit, or in fact
ARMS AND INFLUENCE
THE ART OF COMMITMENT 85
we can suppose that the United States government did not know in detailjust how much control or influence the North Vietnam- eseregimehadovertheVietcong;andwecanevensupposethat the North Vietnamese regime itself might not have been alto- gether sure how much influence it would have in commanding a withdrawal or in sabotaging the movement that had received its moral and material support. The United States government-may not have been altogether clear on which kinds of North Viet- namese help- logistical help, training facilities, sanctuary for the wounded, sanctuary for intelligence and planning activities, communications relay facilities, technical assistance, advisors and combat leaders in the field, political and doctrinal assis- tance, propaganda, moral support or anything else- were most effective and essential, or most able to be withdrawn on short notice with decisive effects. And possibly the North Vietnam- ese did not know. The American government may have been in the position of demanding results not specific actions, leav- ing it to the North Vietnamese through overt acts, or merely through reduced support and enthusiasm, to weaken the Viet- cong or to let it lose strength. Not enough is known publicly to permitustojudgethisVietnameseinstance;butitpointsupthe important possibility that a compellent threat may have to be focused on results rather than contributory deeds, like the fa- ther's demand that his son's school grades be improved, or the extortionist's demand, "Get me money. I don't care how you get it,just get it. " A difficulty, of course, is that results are more a matter of interpretations than deeds usually are. Whenever a recipient of foreign aid, for example, is told that it must elimi- nate domestic corruption, improve its balance of payments, or raise the quality of its civil service, the results tend to be uncer- tain, protracted, and hard to attribute. The country may try to comply and fail; with luck it may succeed without trying; it may have indifferent success that is hard to judge; in any case com- pliance is usually arguable and often visible only in retrospect.
Even more than deterrence, compellence requires that we recognize the difference between an individual and a govern- ment. To coerce an individual it may be enough to persuade
? 84
urgently desired was to reduce the support for the Vietcong from the North, and any tendency for the compellent pressure of bombing to produce a corresponding resistance would have been deprecated. But it cannot always be avoided, and if it cannot, the compellent threat defeats itself.
? was explicit but only privately- an important possibility is that vague demands, though hard to understand, can be less embar- rassing to comply with. If the President had to be so explicit that any Europeanjournalist knew exactly what he demanded, and if the demands were concrete enough to make compliance recog- nizable when it occurred, any compliance by the North Viet- namese regime would necessarily have been fully public, perhaps quite embarrassingly so. The action could not be hidden nor the motive so well disguised as if the demands were more privately communicated or left to inference by the North Vietnamese.
Another serious possibility is suggested by the North Viet- namese case: that the initiator of a compellent campaign is not himself altogether sure of what action he wants, or how the result that he wants can be brought about. In the Cuban missile case it was perfectly clear what the United States government wanted, clear that the Soviets had the ability to comply, fairly clear how quickly it could be done, and reasonably clear how compliance might be monitored and verified, though in the end there might be some dispute about whether the Russians had left behind things they were supposed to remove. In the Vietnamese case,
? ? ? ? 86 ARMS AND INFLUENCE
THE ART OF COMMITMENT 87
arises whether they ought to be connected at all. If the object is to harass, to blockade, to scare or to inflict pain or damage until an adversary complies, why cannot the connection be made verbally? If the Russians want Pan-American Airways to stop using the air corridor to Berlin, why can they not harass the airline on its Pacific routes, announcing that harassment will continue until the airline stops flying to Berlin? When the Russians put missiles in Cuba, why cannot the President quar- antineVladivostok,stoppingSovietshipsoutside,say,atwelve- mile limit, or perhaps denying them access to the Suez or Panama Canal? And if the Russians had wanted to counter the President's quarantine of Cuba, why could they not blockade
A hasty answer may be that it just is not done, or is not "justified," as though connectedness implied justice, or as though justice were required for effectiveness. Surely that is part of the answer; there is a legalistic or diplomatic, perhaps a casuistic, propensity to keep things connected, to keep the threat and the demand in the same currency, to do what seems reasonable. But why be reasonable, if results are what one
25. It has often been said that American tactical superiority and ease of access in the Caribbean (coupled with superiority in strategic weaponry) account for the success in inducing evacuation of the Soviet missiles. Surely that was crucial; but equally significantwas the universal tendency- a psychological phenomenon, a tradition or convention shared by Russians and Americans- to define the conflict in Caribbean terms, not as a contest, say, in the blockade of each other's island allies, not as a counterpart of their position in Berlin, not as a war of harassment against strategic weapons outside national borders. The countermeasures and counterpressuresavailable to the Russians might have looked very different tothe "Russian" side if this bad been a game on an abstract board rather than an event in historical time in a particular part of the real world. The Russians tried (as did some unhelpful Americans) to find a connection between Soviet missiles in Cuba and American missiles in Turkey, but the connection was evidently not persuasive enough for the Russians to be confident that, if the dispute led to military action or pressure against Turkey, that definition would hold and things would go no further. The Caribbean definitionhad more coherence or integrity than a Cuban- Turkish definition would have had, or, in terms of reciprocal blockade, a Cuban-United Kingdom definition would have had. The risk of further metastasis must have inhibited any urge to let the crisis break out of its original Caribbean definition.
him to change his mind; to coerce a government it may not be necessary,butitalsomaynotbesufficient,tocauseindividuals to change their minds. What may be required is some change in the complexion of the government itself, in the authority, pres- tige, or bargaining power of particular individuals or factions or parties, some shift in executive or legislative leadership. The Japanese surrender of 1945was marked as much by changes in the structure of authority and influence within the government as by changes in attitude on the part of individuals. The victims of coercion, or the individuals most sensitive to coercive threats, may not be directly in authority; or they may be hopelessly
committed to non-compliant policies. They may have to bring bureaucratic skill or political pressure to bear on individuals who do exercise authority, or go through processes that shift authority or blame to others. In the extreme case governing authorities may be wholly unsusceptible to coercion- may, as a party or as individuals, have everything to lose and little to save by yielding to coercive threats- and actual revolt may be essential to the process of compliance, or sabotage or assassina- tion. Hitler was uncoercible; some of his generals were not, but they lacked organization and skill and failed in their plot. For
working out the incentive structure of a threat, its communica- tion requirements and its mechanism, analogies with individu- als are helpful; but they are counterproductive if they make us forget that a government does not reach a decision in the same way as an individual in a government. Collective decision depends on the internal politics and bureaucracy of govern- ment, on the chain of command and on lines of communication, on party structures and pressure groups, as well as on individual values and careers. This affects the speed of decision, too.
