Working with Statutes

Introduction

In its 2024 decision overruling the decades-old Chevron^[1] doctrine directing judges to accept an agency’s reasonable interpretation of ambiguous statutory language,^[2] the Supreme Court declared: “[A]gencies have no special competence in resolving statutory ambiguities. Courts do.”^[3] In this Article, we reveal this statement to be a profoundly blinkered assertion of judicial hubris. We present findings from a qualitative empirical study of how agencies work with the statutes that give them authority to act, showing that interpreting statutes to ensure that they have effect in the world is at the heart of agency competence. Loper Bright v. Raimondo,^[4] much like the Court’s recent supercharging of the “major questions doctrine,”^[5] rests on two flawed assumptions. The first separates legal interpretation and policymaking into two distinct enterprises—the former belonging to courts, the latter to political actors. The second posits that only Congress, not agencies, should make the important decisions that policymaking requires, on the theory that bureaucrats lack accountability or undermine democracy.^[6]

But these abstractions present a mirage with little grounding in reality and slim normative appeal to boot. Our study of the administrative state suggests not only that the line between law and policy cannot be consistently and constructively drawn, but also that values-driven interpretation and policymaking by agencies are central to our democratic system. Administrative agencies bring statutes to life; their policymaking represents our primary conduit for ensuring that the decisions of democratically elected officials are effectuated in practice.

This basic aspect of the administrative state should be obvious. Indeed, the available evidence suggests that statute drafters, aware of this reality, address agencies with their products.^[7] But even though a nascent scholarly literature has begun to emphasize the centrality of agencies to American democracy,^[8] neither doctrine nor scholarship has paid sufficient attention to the everyday realities of how agencies themselves work with statutes.

To address this deficiency, we embarked on our study, which revolved around extensive interviews with dozens of officials across the administrative state. In so doing, we found inspiration in Jerry Mashaw’s recognition, two decades ago, that “[t]here are persuasive grounds for believing that legitimate techniques and standards for agency statutory interpretation diverge sharply from the legitimate techniques and standards for judicial statutory interpretation.”^[9] But while Mashaw urged scholars to investigate those differences,^[10] much work in this area remains theoretical and adopts prescriptive frameworks developed for courts as a baseline against which to evaluate, or imagine, agency action.^[11] The relatively few empirically oriented studies that investigate agency statutory interpretation also typically accept judicial doctrines and interpretive tenets as the points of departure.^[12] But judicial interpretation theories tend to be prescriptive rather than empirical.^[13] A juricentric orientation also implies that courts sit at the heart of the democratic administration of the laws—a highly debatable proposition. The dominance of the juricentric model has meant that courts and commentators continue to opine on agency decisionmaking without confronting the realities of administration; it has also arguably contributed to judges’ blithe assertions that determining the meaning of statutes amounts to a legal question belonging primarily, even exclusively, to courts.^[14]

We designed our study to escape the juricentric model by simply asking agency officials to explain in their own words how they work with the statutory frameworks that give them authority and responsibility to act.^[15] Accordingly, in this Article, we treat agency practice as a social phenomenon worth studying in its own right; without direct input from the people who perform it, we cannot imagine all of the factors, constraints, and considerations that shape any practice. By casting a light on agencies’ everyday work with statutes, and agency officials’ own internal points of view, we join a small but growing group of scholars adding vital new perspective to a literature dominated by courts, doctrines, and judicial voices.^[16]

Through our study, we show that agencies’ work with statutes is more complex and multidimensional than courts’.^[17] Our interviews ultimately bring to light the pluralistic, deliberative ways agencies construct the regimes elected officials have called for—how, as one interviewee put it, agencies “build [the] thing” their statutes demand.^[18] Specifically, our empirical research supports an understanding of agencies as custodians of the statutes they administer.^[19] Regulatory statutes authorize, mandate, and constrain agency action. Agencies, in turn, construct regimes to effectuate their statutes, maintaining them over time even as the enacting political coalition fades away.

Other scholars, such as Mashaw, have described agencies as statutory custodians. Our study sheds light on what this custodial role entails in reality. We find that agencies’ self-conception and behavior can be constructively described as adhering to a pair of concepts from private law: the duty of loyalty and the duty of care. As custodians of a statute, agencies understand themselves to have a general duty to act in the statute’s interest: They are loyal to their statutory regimes, rather than to an imagined unitary Congress or public. And agencies act on their duty of care by bringing together the perspectives and information offered by multiple actors in and out of government to produce reasonable, workable policies. These concepts, though stylized, help capture the aspirations and actions our interviewees discussed.

Our conclusions resonate with recent normative elaborations of a “‘fiduciary model’ in administrative law,” which proposes that we understand statute-enacting electoral coalitions to “entrust [agencies] with regulatory discretion over an area of public concern,” treating agencies as bound by “a duty of fidelity to their statutory mandates, and duties of care and loyalty to their statutory beneficiaries,” while also subjecting agencies to monitoring.^[20] For instance, Evan Criddle has argued that agencies’ “fidelity to their legal mandates and the public welfare” offers a form of accountability and responsibility independent of “electoral authorization” or majoritarianism.^[21] Others have argued that the Constitution itself imposes a fiduciary imperative on the Executive.^[22] Most of this scholarship makes a case that agencies should be understood as having fiduciary duties based on constitutional, normative, or public law principles. We take a more empirical tack. Our interviews indicate that many agency personnel describe their work in ways that comport with the fiduciary model without using that terminology. Numerous agency personnel share a role morality that befits a public fiduciary, giving an empirical basis for public fiduciary theory’s normative arguments.

From our empirical findings, we conclude that democratic governance depends on the work agencies do. Our study offers a rejoinder to abstract battles over constitutional principle—abstractions that obscure political conflicts over the very existence and extent of regulation.^[23] By looking at agency practices, we explore how government institutions can in fact serve democratic values, giving an empirical foothold to normative ideals like representation, accountability, and the public interest. In other words, our work helps show that administrative agencies and the work they do with statutes are democratically vital and necessary. Recognizing these features seems especially important today, in the face of tectonic doctrinal and political changes that may dramatically transform how agencies function.

We elaborate our findings in Parts I and II of this Article and offer our normative conclusions in Part III. In Part I, we show that officials’ main loyalty is not so much to Congress or the President as to their statutory regimes. This loyalty entails attention to statutory text. Not everyone involved in producing a policy combs through the statutory language, but someone does: Agencies’ divisions of labor allow statutory text to play a central role in the policymaking process without being central to every participant. Taking text seriously, however, does not make agencies textualist. The practices our respondents described do not comport with the tenets of textualism as laid out in judicial interpretive theories, nor did interviewees claim adherence to those tenets.^[24] A fair number were not sure what textualism even was. Rather, a wide world beyond the text informed their dealings with statutes.

One central factor we found shaping agencies’ work with statutes was the pursuit of agency mission—a factor we identify and define because it emerged so consistently in our interviews. Mission is a cultural factor in agency work, an enduring yet evolving influence over the agency’s role that channels and shapes discretionary decisionmaking. But while mission has a goal-infused quality, being guided by mission does not mean that agencies are purposivist as understood within juricentric theory.^[25] Purposivism (like textualism) is an “archeological” approach to interpretation that assumes a statute’s “meaning . . . is set in stone on the date of its enactment, and it is the interpreter’s task to uncover and reconstruct that original meaning.”^[26] Agencies, in contrast, seek to ensure that a statute remains relevant and efficacious through social, legal, and political changes, sometimes over lengthy periods.^[27] The sense of mission we found was rooted in the statute but not coterminous with its text or even its enacting context—it evolved over time.

We also found that the concept of mission helps structure the interagency interactions and conflicts that form an integral part of the policymaking process. An agency, after all, cannot make policy on its own: Agencies are not just structured by internal divisions of labor, but subject to external ones as well. More than one agency often has a say in any given process, and agencies with differing missions will bring differing normative considerations and practical concerns to bear. This pluralistic phenomenon may arise by congressional design and create productive counterweights within the policymaking process, but it can also impede agencies’ custodial abilities by enabling one agency’s mission to run roughshod over others’. Text and mission ultimately focus officials on the interests of the statutory regime itself—not quite the same thing as the Congress that enacted the statute, nor the Congress that currently oversees the agency’s work, nor a presidential administration past or present.

In Part II, we show how agencies perform their duty of loyalty by exercising a duty of care, which they understand as making statutes workable. This pursuit of workability requires developing policies that will be effective and implementable.^[28] But it also requires incorporating pluralistic demands into decisionmaking: responding to the emergence of new technologies; absorbing popular and interest-group demands; formulating policy that will survive judicial review so that it can have practical effect; taking input from members of Congress and congressional committees; and reconciling the aims of a presidential administration with longer term agency missions and statutory directives.

In Part III, we argue that the forms of agency decisionmaking we document perform vital democratic work in our system of government. We then spell out the institutional, political, and doctrinal implications of the picture we see. Agencies frequently serve as sites for democratic contestation, making them as central to channeling popular sentiments and promoting social welfare as the political bodies that create and oversee them. This work, our research shows, is pluralistic and contestatory; it takes into account policies’ consequences and workability; it combines expertise and experience with political influence and public pressures; and it is grounded in authority granted by elected officials. But because agencies retain responsibility for their regulatory regimes well beyond the tenure of any statute-enacting coalition, their statutory work also helps connect successive political regimes and real-world developments, balancing interests in stability and change. This balancing gives rise to tensions, even potentially irreconcilable differences, between views of what a statute should be—between historical and contemporary understandings. These tensions are not just pesky incidentals; they are a central fact of democratic governance, which requires respecting the enactments of past governing coalitions as well as the needs and interests of the people statutes govern in the present.^[29] Agency work with statutes thus offers one essential way our system of government addresses the dead-hand problem presented by legal enactments.

Agency work with statutes, of course, is neither perfect nor perfectible; it has biases, blind spots, and limitations, and many aspects of the administrative state may well be in need of reform. But we think that our study helps to substantiate that, from a comparative institutional perspective, no other United States government institution equals agencies’ capacity for the effectuation of statutes. The Supreme Court’s evolving administrative law doctrines, though cloaked in the language of balanced powers, actually channel decisions about both general delegation and specific policy to the courts.^[30] But, lacking capacity, courts are not up to the job. Agencies’ ongoing, pluralistic practices make them the superior institutional choice for the foundational aspects of democratic governance the Court increasingly arrogates to itself. All of this suggests that, while judicial doctrine and political discourse often denigrate and undermine agency work, the capabilities we bring to light should be recognized and strengthened, not dismissed, by future courts and Congresses. Our research helps show just how anti-democratic the administrative law of the Roberts Court turns out to be.

Because our claims diverge from the Supreme Court’s current direction, not to mention the motivations behind the second Trump Administration’s assault on the state’s very existence, it is all the more vital to present the picture we do. We have no illusions about our ability to persuade the current Court to unwind its multifarious assertions of judicial dominance over the administrative state. Some readers might even take our findings to justify the Court’s anti-administrative impulses as articulated in the unitary executive theory, the major questions doctrine, and nondelegation concerns.^[31] In fact, though, our study offers a picture of a state capable of being democratic in an ongoing, day-to-day way. We thus illuminate the realities of the institutional context the Court’s doctrinal trends target but distort. We underscore the political stakes of these supposedly purely legal interventions—political stakes that Congress, presidential administrations, and lower courts, as well as the broader public, should acknowledge. And we offer a picture of the state that should inform the ongoing struggle over how to make government work and, increasingly, over why it should exist at all.

We reach our conclusions through an interpretive empirical process based in the qualitative social sciences—a process worth explaining in more detail before we turn to our findings.^[32] The primary data for this project comprises thirty-nine semi-structured interviews with political appointees and career civil servants from eleven agencies, including smaller and larger, older and newer, primarily benefits-managing and primarily conduct-regulating components.^[33] We asked respondents to discuss how the process of converting a statute into a regulatory regime worked in their agency, both in general terms and in the context of specific initiatives.^[34] Often with a focus on particular projects in which a respondent participated, we asked about the tools their agencies used to make regulations; about the role of Congress, the President, other agencies, and states in that process; and about the influence of courts, judicial doctrine, and the specter of litigation. We mostly asked respondents about their experiences rather than their opinions, using open-ended phrasings to encourage participants to form answers in their own terms, instead of agreeing or disagreeing with ours. At the end of each interview, we then posed several questions to probe respondents’ opinions and understandings of the place of their own work in the larger field of statutory interpretation.^[35]

Although our interviewees differed along many axes—the roles they occupied, the kinds of agencies they worked for, the presidential administrations they served under, the subject matter they dealt with—their descriptions of their work also converged in significant ways. In this Article, we focus on those convergences, presenting a rarely seen agency perspective on statutory work. At the same time, we have not limited ourselves to our interviewees’ words, and we do not merely report what they told us. Rather, we place our interview data in the context of statutory interpretation scholarship, doctrinal developments, and democratic theory. We also evaluate our data’s implications for democratic governance—a phenomenon we did not ask our interviewees to address.

As our interviewees demonstrated, agency work with statutes often involves influences, practices, restrictions, and considerations that doctrinal and theoretical accounts generally do not recognize, much less analyze. Such internal workings are difficult to identify in publicly visible end products like Federal Register notices. When we reviewed the rulemaking publications of many of the specific policies our interviewees discussed, we found little sign of the intensive production process our respondents revealed to us.^[36] That is to be expected. The Federal Register presents the public with the outcomes of agency processes and the reasons that best support those results—it does not purport to provide a policy’s biography. But to understand what administrative agencies actually do within our system requires this very knowledge.

Like any method, our qualitative interview study can only provide a partial picture. We talked to a limited number of people in a limited number of agencies, and we do not purport to report on the practices and views of all two million employees of the federal bureaucracy. We talked only to people who were interested in talking to us, so our sample likely overrepresents certain characteristics like self-reflection, a desire to publicize one’s work, or congeniality to academic research. Respondents may also have wanted, consciously or not, to present themselves in a flattering light, skewing the information they gave us toward the positive. We sought to mitigate such skewing by focusing our questions on facts and experiences, rather than evaluations or opinions, and by finding areas where differently situated interviewees largely overlapped in their descriptions of internal agency workings.

Whatever the limitations of this qualitative methodology, it is worth noting that much research on statutory interpretation already suffers from biases analogous to, and arguably more severe than, those of our interviews.^[37] That research takes as its primary data judicial opinions: curated vehicles for justifying court decisions, written for public consumption.^[38] Yet, scholars have found judicial opinions worth studying because they do reveal some things about how judges interpret statutes. Our research provides an essential corrective to the biases of the juricentric approach by revealing some of the everyday practices of agencies’ work.^[39] It thus presents a perspective central to the effectuation of federal law that nonetheless remains mostly absent from scholarly discussions about it: the perspective of agency personnel.^[40]

I. The Duty of Loyalty: Serving the Statute

Agencies implement and administer the statutes that give them authority to act. Whom, or what, do they serve in that process? A transmission belt model suggests that agencies serve Congress;^[41] a presidentialist approach suggests the President.^[42] Much administrative-law and separation-of-powers scholarship contends at length over which of the two constitutional branches agencies should and do serve. Our study indicates, instead, that agencies serve something else both obvious and overlooked: They bear a duty of loyalty to statutes themselves. Both Congress and the President contribute to enacting statutes, but in our interviews, statutes qua statutes transcended the two branches. Recognizing the centrality of the statutes themselves helps reveal agencies’ crucial role within our democratic system of government—a role increasingly under pressure from Supreme Court doctrine as well as presidential politics.^[43]

Scholarly literature often discusses agencies’ relations with the legislature as though Congress were an enduring body that remained self-same over time. But most agencies, most of the time, work with statutes produced by past Congresses, even as their work is overseen and enabled by the sitting Congress. An agency could simply transfer loyalty from one Congress to the next or retain loyalty to the enacting Congress. Our interviewees painted a more complicated picture, however, of loyalty to a statutory regime—a regime constructed not just at the moment of enactment, but over time through statutory amendments, agency action, and doctrinal and political developments.^[44] “Obviously, you want to get along with Congress,” one interviewee observed, “but at the same time you’re trying to be . . . a good steward to the program and adhere to the statute.”^[45]

This obligation of “stewardship” emerged forcefully in our interviews, reinforcing observations by other scholars, such as Jerry Mashaw, David Spence, and Jody Freeman, who have described agencies as statutes’ congressionally appointed custodians.^[46] When a Congress and a President enact a regulatory statute, they place some socioeconomic arena within an agency’s purview, tasking the agency with addressing matters that arise within that arena. The statute gives the agency legal authority and resources for this work, even as it constrains the agency’s decisionmaking scope through mandates, guidelines, and exceptions. The statute is thus the main backdrop against which agency action takes place. Our study not only shows that agency officials often understand themselves to be playing this custodial role but also reveals what this role actually consists of.

In this Part, we explain how agencies understand their duty of loyalty to their statutes—their obligation to work in the statute’s best interest. We first show that agency officials work closely with their statutory texts, which serve simultaneously as a source of authority, a resource for action, and a constraint. Agencies’ work with statutory text is tied inextricably to their policy objectives. But officials do not pursue those objectives in a freewheeling fashion. Instead, loyalty to statutes is grounded in what we call agency mission: an overarching set of values and objectives served by the regulatory regimes agency officials superintend. When determining what their statutes authorize, obligate, or enable them to do, agency officials frequently understand themselves as serving their agency’s mission. This Part articulates and defines the phenomenon of agency mission—a central influence over agency action not sufficiently appreciated in the literature—and explains how it structures agencies’ loyalty and channels their decisions.

