“Procedures are politics.” P. 201.
Rachel Potter’s book, Bending the Rules: Procedural Politicking in the Bureaucracy, examines agencies’ choices in structuring their rulemaking processes. One might assume that each agency follows a consistent procedure across all its rulemakings, but Prof. Potter suggests otherwise. For instance, comment periods often vary greatly, p. 119, as do choices regarding to public participation, see, pp. 74-78, accord, pp. 62-63. Rather, agencies design the rulemaking process strategically to minimize the risk that the President, Congress, or the Judiciary will reject the particular rule that agency adopts. Potter seeks to describe the reality of the rulemaking process, rather than make normative judgments about it.
According to Potter, scholars have theorized that agencies seek to minimize the risk presidential, congressional, and judicial rejection of a rule by modifying the proposed rule’s substance to accommodate likely presidential and congressional policy preferences as well as potential judicial skepticism. PP. 55, 61-62, 82. But, she argues, agencies can deploy a second gambit — strategically employing procedural devices that discourage or deter review by any of the three institutions. PP. 55, 62, 82.
I. The Means of Procedural Politicking
Prof. Potter enumerates several procedural devices at agencies’ disposal, sorting them into three categories (or “tool kits”): the “writing,” “consultation,” and “timing” tool kits. PP. 69-70. The tools in the writing tool kit consist of “policy framing,” establishing analytical assumptions, and drafting choices, in terms of both readability of the rule abstract and the length of rule preambles. PP. 70-74.
The consultation tool kit involves procedural choices that govern the extent and manner in which the agency invites “open” consultation. Open consultation contrasts with the more shadowy ex parte consultation that often occurs as the agency develops its regulatory proposal internally. PP. 32-33, 77. This tool kit encompasses decisions regarding the length and timing of the formal comment period, as well as decisions regarding whether to use an array of consultation and consultation-avoidance devices, including advance notices of proposed rulemaking (“ANPRM”), negotiated rulemaking (“reg-neg”), advisory committee meetings, outreach fora, promulgating pre-comment interim rules based on a finding of good cause. PP. 74-78. 
The timing tool kit includes decisions regarding promulgating and implementing the final rule. Procedural considerations at this stage involve the timing of the rules publication, determining the rule’s effective date, and deciding upon the advisability of delaying the rule’s implementation. PP. 78-81.
Prof. Potter acknowledges the “sincere” justifications for making certain procedural choices. PP. 97, 145, 149. The length of preambles may reflect the obligations to elucidate an agency’s reasoning imposed by exacting judicial review. See, pp. 51-52. The relative length of time the process takes may relate to differences in the breadth of impact and complexity of rules. P. 143. Prof. Potter acknowledges the difficulty of teasing out the impact of “sincere” and “strategic” motives using statistical analysis. P. 145.
Prof. Potter makes a strong case that agencies exercise discretion with regard to the choice of rules on a case-by-case basis. While she bases her conclusions mostly upon the results of her sophisticated statistical study, she also shows how the FDA employed many of these techniques in its eight-year calorie labelling rulemaking, pp. 169-79.
I have separated my contribution to this symposium into two posts, each with a distinct goal. This first post explores the ambiguity of Prof. Potter’s findings with respect to two related decisions agencies make, regarding the rule abstract readability and preamble length. Do agency choices with regard to these matters reflect “sincere” or “strategic” agency calculations? My second post will discuss the broader normative implications of Prof. Potter’s conclusion that agencies employ procedural devices in a strategic manner. I will suggest that use of some procedural devices in a strategic manner threaten the very legitimacy of agency rulemaking.
II. Abstracts and Preambles: Means to Build a Record or a Devices to Obscure?
Prof. Potter theorizes that when agencies expect potential opposition from the President, Congress, or the Judiciary, they make the abstracts of their proposed rules less readable, to obscure the rule’s import. Similarly, agencies tend to lengthen the preambles to proposed rules when they expect a skeptical reaction from the President, Congress, or the Judiciary. Lengthier preambles can serve two purposes. They allow the agency to “demonstrate to external audiences that [the agency] has done its homework and thought through all the implications of its proposed policy action.” P. 97. But lengthy preambles may also deter OMB analysts, congressional staffers, and judges from reading them closely. PP. 89, 97-98. Potter observes that more detailed preambles are not necessarily more informative. P. 97. Empirically testing the relationship between preambles’ length and their quality in terms of enhancing understanding of the agency’s proposal would be difficult, and Prof. Potter relies on anecdotal support for her observation.