"Connectedness" in Compellent Threats
As mentioned earlier, a deterrent threat usually enjoys some connectedness between the proscribed action and the threatened response. The connection is sometimes a physical one, as when ~OOpasreputinBerlintodefendBerlin. Compellentactionsoften have a less well-defined connectedness; and the question
? ? ? ? ? ? ? ? ? ? ? ? ? 88 ARMS AND INFLUENCE
THE ART OF COMMITMENT 89
ut actions threatened against potential provocation-often need the credibility that connectedness can give them.
Connectedness in fact provides something of a scheme for classifyingcompellentthreatsandactions. Theidealcompellent action would be one that, once initiated, causes minimal harm if compliance is forthcoming and great harm if compliance is not forthcoming, is consistent with the time schedule of feasible compliance, is beyond recall once initiated, and cannot be stopped by the party that started it but automatically stops upon compliance,withallthisfullyunderstoodbytheadversary. Only he can avert the consequences; he can do it only by complying; andcomplianceautomaticallyprecludesthem. Hisisthenthe "last clear chance" to avert the harm or catastrophe; and it would not even matter which of the two most feared the consequences as long as the adversary knew that only he, by complying, could avert them. (Of course, whatever is demanded of him must be less unattractive to him than the threatened consequences, and the manner of threatened compliance must not entail costs in prestige, reputation, or self-respect that outweigh the threat. )
It is hard to find significant international events that have this perfectionist quality. There are situations, among cars on high- ways or in bureaucratic bargaining or domestic politics, where one comes across such ideal compellent threats; but they usually involve physical constraints or legal arrangements that tie the hand of the initiator in a way that is usually not possible in international relations. Still, if we include actions that the initia- tor can physically recall but not without intolerable cost, so that it is evident he would not go back even if it is equally clear that he could, we can find some instances. An armed convoy on aBerlin Autobahn may sometimes come close to having this quality.
Adegreelesssatisfactoryisthecompellentactionofwhichthe consequences can be averted by either side, by the initiator's changing his mind just in time or by his adversary's compliance. Because he can stop before the consequences mount up, this type of compellent action may be less risky for the party that starts it; there is a means of escape, though it may become a test of nerves, or a test of endurance, each side hoping the other will
wants? Habit, tradition, or some psychological compulsion may explain this connectedness, but it has to be asked whether they make it wise.
There are undoubtedly some good reasons for designing a compellent campaign that is connected with the compliance desired. One is that it helps to communicate the threat itself; it creates less uncertainty about what is demanded, what pressure will be kept up until the demands are complied with and then relaxed once they are. Actions not only speak louder than words on many occasions, but like words they can speak clearly or confusingly. To the extent that actions speak, it helps if they reinforce the message rather than confuse it.
Second, if the object is to induce compliance and not to start a spiral of reprisals and counteractions, it is helpful to show the limits to what one is demanding, and this can often be best shown by designing a campaign that distinguishes what is demanded from all the other objectives that one might have been seeking but is not. To harass aircraft in the Berlin air corridor communicates that polar flights are not at issue; to harass polar flights while saying that it is punishment for flying in the Berlin corridor does not so persuasively communicate that the harassment will stop when the Berlin flights stop, or that the Russians will not think of a few other favors they would like from the airline before they call off their campaign. Most of the problems of defining the threat and the demands that go with it, of offering assurance about what is not demanded and of promising cessation once compliance is forthcoming, are ag- gravated if there is no connection between the compellent action (or the threat of it) and the issue being bargained over.
The same question can arise with deterrent threats; some- times they lack connectedness. To threaten the Chinese main- land in the event of an overland attack on India has a minimum of connectedness. If the threatenedresponse is massive enough, though, it may seem to comprise or to include the local area and not merely to depart from it. But it often lacks some of the credibility,throughautomaticinvolvement,thatcanbeachieved by connecting the response physically to the provocation itself.
not actions initiated to induce compliance,
? Contingent actions -
? ? 90
back down, both sides possibly waiting too long. The escape hatch is an asset if one discovers along the way that the compellent attempt was a mistake after all- one misjudged the adversary, or formulated an impossible demand, or failed to communicate what he was doing and what he was after. The escape hatch is an embarrassment, though, if the adversary knows it is there; he can suppose, or hope, that the initiator will turn aside before the risk or pain mounts up.
Still another type is the action that, though beyond recall by the initiator, does not automatically stop upon the victim's compliance. Compliance is a necessary condition for stopping the damage but not suflicient, and if the damage falls mainly on the adversary,he has to consider what other demands will attach to the same compellent action once he has complied with the initial demands. The initiator may have to promise persuasively that he will stop on compliance, but stoppage is not automatic. Once the missiles are gone from Cuba we may have after- thoughts about antiaircraft batteries and want them removed
too before we call off the quarantine or stop the flights. Finally,thereistheactionthatonlytheinitiatorcanstop,but can stop any time with or without compliance, a quite "uncon-
nected" action.
In all of these cases the facts may be misperceived by one
party or both, with the danger that each may think the other can in fact avert the consequences, or one may fail to do so in the mistaken belief that the other has the last clear chance to avert collision. These different compellent mechanisms, which of course are more blurred and complex in any actual case, usually depend on what the connection is between the threat and the demand- a connection that can be physical, territorial, legal, symbolic, electronic, political, or psychological.
Compellence and Brinkmanship
Another important distinction is between compellent actions that inflict steady pressure over time, with cumulative pain or damage to the adversary (and perhaps to oneself), and actions
that impose risk rather than damage. Turning off the water supply at Guantanamo creates a finite rate of privation over time. Buzzing an airplane in the Berlin corridor does no harm unless the planes collide; they probably will not collide but they may and if they do the result is sudden, dramatic, irreversible, and grave enough to make even a small probability a serious one.