A. Statutory Texts

Agency officials in our study, across the board, clearly understood their need for statutory authority to act. They showed both an interest in and fidelity to their statutes’ texts. As one person put it, “[E]very single thing I worked on was always carefully analyzed at the highest level about, ‘Do we have the authority to do this? Does the statute say this? Has it always been interpreted this way? Can we really say that?’”^[47] Indeed, statutory texts can foil well-considered policy plans. “I came to the agency thinking I was going to do a specific thing, and I was totally convinced that we absolutely had the authority to do it. We better get right on it,” one political appointee told us.^[48] But after looking further into the statute itself, researching “what does this really mean and how did it get there? And how did the agency do this other thing last time?” this person “realized, ‘Oh my god, . . . we have no justification for’” the contemplated policy.^[49] “‘I guess we can’t do it.’ I had no idea.”^[50] As another official noted, sometimes a statute could be “so clear that there’s really nothing to interpret.”^[51]

At the same time, interviewees consistently emphasized the open-ended quality of many statutes they worked with.^[52] As one noted, the “typical framework [is] Congress generally would set . . . broad statutory [goals], but would leave a lot of the technical details to the agency to set up.”^[53] The discretion conferred by open-ended statutes also extends well beyond the filling up of technical details.^[54] Regulatory statutes often do not require one particular response to a given issue; they allow for multiple possible solutions to any one problem.^[55] As one person put it, “Congress writes laws that are these huge balancing acts. They craft these very intricate compromises, and different . . . members of Congress have different points of view of what the law should be . . . . That creates a lot of indecision, and it creates a lot of discretion for the agency.”^[56] Some described this open-endedness as inevitable. “I thought drafting legislation would be a piece of cake,” one respondent who had provided legislative drafting assistance from the agency’s side recalled.^[57] “It’s insanely hard. You think you’ve covered everything and then you give it to five people and they say, well what about this, what about that.”^[58] Another interviewee had drafted legislation as a congressional staffer before going to work in the agency charged with implementing the resulting statute.^[59] “The words end up on the page because they are unclear,” this person told us: “[E]verybody can see their own outcomes . . . [and] goals in them.”^[60]

In its custodial role, the agency must find sense and coherence when implementing open-ended texts and words.^[61] But our subjects generally did not describe themselves as searching for a statutory text’s “best” meaning most of the time: Statutes’ pervasive open-endedness meant that, often, there was no necessarily best interpretation of any given provision. Our respondents agreed that statutory text must be able to bear the meanings a policy ascribes to the text, but as one official put it, “[i]t was very hard to say what would be the best interpretation independent of policy consideration[s].”^[62] As another suggested, “Once Congress unleashed the tools, I think [agency officials] view it as their responsibility to utilize those tools from a policy choice as best as they think is appropriate for the public interest,” and “they’re going to experiment and do that until a court would tell them that they have misinterpreted the powers that were given to them.”^[63] Rather than searching for an eternal, uniquely correct meaning, agencies generally work with statutory text to “advance reasonable interpretations that we thought best served the policy goals of the statute and of the administration.”^[64] And in the case of “a really badly drafted statute,” with which “you were sort of trying to make a set of words [work] in the context that it wouldn’t normally work in,” there might be no other choice than asking, “[H]ere’s our broader goal. How do we fit that within what these [words] mean[]?”^[65]

Other officials were even more explicit that their use of statutes was goal driven—motivated by a pragmatic impulse to create effective policy grounded in statutory authority. As interviewees underscored in describing their work with statutes, agencies (especially lawyers) often ask: Does the statute support a desired policy? Does it offer underutilized authorities?^[66] Can we reconfigure existing rules for new purposes? Some respondents described themselves as turning to statutes in search of authority to support or enable what they sought to do, not always to understand in the first instance what they were supposed to do: “[T]he first question isn’t ‘what do these statutory words mean?,’ because you’re in a space. You’re regulating within a space . . . and then it becomes, you want to do this thing . . . and then you’re going back in statutes to see if you can do it.”^[67] As another subject said, “I can literally remember being in the room and who the colleague was who had crawled through every part of the statute to say, well, where’s a thing that we could waive that would provide some flexibility on this.”^[68] While agency lawyers may help policymakers “figure out congressional intent” and while “courts have their rules about what to turn to first, and which documents have precedence,” one official told us, “my experience just on a practical level, it’s really case-by-case, and it’s driven by the political process, driven by the goals you’re setting for the agency.”^[69] Creative use of statutory text might even become part of the job: “[P]eople are trying to push the boundaries by reading the regulations that are on the books the way that they think they need to do their job, not so much a power grab as you’re a professional, you’re trying to accomplish your goal, trying to be creative.”^[70]

Interviewees also described how statutes could become more ambiguous or open-ended over time because of changing circumstances and statutory amendments.^[71] “[Y]ou could have historical precedents from older language, and now newer language is in conflict with that older language, and the older language has not been removed from the statute,” one person explained.^[72] Although in principle, “the newer language trumps the older language, . . . it’s not always that clear. Sometimes, it’s not as though it’s in conflict or even informing it, it could make what’s already an ambiguous provision even more ambiguous.”^[73] Other legal developments might contribute to this evolution, too: “A lot of times we were working with statutes that had a history, so it would be not only the language of the statute would be relevant or legislative history, but there’d be a fairly substantial case history . . . [making a] precedential body.”^[74] And sometimes “things just aren’t spelled out,” and officials might turn to their own foundational documents for guidance as to how to proceed.^[75]

To be sure, attention to statutory text was unevenly distributed amongst our respondents. As one non-lawyer put it, “I guess you could say the statute’s in the back of our mind . . . [b]ut nobody’s going out and reading the caselaw or anything like that.”^[76] “To be honest with you,” a subject-matter expert told us, “I never read the Act nor did I think most people on a day-to-day basis were aware of it. . . . [A]s a political appointee, . . . [I] had a few different missions and responses.”^[77] For some officials, lawyerly approaches to statutory authority stood in the way of accomplishing the agency’s objectives: “Sometimes there’s an undercurrent of ‘the darn lawyers are slowing us down’ and ‘we care about . . . safety.’ That line gets used in the agency itself against the lawyers” who ensure statutory fidelity.^[78]

But even while some people involved in policymaking may not read the statute or the case law, there are always others who do. It perhaps should come as no surprise that lawyers within agencies play a significant, if not lead, role in tying agency action to fidelity to statutory texts. Officials highlighted general counsel offices’ important role in ensuring concordance between policy and statute. As one interviewee explained, interpretive creativity can generate “pushback” from lawyers, underscoring that agency lawyers (and other career civil servants) often set the bounds of work with statutes.^[79] But even as agencies treat statutory text as foundational to their policymaking process, statutory text in itself does not suffice to resolve the issues agencies tackle. As one person put it, “Once you’ve said, ‘Oh the statute allows us to do this and such . . . ,’ then the statute is no longer answering the question.”^[80] Statutory text thus becomes an adaptable, but not infinitely malleable, resource for an agency addressing the issues that the statute itself places within the agency’s purview. Officials elaborate on the text with goals and values in mind. These goals and values—the overarching sense of mission we turn to in the following subpart—help orient officials’ sense of responsibility for the statutory regime.

B. Agency Mission

Throughout our interviews, we identified a second dimension of the duty of loyalty that shaped officials’ pursuit of policy goals within statutory frameworks. Officials across agencies displayed a commitment to overarching objectives and values that they cited as representing the fundamental purposes of the agency’s work.^[81] This sense of agency mission guided how officials put statutes to work in any given context. Rather than settling the meaning of a particular word or the details of a specific policy, officials spoke of agency mission as providing a general value orientation that influenced how they approached their policy choices, without determining the conclusions they would come to. While certainly not the exclusive source of officials’ purpose or motivation, mission operated as a rudder to steer agencies’ work with their statutes and gave their work a rough coherence over time.

We focus in this subpart on defining the factors that we saw shaping officials’ sense of mission. Crucially, the concept was often grounded in a set of statutory enactments and their particular purposes,^[82] but it also evolved diachronically, taking shape over time under the influence of presidential administrations, subsequent Congresses, litigation, public participation, and changing circumstances in the regulated world.^[83] Whereas statutory background exerted a pull from the past, harking back to the coalition that enacted the statute, ongoing political projects and interactions pulled the agency forward. The commitment to mission thus pushed officials to coordinate values across groups and over time.

1. Sources of Mission

.—The statutes that mandate and constrain agency action are central determinants of an agency’s mission. Specific statutory language or a foundational committee report may define a clear mission that carries through all of the agency’s work. The Federal Coal Mine Health and Safety Act of 1969, which created the Mine Safety and Health Administration (MSHA), itself articulated the Act’s overarching mission: “to protect the health and safety of the Nation’s coal miners.”^[84] Congress passed the statute as a direct response to a massive underground mine explosion, and the statute’s supporting committee report clearly articulated a vision repeated by President Nixon upon signing the bill—that it was “a crucially needed step forward in the protection of America’s coal miners.”^[85] MSHA officials in our study described that mission as permeating the agency’s work and its officials’ self-conception. “[E]verybody I work with is very focused on mine safety. The legislative history specifically says everything should be interpreted, when in doubt, on the side of mine safety. I sort of take that command from Congress. It’s nothing that would need to be said.”^[86] This explicit “unidirectional mandate”^[87] makes MSHA decisionmaking “always about safety, and health, and . . . how best to promote safety and health within . . . the possible range of interpretations.”^[88]

Even when an agency’s mission was not so expressly laid out in text, though, officials described how an implicit understanding of the agency’s basic orientation guided their work with statutes. “[A]t the [end of] the day,” an official from the Small Business Administration told us, “we’re trying to help businesses get contracts, so they can employ people, that’s what our mission is. We want to do it in a way that . . . is as efficient and least burdensome as can be based on the language of the statute.”^[89] We heard that kind of overarching, general, and relatively simple sense of mission repeatedly in our interviews, often in the course of an official explaining some particular process or decision. “Sort of the unifying thing about [a particular agency],” another subject said, is that all parts of it are “very much focused on preventing harm to [a particular group].”^[90] As another person put it, “[W]e’re dealing with a very mission-oriented agency, an agency that thinks if the air is cleaner we’re doing a good job, not like if [some] company has higher profits we’re doing a good job.”^[91] Officials might ask in a “broad sense, what is the statute trying to do? Maybe it’s trying to clean up the air and water and if so, then let’s lean in that direction.”^[92] Another spoke in terms of a statute’s overarching purpose: to “best serve the Medicare beneficiaries because ultimately that’s what Medicare’s there to serve. It’s not for hospitals, or doctors, or other clinicians.”^[93]

Even as agency officials characterized themselves as seeking to advance their mission, they also demonstrated a commitment to pursuing the priorities of the extant presidential administration. Importantly, however, we found that the latter can often be complementary to or shaped by the former. As we have documented elsewhere, we consistently heard about the ways in which political officials defined agencies’ purposes: “[T]he people who work at senior levels . . . , they come in with an agenda,” one official told us—an agenda that drives them to analyze the relevant statute for the “degrees of freedom you have to make changes to regulation.”^[94] Some officials saw their work as necessarily closely aligned with presidential projects: “We’re an executive agency, so it’s hard to think about priorities of the agency that aren’t also the President’s priorities.”^[95] And “a particular philosophy within a given administration,” such as creating “innovation” or “bringing new technology into the system,” might lead officials to “try[] to find interpretations that allow us to do that versus maybe some other kind” of policy.^[96] Political officials might be particularly motivated by public perception and political salience, especially in relation to hot-button issues.^[97]

But the interaction between a particular administration’s philosophy and priorities and the preexisting sense of agency mission could be complex. On the one hand, “there’s a natural hesitancy to just overturn past interpretations. . . . [F]or one thing you want the work you did to live on. . . . [Y]ou don’t want to create a culture where you think it’s okay for every new administration just to undo everything that was done before.”^[98] As a career civil servant told us, “I think most [political appointees] just want to come in and manage the stewardship of the program and propel it forward and make it better for the next generation.”^[99] At the same time, appointees generally were understood to come in with political objectives and ideological predilections—projects they wanted to accomplish and values they wanted to pursue.^[100] Career staff, with generally longer tenures, could have different reactions to that: “[A] lot of times when a politico comes along and says, ‘Hey, we’re going to go 100% recycled,’ sometimes the career staffers are rolling their eyes and sometimes, they’re like, ‘Okay, awesome, here’s my opportunity.’”^[101] Either way, political appointees often needed to integrate their goals with a longer term sense of agency mission to be successful: As we have argued elsewhere, political appointees and career civil servants play complementary, interdependent roles, each crucial to the agency’s work.^[102] Appointees may come in with political objectives and ideological predilections, but they may also join agencies because of some concordance between their own professional objectives and the agency’s work.

Regardless of their intentions, political officials can find their ability to advance preferred objectives constrained by the mission-driven and statutorily defined boundaries of the institutions they enter. “In any change of administration, there are always changes at the edges, and we expect that. But in terms of core principles and activities, they more or less remain the same.”^[103] Indeed, our interviews underscored how agencies can develop a strong sense of internal culture and character—a self-understanding of “how we do things around here” that guides decisions, whether that is a “long history . . . of bipartisan cooperation” in an independent agency,^[104] an understanding that an agency-internal body “does not like to overrule itself,”^[105] or a culture of “everybody staying in their lane.”^[106] Long-serving career officials, many of whom relate deeply to their agency’s mission, are likely to internalize the agency’s culture and purposes, carrying them through the agency’s work across administrations.^[107] One interviewee explained how long-time officials’ commitment to mission might constrain political ambitions: “[C]areer folks tend to be much more conservative or let’s just say risk averse and tend to think of themselves as the keeper of the programs and wanting to keep the bad people out as opposed to the opposite, trying to get more people in.”^[108] Our interviews suggest that we should think of political officials’ ambitions as being incorporated into rather than replacing agency mission.

2. Mission as Rudder

.—As our account of agencies’ work with statutory texts underscores, officials’ pursuit of agency mission did not replace the analysis of texts. Instead, officials’ sense of mission provided them with a rudder that steered their decisions about available policy options. A sense of agency mission can provide a common orientation across multiple decisionmakers, even while it remains somewhat general and subject to gradual change. Statutes, after all, construct complex environments with many competing priorities: Any given provision’s instruction can be read in a number of ways. Most statutes do not specify one single value to hold above all others. Figuring out how to actually proceed requires policy judgment. Making a plan for regulatory action takes more than just identifying what the statute allows.

Interviewees explained how agencies must consider the effects they seek to achieve, the values they hope to serve, and how their choices relate to other policy goals whose relative costs and benefits may be hard to commensurate.^[109] As one interviewee explained, “You have to turn to how much [regulatory] protection you want, how much cost you’re willing to impose, what the politics are.”^[110] Another, in the context of a consumer protection statute, noted that the agency considered “[w]hat’s good for the American consumer . . . ? Also, . . . what works for business? . . . [You must be m]indful for the effects of your legislation on a bunch of stakeholders.”^[111] Because agency action affects different people in different ways, a benefit often comes with its own cost: Simplifying eligible people’s access to entitlements increases opportunities for ineligible people to gain access, but blocking ineligible people increases the obstacles to eligible people too. Statutes rarely spell out the terms of these tradeoffs.^[112] But as we saw in our interviews, a sense of mission can help steer these choices by keeping officials alert to the normative decisions that policymaking inevitably involves.^[113]

We also observed how the sense of agency mission can help agencies adapt to novel circumstances in a coherent way. As one interviewee described updating a rule, the “first step is, we have a problem. The second step is, we can solve this problem . . . by revising . . . and updating our rule, and three is, we will try to rewrite the rule in a way that we think fulfills the mission Congress set out for us [in the statute].”^[114] Another interviewee provided an example of that kind of adaptation. Unmanned aerial vehicles, or drones, did not exist when the Federal Aviation Authority (FAA) was given responsibility for aviation; its “rulemaking . . . process is all designed for safety and certification of things that have people in them, and thus qualifications for pilots.”^[115] To address developments in the area that its statute placed within its purview, the agency found itself “trying to construct old statutes in new ways.”^[116]

Any individual statute managed by a given agency, moreover, is part of a larger statutory and regulatory context involving other agencies and officials. Agencies often find themselves in “shared regulatory space.”^[117] Conflicts over policy priorities and responsibility distribution underscored how different missions influenced agencies’ policy orientations. “Government agencies may have very conflicting objectives,” one person explained.^[118] As one respondent put it, “different agencies . . . come to it with their own biases and their own historical interpretations and their own kind of equities.”^[119] Another observed that “[e]ach department is also sort of producing their own sort of processes for how to interpret congressional intent.”^[120]

Take the example of two components of the Department of Health and Human Services coming up against each other. The Food and Drug Administration (FDA), primarily in charge of medical product safety, “was adamant that . . . there’s a . . . safety risk associated with” a particular medical product prepared in a relatively inexpensive way.^[121] Meanwhile, the Centers for Medicare and Medicaid Services (CMS), primarily in charge of payments for the medical product, “had a study where it said that the risks were very low . . . . [Y]ou had two agencies who had very, very different policy goals because their missions are very different. This is not uncommon.”^[122] Agencies can, in other words, push one another to examine the contours of their statutory authorities by interpreting those authorities in light of their respective missions.^[123] In this sense, agency mission plays a coordinating role among perspectives not just over time, but also across the government at a given time.

Such interactions also emerge because “some [statutory] provisions don’t have a natural home.”^[124] In one example, the National Transportation Safety Board proposed that the FAA, in charge of aircraft, and CMS, in charge of Medicare ambulance payments, jointly produce safety standards for ambulance helicopters. This pushed each agency to determine whose regulatory sphere such vehicles fell into.^[125] CMS sought a constrained, rather than expanded, sphere of power. “That actually is a fairly common thing,” our subject told us, “where one part of the agency is basically trying to avoid responsibility for a particular provision, so they’re trying to get it assigned to somebody else.”^[126] Even statutory provisions that could land in more than one “home” require particular kinds of expertise. “[T]he last person you want to be regulating helicopter safety,” our healthcare system expert told us, “is somebody like me.”^[127] Although ambulance helicopters fell into CMS’s core function of managing healthcare costs, the agency sought to place safety standards for these aircraft elsewhere because such safety considerations fell outside of the agency’s sense of expertise and mission.

* * *

The sense of agency mission we identified helped officials define problems worth addressing and identify the policy experiments or political conflicts worth engaging in. As one official put it, “I think that, in the end, . . . people thought that the statutes were kind of with them. . . . They had the law at their backs and that doesn’t come from particular little words in a statute.”^[128] Indeed, many of our respondents defined their success in terms of the extent to which they advanced the agency’s mission.^[129] Effective policy could mean “maki[ng] sure whatever you do, you’re building enough program integrity into the system so that you don’t make it easy for scoundrels and fraudsters to come and raid the . . . program.”^[130] From the administrator’s perspective, this utility helps explain the pervasive influence of mission we observed in policymaking. From a systemic standpoint, this sense of mission helps agencies mediate differing policy perspectives over historical time and sociological space.^[131]

II. The Duty of Care: Producing Policy that Works

For agencies to fulfill their role as custodians of statutes, they must “build [the] thing”^[132] for which they were given responsibility. In our interviews, we saw pervasive and consistent preoccupation with ensuring that statutory schemes worked—that the rules, guidelines, and policies agencies adopted functioned effectively in the contexts and circumstances within the agencies’ purview. Our respondents made clear that statutes were not so much for understanding as for governing: “So, your statute is incredibly general, and so then you need to make it real.”^[133]

Agency personnel often described themselves as acting incrementally, focusing on the practical effects of their policy choices. One person involved in Affordable Care Act (ACA) implementation described the kinds of considerations that went into any policy decision: “One is just . . . ‘Will this make the marketplace stronger? Will it make it easier for the insurance companies that are participating to . . . price risk?,’” thereby encouraging participation and helping effectuate the statute.^[134] “The second . . . ‘Will it be more efficient?’ Like just eliminate points of friction in the system. And then another dimension is . . . ‘Will it protect people from abuses by other actors in the system?’”^[135] Many respondents shared this overarching practical concern with policy effects. “[Y]ou always know that statutes are enacted for some purpose,” one said.^[136] “They have some goal in mind. . . . So then the question is, whether [the policy] advances the purpose of the statute . . . consistent with what the statute requires, and [considering] what burdens are going to be imposed on the people who have to live with it.”^[137]

Our interviews show agency officials seeking practicable paths to implement statutes in an admittedly provisional way, subject to adjustment down the line. This effort instantiated in them what appeared to us as a duty of care for the statutes that designated the agencies as custodians. In this Part, we describe how administrators in our study acted on this duty of care—how they interpreted and implemented their statutes to make them real and effective. This process involved ongoing interaction and negotiation with numerous actors and factors: external events, other administrators, other government branches, and public pressures. Officials sought to respond to events and realities in the world—to ensure that their policy plans addressed lived experience, changing circumstances, and the goals and needs of the regulated public. Just as important, they incorporated other actors in the democratic system—the political branches (who could also be sources of information about the needs of the regulated public) and the courts. These forms of pluralistic responsiveness showed officials’ awareness of the concrete value and legitimating effect of incorporating the views of entities capable of both enabling and thwarting agency work.

