This is the third and most refined of a series of books emerging from the monumental data-collection enterprise known as the Policy Agendas Project (http:www.policyagendas.org). The Policy Agendas Project and its companion Comparative Agendas Project have lead to a substantial body of important work that goes far beyond the three books themselves. The topic of this particular effort is the conflict between complexity and control in information processing about public policy in the U.S. over the last six decades (1947–2008).

The Politics of Information is divided into three parts. The first—“seek and ye shall find”—lays out the basic nature of the problem: Complexity leads to greater information and search, then discovers more complexity and new problems. Baumgartner and Jones identify the paradox of search in greater detail in both Chapters 1 and 2, arguing that “[g]athering information about complex problems, prioritizing those problems, and understanding the myriad repercussions that current policies may have on different elements of society … require that we organize diversity into the process of gathering information. … But implementing solutions and doing so efficiently requires clarity of organizational design and a clear mission (pp. 6–7).

Chapter 3 formalizes this concept by distinguishing between “expert search” and “entropic search”—with the latter more critical for identifying and prioritizing problems. To measure entropy, the authors employ Shannon’s H, which is an index of the spread of objects across categories, initially developed as a measure of market concentration. In policy agenda terms, entropy is the spread of topics across a large number of different categories, such as a policy considered by a variety of congressional committees. The term “entropy” has much to recommend its usage for describing the general phenomenon, particularly in statistical terms, but it carries with it some negative implications. It might have been useful for the authors to take another step in borrowing from psychology the general idea of the dispositional system versus the surveillance system in information processing. Dispositional search would correspond easily to expert search, while the term “surveillance” might seem less wasteful than entropic search.

The second part of the book deals with the corollary problem that “search implies that if a problem is found the probability of government action will increase” (p. 11).Thus, diversity in search inevitably leads to larger government, “[m]ore often than not” (p. 11). This is the heart of the book, and it is approached with an appealing mixture of history and data analysis. The descriptive history of the rise and fall of institutional information processing in the executive and legislative branches is entertaining for those of us who remember when President Ronald Reagan took Office of Management and Budget Director David Stockman “to the woodshed” for contradicting the “rosy scenario,” and crucial grounding for those who do not. The data (for example, Figure 4.2—GAO Reports by Time; p. 77) trace an arc in the production of information, with its peak in approximately 1978. Limitations on information production suggest limitations on government.

Chapter 5 addresses the nature of entropy as it relates to Congress over time. The authors note that the fact that the number of congressional committees has remained relatively stable since the Legislative Reorganization Act of 1946, while the range of issues considered has grown broader, must lead to greater complexity in congressional jurisdiction over time. Using measures of both committee entropy and topic entropy (and a combined index of the two), they find three phases. The first (postwar years) has a period of stability, followed by a massive increase in entropy in the second phase, from the late 1950s to the mid 1970s, which saw a dramatic rise in attention to new issues, coupled with a corresponding decrease in jurisdictional clarity. The third phase witnesses a temporary drop in entropy corresponding to the Gingrich centralization of the committee structure, followed by a return to the previous highs, but no further increase in entropy since the late 1970s. Thus, the historical expansion of information and complexity can under some circumstances be contained, and even reversed in the short run. The fundamental reason for such reversal is approached in earlier chapters and amounts to censorship of information—generally by defining some aspects of the problem as irrelevant to committee inquiry, or outside the range of congressional action.

Chapter 6 illustrates the growth of information and the growth of government activities in response to the “Great New-Issue Expansion” of the late 1950s to the mid-1970s period. This expansion is facilitated by the growth in information-processing capabilities of the executive and legislative branches, the increasing activity of interest groups, and the discovery or redefinition of problems that are thought to be within the government’s scope to solve. Baumgartner and Jones put the latter very effectively as “the translation of ‘conditions’ into ‘problems’” (p. 121). This results in the “thickening” of government as needs to “fix” existing areas of government policy emerge from oversight and entropic search, and the “broadening” of government as new issues are prioritized on the agenda. Chapter 7 extends the examination of new-issue expansion to identify an arc with its peak in 1981–83. However, as some measures indicated contraction, others suggested increasing complexity, for example, more hearings on subtopics per bill. They ultimately find that broadening—expansion of new issues—dominates over thickening with new issues, for example, taking part of the budget share of traditional ones. The dynamic of this “broadening” is best explained by leaders responding to shifting social problems.

The authors suggest that this expansion of new issues set the stage for a conservative backlash expressed in elections, legislative behavior and interest-group activity, all involving the treatment of information. Further, they suggest that such changes interacted with each other dynamically to produce changes in the institutions, as well as acting upon issues. They also note early-on President Reagan’s disregard for policy analysis and, perhaps, facts more generally. This perspective also apparently became more prevalent in Congress. They find that one way that did seem to slow government expansion was to cut down on the flow of information bringing along with it the problems of error accumulation and punctuated equilibrium described in their previous work, Agendas and Instability (Baumgartner and Jones Reference Baumgartner and Jones1993).

In Part III, the authors return to the general implications of information in government, stating bluntly, “Control the flow of information and you can control the growth of government” (p. 183). At the same time, they note that government growth accelerated during the George W. Bush era without increase in entropic information search. They attribute this to carefully controlled and concentrated search within selected areas, such as domestic security, defense, and education. Carefully controlled policy search inevitably excludes the more diverse entropic search necessary to successfully handle complex problems. Perhaps the policy disasters of the Bush administration, from 9/11 through Hurricane Katrina to the meltdown of the economic system, bear witness to the lack of entropic search. The two phases of the Iraq War parallel the difference between clearly defined goals (get Saddam), and the complexity of the post-Saddam period also illustrates the paradox of search.

In their conclusion, Baumgartner and Jones emphasize information and its censorship, identifying several examples of denial of information—even information that is available and easy to collect. They acknowledge that leadership decisions restricting the flow of information to government may be tempting in the short run. Yet they argue the complexity of current problems requires diverse viewpoints, redundancy in observations, and the ability to tolerate ambiguity. This is better achieved with an understanding that governments are “complex adaptive systems evolving in concert with even more complex environments” (p. 208).