ARMS AND INFLUENCE
THE ART OF COMMITMENT 91
? ? The creation of risk- usually a shared risk- is the technique of compellence that probably best deserves the name of "brink- manship. " It is a competition in risk-taking. It involves setting afoot an activity that may get out of hand, initiating a process that carries some risk of unintended disaster. The risk is in- tended, but not the disaster. One cannot initiate certain disaster as a profitable way of putting compellent pressure on someone, but one can initiate a moderate risk of mutual disaster if the other party's compliance is feasible within a short enough pe- riod to keep the cumulative risk within tolerable bounds. "Rocking the boat" is a good example. If I say, "Row, or I'll tip the boat over and drown us both," you'll not believe me. I can- notactuallytiptheboatovertomakeyourow. ButifIstart rocking the boat so that it may tip over- not because I want it to but because I do not completely control things once I start rocking the boat- you'll be more impressed. I have to be will- ing to take the risk; then I still have to win the war of nerves, unless I can arrange it so that only you can steady the boat by rowing where I want you to. But it does lend itself to compel- lence, because one may be able to create acoercive riskof grave consequences where he could not profitably take a deliberate step to bring about those consequences, or even credibly threaten that he would. This phenomenon is the subject of the chapter that follows.
? ? ? ? ? ? ?
? ? ? ? ? ? ? ? 72 ARMS AND INFLUENCE
nonprovocative. The act that is intrusive, hostile, or provoca- tive is usually the one to be deterred; the deterrent threat only changes the consequences ifthe act in question- the one to be deterred- is then taken. Compellence, in contrast, usually involves initiating an action (or an irrevocable commitment to action) that can cease, or become harmless, only if the opponent responds. The overt act, the first step, is up to the side that makes thecompellentthreat. Todeter,onedigsin,orlaysaminefield, and waits- in the interest of inaction. To compel, one gets up enough momentum (figuratively, but sometimes literally) to maketheotheracttoavoidcollision.
Deterrence tends to be indefinite in its timing. "If you cross the line we shoot in self-defense, or the mines explode. " When? Whenever you cross the line- preferably never, but the timing is up to you. If you cross it, then is when the threat is fulfilled, eitherautomatically,ifwe'veriggeditso,orbyobligationthat immediately becomes due. But we can wait- preferably for- ever; that's our purpose.
Compellence has to be definite: We move, and you must get out of the way. By when? There has to be a deadline, otherwise tomorrow never comes. If the action carries no deadline it is only a posture, or a ceremony with no consequences. If the compellent advance is like Zeno's tortoise that takes infinitely longtoreachtheborderbytraversing,withinfinitepatience,the infinitely small remaining distances that separate him from collision, it creates no inducement to vacate the border. Compellence, to be effective, can't wait forever. Still, it has to wait a little; collision can't be instantaneous. The compellent threat has to be put in motion to be credible, and then the victim must yield. Too little time, and compliance becomes impos- sible; too much time, and compliance becomes unnecessary. Thus compellence involves timing in a way that deterrence typically does not.
In addition to the question of "when," compellence usually involves questions of where, what, and how much. "Do noth- ing" is simple, "Do something" ambiguous. "Stop where you are" is simple; "Go back" leads to "How far? " "Leave me
THE ART OF COMMITMENT 73
alone" is simple; "Cooperate" is inexact and open-ended. A deterrent position- a status quo, in territory or in more figura- tive terms- can often be surveyed and noted; a compellent advance has to be projected as to destination, and the destina- tion can be unclear in intent as well as in momentum and braking power. In a deterrent threat, the objective is often communicated by the very preparations that make the threat credible;thetrip-wireoftendemarcatestheforbiddenterritory. There is usually an inherent connection between what is threat- ened and what it is threatenedabout. Compellent threats tend to communicateonlythegeneraldirectionofcompliance,andare less likely to be self-limiting, less likely to communicate in the very design of the threat just what, or how much, is demanded. ThegarrisoninWestBerlincanhardlybemisunderstoodabout what it is committed to resist; if it ever intruded into East Berlin, though, to induce Soviet or German Democratic Republic forces to give way, there would be no such obvious interpreta- tion of where and how much to give way unless the adventure could be invested with some unmistakable goal or limitation-
a possibility not easily realized.
The Quemoy escapade is again a good example: Chiang's
troops, once on the island, especially if evacuation under fire appeared infeasible, had the static clarity that goes with com- mitmenttoanindefinitestatusquo,whilethecommitmentjust to send troops to defend it (or air and naval support) according to whether a Communist attack there was or was not prelude to anattackonFormosalackedthatpersuasivequality,reminding us that though deterrent threats tend to have the advantages mentioned above they do not always achieve them. (The ambiguous case of Quemoy actually displays the compellent ambiguity, seen in reverse: a "compellent" Communist move against Quemoy was to be accommodated, as long as its extent couldbereliablyprojectedtoaterminusshortofFormosa;ifthe Communiststhought we meant it, it was up to them to design an action that visibly embodied that limitation. ) An American or NATO action to relieve Budapest in 1956- without major engagement but in the hope the Soviets would give way rather
? ? ? ? ? ? ? 74 ARMS AND INFLUENCE
than fight- would have had the dynamic quality of "compellence" in contrast to Berlin: the stopping point would have been a variable, not a constant. Even "Budapest" would have needed a definition, and might have become all of Hun- gary- and after Hungary, what? - if the Soviets initially gave way. The enterprise might have been designed to embody its specificintent,butitwouldhavetakenalotofdesigningbacked up by verbal assurances.
Actually, any coercive threat requires corresponding assur- ances; the object of a threat is to give somebody a choice. To say, "One more step and I shoot," can be a deterrent threat only if accompanied by the implicit assurance, "And if you stop I won't. '' Giving notice of unconditional intent to shoot gives him no choice (unless by behaving as we wish him to behave the opponent puts himself out of range, in which case the effective threat is, "Come closer and my fire will kill you, stay back and it won't''). What was said above about deterrent threats being typically less ambiguous in intent can be restated: the corre- sponding assurances- the ones that, together with the threat- ened response, define the opponent's choice- are clearer than those that can usually be embodied in a compellent action. (Ordinary blackmailers, not just nuclear, find the "assurances" troublesome when their threats are compellent. ) l 8
They are, furthermore, confirmed and demonstrated over time; as long as he stays back, and we don't shoot, we fulfill the assurances and confirm them. The assurances that accompany a compellent action- move back a mile and I won't shoot (other- wise I shall) and I won't then try again for a second mile- are
18. The critical role of "assurances" in completing the structure of a threat, in making the threatened consequences persuasively conditional on behavior so that the victim is offered a choice, shows up in the offers of amnesty, safe passage, or forgiveness that must often be made credible in inducing the surrender of rebels or the capitulation of strikers or protesters. Even libraries and internal revenue agencies depend on parallel offers of forgiveness when they embark on campaigns to coerce the return of books or payment of back taxes. In personal life I have sometimes relied, like King Lea, on the vague threat that my wrath will be aroused (with who knows what awful consequences) if good behavior is not forthcoming, making a tentative impression on one child, only to have the threat utterly nullified by another's pointing out that "Daddy's mad already. "
THE ART OF COMMITMENT 7s
harder to demonstrate in advance, unless it be through a long past record of abiding by one's own verbal assurances.