By bringing these elements of the duty of care to light, our study shows that agencies tend to be self-consciously pragmatic; they are not typically searching for transhistorically or absolutely correct answers.^[138] We show how agencies, as their statute’s custodians, mediate and participate in defining their statutes’ meaning by determining what they will become in practice. Bearing responsibility for the statutory scheme makes agencies more than mechanistic agents of the current Congress, or servants of the sitting president. The statute, after all, belongs to neither. Although scholarship and doctrine often present agencies as working in a fairly clear-cut principal–agent relationship, agency loyalties quickly overflow the discrete, simple vessels thought to contain them.^[139] In some sense, we can say that statutes lack much independent meaning; they act as resources whose significance derives largely from how agencies fulfill their duty of care. Our research thus supports Jerry Mashaw’s proposal to “imagine agency statutes as works-in-progress, to be shaped and molded by continuous interaction among the implementing agency, the political branches and affected interests.”^[140]

In emphasizing the pluralism that undergirds agencies’ efforts to create workable policies, we should underscore that influence over their work is not necessarily equally distributed. Even as our interviewees described their efforts to be responsive to the public and to facts on the ground, we cannot discount the well-established reality that economically powerful voices often dominate the conversation with agencies, just as they do with other government institutions.^[141] But the pluralism we show pervading the policymaking process builds a solid foundation for further democratizing governance. As we explore in Part III, this foundation places agencies in a unique position to keep a statutory scheme relevant and responsive to complex current circumstances, and it pushes them to take into account a broad range of perspectives, claims, and information as they work with statutes.

A. The Regulated Public & the Regulated World

Asked how statutory interpretation worked in her agency, one interviewee explained, “I think the answer is . . . it’s not that simple.”^[142] Rather, this interviewee suggested we consider, “what are the different ways . . . . that acts of statutory interpretation happen.”^[143] The interviewee noted that “[s]ometimes statutory interpretations come up because something happens on the ground”—for instance, an inspector issues a citation to an industry operator—“and somebody has to make decisions fast, and all of a sudden you wind up with an interpretation.”^[144] Other times, interpretive issues “come from top-down mandates because somebody’s thinking about a problem.”^[145] And “sometimes, there’s situations that happen in between, and exactly what those mechanisms will be” depends on the specifics.^[146]

This interviewee directed our attention to the “ways that statutory interpretive questions come to the [agency].”^[147] That insight underscored how agencies often turned to their statutes in search of authority to act because events in the world or members of the public catalyzed them to do so. Throughout our interviews, we were given examples of this sort of responsiveness.^[148] Real-world events; technological developments; social practices and patterns; and emerging ideas and ideologies all made problems evident and salient: “The drive for a change in a rule or the creation of a rule happens when something bad happens or they are afraid of something bad happening from a health or safety perspective. Then they look at the statute.”^[149] This central feature of agency action highlights officials’ sense of responsibility to make their statutory regimes work in practice, underscoring how agencies engage with statutes frequently to solve problems and grapple with live issues. It is in this sense that agencies work with their statutes, looking to them as tools to accomplish their objectives, particularly to keep their statutory regime relevant to a changing world.^[150]

Our interviewees also demonstrated how agencies often develop awareness of the issues they need to address by receiving various forms of feedback from the outside world. “Definitely on the overall policy mandate,” one person observed, “hearing from people doing the work, living under the statutes, regs, [engaging in the regulated activity], dramatically informed our policy, including shifts we made in our own policy over time that were pretty visible and in response to what we were hearing and seeing.”^[151] Another person explained: “[W]e literally got zillions of questions on a daily basis, like any agency, from people who are trying to interpret what they are supposed to do under the law.”^[152] Another respondent recalled that “no one really cared about [a particular issue] in terms of the . . . political coalitions . . . in Congress . . . but somehow the editorial boards of the New York Times and the Washington Post . . . were obsessed with it.”^[153] These “editorials often were just like, ‘Hurry up and get the damn thing done,’ rather than, ‘It should have X, Y, and Z.’”^[154] And “my boss, basically, . . . would see that and say, ‘What’s up? . . . . Let’s try to get this going more quickly.’”^[155] Pressure, publicity, and novel arguments from the media, interest groups, and regulated parties could push agencies to return to their statutory authorities in search of a response.^[156]

The actual work of superintending a regulatory regime also produced information that gave agency officials motivation to act. Enforcement staff, “always briefing up,” alerted other officials to implementation problems or a lack of fit between regulations and social practices or products.^[157] As one interviewee noted, with enough uncertainty, “the agency is like, ‘This is a source of confusion, and we need to—our job is supposed to be to provide certainty to the petitioners and applicants’”—to make decisions enabling the agency to say, “[t]his is what the answer is.”^[158] In one example, a statute provided a reimbursement scale that used rural facilities as a payment floor, on the assumption that rural facilities had relatively fewer resources.^[159] Then a highly resourced facility in a low-density area was reclassified as a rural facility. That change would have garnered its state a windfall of several hundred million dollars from a budget shared across all states.^[160] Members of Congress from that state lobbied the agency to keep the reclassification.^[161] That kind of situation placed the agency between conflicting interpretations and preferences, alerting it to potentially perverse consequences of its decisionmaking.^[162]

Staff also explained to us how they regularly researched the realities of the practices they regulated. As one subject observed, “We engage in a . . . strategic planning process every year” in which, “from the ground up, the teams survey the trends in the marketplace and figure out where they propose to deploy their limited resources and what are the real problems out there that need to be tackled.”^[163] As another mused, “One thing that I sort of wondered about is how anyone can do policy, let alone trying to faithfully implement a statutory mandate . . . without having access to really top-flight technical people, because many of the decisions . . . really depend on one’s understanding of the technology.”^[164]

As others have documented, and as we found in our interviews, agencies use a range of approaches to interact directly with regulated parties as they develop policy ideas.^[165] Many of our respondents described proactive efforts to glean information about issues in the field, the effects of existing rules, and the probable effects of prospective policies from the public.^[166] Numerous officials emphasized the importance of “invit[ing] public comment,”^[167] holding “listening sessions or round tables” where an agency could “hear[] of the challenges of applying for the program,”^[168] and getting input from numerous sources to ensure that agencies were “not advocates just for [particular interests]” but for “the public interest, the sweet spot somewhere in the middle.”^[169] Such interactions often focused on a policy option’s real-world consequences: “The most useful stuff was . . . when someone in the regulat[ed] community would point out an impact that we hadn’t thought of.”^[170] As one person put it, “[W]e use a [] multi-layered approach to rule making. . . . We may have a workshop. We may have a round table inviting in stakeholders from all elements. We do this as much to enlighten the staff . . . [as] to make it as open a process as possible.”^[171] Others might “contract with some polling firms that would do questionnaires or do focus groups of beneficiaries, so we could understand what their challenges were with the [benefits] program” or do “a lot of surveillance activities for us to make sure that the programs are working.”^[172]

Importantly, as we saw in our interviews, much of this work occurs before any legally required notice-and-comment on proposed rulemakings begins.^[173] Agencies often develop their policy “drafts” through an interactive, responsive process, in service of their workability objective. Respondents described combing through “various statistical studies or research studies or other policy documents,”^[174] “peer review publications, . . . legal opinions” and other “external inputs to help guide and shape” policy consideration.^[175] “[I]t typically is an enormous amount of pre-proposal preparatory work. A lot of compiling information. A lot of work [with] the affected industries.”^[176] As one person recalled, “During the ramp-up to [a particular] rule and during the rulemaking process, I probably met with stakeholders of every stripe, and they, of course, had their own ideas about how the statute ought to be interpreted, so that is a source of information.”^[177] Often, such interactions, where an agency “tried to listen to a lot of stakeholders,” were not “formal in the sense it was mandated by anything. We just thought it made sense to do.”^[178]

Not every agency engaged in the same level of public consultation. For example, one respondent commented “that a lot of the rulemaking in the immigration space would benefit from some ability, before they spend months and months developing the proposal, to get feedback from stakeholders to understand the unexpected.”^[179] This political appointee reported being “disappointed” at the lack of such practices.^[180] Our research thus accords with others’ findings that, although agencies use a wide variety of agenda-setting public consultation approaches, these efforts “tend to be relatively unstructured, unsystematic, and ad hoc.”^[181] Moreover, although our research does not allow us to compare the relative influence of different groups, our findings are consistent with studies showing that regulated industry and well-resourced civil-society groups have a greater effect on policy development than smaller public interest groups and ordinary private parties.^[182] Nonetheless, our research supports scholarship concluding that many agencies do engage in significant, meaningful public consultation, not just at the agenda-setting stage,^[183] but also after a rule’s enactment.^[184]

The formal notice-and-comment process, of course, also brought outside opinions to the agency, and our study respondents painted a picture consistent with others’ conclusions that this aspect of rulemaking influences agency decisionmaking. As one respondent put it: “For a rule of any complexity, it would be very unusual for comments not to result in certain shifts.”^[185] One agency, for instance, “was in a rush” to make a rule changing a medical reimbursement system “because the chairman of the Ways and Means Committee at the time was very interested in” a particular new approach.^[186] “[T]he public commenters just were uniformly opposed . . . , and then [the agency] didn’t adopt that system, and then a year later, [the agency] adopted . . . what I thought was a much better system in response to the public comments.”^[187] Agencies also used comment periods creatively to float policy trial balloons gauging public reaction to a policy option: “[If we were] struggling with the best interpretation or whether we had authority or whatever, . . . we would sometimes directly ask for comments that built out a case for doing something, a statutory theory for doing something” for use in later policymaking.^[188] “There were a lot of things where we sort of floated an idea [during a rule-making] . . . , got some stuff that felt useful, and then we were in a position to propose with a stronger . . . record” in a later rulemaking.^[189] Agencies can utilize public input in this way by building on existing systems for assessing and responding to comments in rulemaking.^[190]

Joining our research with the work of others yields a nuanced picture of agencies’ interactions with the public. On the one hand, agency information collection is often unsystematic and inefficient. And agency decisionmaking is often influenced most by well-resourced regulated industries serving their own economic interests. On the other hand, agencies can and often do convene a wide range of participants when formulating policy ideas. And they have personnel and systems in place for taking into account multiple diverse interests. Indeed, their capacity to incorporate and negotiate among pluralistic views in any given policy decision process probably outstrips that of other government institutions. As we discuss further in Part III, strengthening and channeling these pluralistic decisionmaking capacities would significantly enhance government’s democratic quality, a possibility that in turn underscores the serious costs of the current presidential administration’s efforts to dismantle bureaucratic capacity, as well as of recent Supreme Court doctrine that has been redistributing decisionmaking power to the far less accountable federal courts.^[191]

* * *

Our interviewees consistently underscored that agency action often requires a catalyst—something that drives an agency to designate some phenomenon as a problem that warrants a policy response.^[192] To a large extent, statutes leave it up to agencies to decide whether to address—and how to even identify—an issue under their purview. Presidential administrations and congressional oversight of course play significant roles in setting agencies’ overall regulatory agendas, especially in the modern context of centralized review.^[193] But throughout our interviews, officials described an ongoing, ground-up form of working with statutes, grounded in their evaluation of workability. Our study thus shows how working with a statute intrinsically involves figuring out the real-world consequences of different implementation options, ensuring that decisions express not just theoretical preferences but workable goals so that the agency can effectively implement its policy choices.^[194] An agency’s purposes and goals were often pragmatically, not abstractly, defined.^[195] Interviewees were keenly aware that agencies do not act on their own: “Like a quarterback can throw the ball down the field, but better have someone [to] catch it.”^[196] As we will explore in Part III, this consequentialist orientation comfortably suits an institution of governance in a democracy by helping to ensure that governance decisions are made not just in the abstract or on principle, but with due consideration for their effects on the governed population.

B. (Members of) Congress

In scholarly literature, judicial doctrine, and common parlance, agencies play the role of Congress’s delegee. At the most basic level, when Congress legislates, the relevant agency springs into action. As put by one interviewee, a statute “was signed into law [on] December 15, and we started talking about regulating the next day.”^[197] From another: When statutes impose schedules, “we get started immediately.”^[198]

But agencies have complex, internally cross-cutting relations even with the seemingly single institution of Congress. First and foremost, the delegating or enacting Congress—and the governing coalition behind a statute—quickly recede into history after the next biennial election.^[199] The bulk of agency work thus entails statutes enacted by past Congresses, even as that work is overseen and enabled by later Congresses, including through ongoing appropriations.^[200] As we note in Part I, agencies balance their desire to “get along with Congress” with their duty of loyalty to a statutory regime that may well predate sitting lawmakers: “[Y]ou’re trying to be . . . a good steward to the program and adhere to the statute.”^[201] Though our respondents identified and understood the preferences of existing members of Congress, they distinguished those from the imperatives of their statutes. “We kind of knew that [Congress was not] gonna be happy with us,” one official said of a particular program.^[202] “We kind of knew that they were probably against it because either my Program Office or General Counsel had had at least some conversation with the committee . . . . But we felt we had authority anyway. So we went ahead and did it.”^[203]

Our respondents’ identification of the tensions between Congresses highlights that a critical component of their duty of care—making statutory regimes workable—involves collaborating with and responding to sitting lawmakers. Throughout our study, officials articulated their efforts to incorporate congressional demands alongside their own understandings of what their statutory regime required of them. These forms of influence were diverse: From pressure and lobbying by individual members of Congress, to consultation with oversight and appropriations committees, to negotiation and joint problem-solving, the sitting Congress helps shape statutory meaning.^[204]

Although our interviewees were attuned to Congress’s legislative powers writ large, the lower-level, individualistic influence they frequently cited resembled collaboration or competition, rather than the principal–agent relationship often attributed to the separation of powers or delegation at high levels of generality.^[205] Scholars such as Christopher Walker and Jarrod Shobe have documented how agency officials provide technical assistance to legislative drafters, reflecting the value of the agencies’ expertise to the lawmaking process.^[206] This collaboration works the other way, too. Members of Congress provide inputs into a multifarious decisionmaking process with agencies, though members’ authority or political preferences, rather than expertise, shape such “assistance.”^[207] As one interviewee put it, when members of Congress lobbied their agency to adopt some particular view of a statutory provision, “we would certainly take it under advisement whether we could do it or not, whether we thought it made sense of what we’re trying to accomplish, and whether it could be done legally with all the issues that we’d have to evaluate.”^[208] In the end, though, the statute took primacy: If some option is not “permissible in terms of certain things that maybe certain members of Congress want you to do, you just have to deal with it and say, ‘The law is the law here. There’s not much more we can do. Basically, you wrote a crappy law and we’re stuck with it.’”^[209]

Officials’ sense of responsibility for effectuating statutes over time, even in the face of opposition from later Congresses, was probably bolstered by administrators’ relative expertise in the statutory regime itself. Members of Congress are fundamentally generalists, while agency employees specialize in the statutes their agency implements, with divisions of labor inside the agency supporting specialization. “I don’t think any of the members of Congress—maybe some of the staff, [but] none of the members of Congress—that I talked to had read the statute. They plainly did not have . . . detailed familiarity with what the statute said . . . or what our brief under the statute was.”^[210] Instead, discussions with Congress as reported by our respondents tended to be “much more general, which is, ‘Yeah, we want to [protect individuals from harms by an industry,] but we don’t want to kill the [industry]. How are you going to do that?’”^[211] Accordingly, our subjects often described members of Congress as less focused on particular (and likely unfamiliar) statutory text than on the practical implications of agency action.

The extent of officials’ interactions with members of Congress varied significantly among our interviewees. A small agency with little direct regulatory power “would be in touch with Congress before [it] decided whether to move forward on a particular action, like in the yes or no stage,” while “we were developing the policy itself, and . . . as appropriate when we were developing the regulatory language . . . . So we’d be in touch with them at every stage through till the end.”^[212] These interactions would involve “the leadership of the [Oversight] Committee, . . . Democratic and Republican staff. And to some degree with the individual members’ staff who had a particular interest in the issue.”^[213] Another interviewee described some collaborative problem solving: “There are certain areas where Congress passed laws that frankly were almost impossible to implement, like the 100% screening requirement for cargo entering the United States.”^[214] The relevant agencies had “gotten waivers from Congress year after year after year, just because it hasn’t been possible [to implement] for resource reasons, for all kinds of technical reasons.”^[215] That led to creative negotiation with members of Congress: “If we’re going to do something to try and meet your goal on an interim basis, would this satisfy the committee?”^[216]

Other respondents described situations in which “executive agencies hav[e] some reluctance to maybe allow Congress into the process, because their motivations were not necessarily the same as [the agency’s].”^[217] For instance, an interviewee discussed a statute requiring the agency to “designate specific airports for [particular kinds of arrival screening].”^[218] The law, which had been enacted decades earlier, evinced “no idea that you could have some kind of electronic submission of information for advanced screening at a central place,” so technological developments had made that law “out of date.” ^[219] But it still had an economic impact: Designated airports could benefit from increased commerce.^[220] As U.S. foreign policy changed, “members of Congress were jockeying to . . . designate certain airports [under the statute] . . . , for purely economic and commercial reasons.”^[221] The interviewee recalled, “[W]e’d look at them like they were aliens, like, ‘We’re sitting here trying to find a way to [improve screening for national security], and you’re going to try and use this old security law to force us to put it at your airport?’”^[222] For agency officials, orienting their action around the statutory scheme provided a way to confront congressional capture by commercial interests or narrow constituencies.^[223]

That orientation also influenced how administrators interacted with individual members of Congress. “[I]t was very, very common,” one long-term administrator told us, that “Congress would enact some piece of legislation and maybe . . . [some legislators] were dissatisfied with what they drafted, and they wanted to do a . . . [p]ost-enactment reinterpretation of what they wanted . . . .”^[224] This interviewee recalled a statute drafted to exclude a certain class of entity from public cost reporting—an exclusion designed to create cost savings. “I know [the statute was supposed to do] that because I was assisting the people who were drafting the bill and [were] trying [to] make sure the criteria would exclude [that kind of entity].”^[225] These entities first learned about their exclusion late, when “the language couldn’t be changed.”^[226] They “went complaining to” two Senators, who “had this colloquy on the floor to say” that the entities should be included in cost reporting after all.^[227] That kind of colloquy, our respondent emphasized, “has zero meaning. . . . The agency can’t go based on what two Senators say on the floor of the Senate one month after the bill was enacted, in place of the . . . statutory language . . . . The agency is obligated to implement the statutory language that is passed by Congress.”^[228]

Other scholars have found that congressional oversight hearings can “mov[e agency action] towards congressional preferences on issues ranging from the level of regulatory enforcement to the creation of programs that stretch agencies’ statutory authority.”^[229] Appropriations committees can also exert significant influence.^[230] Our respondents, however, did not recount all that much experience with congressional hearings. It may be that our respondents, many of whom were civil servants, were not closely connected to the leadership offices or offices of legislative affairs that would have dealt with oversight hearings on a regular basis. But the lack of discussion of oversight hearings could also reflect the fact that such hearings themselves are not the norm; hearings require a political choice on the part of members of Congress, somewhat costly in terms of “time constraints[,] opportunity costs, [and] informational demands.”^[231] Either way, our study supports the conclusion that agency officials formulating policy and overseeing programs consider the views of members of the sitting Congress. But our research also suggests that officials incorporate congressional input to enhance the workability of their statutory regimes, giving primacy not to the opinions of particular legislators, but to what those statutory regimes require.

C. Presidential Administrations

Political officials—the President, Cabinet secretaries, and the thousands of political appointees across the Executive Branch—play an important role in shaping how statutes work. We conducted our study well before the current presidential administration launched its radical efforts to control and remake the bureaucracy, and the conclusions we offer here do not account for the dramatic expansion of centralized control and apparent rejection of numerous statutory contraints currently underway. But our research does demonstrate that achieving workability long has required agencies to attend to the priorities and values of the political officials installed within agency leadership, as well as the directives, pressure, or requests for assistance that might come from the White House and the Executive Office of the President (EOP). After all, much of an agency’s actual decisionmaking authority rests in the hands of political officials. And regardless of the extent to which the President and White House can and should control agency action, EOP support remains vital to the success of many agency policies, including because of the entrenched practices of centralized OMB and OIRA review. That political officials’ priorities influence policymaking also demonstrates a form of democratic responsiveness within administration, allowing elections to have consequences. This influence can contribute to public acceptance and democratic legitimation—important aspects of workability, too.