An alternative thesis, positing more “sincere” agency behavior, focuses on the role preambles play in judicial review. Judicial review focuses on agencies’ failure to address significant comments and their failure to elucidate their reasoning in rejecting comments. See, p. 134. This effect might be more muted with regard to notices of proposed rulemaking as opposed to notices of final rulemaking. Nevertheless, the more searching the judicial review expected, the greater the incentive to provide detailed explanations of regulatory choices by means of the preamble.
So are preambles intended to inform or obscure? Which purpose predominates, the “sincere” or the “strategic” one? While providing detail to inform the public, the political branches of government, and the courts seems laudable, useless detail designed to impede understanding seems fundamentally anti-democratic.
A. Professor Potter’s Statistical Analysis and Results
Prof. Potter compares preamble length and abstract readability when each principal institution of the federal government is aligned and unaligned with the agency. The President is “aligned” with the agency when the President’s political party is ideologically-supportive of the agency, and “unaligned” when the President’s party customarily ideologically opposes the agency. P. 101. Congress is “aligned” with the agency when both houses of Congress are controlled by an ideologically supportive party. Congress is “unaligned” when at least one house is controlled by a party whose legislators are both generally ideologically hostile to the agency and “cohesive.” P. 101-102. Courts are aligned when they decide few cases in the agency’s area, i.e., when they are inactive, and unaligned when they decide numerous cases in the agency’s area. P. 102. Prof. Potter examines the change in preamble length and readability in comparing notices of proposed rulemaking when there is an aligned President and an unaligned President, an aligned Congress and an unaligned Congress, and inactive courts and active courts. The three comparisons are laid out in the chart below.
Prof. Potter finds preamble length higher and abstract readability lower when courts are active in the subject area dealt with by the agency. Anticipated congressional opposition (i.e., an unaligned Congress) results in a statistically significant increase in preamble length, but a statistically-insignificant decrease in summary readability.
Prof. Potter then adds an additional layer of complexity to her study. She explains that interest groups often alert OMB or Congress to controversial regulatory proposals and frequently initiate litigation challenging agency proposals. Therefore she performs each of the three analyses described above in two different scenarios, when interest group opposition is expected and when it is not. The six comparisons are laid out in the chart below.
When moving from an aligned President to unaligned President, if interest groups support the agency, there is little difference in preamble length. Under the same condition, when interest groups oppose the agency, there is a statistically-significant increase in preamble length. But there are no statistically significant results with regard to the readability of rule abstracts.
With respect to moving from an aligned Congress to an unaligned Congress, if interest groups support the agency there is little difference in preamble length. But when interest groups oppose the agency, preambles are substantially longer when Congress is not aligned with the agency. Again there is no statistically significant difference in terms of abstract readability.
When moving from an inactive judiciary to an active one, there is a significant increase in preamble length, even when groups are supportive, and an even greater increase in preamble length when groups are not. In addition, there is a statistically-significant decrease in abstract readability when groups are supportive and a greater decrease when groups are unsupportive. However, while statistically significant, Prof. Potter characterizes the effect as modest.
The results are graphically displayed below. The comparisons in red type, 5/6, 7/8, 9/10, and 11/12 produced a statistically-significant increase in preamble length. The only comparison in italics, 11/12 represents a statistically-significant decrease in preamble readability.
Both preamble length and readability should be affected if an agency’s goal is simply to obscure the significance of its actions. Indeed, decreasing abstract readability would seem to be the easier, less time-consuming route to take. Prof. Potter offers reasons why her results support her thesis with respect to preamble length, but are decided mixed and more modest with regard to abstract readability. These reasons include deficiencies in her readability metrics (their “inherent noisiness”), disadvantages to agencies of decreasing abstract readability (while they obscure the rule from opponents they also obscure it from supporters), and the success of efforts to force agencies to use plain language, see, pp. 109-10.