Because in the West we deal mainly in deterrence, not com- pellence, and deterrent threats tend to convey their assurances implicitly, we often forget that both sides of the choice, the threatened penalty and the proffered avoidance or reward, need to be credible. The need for assurances- not just verbal but fully credible- emerges clearly as part of "deterrence" in discussions of surprise attack and "preemptive war. " An enemy belief that we are about to attack anyway, not after he does but possibly before, merely raises his incentive to do what we wanted to deter and to do it even more quickly. When we do engage in compellence, as in the Cuban crisis or in punitive at- tacks on North Vietnam that are intended to make the North Vietnamese government act affirmatively, the assurances are a critical part of the definition of the compellent threat.
One may deliberately choose to be unclear and to keep the enemy guessing either to keep his defenses less prepared or to enhance his anxiety. But if one wants not to leave him in doubt about what will satisfy us, we have to find credible ways of com- municating, and communicating both what we want and what we do not want. There is a tendency to emphasize the commu- nication of what we shall do if he misbehaves and to give too little emphasis to communicating what behavior will satisfy us. Again, this is natural when deterrence is our business, because the prohibited misbehavior is often approximately defined in the threatened response; but when we must start something that then has to be stopped, as in compellent actions, it is both harder and more important to know our aims and to communi- cate. It is particularly hard because the mere initiation of an en- ergetic coercive campaign, designed for compellence, disturbs the situation, leads to surprises, and provides opportunity and temptation to reexamine our aims and change them in mid course. Deterrence,if wholly successful, can often afford to con- centrate on the initiating events- what happens next if he mis- behaves. Compellence, to be successful, involves an action that must be brought to successful closure. The payoff comes at the end, as does the disaster if the project fails.
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 76 ARMS AND INFLUENCE
THE ART OF COMMITMENT 77
? The compellent action will have a time schedule of its own, and unless it is carefully chosen it may not be reconcilable with the demands that are attached to it. W e cannot usefully threaten to bomb Cuba next Thursday unless the Russians are out by next month, or conduct a six weeks' bombing campaign in North Vietnam and stop it when the Vietcong have been quiescent for six months. There will be limits, probably, to how long the compellent action can be sustained without costing or risking too much, or exhausting itself or the opponent so that he has nothing left to lose. If it cannot induce compliance within that time- and this depends on whether compliance is physi- cally or administratively feasible within that time- it cannot accomplish anything (unless the objective was only an excuse for some act of conquest or punishment). The compellent ac- tion has to be one that can be stopped or reversed when the enemy complies, or else there is no inducement.
surrender statement or acknowledgement of submission, some symbolic knuckling under, will itself achieve the object, verbal compliance may be enough. It is inherent in an intense crisis that the conditions for bringing it to a close have to be of a kind that can be met quickly; that is what we mean by an "intense crisis," one that compresses risk, pain, or cost into a short span of time or that involves actions that cannot be sustained indefi- nitely. If we change our compellent threat from slow pressure to intense, we have to change our demands to make them fit the urgent timing of a crisis.
Notice that to deter continuance of something the opponent is already doing- harassment, overflight, blockade, occupation of some island or territory, electronic disturbance, subversive activity, holding prisoners, or whatever it may be- has some of the character of a compellent threat. This is especially so of the timing, of who has to take the initiative. In the more static case we want him to go on not doing something; in this more dy- namic case we want him to change his behavior. The "when" problem arises in compelling him to stop, and the compellent action may have to be initiated, not held in waiting like the de- terrent threat. The problems of "how much" may not arise if it is some discrete, well defined activity. "At all" may be the obvi- ous answer. For U-2flights or fishing within a twelve-mile limit, that may be the case; for subversive activity or support to in- surgents,"atall"mayitselfbeambiguousbecausetheactivityis complex, ill defined, and hard to observe or attribute.
Blockade, harassment, and "salami tactics" can be interpreted as ways of evading the dangers and difficulties of compel- lence. Blockade in a cold war sets up a tactical "status quo" that is damaging in the long run but momentarily safe for both sides unless the victim tries to run the blockade. President Ken- nedy's overt act of sending the fleet to sea, in "quarantine" of Cuba in October 1962, had some of the quality of deterrent "stage setting"; the Soviet government then had about forty- eight hours to instruct its steamers whether or not to seek collision. Low-levelintrusion,as discussed earlier, can be a way of letting the opponent turn his head and yield a little, or it can be a way of starting a compellent action in low gear, without the
? ? ? If the opponent's compliance necessarily takes time- if it is sustained good behavior, cessation of an activity that he must not resume, evacuation of a place he must not reenter, payment of tribute over an extended period, or someconstructiveactivity that takes time to accomplish- the compellent threat requires some commitment, pledge, or guarantee, or some hostage, or else must be susceptible of being resumed or repeated itself. Particularly in a crisis, a Cuban crisis or a Vietnamese crisis, there is strong incentive to get compliance quickly to limit the risk or damage. Just finding conditions that can be met on the demanding time schedule of a dangerous crisis is not easy. The ultimate demands, the objectives that the compellent threat is really aimed at, may have to be achieved indirectly, by taking pledges or hostages that can be used to coerce compliance after the pressure has been re1ie~ed. IO~f course, if some kind of
? ? ? ? ? 19. Lord Portal's account of the coercive bombing of the villages of recalcitrant Arab tribesmen (after warning to permit evacuation) includes the terms that were
? demanded. Among them were hostages- literal hostages, people -
as well as a fine; otherwise the demand was essentially cessation of the raids or other misbehavior that had brought on the bombing. The hostages were apparently partly to permit subsequent enforcement without repeated bombing, partly to symbolize, together with the fine, the tribe's intent to comply. See Portal, "Air Force Cooperation in
? ? Policing the Empire," pp. 343-58.