Our interviews illuminate how the political layer that governs agencies shapes the production of statutory regimes. In describing this relationship here, we draw on other work, based on the same dataset, in which we analyze this set of political connections.^[232] That work identifies three key features of the role political actors play in agencies’ work with statutes. First, the White House and the EOP play important, though hardly all-encompassing, roles in shaping agency policymaking. An administration may seek to centralize the implementation of a signature initiative within the EOP—a signal feature of the Obama administration’s implementation of the Affordable Care Act. ^[233] But, more often than not, White House involvement in agency policymaking is not “a top-down process displacing agency judgment . . . but rather . . . a collaborative and participatory process.”^[234] Interviewees described White House attention not as demanding specific outcomes, but as requiring exhaustive consideration, explanation, and documentation from regulators.^[235]

Second, the transmission of presidential priorities across the state does not typically come from clear pronouncements by the President himself. In practice, presidential influence tends to be diffuse and general rather than direct or specific.^[236] It is really more appropriate to speak in terms of political rather than presidential control, as the influence of an administration depends a great deal on the particular priorities of the political officials who occupy positions of authority within agencies^[237] and who enter office with their own policy ideas and personal backgrounds, connected to a thick network of political operatives, civil-society groups, and other sources.^[238]

And third, even though such officials play important and granular agenda-setting and decisionmaking roles, they have a complex and often collaborative relationship with career civil servants, who themselves draw on their experience with their statutes to evaluate what policies might be necessary or appropriate and what might be accomplished given the political context.^[239] We have described the relationship between “politicals” and “careers” as integrating different, complementary epistemic orientations, each necessary in the policymaking process.^[240] Much as with congressional influence, agency officials exercising their duty of care seek to reconcile the policy priorities of the current political apparatus with their loyalty to statutory regimes. As we emphasized in Part I, those priorities are themselves shaped and even constrained by a longer standing values orientation embodied in the concept of agency mission.

Our findings ultimately suggest that presidentialist theories misapprehend the President’s actual role in policy development.^[241] Given the complexity of agency work with statutes and the practical limitations on presidential involvement, these theories likely overstate how much even a determined President could plausibly direct regulatory action across the breadth of the administrative state.^[242] But presidential administrations, and the intricate coordination they produce between political officials and civil servants who both pre- and post-date them, heavily shape the course of statutory regimes.

D. Courts, Litigation, and Litigation Risk

Most of administrative law concerns the judicial review of agency action, and most scholarship concerning judicial review takes courts as the starting point: Commentators ask how courts should, and do, evaluate agency decisionmaking, and sometimes consider how legal doctrines interact with, and even impede, responsible policymaking.^[243] Our study illuminates a different and typically ignored perspective on judicial review: agency officials’ own conceptions of courts as decisionmakers and authority figures.

Nearly all of our subjects, when discussing courts and court doctrine, made clear that calculating “litigation risk” was integral to policymaking. Our interviewees functioned within the palimpsest of legal doctrine built up by courts over time.^[244] But officials described to us that they related to doctrine strategically. They did not receive judicial pronouncements as articulating “correct” meanings, but as factors with which they had to contend to get their work done. Whether agency action ends up in court, and whether a court will uphold it, are probabilistic inquiries; many respondents explained how they sought to make these predictions as they went along, shaping statutory implementation in a judicial image. Potential litigation “was always a consideration” in policymaking: Administrators asked “what’s the probability of litigation, and what’s the probability that the federal government would prevail?”^[245] Tolerance for litigation risks varied considerably across agencies and individuals,^[246] but considering the possibility of litigation was clearly a normal part of working with statutes.^[247] As one interviewee put it, “[J]ust part of my DNA at this point, is, I’ve got to ask the question, ‘Well, what happens if we get sued on this? And how do we defend it?’”^[248]

Agencies’ relationship to judicial review is bound up with their pursuit of workability: Producing workable statutory regimes requires staying within the parameters set by evolving judicial doctrine and behavior. As one interviewee put it: “I guess the takeaway . . . is court filings are also a way that policy gets made.”^[249] This finding does not mean that all or even most agency action ends up in court; in fact, we think it likely that large swaths of final agency action are truly final. But the design of judicial review under the APA and relevant case law ensures that agency policymaking occurs in the shadow of litigation. Working with a statute thus requires being ready to translate the agency’s reasoning for courts. “[W]hen you’re in the rulemaking, you’re looking at things in a much more holistic way . . . . [W]hen you’re defending it in court, you’re having to take a more strategic perspective in understanding what the predilections are of the court before which you’re arguing.”^[250] This imperative pushed agencies to create thorough administrative records that made a policy “defensible” in court.^[251] Even “[o]n technical issues, the ultimate judge . . . will be a federal judge. So I would want to make sure, not only had we explained in a way that a non-technical person like me or a federal judge could understand,” but also “that all significant comments be . . . answered in a manner that really deals with their efforts.”^[252]

The effects of litigation could also linger long past the court’s decision, reinforcing the importance of calculating litigation risk with accuracy.^[253] As one official put it, “[W]hen an agency has lost a bunch of cases under a statutory provision, when a question, even a new question, comes up under that provision, the lawyers are very anxious about losing more.”^[254] This position made sense, the official explained, because “they felt like they are repeat players in front of the D.C. Circuit.”^[255] Our subjects frequently described failure in court as a blow to an agency’s reputation, credibility, and ability to get things done.^[256] A bad litigation result could hang over an agency for a long time, affecting the way participants made policy decisions down the line—post-traumatic effects that could linger even when participants were skeptical of the court’s results.^[257]

But as a number of our interviewees also expressed, translating their work for courts was not always straightforward. Part of the challenge of litigation risk assessment arose because courts could be inconsistent, unpredictable, and opportunistic. As one person expressed it,

[C]ourts always are invoking these . . . rules and statutory constructions but . . . to my mind, at least personally, you know, it’s kind of a result-oriented application. You know, there’s no principle of statutory construction that we’re trying to uphold. It’s the opposite. We’re trying to use the statutory principles, the statutory construction to uphold the purpose of [the statute].^[258]

In the words of another, “I can’t find any particular consistency. When courts like our stuff, they just go through it. When they don’t, they don’t.”^[259]

In the actual assessment of litigation risk, an agency’s Office of General Counsel (OGC) loomed large. Many of our interviewees described their reliance on their lawyerly hubs for guidance on the risk/reward tradeoff of policymaking. As one non-attorney told us about sub-regulatory guidance, “Are we thinking about litigation when we write the guidance? No, I think we’re thinking about offering a blueprint [for regulated parties], but the Office of General Counsel is looking at it and saying, okay, what could come back to haunt us.”^[260] As another interviewee explained, OGC’s risk assessment might then occasion negotiation with and among subject-matter experts about whether it was worth proceeding with the planned policy. “It’s hard to say who decided in those cases ’cause we would . . . talk about it for a long time, and eventually someone would back down. So, if [OGC thought] something carried a lot [of] litigation risk, then they would continue saying that . . . , and we would move towards a decision.”^[261]

Across interviews, our subjects identified one judicial doctrine as especially relevant to their risk assessments: the now-overruled framework established by Chevron v. Natural Resources Defense Council to govern review of agency interpretations of statutes. Though lawyers were more likely to describe Chevron’s two steps with doctrinal precision,^[262] nearly all officials whom we asked about the role of judicial doctrine knew the case by name and understood that courts sometimes defer to agency conclusions.^[263] Officials described Chevron as enabling their pursuit of policy goals, or as something agencies “rel[y] on in order to promote the purposes of their mission. You know, the purposes of the Act that they’re tasked with administering and enforcing.”^[264] As another official put it: “Chevron gives a lot of policy discretion, and that’s really the job of the political people, that’s their role. The role of staff attorney would be to indicate what the permissible parameters for making the decision would be.”^[265]

While some officials described using Chevron to remind courts of their proper place,^[266] the doctrine hardly translated into a view that “anything goes” in the face of statutory ambiguity: As one respondent summarized, “We have to have a reason.”^[267] One official emphasized that Chevron was relevant only in a small class of cases involving “meatier appeals.”^[268] Another even described Chevron as irrelevant, because the agency’s job was “try[ing] to do it right.”^[269] Others, moreover, expressed a critical perspective on courts’ use of Chevron, suggesting that a court’s application of the doctrine could be a function of its views on the agency’s policy itself.^[270]

Beyond its strategic dimension, litigation also appeared to offer our subjects a learning opportunity, independent of courts’ final decisions. Some of our interviewees described litigation as pushing the agency to reconsider its approach.^[271] Where a regulated party behaves in ways not foreseen by a regulation or makes arguments the agency did not anticipate, the agency (and potentially the DOJ) must develop a response. One attorney described a situation “where an [industry] operator has raised a legitimate interpretive question, or sort of real-world question about what our interpretation means, and we’re kind of trying to interact with [the policy making part of the agency] about, ‘Is this really a position you want us to defend?,’” or “Is there guidance that you want to put out that would sort of preserve our interpretation but handle the particular troublesome problem that the operator has raised?”^[272] Another interviewee described a situation in which the agency “delved deeper into the regulatory provisions than perhaps we did when we were drafting them, because the litigants were in some measure raising arguments that we hadn’t encountered before.”^[273]

Our respondents’ accounts of their experiences with litigation ultimately highlighted the costs and benefits of judicial review itself. The prospect of litigation, not to mention court invalidation, might impede an agency from performing its custodial role by dissuading it from acting on a statutory provision, particularly if officials’ risk assessment turns to risk aversion, or where lawyers play a particularly heavy-handed role in policy vetting. We might in fact worry that recent doctrinal developments will contribute to such aversion. Agencies must continue interpreting statutes as part of their statutory responsibilities. But the combination of Chevron’s end with the recent supercharging of the major questions doctrine that post-dated our interviews does seem likely to produce greater risk aversion and therefore less policymaking generally, whether or not grounded in novel uses of statutes. Understanding the implications of the Roberts Court’s administrative law precedents is certainly a vital subject for future research, including of the sort we present here.

But litigation, and the assessment of litigation risk, can also turn into a form of public participation in policymaking, underscoring one conventional account of the judicial review provisions of the APA—that they enable public monitoring of agencies through litigants’ identification of policy flaws.^[274] Litigation can alert agencies when regulations might not cohere with reality or with other policies, and provide another way for affected publics to push regulators to take their views into account. Examples like those we heard in our interviews highlight agencies’ intimate connections to the world they regulate and the pragmatic, problem-solving nature of agencies’ work with statutes. Agency officials thus develop strategies for avoiding and surviving contest in the courts; in the process, they negotiate among the competing demands expressed in litigation. These negotiations occur as part of their duty to produce workable statutory regimes.

III. Implications

Our empirical findings support the contention that agencies act as their statutes’ custodians, managing the ongoing effectuation of statutes over their life cycles, which may last decades. We show how, in performing this role, agency officials understand themselves to possess duties of loyalty and care: a duty to act in service of the statutory regime, which in turn yields a duty to implement the statute in a workable fashion. Officials draw on an enduring yet evolving sense of mission to reconcile changing and conflicting circumstances, interests, and values, ameliorating the dead hand problem posed by old statutes that nonetheless continue to govern the public.^[275] As part of this reconciliation, officials negotiate among divergent interests and values at any given point in time, incorporating diverse political and institutional considerations into particular policy decisions. Agencies are thus in a position to ensure that a statute passed by democratic coalitions of the past remains viable over time, at the same time that officials account for the evolving circumstances of the people governed by the statute in the present.^[276] Pluralistic decisionmaking along these two dimensions—what we might call the historical and the sociological—helps keep an agency from becoming a tool of rule in the hands of any one political participant.

In the course of doing this work, agencies may well make mistakes, perform ineffectively, overlook the interests of the non-monied and disadvantaged, or privilege organized interests over the diffuse public. But our work suggests that, despite these shortcomings, enervating agency discretion and capacity, as the sitting Supreme Court doctrine and the current presidential administration both seem set on doing, would actually undermine crucial functions of our system of government. Our findings thus have important implications for legal scholarship, the study of democratic institutions, and doctrinal and political debates over the scope and future of the administrative state. This research also helps us to identify how agencies can be made better, more accountable institutions.

A. Statutory Interpretation Scholarship

Most narrowly, our findings should influence legal scholarship on statutory interpretation, which traditionally has taken the perspective of courts and focused primarily on the judicial review of agency decisions.^[277] While more recent research has gone further afield to study agency interpretation itself, inquiries in this area still have a largely juricentric orientation.^[278] Scholars have asked whether agency statutory interpretation does or should resemble textualism or purposivism, or examined how much administrators know about, and follow, such judicial tenets and canons of interpretation.^[279] Our Article poses a different set of questions focused on how agencies understand what they are supposed to be doing with statutes. Our effort to answer these questions helps bring to light a set of political and legal practices that are crucial to understanding how law is given effect in the world—practices influenced but hardly determined by (and sometimes inconsistent with) those of courts.^[280]

Rather than seeking an abstract or semantic vantage onto statutory meanings, agencies use statutes to address issues or problems. They treat statutes as enactments meant to create effects. Our interviewees described agencies as responsible for the consequences of their actions; policy consequences thus become central to officials’ consideration of statutes. This practical orientation contrasts sharply with, for instance, textualism, which generally rejects basing statutory interpretation on its likely real-world effects.^[281] As agencies work to address the issues that come up in their statutory sphere of concern, the statute serves simultaneously as prod, constraint, and resource—combining with other resources such as a sense of mission, diverse public inputs, real-world circumstances, legislative pressures, and presidential preferences. The measure of success is not semantic correlation so much as the production of effective solutions to problems that a statute lays at the agency’s feet. It is the agency’s job, our interviewees made clear, to make statutory mandates effective, coherent, and implementable. This concern recognizes that practical implications are what make a statute a force in the world. One might even say that consequences constitute the statute.

Our findings suggest that textualism and purposivism, with their attendant canons and rules, are simply not that helpful for understanding how agencies work with statutes. Our interviewees certainly considered both texts and purposes, often in differentiated ways depending on their agencies’ internal divisions of labor. But text and purpose were just two factors of many, some of which the prescriptive theories do not recognize and cannot account for.^[282] Courts’ interpretive theories are tied up with the countermajoritarian difficulty, and their canons of interpretation rest on (often inaccurate) heuristics about the legislative process.^[283] Agencies, authorized by statutory delegations and intimately familiar with legislative production, have other concerns. Even interviewees familiar with judicial tenets did not claim to be guided by them, except to the extent adherence to those tenets might prove necessary to ensure that agency policy could withstand judicial review: These tenets do not help agencies address the issues that statutes place within their purview. As our discussion suggests, agencies cannot accurately be described as textualist, purposivist, or an eclectic mix of the two. They do something different that needs to be studied on its own terms. Indeed, our agency-focused perspective ultimately gives us insight into what goes on inside the bureaucracy and how statutory enactments come to have effects—insights necessary to any efforts to enhance agency accountability and crucial for evaluating the separation of powers and the demands of modern government.

As we elaborate in more detail below, we think the juricentric view that treats courts as the decisive players in giving the law meaning actually ignores the dictates of Congress, which depend on agencies to give statutes effects in the world.^[284] To give Congress its proper place in statutory effectuation and to recognize the realities of agency practice, we propose reorienting the discussion. Pace the Roberts Court, our work shows that grasping the meaning of a statute—understanding what it asks agencies to do—is not a purely legal question. Scholarship should not start with the way courts interpret statutes—a phrasing that itself evokes an abstract, semantic, cognitive understanding. Rather, scholars should start by focusing on how agencies work with statutes: how agencies convene multiple institutional participants to produce a workable plan for addressing issues within a statute’s domain. Freed from inapposite categories of judicial justification, this orientation clears a place to explore how agencies—and statutes—actually work in our democracy. From that point of view, we can then engage and evaluate interpretation by courts.

Even for juricentric scholarship, our insights should raise new questions. It may well be that certain judicial models or theories are preferable to the practices we have uncovered, or that courts should continue what they’re doing irrespective of what we learn about how agencies work with statutes. Perhaps the limited accountability and policy expertise of courts makes them uniquely ill-suited to engage in the sort of consequentialist decisionmaking we have described. But as we explore in more detail in subpart III(C) below, that assumption would still be compatible with (and might even require) deferential judicial review. To some readers with a more juricentric orientation, the divergence we document might suggest that it is agencies that should change. But why should we assume that courts and their particular methods of discerning statutory meaning hold pride of place in interpreting statutes that Congress charges agencies with effectuating? The courts’ divergence from agency models demands justification.

In its recent decision overruling Chevron, the Court provides an answer to this demand—interpreting statutes is just what courts do, they answer legal questions. But this observation is entirely conclusory, and our work shows why effectuating statutes by giving meaning to their texts cannot be reduced to acontextual “legal” question answering. Scholars need not make the same baseless assumptions as the Roberts Court—that agencies have no special expertise in interpreting statutes. It is arguably a burden of the juricentric scholar to explain why courts rather than agencies should play the leading role in ensuring that congressional will has been followed. No matter how hard the Supreme Court insists that agencies have no expertise when it comes to interpreting statutes, agency officials will continue doing so—their job within our system of government requires it.

B. Government Structure and Institutional Capacity

As we have emphasized throughout this Article, our research shows how agency work with statutes helps ameliorate the inherent tension between continuity and change in our democracy. In this subpart we spell out the comparative institutional advantage of relying on agencies to play this role. And once we understand agencies as sites for a particular sort of democratic action, we are in a better position to evaluate how well our system of separation of powers is working, to improve the practices of government and the efficacy of statutes, and to articulate the proper scope for judicial review of agency action.

As we have noted throughout, like any institution, agencies do their work imperfectly.^[285] Many struggle to incorporate the full range of relevant interests in their deliberations.^[286] Industry and other powerful players regularly have the loudest say.^[287] Agency action can ignore or even harm the interests of marginalized and disempowered groups.^[288] As with similar features of our other governing institutions,^[289] these limitations demand attention and amelioration. But it also becomes especially important to consider how agencies compare to other such institutions available to do the historical and sociological work we have described.^[290] We offer this “comparative institutional analysis” to help reveal our institutions’ relative abilities and to identify efficacious ways to bolster the system’s democratic qualities.^[291] Taking this systemic perspective, our research suggests that agencies play a central, and quite likely the primary, role in keeping statutes relevant and responsive.^[292] We have shown not only how statutes represent “works-in-progress,” rather than fixed meanings, but also how their development requires shaping by agencies through interaction with “the political branches and affected interests.”^[293]

Many critics of Chevron deference, and of contemporary administration more generally, insist that Congress itself should, and can, update statutes to address evolving circumstances.^[294] But the work we have shown agencies performing makes them necessary complements to congressional action. Congress does engage with agencies as they work with statutes, through its committees and individual members.^[295] Still, such congressional participation cannot be regarded as authoritative;^[296] treating it as relevant but not dispositive, as our interviewees did, seems appropriate.

