In Cold War deterrence, nothing was more important or disturbing than the credibility problem. Success depended on conveying credible threats, yet it was hard to see how to make retaliatory threats credible in mutual nuclear deterrence, particularly in extended deterrence or to prevent conventional attacks. It was also hard to make nuclear retaliation threats credible against nonnuclear powers; nuclear use would arouse condemnation, set an awful precedent, and stimulate nuclear proliferation. In fact, how were retaliatory threats of any sort made credible?

Various answers included being able to fight and win any kind of prospective war, building the capacity to defeat any limited attack and not have to escalate one to an unbearable level, and resting deterrence on the fact that the deterrer might retaliate if attacked by losing control, reacting irrationally, or setting off uncontrollable escalation. Each was unattractive in various ways. An additional solution was to bolster retaliatory threats by having previously upheld lesser commitments, and to adopt other security measures that projected resolve. This meant treating one's commitments as interdependent; defending one bolsters the credibility of the others. This was an appealing solution and it has been very influential in U.S. national security policy ever since. However, it was hard to show a logical or empirical connection between upholding commitments with bearable costs and ones with potentially unbearable results. This approach could also readily become open-ended, inviting military overextension, declining public support, and eventual military failures (as in Vietnam) that reduce, not maintain, credibility.

The Cold War vanished but the credibility problem did not. Doubts about deterrence just shifted focus: Can deterrence be credible against terrorists, rogue states, or a China bent on seizing Taiwan? The books by Anne Sartori and Daryl Press reexamine the credibility problem by suggesting theoretical adjustments and providing new evidence. While not asking precisely the same question, they overlap considerably. Each offers a concise, tightly argued analysis, with a clear theoretical position tested via elegant and creative research. Each clearly covers a complex subject. And their carefully developed conclusions point in opposite directions!

Does a defender's prior behavior shape a challenger's assessment of his threat credibility in a crisis? Press says “no.” A challenger (or deterrer) does not assess credibility on the basis of the deterrer's (or challenger's) reactions to prior challenges. Each weighs whether the opponent has serious interests at stake and the military power to carry out threats. Citing the importance of interests and available power fits classic deterrence thinking; dismissing the deterrer's prior behavior as irrelevant does not. The common view, particularly among officials, has been that appearing soft now makes for trouble later.

Press rejects numerous variants of a “past action theory” of deterrence credibility in favor of a “current calculus theory.” Theories of deterrence and, more broadly, coercive diplomacy explain how to manipulate an opponent's cognitive processes, yet studies of deterrence rarely explore those processes. Press does so, using official records from crises, ones often cited on how past behavior shapes credibility, to trace the internal discussions of officials. In 1939, he reports, German policymakers largely ignored prior British and French “appeasement” behavior: “German leaders overwhelmingly tied their assessments of Allied credibility to their assessments of the balance of interests and power” (pp. 74–75). In the Berlin crises of 1958–1961, Nikita Khrushchev repeatedly backed off his threats but Soviet credibility with British and American officials did not shrink. In the Cuban Missile Crisis, those retreats over Berlin again had no effect on U.S. officials; the rising Soviet stake in Cuba and the growing Soviet ability to hit the United States with nuclear weapons were the important considerations.

Press concludes that “less expensive and symbolic acts of diplomacy” can help bolster credibility but are not critical, while fighting wars to preserve credibility is unnecessary, even damaging if they turn out badly. Credibility stems from having vital interests at stake and a powerful military capability. He asserts that a preoccupation with reputation has rigidified American foreign policy, instigating more fighting than is necessary for deterrence credibility.

That prior behavior is not the key to credibility is not a new view. Press's main contribution is to provide more empirical backing for it. The absence in internal discussions of credibility being assessed by officials on the basis of prior behavior is very interesting. So is his finding that the same officials saw their government's future credibility as resting on its current actions (“… investing in their own reputation but ignoring everyone else's—is puzzling”; p. 158)!