B. Can “Sincere” Motivations Explain Prof. Potter’s Results?
What results would be expected under the alternative theory positing “sincere” agency motivations, i.e., that agencies elucidate their reasoning in detail to satisfy exacting judicial review? First, we would expect an increase in preamble length when courts are active in an area, both when interest groups are supportive and when they are not. In either scenario, the agency will need a detailed rulemaking record to survive judicial review. Prof. Potter finds just such increases in preamble length.
Second, if the length of preambles explaining proposed rules are driven primarily by the necessities of surviving judicial review, one would expect an increase in judicial activity to have a greater impact upon preamble length than potential congressional or presidential opposition. Prof. Potter’s figures appear to support that conclusion as well.
Third, we would not expect corresponding decreases in the readability of abstracts, which serve little function in terms of judicial review and would not assist in elucidating the agency’s reasoning. Though there is a statically-significant decrease in abstract readability when moving from inactive courts to active courts, Prof. Potter herself characterizes the decrease as modest. This statistically-significant decrease in abstract readability contradicts the alternative “sincere agency” hypothesis and supports Prof. Potter’s. However, the comparatively minor nature of the change may well support the “sincere agency” hypothesis, namely that agencies do not use notices of proposed rulemaking to confuse, but to enhance the foundation of the agency’s rule in order to survive judicial review.
Notably, decreases in abstract readability are not even statistically significant when the agency faces an unaligned President or an unaligned Congress, even though increases in preamble length in those circumstances are significant when unsympathetic interest groups are dominant. This would suggest that abstract obscurity is a device solely used to avoid judicial review. Yet abstract obscurity would seem a most inefficacious strategy for avoiding demanding judicial review. An appellate court does not choose the rules it will review. Moreover, the arguments regarding a rule’s deficiencies are honed for the reviewing court by the parties challenging the rule. Thus, it seems unlikely that a court would be put off from engaging in rigorous judicial review by even the most incomprehensible of abstracts.
If preamble length is primarily related to the requirements of justifying the agency’s decisions to courts, we might question why we should find increases in preamble length when moving from an aligned President to and unaligned President, and when moving from an aligned Congress to an unaligned Congress. Prof. Potter finds both results, albeit only when interest groups are in opposition. I will offer one potential explanation. Some procedural politicking devices can be segmented, they can be used for OIRA review or potential congressional review, but then abandoned for judicial review. But, presumably, manipulation of preamble length cannot be segmented in such a way. The same preamble is available during OIRA review, potential congressional review, and judicial review. This means that if agencies write longer preambles primarily for judicial review, the preamble will also be longer during the OIRA review and potential congressional review phases.
Given the potential overlap between periods in which Congress is unaligned and periods in which courts are active in the agency’s area, it hardly seems surprising that preambles primarily used for judicial review are longer even under the unaligned Congress condition (albeit the increase in length in the unaligned Congress condition is much weaker). But to confirm that hypothesis, we would need to determine the increase in preamble length when there is congressional opposition and active courts, and compare it with any increase in preamble length when there is congressional opposition but inactive courts. If there is an increase in preamble length in the former, but either a lesser or no increase in preamble length in the latter, such a result would suggest that the length of preambles is more attributable to potential judicial scrutiny rather than to potential congressional opposition. In other words, the statistically-significant relationship between a movement from an aligned to an unaligned Congress (when interest groups are active) and longer preambles would be spurious, because it would really be an artifact of the increase in preamble length due to concerns about intensive judicial scrutiny.
In short, Professor Potter’s data is ambiguous with respect to whether agency choices regarding preamble length and abstract readability are sincere or strategic (or some combination of the two).
While the data with regard to preamble length and abstract readability are not decisive, Prof. Potter presents statistical analysis of agency choice of procedural devices from other tool kits. Those analysis do seem to provide convincing support for the proposition that agencies sometimes do make the strategic choice of procedural devices that Prof. Potter posits. He statistical analyses make a valuable contribution to the field.