? ? ? ? 78 ARMS AND INFLUENCE
THE ART OF COMMITMENT 79
or costly,wecancallita"coercive"or"deterrent"defense. The language is clumsy but the distinction is valid. Resistance that might otherwise seem futile can be worthwhile if, though in- capable of blocking aggression, it can nevertheless threaten to make the cost too high. This is "active" or "dynamic" deter- rence, deterrence in which the threat is communicated by pro- gressive fulfillment. At the other extreme is forcible defense with good prospect of blocking the opponent but little promise of hurting; this would be purely defensive.
Defensive action may even be undertaken with no serious hope of repelling or deterring enemy action but with a view to making a "successful" conquest costly enough to deter repeti- tion by the same opponent or anyone else. This is of course the rationale for reprisals after the fact; they cannot undo the deed but can make the books show a net loss and reduce the incen- tive next time. Defense can sometimes get the same point across, as the Swiss demonstrated in the fifteenth century by the manner in which they lost battles as well as by the way they sometimes won them. "The [Swiss] Confederates were able to reckon their reputation for obstinate and invincible courage as one of the chief causes which gave them political importance. . . . It was no light matter to engage with an enemy who would not retire before any superiority in numbers, who was always ready for a fight, who would neither give nor take quarter. " 2o The Finns demonstrated five hundred years later that the princi- ple still works. The value of local resistance is not measured solely by local success. This idea of what we might call "punitive resistance" could have been part of the rationale for the American commitment of forces in Vietnam. *'
"Compellence" is more like "offense. " Forcible offense is taking something,occupyingaplace,ordisarminganenemyor a territory, by some direct action that the enemy is unable to block. "Compellence" is inducing his withdrawal, or his ac-
20. C. W. C. Oman, The Art of War in the Middle Ages (Ithaca, Cornell University Press, 1953), p. 96.
21. An alternative, but not inconsistent, treatment of some of these distinctions is in Glenn H. Snyder, Deterrence and Defense (Princeton, Princeton University Press, 1961), pp. 5-7, 9-16, 24-40.
conviction that goes with greater momentum but also without the greater risk. Instead of speeding out of control toward our car that blocks his way, risking our inability to see him and get our engines started in time to clear his path, he approaches slowly and nudges fenders, crushing a few lights and cracking some paint. If we yield he can keep it up, if not he can cut his losses. And if he makes it look accidental, or can blame it on an impetuous chauffeur, he may not even lose countenance in the unsuccessful try.
Defense and Deterrence, Offense and Compellence
The observation that deterrent threats are often passive, while compellent threats often have to be active, should not be pressed too far. Sometimes a deterrent threat cannot be made credible in advance, and the threat has to be made lively when the prohibited action is undertaken. This is where defense and deterrence may merge, forcible defense being undertaken in the hope, perhaps with the main purpose, of demonstrating by resistance that the conquest will be costly, even if successful, too costly to be worthwhile. The idea of "graduated deterrence" and much of the argument for a conventional warfare capability in Europe are based on the notion that if passive deterrence initially fails, the more active kind may yet work. If the enemy act to be deterred is a once-for-all action, incapable of with- drawal, rather than progressive over time, any failure of deter- rence is complete and final; there is no second chance. But if the aggressive move takes time, if the adversary did not believe he would meet resistance or did not appreciate how costly it would be, one can still hope to demonstrate that the threat is in force, after he begins. If he expected no opposition, encountering
somemaycausehimtochangehis mind.
There is still a distinction here between forcible defense and
defensive action intended to deter. If the object, and the only hope, is to resist successfully, so that the enemy cannot succeed even if he tries, we can call it pure defense. If the object is to induce him not to proceed, by making his encroachment painful
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ARMS AND INFLUENCE
quiescence, or his collaboration by an action that threatens to hurt, often one that could not forcibly accomplish its aim but that, nevertheless, can hurt enough to induce compliance. The forcible and the coercive are both present in a campaign that could reach its goal against resistance, and would be worth the cost, but whose cost is nevertheless high enough so that one hopes to induce compliance, or to deter resistance, by making evident the intent to proceed. Forcible action, as mentioned in Chapter 1, is limited to what can be accomplished without enemy collaboration; compellent threats can try to induce more affirmative action, including the exercise of authority by an enemy to bring about the desired results.
War itself, then, can have deterrent or compellent intent, just as it can have defensive or offensive aims. A war in which both sides can hurt each other but neither can forcibly accomplish its purpose could be compellent on one side, deterrent on the other.
Once an engagement starts, though, the difference between deterrence and compellence, like the difference between de- fense and offense, may disappear. There can be legal and moral reasons, as well as historical reasons, for recalling the status quo ante; but if territory is in dispute, the strategies for taking it, holding it, or recovering it may not much differ as between the sidethatoriginallypossesseditandthesidethatcovetedit,once the situation has become fluid. (In a local tactical sense, American forces were often on the "defensive" in North Korea and on the "offensive" in South Korea. ) The coercive aspect of warfare may be equally compellent on both sides, the only difference perhaps being that the demands of the defender, the one who originally possessed what is in dispute, may be clearly defined by the original boundaries, whereas the aggressor's demands may have no such obvious definition.
The Cuban crisis is a good illustration of the fluidity that sets in once passive deterrence has failed. The United States made verbal threats against the installation of weapons in Cuba but apparently some part of the threat was unclear or lacked credibility and it was transgressed. The threat lacked the autom- aticity that would make it fully credible, and without some
THE ART OF COMMITMENT 81
80
automaticity it may not be clear to either side just where the threshold is. Nor was it physically easy to begin moderate resistance after the Russians had crossed the line, and to increase the resistance progressively to show that the United States meant it. By the time the President determined to resist, he was no longer in a deterrent position and had to embark on the more complicated business of compellence. The Russian missiles could sit waiting, and so could Cuban defense forces; the next overt act was up to the President. The problem was to prove to the Russians that a potentially dangerous action was forthcoming, without any confidence that verbal threats would be persuasive and without any desire to initiate some irrevers- ible process just to prove, to everybody's grief, that the United States meant what it said.
