1 As important as trust has been in realist theories, it is perhaps surprising that this is in many ways the first book to rigorously analyze the role of trust in international relations. Andrew Kydd provides a welcome explanation for how tragic spirals may occur that do not rest on cognitive errors, in contrast with previous analyses of the security dilemma. He also illustrates nicely how, despite the possibility of tragic spirals, most spiral behavior is likely to be non-tragic in nature. Many concepts central to realist arguments about conflict are clarified by this analysis.

2 The book essentially consists of three parts. After setting up the basic framework he uses in subsequent analyses, Kydd develops formal models to analyze the foundations of trust, the impact of trust on multilateral cooperation, and the relationship between previous interactions on current attempts at cooperation. He then applies the insights of each model to the Cold War.

3 The formal models provide much-needed clarity and precision to longstanding debates in international relations. The relationship between the models and previous approaches is clear. The results are explained in straightforward language, and Kydd takes care to provide strong intuition behind the impact of almost every parameter on equilibrium behavior. The key results are well illustrated with lucid graphs. However, in some respects the interpretation of the models does create some unnecessary confusion. For example, Kydd’s characterization of the argument represented in the models as comprising a new variant of realism, which he terms Bayesian realism, strikes this reviewer as unproductive. While defensive realists have historically asserted that all states are security seeking, they have for some reason insisted that these states do not trust that other states are security seeking. Kydd might instead have simply argued that this is a rather strange assumption and that defensive realism would benefit greatly from positing that some states may in fact be expansionist, and therefore states’ fears that their rivals may be expansionist are well-founded. Had he done so, he could easily argue that his analysis provides a logically consistent reformulation of defensive realism rather than adding Bayesian realism to the already crowded realist camp.

4 Kydd also uses the term “costly signaling” in a rather different manner than most of the formal literature. In Kydd’s setup, cooperation provides a signal of a state’s trustworthiness. The relative value of that signal is a function of the degree to which states value payoffs from the first round of a game with two stages. Yet Kydd often refers to the parameter that measures the degree to which states value first-round payoffs as if this parameter is itself a costly signal. Typically, costly signals are actions taken by states primarily in order to change the beliefs of the opponent. In Kydd’s model, cooperating in the first round serves to reassure the other player that one is trustworthy, but it also carries a very real prospect of delivering a non-trivial reward in the short run should the other state cooperate. Indeed, states are more likely to cooperate in the first round when they expect that this cooperation will be reciprocated. True, cooperation is possible at lower levels of trust than would be true if the game was only played once, and the desire to change the other states’ beliefs about one’s type drives this. Yet, it is nonetheless possible that the action Kydd labels a costly signal will not impose any cost at all. Cooperation is more of a risky action than a costly action. The difference may be subtle but the value of formal models is the ability to make such fine distinctions transparent. There are certainly many obvious parallels between the logic of costly signaling as it is conventionally understood and the behavior Kydd describes. But there are dissimilarities as well, and it is left to the reader to realize this.

5 At times, the links between the formal models and historical applications are a bit loose. Trust is a dyadic characteristic in two of the three models. A state that behaves aggressively toward one state but cooperatively toward another would presumably engender very different beliefs in those two states. Yet Kydd often speaks of the United States as being understood to be trustworthy in some generic sense, rarely discussing the possibility that the UK and the USSR might have had good reason to have very different views of whether the US could be trusted. Likewise, Russian actions toward other states are often interpreted as indications that the US could not trust Russia to cooperate with the US elsewhere. This may be reasonable and may in fact accurately describe the way US policy makers actually interpreted Russian actions. But it is difficult to say this is consistent with the models, which are incapable of speaking to such dynamics. In some ways the models provide a strange reading of the cases. Stronger states are more likely to be expansionist, irrespective of their actual value for taking advantage of a cooperative rival. Given the unprecedented advantage in military capabilities enjoyed by the US in the immediate aftermath of World War II, the model would seem to be likely to predict that the US would be expansionist. States are more likely to have accurate perceptions of their rivals when those states are relatively transparent. Kydd presents some evidence that the USSR did not believe the US to be security seeking at the dawn of the Cold War. It is therefore surprising that Kydd argues that the historical record suggests the US was trustworthy and the USSR was not, and that this is explained by his model. Further, because Kydd presents the expectations of his models as being essentially the same as those of the well-developed traditionalist interpretation of the Cold War, it is not clear that the historical analysis presented would have suffered much if the formal models were removed from the book. That is not to suggest that the formal models are not valuable for other reasons. They certainly are. Nor is this to suggest that Kydd’s analysis of the Cold War is not a valuable contribution to that literature. Rather, this reviewer found that the combination of the two methodological approaches failed to improve one another the way they often do in the best multi-method research.

6 All things considered, Trust and Mistrust in International Relations provides an important contribution both to the analysis of the security dilemma and to the interpretation of the origins and resolution of the Cold War. One might wish the two halves fit together more seamlessly, and one might at times question the interpretations of the models and of the historical evidence. Yet, both approaches provide strong challenges to existing arguments. The relative plausibility of tragic versus non-tragic spirals, the role of trustworthiness in mitigating the relationship between hegemony and multilateral cooperation, and the conditions under which reassuring gestures are likely to facilitate future cooperation are spelled out much more precisely than in the past. The formal analysis is also quite accessible relative to many other recent works in international relations. Kydd’s strong defense of the traditionalist interpretation of the origins of the Cold War may also reinvigorate that debate. This book is likely to inform future research in many ways.

7 Philip Arena is an Assistant Professor in the Department of Political Science at the State University of New York at Buffalo.