More importantly, we think it both unrealistic and undesirable to demand that Congress constantly legislate anew. Such a burden would quickly outstrip congressional capacity and exceed the limits of a reasonable legislative agenda.^[297] Members of Congress are generalists with limited staffs facing regular elections while governing an enormously complex economy and society.^[298] Congress as a whole lacks the diversified expertise of the agencies it creates.^[299] Individual members of Congress, concerned with their geographic constituencies or donors, lack both the incentives and the institutional capacity to undertake the kind of broad consultation characteristic of agency work.^[300] Our findings suggest that several key features of agencies—specified focus, divisions of labor, multiple ongoing communication channels, and attenuated dependence on elections—allow them to engage in ongoing policy development through deliberation and interaction with relevant parties and publics in a way that Congress and its members simply cannot. Agencies thus possess a clear comparative advantage when addressing day-to-day problems and evolving demands for government action. We can, of course, point to instances in which new legislation would be far more effective at addressing novel challenges than regulation under old statutes. But that does not mean that Congress can or should legislate each time change is required. To base a theory of administrative power on that assumption is glib at best.

Critics of agency power also make a related claim: that agency decisionmaking is undemocratic and therefore must be circumscribed. Justices of the Supreme Court have expressed concern about agencies’ role in a government of separated powers, decrying administrative “edicts” as enforcing “unbounded policy choices,” or painting the administrative state as a rule-free zone where agencies issue commands at whim.^[301] Others impugn agencies for privileging the technocratic expertise of unelected bureaucrats, undermining Congress’s democratic will.^[302] As we have discussed at length elsewhere, our research shows how these criticisms are misplaced, underscoring that such claims are often motivated by a suspicion of regulation generally.^[303] The characterization of agency action as despotic bears little resemblance to the working world our interviewees described; the myriad dimensions of the duty of care underscore the deliberative, responsive, and consultative nature of agency work.

Critics of agencies’ role in a government of separated powers must also confront the fact that delegation to agencies embodies the choice of the legislature itself. Delegation is the means by which a sitting Congress projects its democratic choices into the world and the future.^[304] The agency practices our research reveals highlight that government is for governing: Workability itself represents a core constitutional value.^[305] Our research ultimately shows how agencies’ work with statutes can reinforce democratic responsiveness, providing an important, likely unique, site for pluralistic contestation over the specific policy choices the federal government must make for statutes to have concrete meaning.^[306] Prioritizing juricentric forms of interpretation, particularly those that would circumscribe delegation, will also often mean depriving democratically enacted statutes of their efficacy.^[307]

The idea that agencies co-create statutory meaning through practical implementation may raise concerns that agencies will distend statutes, manipulating them to achieve idiosyncratic policy objectives not contemplated or desired by elected officials.^[308] And indeed, we do not mean to suggest that agencies always produce desirable or even workable programs, nor that agency policymaking through existing statutory frameworks will always be preferable to fresh congressional action. Sometimes agency action will be insufficient to address new problems precisely because old laws addressed to old problems do not provide the requisite legal authority, or because old statutory regimes are simply too disconnected from an emergent reality to function well.^[309] Agency decisions can also be inefficient, misguided, distorted by ideological objectives, and as neglectful of society’s disadvantaged as the U.S. Code itself.^[310] Even the best procedures cannot eliminate the risk that agencies will bend statutes to the will of aggrandizing officials.^[311]

But given the relative institutional capacities we have emphasized, we think our best hope for minimizing the risks associated with agency action lies in strengthening the kinds of responsive practices we have described. We should facilitate broader participation at earlier stages of the policymaking process and ensure that agencies are both empowered and required to consider important normative values beyond economic costs.^[312] Insisting that Congress legislate with alacrity to meet every moment that demands a response, in contrast, undercuts workable, accountable governance.^[313]

As we intimated in subpart III(A), some might prefer to assign the historical and sociological mediation of policy preferences to courts.^[314] While we think courts can and should play a part in keeping statutes relevant to evolving circumstances (in ways we describe in more detail in subpart III(C)), our research suggests that the judiciary will have less capacity and competence to perform this role than agencies.^[315] This observation, in turn, suggests that courts should regard agencies at least as equals, if not leaders, in the interpretation of statutes—agencies do have special competence in interpreting statutes. For one thing, courts confront statutes sporadically, and only at the behest of litigants seeking specific relief. Agencies, by contrast, must maintain the statutory regime continuously, independently of the demands of aggrieved parties. Courts, limited by standing doctrines, can neither solicit nor process broad input on the myriad factors that relate to the purposes or efficacy of a statutory regime; they lack institutional mechanisms for taking into account the range of interests and views a given statute implicates. Agencies, in contrast, solicit such input both as a legal requirement for rulemaking and as a discretionary practice.^[316]

Perhaps most important, courts are staffed by legal generalists. They lack the capacity to fully assess the implications of real-world changes, evaluate options within the broader policy landscape, or assess how a given decision will affect other regulatory policies.^[317] In contrast, as our research shows, these kinds of assessments form a large part of the work of agencies, which are staffed by many people with diverse expertise who must work together to produce policy plans.^[318] Unlike agencies, moreover, courts do not participate in making legislation; most judges lack an understanding of the concerns that animate any given statute or statutory drafting in general.^[319] Agencies are in conversation with members of Congress, regulated parties, and affected publics.^[320] Courts, in other words, cannot match agencies’ capacity for reconciling directives from the past with the pluralistic demands of the present.^[321] Unlike courts, agencies do not have the luxury of merely interpreting some statutory term a litigant has placed at issue. Agencies must develop plans to effectuate the whole statute. And they must adapt those paths over time as the world to which the statute applies changes. Agencies’ work with statutes is thus more complex—both conceptually and in terms of the number of people and views involved—than courts’.

C. Doctrine

Contemporary Supreme Court jurisprudence is in stark tension with the conception of agencies’ role that we have just offered, increasingly placing courts at the statutory helm. Take Chevron deference. Even before the Court overruled Chevron during its 2023 Term, the Court had mostly abandoned the doctrinal regime that instructs courts to accept reasonable agency understandings of open-ended statutory terms.^[322] Courts that have either ignored or eschewed Chevron deference often have painted themselves as returning power usurped by the agency back to Congress.^[323] In fact, though, such moves undermine Congress, negating its ability to decide who should effectuate its statutes.^[324] Courts operating in this vein effectively claim for themselves the power to decide how regulatory statutes should be effectuated—decisions that, as we have noted, they have comparatively less capacity than agencies to make. These most recent decisions only promise to heighten the agency risk aversion we bring to light in Part II, which almost certainly will slow policymaking down. This very result may be precisely what Chevron’s foes have in mind, but it simply underscores how recent doctrinal developments increasingly put the judiciary in charge of how—and whether—statutes can have concrete meaning in the world.

The Court’s “major questions doctrine,” which it has broadened considerably in recent years, exacerbates this situation by imposing unrealistic, unpredictable, and ahistorical conditions on Congress’s choices as embodied in legislation.^[325] This doctrine, too, purports to serve Congress while instead aggrandizing power to courts, which now get to determine which agency actions will be allowed to go into effect absent a specific statutory authorization, with courts determining ad hoc how much specificity is needed—even where a broadly worded delegation comfortably encompasses the agency action in question.^[326] A growing literature rightly criticizes this new doctrine from a number of perspectives.^[327] Justice Kagan noted the irony well: “[T]he Court substitutes its own ideas about delegations for Congress’s. And that means the Court substitutes its own ideas about policymaking for Congress’s. The Court will not allow the [statute] to work as Congress instructed. The Court, rather than Congress, will decide how much regulation is too much.”^[328] Our research highlights how the doctrine undermines agencies’ pluralistic decisionmaking, undertaken with the people it will affect in mind. It also depletes Congress’s power to project its commands into the future.^[329] The major questions doctrine puts courts in charge of deciding how, if at all, multivalent statutes should apply to novel regulatory challenges and changed circumstances.^[330] The Court thus enervates agencies’ ability to perform a function necessary in a democracy—integrating the political will of past generations with the needs of the day.

The systemic understanding of statutory implementation we offer, and the concomitant critique of existing doctrine, do not mean that courts have no meaningful role to play in overseeing the administrative state. On the contrary, courts ought to recognize the key functions agencies serve in our democracy and push agencies to play their role well. We have shown, for instance, that agencies have the capacity to engage in regular pluralistic contestation about specific policy choices over time. Courts should acknowledge that competence and encourage agencies to use it. The drive to eliminate Chevron rests on the contention that courts are best positioned to identify a statute’s true, or best, meaning.^[331] But as our discussion here indicates, regulatory statutes often have no objectively best meaning; judges who find one without accounting for differentiated effects, competing interests, and workability considerations will often merely substitute their personal ideas for an agency’s more informed, pluralistic process—one that also has the benefit of congressional authorization.^[332]

The current Court has made clear its disdain for deference doctrines and embraced a conception of statutes as law with fixed meanings that courts must discern by their own lights. We offer an alternative approach to judicial review for the future—one that entails a more radical appreciation of agency work than even Chevron itself. When confronted with an agency interpretation of a statute, a court should ask whether the statutory regime supports the agency’s view of what the statute authorizes. In other words, we would not return to the so-called step 1 of Chevron and the idea that courts should attempt to use all of the tools of statutory construction to determine whether a statute is ambiguous. The pervasively open-ended nature of statutes means that, in many cases, such analysis is little more than an opportunity for judges to incorporate their own intuitions into their supposedly legalistic practice. And the varied means by which agencies implement statutes underscore that determining a statute’s meaning removed from context belies the very thing statutes are for—to provide frameworks for policy or action.

This picture could, in theory, emerge from a regime that applies Skidmore^[333] deference and therefore in the aftermath of Loper-Bright. Courts could begin with the agency’s account of what its statute authorizes and remain open to understanding the statute in light of what the statutory custodian has come up with. The demise of Chevron does not require otherwise. Such an approach respects Congress’s decision to instruct agencies to “build the thing” its statutes call for, and acknowledges that regulatory decisions are largely provisional, allowing and even requiring changes as circumstances develop.^[334] A court’s role would be to ensure that agencies exercise their duties of loyalty and care, establishing a sound legal basis for taking action. In this vein, in place of the major questions doctrine, we might embrace a cabined elephants-in-mouseholes canon, in which courts ensure that agencies do not promulgate unexpected policies that exceed the scope of statutory authorization.^[335] Instead of unrealistically claiming that agencies need specific direction from Congress to address whatever a court deems important, this inquiry would recognize that Congress delegates broadly and relies on agencies for intertemporal policy cohesion, in part because lawmakers cannot predict the future.

Our research also speaks to how courts might conduct arbitrariness review. Given how agencies work with statutes, they should not have to consider all conceivable policy options, nor eliminate all uncertainty about a policy’s likely effects.^[336] Policymaking is inevitably, and appropriately, informed by value judgments, which can provide adequate foundation for agency policy, including policy changes within an unchanged statutory regime. Judicial review should not seek to ferret out or eradicate such normative orientations, or require an agency to be so comprehensive in its policy work that it becomes sluggish at best, but should focus instead on whether a policy lacks a reasonable foundation or appears to result from untested whims.^[337]

We do not expect the current Court to suddenly reverse course based on our evidence-based critique of its dominant assumptions about the administrative state, any more than it has in response to arguments based in doctrine, interpretive theory, formalist concerns, and so on.^[338] Still, the Court, like any government institution, will change over time; future members may be more receptive to doctrines grounded in a reality-based approach that tolerates and even appreciates law’s indeterminacy. The lower courts, confronted with new challenges to agency policymaking based on open-ended statutory language, may begin to develop the sort of deference regime we describe above. Moreover, by showing how the Court’s doctrinal disruptions undermine the very democratic accountability it claims to promote, we hope to inform how scholars and other interested observers credit the Court’s pronouncements and seek to ameliorate their antidemocratic effects.

Even more important, Supreme Court Justices are not the only people who get a say in how government runs; democracy is fundamentally characterized by dispute about both the distribution and the content of power.^[339] We aim here to illuminate the everyday practices of agencies working with statutes, to inform all manner of officials, groups, and citizens operating in a policymaking world with a view to channeling popular will into concrete action. We believe that the pluralistic practices we uncover help make governance democratic, as well as effective. That should provide political support for the administrative state at a time when its greatest threat comes not from courts but from interest groups, lawmakers, and political officials who seek to dismantle the state and its capacity to govern.^[340]

Conclusion

The empirical realities our study reveals can help ground normative evaluations of administration. Our research supports the contention that agency action is ineluctably political: permeated not just with subject-matter and legal expertise but also with considerations of competing values, diverse interests, and public palatability. With one single exception, none of our interviewees claimed that working with statutes was simple or followed standardized rules. None claimed—or even claimed to strive for—the kind of neutrality sometimes attributed to idealized administration. In our view, this open acceptance of normative, political considerations is a strength, as is agencies’ concern about consequences. The implementation of statutes involves choices that inevitably distribute public goods and constraints unevenly; there is no neutral outcome. Claims to neutrality only obfuscate the value judgments and political considerations that inevitably attend statutory effectuation. The agency practices we present here do not eliminate the possibility of inequitable, unjust, or unwise outcomes.^[341] But they provide opportunities for ameliorating them. Our work shows that the way forward requires strengthening, not limiting, the tools agencies have to create policies that benefit the public interest.

Methods Appendix^[342]

This Appendix describes how we enrolled participants in the study, how we interviewed them, and how we addressed the resulting data. The interviews were oriented around investigating how agency personnel work with statutes.^[343]

A. Subject Population

Our aim with this project was not to interview a representative sample of government employees, nor to gather answers that could easily be tabulated and compared, but to conduct in-depth interviews that could reveal realities not easily captured by larger-scale methods. We conducted thirty-nine open-ended, semi-structured interviews with current and former agency employees, both political appointees and career civil servants, between July 2016 and May 2018. Our subjects came from eleven agencies: executive agencies, independent agencies, and the Executive Office of the President. Some focused on benefits management, others on industry regulation, and a couple on the governance of the government itself. A majority of our interviewees were no longer serving in the government.

The majority of the political appointees we interviewed served during the Obama Administration; several of those had also served during the Clinton Administration. Several served during the George W. Bush Administration, and one in the Trump Administration. The significant representation by Obama-era officials was primarily due to the ease of enrolling subjects, but also to the recency of their experience, which made their accounts fresher and more specific.

The career civil servants we interviewed spanned presidential administrations from George H.W. Bush to Trump. The bulk of their experience was during the George W. Bush and Obama years. Figure 1 lists the agencies where our subjects had worked over the course of their careers. Some subjects served in more than one administration, and some had moved between career and political roles over their careers.

Agencies Represented^[344]

1. Centers for Medicare and Medicaid Services (CMS)
2. Department of Education
3. Department of Homeland Security (DHS)
4. Department of Justice (DOJ)
5. Environmental Protection Agency (EPA)
6. Equal Employment Opportunity Commission (EEOC)
7. Federal Trade Commission (FTC)
8. Health and Human Services (HHS)
9. Housing and Urban Development (HUD)
10. Mine Safety and Health Administration (MSHA)
11. Small Business Administration (SBA)

B. Enrolling Participants

To recruit subjects, we primarily relied on a “snowball” approach, in which we contacted acquaintances in our personal and professional networks who worked in agencies, requested interviews, and asked for recommendations of or introductions to others who might be interested in participating in the study. In a broader solicitation approach, two of our contacts who had worked in the Obama Administration published our project solicitation email on listservs for former colleagues. We later learned that this email was republished on at least one other listserv. (This is one reason Obama-era subjects are highly represented in our interviews.)

In addition, having drawn several participants from one specific agency through personal contacts and snowball methods, we sent personal solicitations to a number of other colleagues in that agency and followed up with phone calls. (We identified possible participants by looking at agency websites and Federal Register publications, as well as through the interviews we had already conducted.) Although a number of people initially responded to this solicitation, most canceled before we could conduct the interviews after a department-wide email from a high-level official prohibited employees from participating in our study.^[345] After this, we relied only on the first two, informal, methods to solicit subjects. Every person who agreed to participate was interviewed.

Our recruitment email read as follows:

We’d like to invite you to participate in a research project on statutory interpretation and policymaking in federal agencies. We are two law professors ([redacted]) who specialize in administrative law and statutory interpretation. For this project, we are conducting confidential interviews with current and former agency personnel. We would be grateful for the chance to interview you about your work.

Our interviews explore how agency employees interpret unclear statutory terms, and how they formulate policies in the face of multiple pressures and constraints. The point is not to expose deficiencies in the regulatory process, but to educate scholars, judges, and the public about how our government actually works. It has never been more important to understand and explain the complex and vital work that agencies do. We hope illuminating the agency’s perspective will make an important contribution to the national conversation about regulation.

We are conducting interviews by phone, and can arrange to talk at a time convenient to you. Both the [redacted] Internal Review Boards have approved the project. (The attached “informed consent” document tells you more about the research and our strict confidentiality measures.) We plan to use material from these interviews in articles for publication in law reviews.

If you would be interested in participating, please contact us. We’d also be happy to talk about the project further. Thanks for considering it, and we hope to talk to you soon.

[Signatures]

The informed-consent document referenced in the recruitment email, which we sent to every prospective participant, included both an Institutional Review Board-mandated disclosure and a brief introduction to the study. It read as follows:

Research Study: Agency Statutory Interpretation and Policymaking

Thank you for agreeing to participate in the Agency Statutory Interpretation and Policymaking Study run by [redacted]. For this study, we are conducting confidential interviews about rulemaking practices with current and former federal agency personnel. In particular, we hope to explore how agency employees interpret unclear terms in the statutes they implement, and how they formulate policies in the face of multiple pressures and constraints. We hope that our results will add to knowledge about how government functions and how statutes are implemented over time. We also anticipate that this research will illuminate how administrators’ statutory interpretation practices differ from judges’. We believe the lack of attention to this topic has been detrimental both to legal scholarship and to public discourse, and we hope to make a contribution to both.

The human subjects research review boards at [redacted] have approved this study. All of your responses will be held in the strictest confidence. Only the researchers involved in this study and those responsible for research oversight at [redacted] will know your identity or have access to the information you provide. We plan to audio tape the interview. If you prefer not to be recorded, please advise us and we will note your responses by hand instead. Notes or transcripts of your interview will be numbered, and the code linking your number with your name will be stored in a separate file from the transcript, on a secure server and password protected computer. After publication of the study, transcripts will be stored on a secure server without any identifying information connected to them. In addition to avoiding any direct links between your identity and what you tell us, we will do our utmost to ensure that nothing we publish based on this study can be traced back to you individually.

Participation in this study will involve an interview, by phone or in person, of approximately one hour. There is no compensation for participation. There are no known or anticipated risks to you for participating. Participation in this study is completely voluntary. You are free to decline to participate, to end participation at any time for any reason, or to refuse to answer any individual questions. Your decision whether to participate in this study will not affect your relationship with [redacted].

If you would like to speak with someone other than the researchers to discuss problems or concerns, to discuss situations in the event that a member of the research team is not available, or to discuss your rights as a research participant, you may contact the [redacted].

If you would like to speak with the researchers, please feel free to contact us.^[346]

Brief Study Synopsis

Modern government depends in large part on how agencies interpret statutory provisions and make policy, but legal scholarship generally focuses instead on judges and legislators. This study will be among the first to illuminate how agency personnel interpret statutory provisions. We hope to highlight the creativity and difficulty of making a statutory scheme work on the ground. We intend to interview a range of agency administrators about the tools, assumptions, approaches, methods, and reflections that guide administrators’ interpretation and implementation of statutes. We hope to give the administrator’s point of view a voice that is missing from scholarship and doctrine.

We will ask administrators to discuss, in their own words, what they do when confronted with a complex or ambiguous statute, how they formulate policy based on the statute, and what their work means to them. We will also ask how agencies structure the interpretive and policymaking process: What sorts of officials make which decisions during the different phases of developing a guidance or a rule?