The problem was to find some action that would communi- cate the threat, an action that would promise damage if the Russians did not comply but minimum damage if they com- plied quickly enough, and an action that involved enough momentum or commitment to put the next move clearly up to the Russians. Any overt act against a well-defended island would be abrupt and dramatic; various alternatives were appar- ently considered, and in the end an action was devised that had many of the virtues of static deterrence. A blockade was thrown around the island, a blockade that by itself could not make the missiles go away. The blockade did, however, threaten aminor military confrontation with major diplomatic stakes- an en- counter between American naval vessels and Soviet merchant ships bound for Cuba. Once in place, the Navy was in a position to wait; it was up to the Russians to decide whether to continue. If Soviet ships had been beyondrecall, the blockade would have been a preparation for inevitable engagement; with modem communications the ships were not beyond recall, and the Russians were given the last clear chance to turn aside. Physi- cally the Navy could have avoided an encounter; diplomati- cally, the declaration of quarantine and the dispatch of the Navy meant that American evasion of the encounter was virtually out of the question. For theRussians, the diplomatic cost of turning
? ? ? ? ? ? ? ? 82 ARMS AND INFLUENCE
THE ART OF COMMITMENT 83
threats of more violent action, had no effect. 23If the North Vietnamese dramatically issue a call to the Vietcong to cease activity and to evacuate South Vietnam, it is a conspicuous act of submission. If the Americans had evacuated Guantanamo when Castro turned off the water, it would have been a con- spicuous act of submission. If an earthquake or change in the weather had caused the water supply to dry up at Guantanamo, and if the Americans had found it wholly uneconomical to supply the base by tanker, they might have quit the place without seeming to submit to Castro's cleverness or seeming afraid to take reprisals against their ungracious host. Similarly, the mere act of bombing North Vietnam changed the status of any steps that the North Vietnamese might take to comply with American wishes. It can increase their desire, if the tactic is successful, to reduce support for the Vietcong; but it also increases the cost of doing so. Secretary Dulles used to say that while we had no vital interest in Quemoy we could not afford to evacuate under duress; intensified Chinese pressure always led to intensified determination to resist it. "
If the object is actually to impose humiliation, to force a showdown and to get an acknowledgement of submission, then the "challenge" that is often embodied in an active compellent threat is something to be exploited. President Kennedy un- doubtedly wanted some conspicuous compliance by the Soviet Union during the Cuban missile crisis, if only to make clear to the Russians themselves that there were risks in testing how much the American government would absorb such ventures. In Vietnam the problem appeared the opposite; what was most
freighters around, or even letting one be examined, proved not to be prohibitive.
Thus an initial deterrent threat failed, acompellent threat was called for, and by good fortune one could be found that had some of the static qualities of a deterrent threat. 22
There is another characteristic of compellent threats, arising in the need for affirmative action, that often distinguishes them from deterrent threats. It is that the very act of compliance- of doing what is demanded- is more conspicuously compliant, more recognizable as submission under duress, than when an act is merely withheld in the face of a deterrent threat. Compli- anceislikelytobelesscasual,lesscapableofbeingrationalized as something that one was going to do anyhow. The Chinese did not need to acknowledge that they shied away from Quemoy or Formosa because of American threats, and the Russians need not have agreed that it was NATO that deterred them from conquering Western Europe, and no one can be sure. Indeed, if a deterrent threat is created before the proscribed act is even contemplated, there need never be an explicit decision not to transgress, just an absence of any temptation to do the thing prohibited. The Chinese still say they will take Quemoy in their own good time; and the Russians go on saying that their intentions against Western Europe were never aggressive.
The Russians cannot, though, claim that they were on the point of removing their missiles from Cuba anyway, and that the President's television broadcast, the naval quarantine and
22. Arnold Horelick agrees with this description. "As an initial response the quarantine was considerably less than a direct application of violence, but considerably more than a mere protest or verbal threat. The U. S. Navy placed itself physically between Cuba and Soviet ships bound for Cuban ports. Technically, it might still have been necessary for the United States to fire the first shot had Khrushchev chosen to defy the quarantine, though other means of preventing Soviet penetration might have been employed. But once the quarantine was effectively established- which was done with great speed- it was Khrushchev who had to make the next key decision: whether or not to risk releasing the trip-wire. '' "The Cuban Missile Crisis," World Politics, 16 (1964), 385. This article and the Adelphi Paper of Albert and Roberta Wohlstetter mentioned in an earlier note are the best strategic evaluations of the Cuban affair that I have discovered.
23. The tendency for affirmative action to appear compliant is vividly illustrated
? ? by the widespread suspicion -
one that could not he effectively dispelled- that the U. S. missiles removed from Turkey in the wake of the Cuban crisis were
? part of a bargain, tacit if not explicit.
24. Almost everyone in America, surely including the President and the Secretary
of State, would have been relieved in the late 1950s if an earthquake or volcanic action had caused Quemoy to sink slowly beneath the surface of the sea. Evacuation would then not have been retreat, and an unsought commitment that had proved peculiarly susceptible to Communist China's manipulation would have been disposed of. Such is the intrinsic value of some territories that have to he defended!
? ? ? ? ? ? ? ? ? Skill is required to devise a compellent action that does not have this self-defeating quality. There is an argument here for sometimes not being too explicit or too open about precisely what is demanded, if the demands can be communicated more privately and noncommittally. President Johnson was widely criticized in the press, shortly after the bombing attacks began in early 1965,for not having made his objectives entirely clear. How could the North Vietnamese comply if they did not know exactly what was wanted? Whatever the reason for the Ameri- can Administration's being somewhat inexplicit- whether it chose to be inexplicit, did not know how to be explicit, or in fact
ARMS AND INFLUENCE
THE ART OF COMMITMENT 85
we can suppose that the United States government did not know in detailjust how much control or influence the North Vietnam- eseregimehadovertheVietcong;andwecanevensupposethat the North Vietnamese regime itself might not have been alto- gether sure how much influence it would have in commanding a withdrawal or in sabotaging the movement that had received its moral and material support. The United States government-may not have been altogether clear on which kinds of North Viet- namese help- logistical help, training facilities, sanctuary for the wounded, sanctuary for intelligence and planning activities, communications relay facilities, technical assistance, advisors and combat leaders in the field, political and doctrinal assis- tance, propaganda, moral support or anything else- were most effective and essential, or most able to be withdrawn on short notice with decisive effects. And possibly the North Vietnam- ese did not know. The American government may have been in the position of demanding results not specific actions, leav- ing it to the North Vietnamese through overt acts, or merely through reduced support and enthusiasm, to weaken the Viet- cong or to let it lose strength. Not enough is known publicly to permitustojudgethisVietnameseinstance;butitpointsupthe important possibility that a compellent threat may have to be focused on results rather than contributory deeds, like the fa- ther's demand that his son's school grades be improved, or the extortionist's demand, "Get me money. I don't care how you get it,just get it. " A difficulty, of course, is that results are more a matter of interpretations than deeds usually are. Whenever a recipient of foreign aid, for example, is told that it must elimi- nate domestic corruption, improve its balance of payments, or raise the quality of its civil service, the results tend to be uncer- tain, protracted, and hard to attribute. The country may try to comply and fail; with luck it may succeed without trying; it may have indifferent success that is hard to judge; in any case com- pliance is usually arguable and often visible only in retrospect.
