Secrecy benefits and harms democracy. Secrecy is necessary for democracies to protect their security. At the same time, secrecy limits accountability, which is also necessary for democracy. If democracies were to conduct their foreign policy in the open, they would suffer transparency costs. If they did not keep secret the intelligence they have on their enemies, they would not be able to anticipate their actions on the battlefield or their ploys in negotiations. Neither would they be able to deceive their enemies. Finally, if they could not keep their own technological advances secret, they would contribute to the capabilities of their opponents (pp. 45–65). Yet once the necessity of secrecy is accepted, the executive can use secrecy to hide its incompetence or malfeasance. If citizens cannot inform themselves about the executive, how can they hold the executive accountable? If they cannot, voting and democracy become a farce and, if the lack of accountability leads to disaster, a tragedy. In either case, democracy is devalued by what is necessary to sustain it. This is the democratic secrecy dilemma, which Michael Colaresi explores in detail in this fine book.

The exploration is methodical and clear. Colaresi presents his argument and evidence in five parts. First, he establishes that the secrecy dilemma exists by critically reviewing the literature with the aid of some historical examples. Second, he discusses the consequences of executive abuse of secrecy. He highlights the possibility that such abuse may lead the public to doubt executive claims about the requirements of national security, which in turn may lead to underinvestment in this public good. He uses the case of France from the Dreyfus Affair to the outbreak of World War II to illustrate and support this claim. Third, the author offers oversight (freedom of information laws, press freedoms, and legislative investigations), particularly retrospective oversight, as a solution to the secrecy dilemma, while explaining why at best it can only reduce abuse of secrecy. Fourth, he argues that the focus on oversight explains better than do simpler theories the support for executive security policies, spending to carry them out, and their success. To support his argument, he analyzes “a new catalog of national security institutions in democracies from 1970 to 2006” (p. 18), as well as other data. (The details appear in a 27 page appendix, “Formal Models, Data, and Statistical Results.”) Fifth is a summary of his results and discussion of their implications.

Colaresi divides the authors of the literature into three categories: democratic irrelevantists, democratic doubters, and liberal electioneers. The irrelevantists (realists) claim that democracy and, even more so, democratic secrecy dilemmas do not matter because the executive makes foreign policy decisions based solely on the actions of foreign adversaries. Doing otherwise incurs unbearable costs. If this is true, Colaresi responds, then the existence of varying levels of oversight in democracies should have no consistent effect on support, spending, and success. The doubters claim that the transparency requirement of democracy imposes costs that lead to less support, spending, and success. If this is true, then higher levels of oversight should lead to lower levels of support, spending, and success. The electioneers claim that elections are sufficient for accountability, without the “bells and whistles” (p. 174) of oversight. If this is so, then, as with the irrelevantists but for a different reason, we should not find covariance between oversight and support, spending, and success.

The author’s analysis shows that varying levels of oversight are correlated with varying levels of support, spending, and success, contrary to the predictions of the irrelevantists. It shows also that more oversight does not necessarily lead to less support, spending, and success, contrary to the predictions of the doubters. Finally, his analysis shows that support, spending, and success is greater if there is oversight in addition to elections, contrary to the claims of the electioneers. The exception to these straightforward results is with spending, where Colaresi’s analysis is more complicated. Moderate levels of oversight mobilize public support for spending because oversight builds trust. Because of the mobilization, the executive can raise funds for security. Yet because oversight is moderate, some nonsecurity spending (corruption) may occur. If oversight increases to a higher level, then the executive has greater concern about the exposure of nonsecurity spending and so makes a greater effort to curtail it. Security spending thus decreases as oversight rises past a moderate level. Oversight both mobilizes support for spending and constrains it. Colaresi’s analysis of the data supports this claim (pp. 207, 214–15).

The analysis is more thorough and nuanced than a summary can suggest. Colaresi’s key contribution is to make the case for oversight in explaining the success of democracies in security matters, and the public support and spending necessary to bring it about. This oversight must be retrospective, however, since transparency costs make oversight concurrent with events too costly. The value of secrets decreases over time, he contends, and thus so do transparency costs. This makes retrospective oversight more likely. But how does retrospective oversight cause the effects that he finds?

Using an analogy from deterrence theory, Colaresi argues that “if the public can target items of value to a leader, then retrospective accountability is possible” (p. 141). He lists four items valuable to leaders: their power, their policies, their party, and their reputations. The first is most valuable but the hardest for retrospective oversight to touch, he acknowledges, given the lag time for information to become available. More generally, we might argue that a rational leader would prefer present and likely benefits over distant and less-likely harm, undermining retrospective oversight as a constraint on executive action. Also, we might question whether the public responds in the same way to all revelations produced by retrospective oversight. If there is variation, does this affect the calculations of leaders? What the executive fears and the public expects from retrospective oversight is critical in making Colaresi’s explanation work. More needs to be done to make this mechanism credible.

The author’s contention that the value of secrecy decreases over time is also worth considering more closely. This value does decrease, but not in every case at the same rate. Spending on new military technology might be revealed once the technology is in the field; intelligence sources and methods, and, hence, spending on intelligence, might need to be secret for much longer. Does one set of secrets affect support, spending, and success more than the other? Not all secrets are alike, either in themselves or, perhaps, in their political effects if revealed. Do revelations on cost overruns in defense spending have the same effect as revelations of criminal or moral transgressions by intelligence services?

Such questions and others arise, of course, only because of Colaresi’s work in opening up so carefully and thoroughly the question of the role of oversight in democracies. As the executive presses for more power to deal with emerging technologies and forms of violence, this question becomes more important.