We do not aim to expose inadequacies in agency practices, nor do we have a predetermined view about how agencies ought to do their work. On the contrary, we feel that the important work administrators do to bring statutes to life has been obscured by the scholarly focus on courts. We aim to inject a pragmatic, real-world understanding of agency practices into both scholarship and doctrine, to complement and perhaps change the theories on which courts rely when evaluating agency action.

C. The Interviews

A few interviews took place in person, but most were conducted over Zoom (usually using audio only). We conducted almost all interviews together, usually each asking about half of the questions. Sometimes, with participants’ permission, we had research assistants listening on the line as well. Most interviews lasted between one and two hours, though several lasted longer or even spanned two separate calls, at the interviewee’s request. Almost all interviewees agreed to record the interviews; for the few that declined, we took notes by hand. Recorded interviews were professionally transcribed.^[347]

To develop an interview guide to structure the conversations, we considered our own primary interests and reviewed scholarly literature to assess where we could add to existing knowledge. We also consulted with colleagues in academia and contacts with experience in the federal administration to ensure our questions were relevant and comprehensible. Because we were not aiming for a statistical comparison of participant answers, we did not treat the guide as a survey instrument. Rather, it provided a reminder of the topics we wanted to cover, while still encouraging participants to focus on the matters of greatest concern or interest to them. This guide did not exhaust the questions asked. We generally asked follow-up questions in each section to explore the interviewee’s responses and experiences further.

Our interview protocol started with open-ended questions about the interviewee’s main roles or projects in government and the normal process of producing policy (through rules or guidance) in their agency. We then asked about the role of Congress (including congressional intent, legislative history, and consultation with Congress); other sources or methods for statutory interpretation; and the roles of other agencies, the White House, states, and the public. We also asked about the role of courts (including prospective consideration of litigation risk, litigation practices and outcomes, judicial doctrines like Chevron, and interpretive theories like textualism and purposivism). We invited interviewees to explain in their own terms how they thought about statutory interpretation and implementation. We also asked whether our questions elicited what they thought was relevant about the work they did and whether there were other questions we should ask.

Interview Guide Overview

Our aim was to encourage our subjects to discuss their particular experiences and views without confining them to a set of predetermined options or answers. We adjusted the interview protocol somewhat for each interview, pursuing the leads interviewees offered and spending more or less time on particular topics depending on what interviewees regarded as important in their work. Where possible, we also tailored interviews to interviewees’ particular backgrounds and the projects they had worked on (which we asked about when setting up interviews). This helped us ground our questions in specific interviewee experiences, which are often easier to talk about and more revealing of actual practices than abstract or conceptual discussions. Our interview protocol also developed over time as subjects suggested relevant questions or pointed us to more productive ways of pursuing an inquiry.

Framing questions in terms of an interviewee’s specific background grounds the conversation in their personal experiences and creates a specific joint object of focus for interviewees and interviewers. It allows interviewees to focus on a particular experience they have gone through and describe it, then examine whether they feel it was unusual in the context of their broader experience. That takes pressure off the interviewee to provide a neat or universal description of a process that may be complex or variable, and it helps us get at the particularities of participants’ experiences of working in the agency. It also lays the ground for follow-up questions that are easier to formulate and to answer, since we can ask about the particular experiences the interviewee has already described.^[348] Often, we would come back to particular situations an interviewee had discussed as we went through the interview, using interviewees’ descriptions as an anchor for further conversation.

D. Synthesizing the Data

Using both our interview protocol and our emerging sense of subject responses from both interviews and transcripts, we, along with our research assistants, developed a code book for coding responses.^[349] We had several aims in constructing the code book: to capture the key issues we asked about; to reflect the key themes emerging in interviewees’ responses; and to be easily usable. Together with our research assistants, we developed codes, tested them on a few transcripts, and further refined them in subsequent discussions. The following table presents the coding categories in our code book, along with the shorthand explanations that we developed for all coders to use.

Code book: Coding categories and explanations (as listed on a “cheat sheet” used by all coders)

Along with two research assistants per interview transcript, we each independently coded each transcript, then talked through discrepancies to come to a consensus on appropriate categories, creating a master-coded transcript for each interview. Research assistants then collated the coded text into spreadsheets showing all the text for each coding label. Each entry indicates the speaker (by pseudonym) and shows what other coding labels we applied to the same text passage (since a single text passage often fit several coding labels). Each text entry has a unique number; this Article cites interview quotations by reference to this number. The coding process also gave us an ambient sense of the material and helped us identify the topic areas where our diverse respondents’ discussions converged to reveal underappreciated aspects of agencies’ work with statutes. Of course, no single methodology or study could reveal all that is important about administration. We hope our research spurs more empirical work, using diverse methodologies, to illuminate the complex operations of our government.