Even more than deterrence, compellence requires that we recognize the difference between an individual and a govern- ment. To coerce an individual it may be enough to persuade
? 84
urgently desired was to reduce the support for the Vietcong from the North, and any tendency for the compellent pressure of bombing to produce a corresponding resistance would have been deprecated. But it cannot always be avoided, and if it cannot, the compellent threat defeats itself.
? was explicit but only privately- an important possibility is that vague demands, though hard to understand, can be less embar- rassing to comply with. If the President had to be so explicit that any Europeanjournalist knew exactly what he demanded, and if the demands were concrete enough to make compliance recog- nizable when it occurred, any compliance by the North Viet- namese regime would necessarily have been fully public, perhaps quite embarrassingly so. The action could not be hidden nor the motive so well disguised as if the demands were more privately communicated or left to inference by the North Vietnamese.
Another serious possibility is suggested by the North Viet- namese case: that the initiator of a compellent campaign is not himself altogether sure of what action he wants, or how the result that he wants can be brought about. In the Cuban missile case it was perfectly clear what the United States government wanted, clear that the Soviets had the ability to comply, fairly clear how quickly it could be done, and reasonably clear how compliance might be monitored and verified, though in the end there might be some dispute about whether the Russians had left behind things they were supposed to remove. In the Vietnamese case,
? ? ? ? 86 ARMS AND INFLUENCE
THE ART OF COMMITMENT 87
arises whether they ought to be connected at all. If the object is to harass, to blockade, to scare or to inflict pain or damage until an adversary complies, why cannot the connection be made verbally? If the Russians want Pan-American Airways to stop using the air corridor to Berlin, why can they not harass the airline on its Pacific routes, announcing that harassment will continue until the airline stops flying to Berlin? When the Russians put missiles in Cuba, why cannot the President quar- antineVladivostok,stoppingSovietshipsoutside,say,atwelve- mile limit, or perhaps denying them access to the Suez or Panama Canal? And if the Russians had wanted to counter the President's quarantine of Cuba, why could they not blockade
A hasty answer may be that it just is not done, or is not "justified," as though connectedness implied justice, or as though justice were required for effectiveness. Surely that is part of the answer; there is a legalistic or diplomatic, perhaps a casuistic, propensity to keep things connected, to keep the threat and the demand in the same currency, to do what seems reasonable. But why be reasonable, if results are what one
25. It has often been said that American tactical superiority and ease of access in the Caribbean (coupled with superiority in strategic weaponry) account for the success in inducing evacuation of the Soviet missiles. Surely that was crucial; but equally significantwas the universal tendency- a psychological phenomenon, a tradition or convention shared by Russians and Americans- to define the conflict in Caribbean terms, not as a contest, say, in the blockade of each other's island allies, not as a counterpart of their position in Berlin, not as a war of harassment against strategic weapons outside national borders. The countermeasures and counterpressuresavailable to the Russians might have looked very different tothe "Russian" side if this bad been a game on an abstract board rather than an event in historical time in a particular part of the real world. The Russians tried (as did some unhelpful Americans) to find a connection between Soviet missiles in Cuba and American missiles in Turkey, but the connection was evidently not persuasive enough for the Russians to be confident that, if the dispute led to military action or pressure against Turkey, that definition would hold and things would go no further. The Caribbean definitionhad more coherence or integrity than a Cuban- Turkish definition would have had, or, in terms of reciprocal blockade, a Cuban-United Kingdom definition would have had. The risk of further metastasis must have inhibited any urge to let the crisis break out of its original Caribbean definition.
him to change his mind; to coerce a government it may not be necessary,butitalsomaynotbesufficient,tocauseindividuals to change their minds. What may be required is some change in the complexion of the government itself, in the authority, pres- tige, or bargaining power of particular individuals or factions or parties, some shift in executive or legislative leadership. The Japanese surrender of 1945was marked as much by changes in the structure of authority and influence within the government as by changes in attitude on the part of individuals. The victims of coercion, or the individuals most sensitive to coercive threats, may not be directly in authority; or they may be hopelessly
committed to non-compliant policies. They may have to bring bureaucratic skill or political pressure to bear on individuals who do exercise authority, or go through processes that shift authority or blame to others. In the extreme case governing authorities may be wholly unsusceptible to coercion- may, as a party or as individuals, have everything to lose and little to save by yielding to coercive threats- and actual revolt may be essential to the process of compliance, or sabotage or assassina- tion. Hitler was uncoercible; some of his generals were not, but they lacked organization and skill and failed in their plot. For
working out the incentive structure of a threat, its communica- tion requirements and its mechanism, analogies with individu- als are helpful; but they are counterproductive if they make us forget that a government does not reach a decision in the same way as an individual in a government. Collective decision depends on the internal politics and bureaucracy of govern- ment, on the chain of command and on lines of communication, on party structures and pressure groups, as well as on individual values and careers. This affects the speed of decision, too.
"Connectedness" in Compellent Threats
As mentioned earlier, a deterrent threat usually enjoys some connectedness between the proscribed action and the threatened response. The connection is sometimes a physical one, as when ~OOpasreputinBerlintodefendBerlin. Compellentactionsoften have a less well-defined connectedness; and the question
? ? ? ? ? ? ? ? ? ? ? ? ? 88 ARMS AND INFLUENCE
THE ART OF COMMITMENT 89
ut actions threatened against potential provocation-often need the credibility that connectedness can give them.