Can a deterrer's earlier behavior affect estimates of its credibility? Sartori says “yes.” The prior behavior usually cited is from prior confrontations, but she innovatively broadens this. She contends that a reputation for honesty in diplomatic communication is an important asset, particularly for effectively communicating threats. She asserts that “when a state is caught bluffing, it acquires a reputation for doing so” (p. 13), and for a time this inhibits its ability to communicate persuasively. States know this and so in conflicts they often honestly indicate whether they have important interests at stake in order to have more credibility on this later. Deterrence theory emphasizes having a reputation for resolve, but in fact one for honesty is more important. In deterrence theory, backing down is dangerous, conveying weakness. However, acquiescing often honestly shows that one's vital interests are not engaged, and thus invoking those interests in a future conflict is taken more seriously. A confrontation forces participants to assess what each values and how strongly, and a reputation for honesty is an important element in those assessments. One implication is that, contrary to suggestions in rational choice literature, signals in confrontations may not have to be costly to be effective.

Sartori offers a case study and a model. The case study is China's failure in the Korean War to convince the United States that it would intervene if United Nations forces approached the Yalu River. This is traced to China's not carrying out earlier threats to invade Taiwan. The United States downplayed Beijing's threats, not because it thought China's military capabilities were inadequate but because it felt China had no major interest at stake—UN occupation of North Korea would pose no threat—and because of the past bluffing over Taiwan. However, she cites U.S. scholars on what U.S. officials thought. Only one official from that time is quoted on this point (p. 37, n. 31). Thus, the case insufficiently illustrates her analysis. It might even fit the Press analysis—the United States did see China as weak militarily and as having no major interests at stake. These were serious misjudgments, which Press claims are less likely in major confrontations (p. 23), but at least they were the sort of judgments he asserts take place.

Sartori bolsters her argument in additional ways. She presents a “reputational theory of diplomacy,” modeled via a game theory format. The model's assumptions are that war is costly, that leaders care unevenly about issues and thus in confrontations they must assess how their opponents value specific issues, and that such situations and interactions are iterated. As leaders rationally update their conclusions in a crisis, evidence of past bluffing helps determine opponents' credibility, and so states typically avoid bluffing in their diplomacy.

Sartori also analyzes Correlates of War data on more than 1,300 crises in militarized international disputes. The concept of a bluff and the way its impact on a state's reputation declines over time are carefully operationalized. Being caught bluffing in the years before a confrontation makes a state about 40 percent more likely to be threatened with force and 20 percent more likely to fail in attempting deterrence (p. 113). Thus, “reputations for honesty—as distinct from reputations for resolve—affect the course of international disputes” (p. 107). Paralleling classic deterrence analysis about credibility, she concludes that to gain a reputation for honesty, acquiescing when appropriate must be supplemented; “the defender must actually be willing to fight more often if deterrence fails in order to obtain this greater credibility” (p. 110).

This finding that the balance of forces matters—deterrence failure is about 9 percent more likely when the challenger has a significant advantage—agrees with the Press analysis and other empirical studies. However, “[W]hen the defender has recently been using its diplomacy honestly, it is more likely to find deterrence success—regardless of whether it is strong or weak compared to its challenger” (p. 119). While Sartori does not directly say so, her analysis implies that American policy should be adjusted differently. Fighting to maintain credibility makes some sense when it honestly reflects a government's interests, but this can be overdone since there are cheaper ways to get the same result. However, backing down now is harmful later if it displays a bluff.

In his analysis, Press neglects the distinction between general and immediate deterrence. Leaders' concern about not looking weak has reflected a desire to build credibility in advance of crises and, in particular, to discourage challenges, that is, to reinforce general deterrence. This suggests an alternative explanation for the intriguing Press findings. In a crisis, the other side may have credibility, regardless of its past behavior, because of its dangerous, provocative steps that helped create the crisis. Those steps suggests that it thinks it has important interests at stake and the military ability to defend them; its having generating a dangerous situation shrinks the impact of any prior backing down and the utility for observers of relying on that to predict what it will do. Yet participants can still see backing down as dangerous in making their general deterrence less reliable. Thus, his case studies may be of limited value in assessing the impact of prior behavior in noncrisis deterrence.

The Sartori analysis does not fully consider the logic of the interdependence-of-commitments position. Being honest where one has a commitment but no interests at stake worth fighting for sounds sensible. But if commitments are believed to be interdependent, there is little difference in the interests at stake among them—the credibility of one's “major” commitments are at stake in challenges to its “minor” ones. The honest stance is, therefore, that all commitments are so important you will fight. What that builds is less a reputation for honesty than a history of consistency, making nondefense of any prior commitments damaging to future credibility.

These very interesting studies offer challenging new approaches to a classic deterrence problem. It is good that the problem continues attracting such able analysts. Each book deserves careful attention and will certainly stimulate further discussion.