1. . Chevron, U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984). ↑
2. . Id. at 843 & n.9. ↑
3. . Loper Bright Enters. v. Raimondo, 144 S. Ct. 2244, 2266 (2024). ↑
4. . Loper Bright Enters. v. Raimondo, 144 S. Ct. 2244 (2024). ↑
5. . West Virginia v. EPA, 142 S. Ct. 2587, 2609–10 (2022) (demanding highly specific statutory authorization for an administrative policy a court deems particularly important); Biden v. Nebraska, 143 S. Ct. 2355, 2374 (2023) (“[W]hile the major questions ‘label’ may be relatively recent, it refers to ‘an identifiable body of law that has developed over a series of significant cases’ spanning decades.” (quoting West Virginia, 142 S. Ct. at 2609)); see also Beau J. Baumann, The Major Questions Doctrine Reading List, Yale J. on Regul.: Notice & Comment (Mar. 18, 2022), https://www.yalejreg.com/nc/the-major-questions-doctrine-reading-list-by-beau-j-baumann [https://perma.cc/CJL7-Z335] (tracking academic work on the major questions doctrine); N.C. Coastal Fisheries Reform Grp. v. Capt. Gaston LLC, 76 F.4th 291, 297 (4th Cir. 2023) (holding that “whether returning bycatch qualifies as a ‘discharge’ of a ‘pollutant’ under the [Clean Water] Act is a major question”); Georgia v. President of the U.S., 46 F.4th 1283, 1296 (11th Cir. 2022) (holding that the decision to require government contractors to be vaccinated against COVID-19 was major and required “clear congressional authorization” (quoting West Virginia, 142 U.S. at 2609)). See generally Blake Emerson, Administrative Answers to Major Questions: On the Democratic Legitimacy of Agency Statutory Interpretation, 102 Minn. L. Rev. 2019 (2018) (detailing and evaluating the underpinnings of the major questions doctrine); Michael Coenen & Seth Davis, Minor Courts, Major Questions, 70 Vand. L. Rev. 777 (2017) (arguing that only the Supreme Court should have the authority to decide what constitutes a “major” question); Daniel T. Deacon & Leah M. Litman, The New Major Questions Doctrine, 109 Va. L. Rev. 1009 (2023) (analyzing the shifting in role of “majorness” in statutory interpretation); Anya Bernstein & Glen Staszewski, Populist Constitutionalism, 101 N.C. L. Rev. 1763 (2023) (discussing the major questions doctrine’s role in shifting power from institutions most responsive to pluralistic contestation to those least responsive to it); Allison Orr Larsen, Becoming a Doctrine, 76 Fla. L. Rev. 1 (2024) (exploring how doctrines develop by analyzing the major questions doctrine). ↑
6. . See, e.g., Nat’l Fed’n of Indep. Bus. v. OSHA, 142 S. Ct. 661, 669 (2022) (Gorsuch, J., concurring) (praising the Court’s enjoining of the Biden Administration’s vaccine-or-test regulation for helping “to prevent ‘government by bureaucracy supplanting government by the people’” (quoting Antonin Scalia, A Note on the Benzene Case, Regul., July/Aug. 1980, at 25, 27)); Seila L. LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2204 (2020) (expressing worry that removal protections for agency officials can cause them to “slip from the Executive’s control, and thus from that of the people” (quoting Free Enter. Fund v. Pub. Co. Acct. Oversight Bd., 561 U.S. 477, 499 (2010))); cf. Ala. Ass’n of Realtors v. Dep’t of Health & Hum. Servs., 141 S. Ct. 2485, 2490 (2021) (“It is up to Congress, not the CDC, to decide whether the public interest merits further action . . . .”). ↑
7. . See, e.g., Lisa Schultz Bressman & Abbe R. Gluck, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part II, 66 Stan. L. Rev. 725, 774 (2014) (quoting a legislative drafter as saying, “The last thing we want is for courts to decide what your law means”); Jesse M. Cross, Legislative History in the Modern Congress, 57 Harv. J. on Legis. 91, 109 (2020) (reporting that “committee legislative staff” who produce legislative history “consistently target a specific intended audience with their documents and statements: legal implementers,” and “[t]his includes both courts and executive agencies”); Jarrod Shobe, Agencies as Legislators: An Empirical Study of the Role of Agencies in the Legislative Process, 85 Geo. Wash. L. Rev. 451, 478 (2017) (detailing extensive agency involvement in statute drafting). ↑
8. . See, e.g., Blake Emerson, The Public’s Law: Origins and Architecture of Progressive Democracy 1–2 (2019) (noting that “[t]he modern democratic state is an administrative state” and arguing that administration should be “structure[d] . . . to empower the public sphere”); Daniel E. Walters, The Administrative Agon: A Democratic Theory for a Conflictual Regulatory State, 132 Yale L.J. 1, 8 (2022) (identifying the administrative state as “the most practically important institution for making and implementing government policy”); Glen Staszewski, Justice by Means of the Administrative State, 122 Mich. L. Rev. 1231, 1231 (2024) (reviewing Danielle Allen, Justice by Means of Democracy (2023)) (arguing that regulatory agencies should be recognized as the “primary institutional site” for implementing certain design principles and “securing justice by means of democracy”); K. Sabeel Rahman, Democracy Against Domination 3 (2016) (noting that administrative “agencies [are] the institutions most responsible for the daily business of governing”); Katharine Jackson, The Public Trust: Administrative Legitimacy and Democratic Lawmaking, 56 Conn. L. Rev. 1, 51 (2023) (legitimating administrative agencies through a “trustee model of democratic representation,” in which an authorized trustee acts “in the public interest,” making discretionary decisions “that . . . can provide representation for those otherwise shut out of the collective decision-making process”); Bernstein & Staszewski, supra note 5, at 1778–82 (explaining that “agencies are a primary site for democratic governance” because of “their multiple channels for diverse inputs and their strong structural safeguards against arbitrary decisions”); Bernardo Zacka, Political Theory Rediscovers Public Administration, 25 Ann. Rev. Pol. Sci. 21, 22 (2022) (describing public administration as the “stable core of the executive branch” while noting that scholars’ attention to it has been uneven). ↑
9. . Jerry L. Mashaw, Norms, Practices, and the Paradox of Deference: A Preliminary Inquiry into Agency Statutory Interpretation, 57 Admin. L. Rev. 501, 504 (2005). ↑
10. . Id. at 536–37. ↑
11. . For instance, scholars have argued that agencies are, or could reasonably be expected to be, purposivist statutory interpreters. See, e.g., Michael Herz, Purposivism and Institutional Competence in Statutory Interpretation, 2009 Mich. St. L. Rev. 89, 93–106 (emphasizing the institutional dimension of purposivism and evaluating purposivism from the perspective of agencies, rather than courts); Kevin M. Stack, Purposivism in the Executive Branch: How Agencies Interpret Statutes, 109 Nw. U. L. Rev. 871, 875–76 (2015) (arguing that Congress “directs agencies to adopt a purposive interpretive method” so that agencies conform their conduct to the purposes and principles established in their respective regulatory statutes). That makes sense if we assume that the relevant styles of interpretation are sufficiently captured by the terms set by judges who express adherence to the competing prescriptive tenets of purposivism and textualism. Our work, in contrast, suggests that neither textualism nor purposivism suffices to describe what agencies actually do. ↑
12. . See, e.g., Amy Semet, An Empirical Examination of Agency Statutory Interpretation, 103 Minn. L. Rev. 2255, 2306–11 (2019) (concluding that NLRB decisions rely on legislative history to mask policy choices); Christopher J. Walker, Inside Agency Statutory Interpretation, 67 Stan. L. Rev. 999, 1012, 1025 (2015) (surveying agencies’ use of tools of judicial statutory interpretation); Christopher J. Walker, Chevron Inside the Regulatory State: An Empirical Assessment, 83 Fordham L. Rev. 703, 704, 716 (2014) (surveying regulation drafters’ understanding and use of judicial deference doctrines). ↑
13. . See Anya Bernstein, Before Interpretation, 84 U. Chi. L. Rev. 567, 635 (2017) (“Despite their explicitly prescriptive, ideological character, . . . theories of statutory interpretation are sometimes treated as though they were tools for analysis, rather than for normative evaluation[, which] . . . allows the infrastructure of interpretation to go unrecognized.”). ↑
14. . See Elizabeth V. Foote, Statutory Interpretation or Public Administration: How Chevron Misconceives the Function of Agencies and Why It Matters, 59 Admin. L. Rev. 673, 675 (2007) (critiquing the Supreme Court’s equation of public administration and statutory construction in Chevron). ↑
15. . This research involved thirty-nine in-depth, open-ended interviews with career civil servants and political appointees from eleven agencies that differed in size, age, and task. We asked about participants’ experiences with interpretive approaches and tools; relations to Congress, the President, the judiciary, other agencies, and the states; and participants’ own views about statutory interpretation. Semi-structured interviews like ours allow participants’ experiences and perspectives to produce information and views unfamiliar to the interviewer. The Methods Appendix discusses our methodology in depth. See also Anya Bernstein & Cristina Rodríguez, The Accountable Bureaucrat, 132 Yale L.J. 1600, 1612–14 (2023) (briefly describing methodology). ↑
16. . See, e.g., Nicholas R. Parrillo, Federal Agency Guidance and the Power to Bind: An Empirical Study of Agencies and Industries, 36 Yale J. on Regul. 165, 173 (2019) (relying on interviews with administrators to analyze the effects of guidance documents); Michael Sant’Ambrogio & Glen Staszewski, Democratizing Rule Development, 98 Wash. U. L. Rev. 793, 799 (2021) (interviewing administrative officials to identify how agencies involve the public in rulemaking); Shobe, supra note 7, at 454 (investigating agencies’ role in the legislative process through interviews with agency counsel and staff); Christopher J. Walker, Federal Agencies in the Legislative Process: Technical Assistance in Statutory Drafting 2 (2015) (interviewing “agency rule drafters” to “reinforce that federal agencies play an important and substantial role in the legislative process”). ↑
17. . See infra Parts I–II. ↑
18. . Interview Comment No. 110. Interview Comment citation numbers indicate the unique numerical identifier we gave a given comment in our compilation of coded interview transcripts. ↑
19. . See Jerry L. Mashaw, Between Facts and Norms: Agency Statutory Interpretation as an Autonomous Enterprise, 55 U. Toronto L.J. 497, 508 (2005) (“Statutes persist while presidents and congresses change. In this context, the agency becomes the guardian or custodian of the legislative scheme as enacted.”); Jody Freeman & David B. Spence, Old Statutes, New Problems, 163 U. Pa. L. Rev. 1, 7 (2014) (“[B]ecause the agency is the legally designated custodian of the statute (so designated by the enacting Congress), the agency has the superior claim to interpret the statute’s application to new problems during periods of congressional quiescence.”). ↑
20. . Evan J. Criddle, Fiduciary Foundations of Administrative Law, 54 UCLA L. Rev. 117, 121 (2006). Recent scholarship has offered a number of articulations of the public fiduciary idea. See, e.g., Ethan J. Leib, Three Modalities of (Originalist) Fiduciary Constitutionalism, 63 Am. J. Legal Hist. 183, 185–88 (2023) (offering an overview and typology of the literature); Ethan J. Leib & Stephen R. Galoob, Fiduciary Political Theory: A Critique, 125 Yale L.J. 1820, 1854–55 (2016) (describing Evan Criddle’s interpretations of fiduciary theories of administrative governance and administrative rulemaking); Aaron Saiger, Agencies’ Obligation to Interpret the Statute, 69 Vand. L. Rev. 1231, 1232 (2016) (“[T]he agency bears a legal and ethical duty to select the best interpretation of its governing statute.”); D. Theodore Rave, Politicians as Fiduciaries, 126 Harv. L. Rev. 671, 712 (2013) (figuring Congress as the fiduciary); Ethan J. Leib, David L. Ponet & Michael Serota, A Fiduciary Theory of Judging, 101 Calif. L. Rev. 699, 701 (2013) (figuring the judiciary as the fiduciary); Joshua Segev, The (Unified?) Fiduciary Theory of Judging: Hedgehogs, Foxes, and Chameleons, 8 Faulkner L. Rev. 73, 74 (2016) (examining the judge-as-fiduciary model’s relationship to Anglo-American tradition); see also Jackson, supra note 8, at 35 (proposing the trustee model as an alternative to recent theories of fiduciary government). This literature has also prompted some dissent. See, e.g., Samuel L. Bray & Paul B. Miller, Against Fiduciary Constitutionalism, 106 Va. L. Rev. 1479, 1481–83 (2020) (offering three critiques of fiduciary constitutionalism); Seth Davis, The False Promise of Fiduciary Government, 89 Notre Dame L. Rev. 1145, 1148 (2014) (arguing that “the promise of fiduciary government is a false one”); Kelli Alces Williams, Leaders Are Not Fiduciaries, 72 Ala. L. Rev. 363, 368 (2020) (arguing that “[l]eadership of large, diverse groups is, by its very nature, incompatible” with a theory of fiduciary government); Suzanna Sherry, The Imaginary Constitution, 17 Geo. J.L. & Pub. Pol’y 441, 442–43 (2019) (summarizing and subsequently criticizing fiduciary constitutionalism as resting on an “imaginary” understanding of the Constitution); Katharine Jackson, Administration as Democratic Trustee Representation, 29 Legal Theory 314, 315–16 (2023) (rejecting the “transmission belt” model of constitutionalism and the idea that “the people’s will travels on the back of the franchise from voter to administrator”). ↑
21. . Evan J. Criddle, Fiduciary Administration: Rethinking Popular Representation in Agency Rulemaking, 88 Texas L. Rev. 441, 473 (2010). ↑
22. . Andrew Kent, Ethan J. Leib & Jed Handelsman Shugerman, Faithful Execution and Article II, 132 Harv. L. Rev. 2111, 2119 (2019) (arguing that the Constitution’s imperative of faithful execution imposes a form of fiduciary duty on the executive); Gary Lawson & Guy Seidman, “A Great Power of Attorney”: Understanding the Fiduciary Constitution 28 (2017) (arguing that the Framers were informed by fiduciary law in writing and understanding the Constitution); Evan D. Bernick, Faithful Execution: Where Administrative Law Meets the Constitution, 108 Geo. L.J. 1, 6 (2019) (tracing the “fiduciary roots of the Take Care Clause” in the Constitution); see also Ethan J. Leib & Jed Handelsman Shugerman, Fiduciary Constitutionalism: Implications for Self-Pardons and Non-Delegation, 17 Geo. J.L. & Pub. Pol’y 463, 465–71 (2019) (grounding fiduciary constitutionalism in the Take Care Clause and the presidential pardon power). ↑
23. . See Gregory A. Elinson & Jonathan S. Gould, The Politics of Deference, 75 Vand. L. Rev. 475, 484 (2022) (discussing how “partisans and interest groups often use doctrine to achieve substantive political ends”); id. at 485 (“[I]n the aggregate, deference advantages the forces of regulatory initiative while de novo review advantages the forces of regulatory inertia and deregulation.”); cf. Jerry L. Mashaw & David L. Harfst, The Struggle for Auto Safety 21 (1990) (“[Our] legal culture . . . denies the need for or legitimacy of administrative regulation.”). ↑
24. . Broadly speaking, we take textualism as some combination of “a commitment to rely on a limited universe of text-based interpretive tools such as plain meaning, dictionary definitions, the whole act and whole code rules, and language canons—and to avoid consulting purposive tools such as legislative history, statutory purpose, legislative intent, and practical consequences”; “a commitment to identify a statute’s ‘original public meaning’ as of the date the statute was enacted”; and a “commitment . . . to identify the ordinary meaning that a statute would have to average citizens or laypersons.” Anita S. Krishnakumar, Textualism in Practice, 74 Duke L.J. 573, 577 (2024). ↑
25. . Broadly speaking, we take purposivism to hold that “how Congress makes its purposes known, through text and reliable accompanying materials constituting legislative history, should be respected, lest the integrity of legislation be undermined.” Robert A. Katzmann, Judging Statutes 4 (2014). ↑
26. . T. Alexander Aleinikoff, Updating Statutory Interpretation, 87 Mich. L. Rev. 20, 21–22 (1988). ↑
27. . Agency mission thus comes closer to a “nautical” approach that views statutory meaning as “an on-going process (a voyage) in which both the” enacting Congress and other participants “play a role.” Id. at 21. Purposivist tenets tend to operate with some specificity: they are used when considering particular statutory terms. See John F. Manning, The New Purposivism, 2011 Sup. Ct. Rev. 113, 116–19 (2012) (highlighting converging justifications and applications of purposivism and textualism). The general kinds of orienting mission concepts we heard—keeping the air clean, helping small businesses get contracts, maintaining the public healthcare budget—rarely appear in purposivist analysis. And probably for good reason: These overarching missions would hardly tell a court how to define some individual term in a statute’s numerous provisions. ↑
28. . Cf. John R. Hibbing & Elizabeth Theiss-Morse, Stealth Democracy: Americans’ Beliefs About How Government Should Work 7 (2002) (arguing that a broad American public prefers that political “decisions [be] made by neutral decision makers who do not require sustained input from the people”). ↑
29. . The relationship between laws made in the past and people governed by them in the present is an enduringly complex aspect of democratic governance. On the one hand, one can sense that “[w]e are indebted to those who came before us, for it is through their efforts that the world of culture we inhabit now exists,” giving social participants “a duty to respect and to conserve the achievements of their predecessors.” Anthony T. Kronman, Precedent and Tradition, 99 Yale L.J. 1029, 1057, 1067 (1990). On the other, one can declare independence: “by the law of nature, one generation is to another as one independant nation to another.” Letter from Thomas Jefferson to James Madison (Sept. 6, 1789), in 15 The Papers of Thomas Jefferson 392, 395 (Julian P. Boyd ed., 1958). Intermediating institutions can help find balance between the two: “the judge in shaping the rules of law must heed the mores of his day.” Benjamin N. Cardozo, The Nature of the Judicial Process 103–05 (1921). Even more so, we argue, can the agency, which is particularly suited to addressing “polycentric” problems. See Lon L. Fuller, The Forms and Limits of Adjudication, 92 Harv. L. Rev. 353, 394–95, 398 (1978) (explaining the concept of polycentric tasks, which are best resolved by “managerial direction”). ↑
30. . For discussion of how the assumptions made in recent Supreme Court decisions about agency action are unmoored from actual practice, see infra Part III. ↑
31. . See West Virginia v. EPA, 142 S. Ct. 2587, 2609–10 (2022) (“[T]he major questions doctrine . . . address[es] a particular and recurring problem: agencies asserting highly consequential power . . . .”); Biden v. Nebraska, 143 S. Ct. 2355, 2374 (2023) (“[W]hile the major questions ‘label’ may be relatively recent, it refers to ‘an identifiable body of law that has developed over a series of significant cases’ spanning decades.” (quoting West Virginia, 142 S. Ct. at 2609)). For examples of the burgeoning work on the major questions doctrine, see supra note 5. There has also been a rich scholarly discussion of the unitary executive theory. See Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2246, 2248–50 (2001) (proposing a “presidential administration” theory); Steven G. Calabresi, Some Normative Arguments for the Unitary Executive, 48 Ark. L. Rev. 23, 86 (1995) (“The best we can do is to let the spokesman for a quadrennial, energized national majority administer the laws.”); Anya Bernstein & Cristina Rodríguez, The Diffuse Executive, 92 Fordham L. Rev. 363, 371 (2023) (“[A] diffused executive, which can accommodate presidential leadership without insisting on its dominance, helps us as a nation to better approximate the democratic ideal.”). ↑
32. . The Methods Appendix discusses our methodology in more detail. ↑
33. . By “semi-structured,” we mean that the interviews were guided by an overarching set of questions explored with each participant but were not limited to that particular set of questions. We asked follow-up questions and allowed the conversation to develop somewhat naturally within the basic overall structure we had set for ourselves. See, e.g., Barbara DiCicco-Bloom & Benjamin F. Crabtree, The Qualitative Research Interview, 40 Med. Educ. 314, 315 (2006) (describing semi-structured interviews). This approach gives interviewers insight into the things respondents find important and allows respondents to present interviewers with new information or perspectives that interviewers might not have known to inquire about. ↑
34. . Interview quotations in this Article often elide specifics to help keep interviewees anonymous. ↑
35. . The Methods Appendix presents our basic interview outline. ↑
36. . See generally Telemarketing Sales Rule, 60 Fed. Reg. 43842 (Aug. 23, 1995) (codified at 16 C.F.R. pt. 310) (making no reference to the influences and processes mentioned in interview comments); Adverse Effect Wage Rate Methodology for the Temporary Employment of H-2A Nonimmigrants in Non-Range Occupations in the United States, 85 Fed. Reg. 70445 (Nov. 5, 2020) (codified at 20 C.F.R. pt. 655) (likewise); National Standards to Prevent, Detect, and Respond to Prison Rape, 76 Fed. Reg. 6248 (Feb. 3, 2011) (codified at 28 C.F.R. pt. 115) (likewise). ↑
37. . It is difficult, for instance, to learn about the on-the-ground, backstage practices that go into producing a particular opinion. See, e.g., Abbe R. Gluck & Richard A. Posner, Statutory Interpretation on the Bench: A Survey of Forty-Two Judges on the Federal Courts of Appeals, 131 Harv. L. Rev. 1298, 1301, 1310 (2018) (reporting on a survey asking judges to self-report their general views and practices regarding statutory interpretation, without inquiring into the production of any particular interpretive outcome). ↑
38. . See, e.g., Krishnakumar, supra note 24, at 596 (showing that Supreme Court Justices who publicly identify as textualist do not confine themselves to textualist interpretive methodology in practice); Jane S. Schacter, Text or Consequences?, 76 Brook. L. Rev. 1007, 1013–14 (2011) (showing the same). ↑
39. . Our work thus contributes to the small but important body of qualitative research illuminating the workings of the American state and its effects on regulated parties. See, e.g., Anna Offit, The Imagined Juror 23, 34 (2022) (presenting an ethnography of how the idea of the jury figures in everyday prosecutorial decisions); Wendy Nelson Espeland, The Struggle for Water 44–45 (1998) (providing an ethnography of Bureau of Land Reclamation administrators’ decision-making process); Justin B. Richland, Jurisdictions of Significance: Narrating Time‐Space in a Hopi‐US Tribal Consultation, 45 Am. Ethnologist 268, 278 (2018) (providing an ethnography of an encounter between Hopi elders and U.S. Forest Service officials); Brittany Kiessling, Keely B. Maxwell & Jenifer Buckley, The Sedimented Social Histories of Environmental Cleanups: An Ethnography of Social and Institutional Dynamics, J. Env’t. Mgmt., 2021, at 1, 6 (providing an ethnography of Environmental Protection Agency employees’ “decision-making power in environmental cleanups”); Josiah McC. Heyman, Putting Power in the Anthropology of Bureaucracy: The Immigration and Naturalization Service at the Mexico–United States Border, 36 Current Anthropology 261, 261, 266 (1995) (analyzing how the U.S. Immigration and Naturalization Service’s worldviews affected its organizational power); Parrillo, supra note 16, at 173 (analyzing agency guidance through interviews with a range of individuals involved in their formulation and production); U.S. Gov’t Accountability Off., GAO-03-455, Federal Programs: Ethnographic Studies Can Inform Agencies’ Actions 3–4 (2003) (finding that ethnographic studies can inform agencies’ actions). ↑
40. . Cf. Clifford Geertz, “From the Native’s Point of View”: On the Nature of Anthropological Understanding, Bull. Am. Acad. Arts & Scis., Oct. 1974, at 26, 29 (“The trick is to figure out what the devil they think they are up to.”). While we did not engage in traditional anthropological participant-observation, we did take an “ethnographic attitude” in this project—that is, we strove to understand how systems and practices make sense from the inside, as well as evaluating them from the outside. See Anya Bernstein, Saying What the Law Is, 48 L. & Soc. Inquiry 14, 15 (2023) (“I take an ethnographic attitude, using anthropological tools to analyze legal artifact.”); Anya Bernstein, Bureaucratic Speech: Language Choice and Democratic Identity in the Taipei Bureaucracy, 40 Pol. & Legal Anthropology Rev. 28, 31, 43 n.1 (2017) (providing an ethnographically based analysis of administrative practices). ↑
41. . See Richard B. Stewart, The Reformation of American Administrative Law, 88 Harv. L. Rev. 1667, 1675 (1975) (“The traditional model of administrative law thus conceives of the agency as a mere transmission belt for implementing legislative directives in particular cases.”). ↑
42. . See Kagan, supra note 31, at 2246 (“We live today in an era of presidential administration.”). ↑
43. . See infra Part III. ↑
44. . For discussion of scholarly literature exploring such intertemporal dynamics, see infra notes 275–276, 321 and accompanying text. ↑
45. . Interview Comment No. 279. ↑
46. . Mashaw, supra note 19, at 508 (“Statutes persist while presidents and congresses change. In this context, the agency becomes the guardian or custodian of the legislative scheme as enacted.”); Freeman & Spence, supra note 19, at 7 (“[B]ecause the agency is the legally designated custodian of the statute (so designated by the enacting Congress), the agency has the superior claim to interpret the statute’s application to new problems during periods of congressional quiescence.”). ↑
47. . Interview Comment No. 1775. ↑
48. . Interview Comment No. 1815. ↑
49. . Id. ↑
50. . Id. ↑
51. . Interview Comment No. 1380; see also Interview Comment No. 269 (“I mean, sometimes the statute would be pretty clear. They would say, ‘Okay, you got to pay at this rate, 50% or 80% of X and that’s it.’ There was no question what they intended and what they wanted.”). ↑
52. . See, e.g., Interview Comment No. 1307 (“[S]tatute[s] can be pretty vague.”). ↑
53. . Interview Comment No. 203. ↑
54. . Even Justice Neil Gorsuch, generally critical of delegation to agencies, has sanctioned allowing agencies to decide on technical details. See Gundy v. United States, 139 S. Ct. 2116, 2136 (2019) (Gorsuch, J., dissenting) (“[A]s long as Congress makes the policy decisions . . . it may authorize another branch to ‘fill up the details.’”). But even supposedly technical questions leave agencies making important policy tradeoffs. See, e.g., Interview Comment No. 371 (“[Some] laws [Congress has passed] frankly were almost impossible to implement, like the 100% screening requirement for cargo entering the United States. . . . It has been incredibly challenging, and we’ve [consistently] gotten waivers from Congress . . . because it hasn’t been possible for . . . all kinds of [resource and] technical reasons.”); Interview Comment No. 1850 (“[T]here really wasn’t . . . much textual guidance in the [Children’s Online Privacy Protection Act]. To the extent we had a North Star, Congress didn’t want third parties tracking 12-and-unders without parental consent, and could this piece of information that was being collected be used to do that. Sometimes the technology didn’t yield a particularly clearcut answer.”). ↑
55. . See Joseph A. Grundfest & A.C. Pritchard, Statutes with Multiple Personality Disorders: The Value of Ambiguity in Statutory Design and Interpretation, 54 Stan. L. Rev. 627, 640–42 (2002) (describing legislative incentives for creating statutory ambiguity). ↑
56. . Interview Comment No. 213. One interviewee explained that the resolution of ambiguous provisionsis entirely dependent on who happens to be sitting in those seats at that time. . . . [I]t really is left up to the agency and the person running the program and how they work with the General Counsel’s Office to determine if they’re gonna look at this really conservatively and risk averse or are they gonna look at this broadly and liberally. And really there’s a lot of discretion left to the agency.Interview Comment No. 1307. ↑
57. . Interview Comment No. 1708. ↑
58. . Id. ↑
59. . Interview Comment No. 509. ↑