As noted earlier, in a later contribution to this symposium I will discuss the broader normative implications of Prof. Potter’s theory that agencies employ procedural devices strategically.
Bernard Bell is a Professor of Law and Herbert Hannoch Scholar at Rutgers Law School.
This piece is part one of a two-part series. Part two is available here. This post is part of a symposium reviewing Bending the Rules: Procedural Politicking in the Bureaucracy, a new book by Dr. Rachel A. Potter, Assistant Professor of Politics at the University of Virginia. All of the posts can be read here.
 Citations to Bending the Rules in this post are indicated by “P.” or “PP.”
 Granted, she offers a few normative observations in the book’s concluding chapter. PP. 189-93, 200-201.
 She illustrates the first by discussing the FDA’s framing of it mid-1990’s proposed tobacco regulation. The agency consciously framed the issue as one of children’s health rather a broader public health issue primarily manifesting its effects in chronic adult smokers. P. 71. She illustrates agency choice of contestable assumptions by discussing the Department of Transportation’s Hours of Service rule. In particular, the agency assumed that drivers would use the mandatory breaks to sleep rather than engage in other activities. PP. 72-73.
 There are other options for forgoing notice and comment rules as well, see 5 U.S.C. §553.
 As she explains: “Evidence of strategic behavior is notoriously illusive.” P. 145. For example, she observes, “[w]ith respect to regulatory pacing, it is not always easy to tell whether an agency is taking a longer time because it is slow-rolling the current political regime or whether more innocuous factors such as an authentic desire for information and consensus are producing delays.” P. 145. And, say says: “Of course, empirically detangling strategic delay from sincere delay is not perfectly straightforward . . .” P. 149.
 Her data set consists of approximately 11,000 rules proposed from 1995 through 2014. P. 91.
 She quotes Richard Parker: “Overworked journalists, OMB economists, and Hill staff are not going to slog through forty to fifty Federal Register-length pages of technical jargon for each major rule in an effort to tease out the agency rationale for its action.” Richard Parker, The Empirical Roots of the Regulatory Reform Movement: A Critical Appraisal, 58 ADM. L. REV. 359, 398 (2006).
 See, JEFFREY S. LUBBERS, A GUIDE TO FEDERAL AGENCY RULEMAKING 385-389, 493, 500-503 (6th ed. 2018).
 Peter L. Strauss, From Expertise to Politics: The Transformation of American Rulemaking, 31 WAKE FOREST L. REV. 745, 770 (1996)(“[m]any attribute much of the ossification of rulemaking at the agency level to th[e] intensification of judicial review, leading agencies to practice “defensive medicine” against the quite exquisite assessments of factual support and judgment “hard look” review can often entail.”)
 Prof. Potter measures the change in abstract readability and preamble length when moving from conditions 1 to 2, conditions 3 to 4, and conditions 5 to 6.
 For this portion of her study Prof. Potter used a reduced data set of approximately 2900 proposed rules. P. 96.
 The difference between 1 and 2 measures the change from the condition of an aligned President to an unaligned President when interest groups are in support. The difference between 7 and 8 measures the change from the condition of an aligned President to an unaligned President when interest groups are in oppositions. Prof. Potter measures the differences between the following conditions: 1 and 2, 3 and 4, 5 and 6, 7 and 8, 9 and 10, and 11 and 12.
 See pages 95-96 for a description of the efforts to require agencies to use plain language.
 See, p. 186 (“[n]ot every tool is used to forestall each principal at every stage of the process; rather, agencies employ different tools to target specific principals throughout the rulemaking process”).
 Timing the comment period for congressional recesses, pp. 124-26, is a device that impacts only congressional consideration. Similarly, waiting out a particular OIRA head until a more favorable one comes along, p. 147, impacts only the OMB review stage of the rulemaking process.
 While I discuss only unaligned Congresses, the same observations would apply to the increase in preamble length when there is an unaligned Presidential Administration.