Connectedness in fact provides something of a scheme for classifyingcompellentthreatsandactions. Theidealcompellent action would be one that, once initiated, causes minimal harm if compliance is forthcoming and great harm if compliance is not forthcoming, is consistent with the time schedule of feasible compliance, is beyond recall once initiated, and cannot be stopped by the party that started it but automatically stops upon compliance,withallthisfullyunderstoodbytheadversary. Only he can avert the consequences; he can do it only by complying; andcomplianceautomaticallyprecludesthem. Hisisthenthe "last clear chance" to avert the harm or catastrophe; and it would not even matter which of the two most feared the consequences as long as the adversary knew that only he, by complying, could avert them. (Of course, whatever is demanded of him must be less unattractive to him than the threatened consequences, and the manner of threatened compliance must not entail costs in prestige, reputation, or self-respect that outweigh the threat. )
It is hard to find significant international events that have this perfectionist quality. There are situations, among cars on high- ways or in bureaucratic bargaining or domestic politics, where one comes across such ideal compellent threats; but they usually involve physical constraints or legal arrangements that tie the hand of the initiator in a way that is usually not possible in international relations. Still, if we include actions that the initia- tor can physically recall but not without intolerable cost, so that it is evident he would not go back even if it is equally clear that he could, we can find some instances. An armed convoy on aBerlin Autobahn may sometimes come close to having this quality.
Adegreelesssatisfactoryisthecompellentactionofwhichthe consequences can be averted by either side, by the initiator's changing his mind just in time or by his adversary's compliance. Because he can stop before the consequences mount up, this type of compellent action may be less risky for the party that starts it; there is a means of escape, though it may become a test of nerves, or a test of endurance, each side hoping the other will
wants? Habit, tradition, or some psychological compulsion may explain this connectedness, but it has to be asked whether they make it wise.
There are undoubtedly some good reasons for designing a compellent campaign that is connected with the compliance desired. One is that it helps to communicate the threat itself; it creates less uncertainty about what is demanded, what pressure will be kept up until the demands are complied with and then relaxed once they are. Actions not only speak louder than words on many occasions, but like words they can speak clearly or confusingly. To the extent that actions speak, it helps if they reinforce the message rather than confuse it.
Second, if the object is to induce compliance and not to start a spiral of reprisals and counteractions, it is helpful to show the limits to what one is demanding, and this can often be best shown by designing a campaign that distinguishes what is demanded from all the other objectives that one might have been seeking but is not. To harass aircraft in the Berlin air corridor communicates that polar flights are not at issue; to harass polar flights while saying that it is punishment for flying in the Berlin corridor does not so persuasively communicate that the harassment will stop when the Berlin flights stop, or that the Russians will not think of a few other favors they would like from the airline before they call off their campaign. Most of the problems of defining the threat and the demands that go with it, of offering assurance about what is not demanded and of promising cessation once compliance is forthcoming, are ag- gravated if there is no connection between the compellent action (or the threat of it) and the issue being bargained over.
The same question can arise with deterrent threats; some- times they lack connectedness. To threaten the Chinese main- land in the event of an overland attack on India has a minimum of connectedness. If the threatenedresponse is massive enough, though, it may seem to comprise or to include the local area and not merely to depart from it. But it often lacks some of the credibility,throughautomaticinvolvement,thatcanbeachieved by connecting the response physically to the provocation itself.
not actions initiated to induce compliance,
? Contingent actions -
? ? 90
back down, both sides possibly waiting too long. The escape hatch is an asset if one discovers along the way that the compellent attempt was a mistake after all- one misjudged the adversary, or formulated an impossible demand, or failed to communicate what he was doing and what he was after. The escape hatch is an embarrassment, though, if the adversary knows it is there; he can suppose, or hope, that the initiator will turn aside before the risk or pain mounts up.
Still another type is the action that, though beyond recall by the initiator, does not automatically stop upon the victim's compliance. Compliance is a necessary condition for stopping the damage but not suflicient, and if the damage falls mainly on the adversary,he has to consider what other demands will attach to the same compellent action once he has complied with the initial demands. The initiator may have to promise persuasively that he will stop on compliance, but stoppage is not automatic. Once the missiles are gone from Cuba we may have after- thoughts about antiaircraft batteries and want them removed
too before we call off the quarantine or stop the flights. Finally,thereistheactionthatonlytheinitiatorcanstop,but can stop any time with or without compliance, a quite "uncon-
nected" action.
In all of these cases the facts may be misperceived by one
party or both, with the danger that each may think the other can in fact avert the consequences, or one may fail to do so in the mistaken belief that the other has the last clear chance to avert collision. These different compellent mechanisms, which of course are more blurred and complex in any actual case, usually depend on what the connection is between the threat and the demand- a connection that can be physical, territorial, legal, symbolic, electronic, political, or psychological.
Compellence and Brinkmanship
Another important distinction is between compellent actions that inflict steady pressure over time, with cumulative pain or damage to the adversary (and perhaps to oneself), and actions
that impose risk rather than damage. Turning off the water supply at Guantanamo creates a finite rate of privation over time. Buzzing an airplane in the Berlin corridor does no harm unless the planes collide; they probably will not collide but they may and if they do the result is sudden, dramatic, irreversible, and grave enough to make even a small probability a serious one.
ARMS AND INFLUENCE
THE ART OF COMMITMENT 91
? ? The creation of risk- usually a shared risk- is the technique of compellence that probably best deserves the name of "brink- manship. " It is a competition in risk-taking. It involves setting afoot an activity that may get out of hand, initiating a process that carries some risk of unintended disaster. The risk is in- tended, but not the disaster. One cannot initiate certain disaster as a profitable way of putting compellent pressure on someone, but one can initiate a moderate risk of mutual disaster if the other party's compliance is feasible within a short enough pe- riod to keep the cumulative risk within tolerable bounds. "Rocking the boat" is a good example. If I say, "Row, or I'll tip the boat over and drown us both," you'll not believe me. I can- notactuallytiptheboatovertomakeyourow. ButifIstart rocking the boat so that it may tip over- not because I want it to but because I do not completely control things once I start rocking the boat- you'll be more impressed. I have to be will- ing to take the risk; then I still have to win the war of nerves, unless I can arrange it so that only you can steady the boat by rowing where I want you to. But it does lend itself to compel- lence, because one may be able to create acoercive riskof grave consequences where he could not profitably take a deliberate step to bring about those consequences, or even credibly threaten that he would. This phenomenon is the subject of the chapter that follows.
? ? ? ? ? ? ?