60. . Interview Comment No. 507 (emphasis added); see also Wendy Wagner & Will Walker, Incomprehensible!: A Study of How Our Legal System Encourages Incomprehensibility, Why It Matters, and What We Can Do About It 205 (2019) (noting that the federal “legislative system . . . sometimes tolerates and sometimes rewards incomprehensibility”). ↑
61. . If nothing else, agencies have to fulfill the reasonableness standards that apply to agency action, but not to statutes. See Edward Stiglitz, The Reasoning State 137–77 (2022) (examining the role of “[r]easoning [c]onstraint[s]” on administrative decisionmaking). ↑
62. . Interview Comment No. 1602. ↑
63. . Interview Comment No. 1501; accord INS v. Chadha, 462 U.S. 919, 954 (1983) (rejecting the legislative veto over agency action as unconstitutional). ↑
64. . Interview Comment No. 1602. ↑
65. . Interview Comment No. 72 (“Nobody would’ve said that that’s what Congress meant when they wrote it, but there was very much a ‘Do the words give us authority to do this, and how can we argue that they do?’”). ↑
66. . Interview Comment No. 1522 (“Basically people sat around and looked at the statute and said, ‘[T]his term hasn’t been used in 40 years.’ They said, ‘How might we use it?’”). ↑
67. . Interview Comment No. 85. ↑
68. . Interview Comment No. 551; see also Interview Comment No. 1204 (“Yeah, we’re very much aware of [canons of statutory construction], and we always use them but I think we use them as inconsistently and as results-orientedly as the courts do, you know, . . . ultimately our goal is to promote the purposes of the Mine Act, and the safety and health of miners.”); Interview Comment No. 1813 (“I think the work that we were doing at that time, they were really looking for the best answer. . . . On [one rule], there really was an effort to get the best answer. . . . [On another rule], we were just trying to come up with enough stuff that looked like there was some meat on the bones that would be defendable . . . .”). One interviewee explained,There’s some issues where we just didn’t really care that much . . . and it seemed that there was a clear answer, and we could dispose of it that way. But then there were things where we told ourselves, look, if we interpret it one way we’re not going to be able to do the thing we want, or we’re going to run into huge implementation problems, and so let’s really push to do it the other way.Interview Comment No. 463. ↑
69. . Interview Comment No. 224. ↑
70. . Interview Comment No. 417 (“[F]ederal agencies are trying to find ways to annex as much authority as they reasonably can within the boundaries set for them by Congress in order to do new and expansive things, until they’re told not to.”). ↑
71. . See, e.g., Interview Comment No. 1409. The interviewee gave an example of how a statute might become ambiguous:Some IT policies or computer policies that are put in place 20 years ago, clearly outdated. That being the case, let’s retake a look again at the legal language and see like what is and is not feasible. Is it written broadly enough such that we can apply, let’s say if it’s about IT security policy, it’s not just the desktop, but now we should clearly apply it to laptops, we should clearly apply to smart phones and tablet devices and Internet routers and buildings and wireless, that kind of thing.Id. (footnote omitted); see also Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part I, 65 Stan. L. Rev. 901, 930–32 (2013) (showing that statute drafters know about the “Whole Act” rule, but do not adhere to it, because it does not express a primary concern); Jesse M. Cross, Where Is Statutory Law?, 108 Cornell L. Rev. 1041, 1045 (2023) (“The text of our statutory law is not found in any published document . . . . Rather, statutory text is something that must be imaginatively reconstructed by interpreters, time and again, in the act of statutory interpretation.”). ↑
72. . Interview Comment No. 135. ↑
73. . Id. ↑
74. . Interview Comment No. 690; see also Interview Comment No. 1492 (“I guess I would say the case law being developed and decisions by the courts are probably used more than legislative history, because as you know, you wouldn’t really look to that unless there was something that was ambiguous or unclear in the statute itself.”). ↑
75. . Interview Comment No. 390 (“Obviously, you have to follow the law, and you want to follow the intent of Congress, but a lot of things just aren’t spelled out, especially as laws age, and so I found myself looking to those sorts of foundational documents like the quadrennial . . . review, . . . strategy [documents], different speeches by senior officials . . . .”). ↑
76. . Interview Comment No. 925. This comment underscores the division of functions within agencies; non-lawyers rely on lawyers to engage in conventional statutory interpretation and legal risk assessment. ↑
77. . Interview Comment No. 1659. ↑
78. . Interview Comment No. 1555; see also Interview Comment No. 1547 (“The agency mostly knows what it wants based on a health and safety and enforcement perspective and then the lawyers are brought in to tell them how to make that work.”). ↑
79. . Interview Comment No. 209. ↑
80. . Interview Comment No. 704. ↑
81. . Cf. Philip Selznick, Leadership in Administration 26 (1957) (“Truly accepted values must infuse the organization at many levels, affecting the perspectives and attitudes of personnel, the relative importance of staff activities, the distribution of authority, relations with outside groups, and many other matters.”). ↑
82. . As we have noted, the idea of agency mission is different from and broader than that of statutory purpose as conventionally understood in the legal literature. it. See supra notes 26–27 and accompanying text. Officials pursuing agency mission may recognize “archaeological” purposes but subsume them into a larger “nautical” process that is continually evolving. Aleinikoff, supra note 26, at 21–22 (discussing archeological and nautical approaches to statutory interpretation). ↑
83. . Cf. Selznick, supra note 81, at 16 (“[O]rganizations have a history; and this history is compounded of discernible and repetitive modes of responding to internal and external pressures . . . . The more fully developed its social structure, the more will the organization become valued for itself, . . . as an institutional fulfillment of group integrity and aspiration.”). ↑
84. . Federal Coal Mine Health and Safety Act of 1969, Pub. L. No. 91-173, § 2(g), 83 Stat. 742, 743 (codified as amended at 30 U.S.C. § 801(g)). ↑
85. . Richard Nixon, Statement on Signing the Federal Coal Mine Health and Safety Act of 1969, The Am. Presidency Project (Dec. 30, 1969), https://www.presidency.ucsb.edu/documents
    /statement-signing-the-federal-coal-mine-health-and-safety-act-1969 [https://perma.cc/8EPP-N3FU]. ↑
86. . Interview Comment No. 1128 (“One of the things that I actually really love about it is we do have this awesome unidirectional statute that if you want to do . . . progressive impact lawyering, it’s a good place to do it.”). See Interview Comment No. 895 for a similar discussion from the perspective of the FTC: “[T]he FTC’s priorities don’t turn on nuances of statutory interpretation. There’s just so much obvious stuff that needs to be done like out-and-out fraud where there’s not a lot of dispute about what the statute means.” ↑
87. . Interview Comment No. 1126. ↑
88. . Interview Comment No. 1191. ↑
89. . Interview Comment No. 1345; see also Interview Comment No. 1258 (“I sort of viewed our approach as we’re serving small businesses.”). ↑
90. . Interview Comment No. 950. ↑
91. . Interview Comment No. 732. ↑
92. . Id. ↑
93. . Interview Comment No. 274. ↑
94. . Interview Comment No. 209; see also Bernstein & Rodríguez, supra note 15, at 1614–16 (describing the Executive Branch’s influence over agency policies). ↑
95. . Interview Comment No. 1921. ↑
96. . Interview Comment No. 274. ↑
97. . See, e.g., Interview Comment No. 163 (“Typically, the career staff [are] going to be less influenced by the political dynamics of an issue.”); Interview Comment No. 310 (“[P]olitical officers . . . have to have a sense of self-awareness about how all this would be [publicly perceived] . . . . It might make sense in a closed room [i]n some office building on Independence Avenue, but in the public when they hear what you are talking about, they think you are crazy.”); Interview Comment No. 277 (“I think where you see [ideological influence] tend[s] to be more in the hot-button issues, maybe women’s reproductive issues, abortion, and things like that.”). ↑
98. . Interview Comment No. 1067. ↑
99. . Interview Comment No. 286. ↑
100. . Interview Comment No. 781 (“Every administration should have an agenda, otherwise they’re just sorta tacking with the wind.”). ↑
101. . Interview Comment No. 1406. ↑
102. . See Bernstein & Rodríguez, supra note 15, at 1627–36 (describing political and career public servants’ interdependent roles in creating agency policies). ↑
103. . Interview Comment No. 864 (describing an independent agency). ↑
104. . Interview Comment No. 1483 (“[A particular independent agency] has a long history, really, of bipartisan cooperation. Whether it’s Democrats or Republicans, they tend to like the idea of being able to go to Congress, or to a court, and say we have unanimity.”). ↑
105. . Interview Comment No. 1111. ↑
106. . Interview Comment No. 1739. ↑
107. . One political appointee recalled asking a career official to take on a major new initiative:[H]e came into my office the next morning and he said that I talked to my wife and we do have some plans coming up, but we could put them aside. I guess that when you work at a government agency . . . and the agency tells you that you’re needed, you can only give one answer . . . yes.Interview Comment No. 1003. ↑
108. . Interview Comment No. 1275. ↑
109. . See, e.g., Interview Comment No. 1469 (“[If] it’s fair to read [a statute] either of two ways, and way one imposes a huge burden on some stakeholder, and interpretation B doesn’t, and yet both will accomplish the same overall goal . . . , then I think you would intuitively and necessarily prefer the second outcome, the less burdensome one.”). ↑
110. . Interview Comment No. 704. ↑
111. . Interview Comment No. 1458 (“You try to come up with that which is reasonable that won’t unduly burden business, because business has to be able to function, but will achieve your goal of protecting . . . consumers.”). ↑
112. . See Interview Comment No. 1320 (describing fraud conducted by large businesses who once, but no longer, qualified for small-business benefits). ↑
113. . See Emerson, supra note 5, at 2025 (arguing that characterizing agencies as neutral implementers of congressional will is both “inaccurate” and “normatively unappealing”); Bernard E. Harcourt, The Systems Fallacy: A Genealogy and Critique of Public Policy and Cost-Benefit Analysis, 47 J. Legal Stud. 419, 422 (2018) (critiquing the “systems fallacy: the mistaken belief that systems-analytic methods [like cost-benefit analysis] are neutral, when in fact they normatively shape politics”). ↑
114. . Interview Comment No. 1870 (“When we took a fresh look at the rule . . . we thought . . . , ‘Look, there’s technology that is not adequately covered by our rule. The technology is evolving very quickly. We need a rule that’s going to be flexible enough . . . so it won’t be out of date three years from now.’”). ↑
115. . Interview Comment No. 412. ↑
116. . Id.; see also Interview Comment No. 1118 (noting that consulting with technical experts is necessary “sometimes just to understand what the problem is”). ↑
117. . See Jody Freeman & Jim Rossi, Agency Coordination in Shared Regulatory Space, 125 Harv. L. Rev. 1131, 1134–36 (2012) (explaining that Congress often assigns multiple agencies overlapping functions). ↑
118. . See, e.g., Interview Comment No. 140 (“Within a given agency, you could have conflicting objectives, and certainly between agencies . . . . For instance, the FDA and CMS would often have conflicting objectives. The FDA, their concern is drug device safety, exclusively . . . they are not particularly concerned about how much medical care costs.”); Interview Comment No. 439 (describing a situation in which the DOJ’s “Civil Rights Division and the Federal Bureau of Prisons were institutionally kind of pushing in opposite directions just by virtue of their role,” where “[i]nevitably the Civil Rights Division will be wanting to do more . . . and the Bureau of Prisons will be wanting to do [] less for cost reasons and . . . operational concerns,” though not necessarily because of different normative positions); Interview Comment No. 95 (“[S]ome of my most fun conversations are just us having a really brainstormy type conversation among tri-department policymakers. . . . I think that’s a real benefit of the process is we all worked really closely together but also had our own institutions that we were representing.”). ↑
119. . Interview Comment No. 226; see also Interview Comment No. 228 (“[W]hat [one agency] thought was congressional intent, [another agency that] wasn’t part of the initial conversations, they’re looking at the world saying, ‘Well, this idea potentially creates more harm than good to the [regulated] system.’”). ↑
120. . Interview Comment No. 228. ↑
121. . Interview Comment No. 141. ↑
122. . Id. ↑
123. . See, e.g., Interview Comment No. 1586 (“We had lots of internal conversations with others [across the broader Department] about their interpretation of analogous rules and the impact that our interpretations would have on their stakeholders.”); Interview Comment No. 227 (detailing “a fundamental[ly] different point of view for [a] provision by different parts of the federal government,” where CMS supported healthcare organizations collaborating in order to create “stronger incentives to care coordinate and to improve quality,” while DOJ feared it would lead to “more negotiating power” for consolidated economic actors). ↑
124. . Interview Comment No. 150. ↑
125. . Interview Comment Nos. 149–50. ↑
126. . Interview Comment No. 150 (noting that this arises not necessarily because “they’re adverse [sic] to doing work, although that could be the situation,” but rather, “it’s probably more a survival mechanism because they’re overwhelmed”). ↑
127. . Interview Comment No. 149. ↑
128. . Interview Comment No. 732. ↑
129. . See Bernstein & Rodríguez, supra note 15, at 1663 (noting that interviewed administrators “consistently emphasized the importance of ensuring the effectiveness of rules or policies,” connecting “efficacy to . . . the realities of the situations they regulated” and observing that “policymaking reflected not the preferences of congressional appropriators or political supervisors in the White House but administrators’ own sense of what the realities on the ground required in order for them to realize their agency mission”); Interview Comment No. 77 (distinguishing “big picture political compromises that were driven totally by the members like around abortion, or even how generous the subsidy structure was gonna be” from intricate implementation questions where “[t]hose kinds of things were just, on the ground totally abstracting from what Congress as an institution cared about,” since “[t]hey just wanted [it] to work”). ↑
130. . Interview Comment No. 274. ↑
131. . See infra Part III. ↑
132. . Interview Comment No. 110 (describing agency responsibility as “we had to build [a] thing”). ↑
133. . Interview Comment No. 976; see also Interview Comment No. 196 (noting that the agency often “has to make a choice on how to interpret” statutory provisions). ↑
134. . Interview Comment No. 86. ↑
135. . Id. ↑
136. . Interview Comment No. 1443; Interview Comment No. 1467 (“I don’t think I’ve ever had any experience where we’ve looked at a statute or developed a policy saying, ‘Oh boy, this is blatantly unreasonable and outrageously stupid, but we’re going to do it because the statute says.’”); see also Jesse M. Cross & Abbe R. Gluck, The Congressional Bureaucracy, 168 U. Pa. L. Rev. 1541, 1681–82 (2020) (“Congress shares three beliefs that are central to purposivist theory: (1) that Congress has an intent or purpose when enacting laws; (2) that statutes are, above all, an expression of intent and purpose; and (3) that Congress’s overriding desire is to see that intent or purpose carried forward.” (emphasis omitted)); Gluck & Bressman, supra note 71, at 972 (noting that “94% [of the authors’ study’s respondents] told [the authors] that the purpose of legislative history is to shape the way that agencies interpret statutes”). ↑
137. . Interview Comment No. 1443; Interview Comment No. 1467. ↑
138. . From a more philosophical perspective, our interviews indicate that agencies tend to focus on real-world situations over ideal potentialities—that is, they are pragmatic in the style of the American Pragmatist philosophers. See, e.g., Charles S. Peirce, How to Make Our Ideas Clear, in 1 The Essential Peirce: Selected Philosophical Writings 124, 132 (Nathan Houser & Christian Kloesel eds., 1992) (“Consider what effects, which might conceivably have practical bearings, we conceive the object of our conception to have. Then, our conception of these effects is the whole of our conception of the object.”). ↑
139. . Bernstein & Rodríguez, supra note 15, at 1647–50 (describing the insufficiency of the principal–agent model). ↑
140. . Jerry L. Mashaw, Agency Statutory Interpretation, Issues Legal Scholarship, 2002, at 1, 17. ↑
141. . See, e.g., Emily Bernstein, Note, Navigating Campaign Finance Reform Through Publicly Funded Elections on the Local Level, 45 Cardozo L. Rev. 233, 242 (2023) (noting that Super PACs’ ability to raise “unlimited amounts of money from both corporations and individuals” has “immeasurably changed the political landscape”); Jennifer Ahearn & Beatrice Frum, What Gifts Must Supreme Court Justices Disclose?, Brennan Ctr. for Just. (Aug. 11, 2023), https://www.brennancenter.org/our-work/research-reports/what-gifts-must-supreme-court-justices-disclose [https://perma.cc/2U9J-YMF5] (expressing concern over gifts given to Justices of the Supreme Court by wealthy individuals); Wendy Wagner, Katherine Barnes & Lisa Peters, Rulemaking in the Shade: An Empirical Study of EPA’s Air Toxic Emission Standards, 63 Admin. L. Rev. 99, 130–31 (2011) (arguing that results of a study of third-party participation in EPA notice and comment rulemaking procedures were “[c]onsistent with dominant participation by industry”); Brian Libgober, Meetings, Comments, and the Distributive Politics of Rulemaking, 15 Q.J. Pol. Sci. 449, 479 (2020) (“[C]omments are a vehicle available to all members of society [participating in the notice and comment process] while meetings [between agency officials and regulated parties] are generally only available to those interests that the regulator is willing to meet with.”); Marianne Bertrand, Matilde Bombardini, Raymond Fisman, Brad Hackinen & Francesco Trebbi, Hall of Mirrors: Corporate Philanthropy and Strategic Advocacy, 136 Q.J. Econ. 2413, 2414–16 (2021) (describing how industry can co-opt seemingly neutral nonprofit groups to provide comments sympathetic to industry interests on pending regulations); Richard L. Hall & Frank W. Wayman, Buying Time: Moneyed Interests and the Mobilization of Bias in Congressional Committees, 84 Am. Pol. Sci. Rev. 797, 799, 815 (1990) (concluding that moneyed interests mobilize bias in committee decisionmaking); Deniz Igan & Thomas Lambert, Bank Lobbying: Regulatory Capture and Beyond 12 (Int’l Monetary Fund, Working Paper No. 171, 2019), https://www.imf.org
     /en/Publications/WP/Issues/2019/08/09/Bank-Lobbying-Regulatory-Capture-and-Beyond-45735 [https://perma.cc/B9GB-MJYC] (noting that the empirical literature on political connections confirms that “politically-connected firms have an influence on the regulatory and supervisory framework that affects their industry”). ↑
142. . Interview Comment No. 1146. ↑
143. . Id ↑
144. . Id. ↑
145. . Id. ↑
146. . Id. ↑
147. . Interview Comment No. 1093; see also Interview Comment Nos. 502–03 (“[T]here’s probably a proactive reason and a reactive reason [to interpret a statutory provision, where in a proactive mode,] Congress didn’t write a new statute, we’re probably getting some questions . . . , but we believe there is an important policy priority to move forward that is a correct interpretation of the law.”). At a certain level of generality, the agency may have similar concerns or approaches across different modalities: “[Whether acting proactively or reactively], the first step would be to spend a whole bunch of time with our lawyers to figure out what is the box around this, what is clear legally, what is not clear legally, where is our flexibility, could it be this way, this way, . . . blah.” Interview Comment No. 504. But our interviews made it clear that modality mattered. ↑
148. . Bernstein & Rodríguez, supra note 15, at 1651 (noting that interviews with agency officials demonstrated that “[i]nterviewees across agencies . . . actively learned about the regulated world to ensure that their policies served their agency’s mission”). ↑
149. . Interview Comment No. 1550; see also Interview Comment Nos. 1195–97 (describing mine explosions as an impetus for the agency to find and exercise previously unused statutory authority). ↑
150. . This picture of agency action is in stark contrast to a conventional transmission-belt model, which holds that agencies act in direct response to congressional mandates. See, e.g., Stewart, supra note 41, at 1675 (“The traditional model of administrative law thus conceives of the agency as a mere transmission belt for implementing legislative directives in particular cases.”); Lisa Schultz Bressman, Beyond Accountability: Arbitrariness and Legitimacy in the Administrative State, 78 N.Y.U. L. Rev. 461, 470 (2003) (“[According to the transmission-belt model,] [l]egislative directives protect individual liberty by confining administrative decision-making within identifiable and determinate bounds. Simply put, they reduce opportunities for arbitrariness by providing agencies with specific instructions rather than general licenses.”). ↑
151. . Interview Comment No. 529. ↑
152. . Interview Comment No. 502 (“It’s not just that your guidance isn’t good enough, it’s actually unclear, and it’s important enough that there might be a reason to move forward.”). ↑
153. . Interview Comment No. 481. ↑
154. . Id. ↑
155. . Id.; see also Interview Comment No. 482 (“[M]edia attention to an issue, I think, definitely can affect how an agency deals with a rule.”). ↑
156. . See Interview Comment No. 1530 (describing MSHA’s clarification about what a “pattern of violations” was under the Mine Act in response to a regulated party’s legal strategy); Interview Comment No. 846 (explaining that when a statutory text has multiple different meanings, the agency considers “what is the best way . . . , without making compliance overly burdensome,” and the agency would get “the views of both industry and consumer advocacy groups, and [would] tak[e] all of these things into consideration”). While regulated industry plays a dominating role in agency policy development, civil-society groups can also have an influence. See William F. West & Connor Raso, Who Shapes the Rulemaking Agenda? Implications for Bureaucratic Responsiveness and Bureaucratic Control, 23 J. Pub. Admin. Rsch. & Theory 495, 496 (2013) (“[A]ccess was heavily weighted in favor of the business and professional groups with whom bureaucrats interacted in carrying out their responsibilities.”); Daniel Walters, Capturing the Regulatory Agenda: An Empirical Study of Agency Responsiveness to Rulemaking Petitions, 43 Harv. Env’t L. Rev. 175, 201 (2019) (reporting a “21.72% chance of a grant ex ante” for a petition for rulemaking in the study’s data set); Eric Biber & Berry Brosi, Officious Intermeddlers or Citizen Experts? Petitions and Public Production of Information in Environmental Law, 58 UCLA L. Rev. 321, 363 (2010) (concluding that, in the case of the Endangered Species Act, “the groups involved with petitioning are overwhelmingly environmental or scientific organizations”); Maureen L. Cropper, William N. Evans, Stephen J. Berardi, Maria M. Ducla-Soares & Paul R. Portney, The Determinants of Pesticide Regulation: A Statistical Analysis of EPA Decision Making, 100 J. Pol. Econ. 175, 187 (1992) (finding that environmental groups commented on 49% of EPA decisions included in the sample, compared to 10% of decisions commented on by industry groups). As we discuss in Part III, because agencies have greater capacity to engage in pluralistic policy contestation than other branches, reform efforts should focus on strengthening their ability and incentives to respond to truly pluralistic concerns. ↑
157. . Interview Comment No. 395; see also West & Raso, supra note 156, at 506 (“Many rules are responses to changes in the conditions addressed by rules already in effect.”). ↑
158. . Interview Comment No. 1789. Agencies increasingly perform this function by posting “Frequently Asked Questions” on their websites to explain application processes, enforcement priorities, or other programs with which a regulated public directly interacts. ↑
159. . Interview Comment No. 19. ↑
160. . Id. ↑
161. . Id. ↑
162. . See Interview Comment No. 438 (noting that in the context of one rulemaking, all agencies involved in crafting the policy agreed on their end goal, but each agency was “institutionally kind of pushing in opposite directions just by virtue of their role”). ↑
163. . Interview Comment No. 895. ↑
164. . Interview Comment No. 1861. ↑
165. . See Sant’Ambrogio & Staszewski, supra note 16, 801 (surveying the ways that agencies involve the public in agenda-setting, including “rulemaking petitions, federal advisory committees, focus groups, requests for information, listening sessions and other public hearings, hotlines or suggestion boxes, public complaints, various forms of web-based outreach, negotiated rulemaking, and advance notices of proposed rulemaking”); Michael Herz, Using Social Media in Rulemaking 29 (2013), https://www.acus.gov/sites/default/files/documents/Herz%20Social
     %20Media%20Final%20Report.pdf [https://perma.cc/PKR2-TP8E] (noting differences between the value of social media in rulemaking and public participation generally); Stephen M. Johnson, #BetterRules: The Appropriate Use of Social Media in Rulemaking, 44 Fla. St. U. L. Rev. 1379, 1379–81 (2017) (comparing proper and improper uses of social media in agency rulemaking); Wendy Wagner, William West, Thomas McGarity & Lisa Peters, Deliberative Rulemaking: An Empirical Study of Participation in Three Agency Programs, 73 Admin. L. Rev. 609, 666, 685 (2021) (describing how the Federal Communications Commission invited public input to “define problems and propose solutions before the Commission has come up with a definite proposal,” which may “facilitate greater engagement with all stakeholders, including thinly financed groups”). ↑
166. . See, e.g., Interview Comment Nos. 1642, 1661 (explaining that staff “were very close with the community organizations that partnered with [the agency] on [a particular] program” and were “[l]ooking at ways to be responsive to what the community organizations were telling them, which is that, you need to change the language on the . . . program” to make it more comprehensible to the target populations, because “an agency like mine is based on . . . outreach, . . . because if a hundred people get government [benefits], a lot of that is because they know to ask”). ↑
167. . Interview Comment No. 981. ↑
168. . Interview Comment No. 1255. ↑
169. . Interview Comment No. 634; see also Interview Comment No. 284 (noting numerous ways a benefits-managing agency would “make sure that programs are working”); Interview Comment No. 1388 (explaining that a government-oversight agency “implemented . . . a performance scorecard for sustainability-related goals”); Interview Comment No. 1575 (“[W]e were having meetings with stakeholders from across the universe of affected interests, . . . the civil rights community, the women’s community, the hospital association, the National Association of Insurance Commissioners, employer groups. . . . I think there had been meetings going on for the preceding four years before I arrived.”); Interview Comment No. 1377 (recalling that agency personnel considering a proposal “went around the country to have hearings, . . . and [] ended up deciding not to do that [proposal] because of the comments [they] got”). Other rulemakings might involve more targeted outreach. “The people who are actually gonna have useful things to say about a [niche] rule is gonna be pretty darn small. . . . [I]t’s a very small universe of people and it’s much more detailed.” Interview Comment No. 983. ↑
170. . Interview Comment No. 91. ↑
171. . Interview Comment No. 819. Administrators often also took pains to ensure that they understood the industries they worked with. See, e.g., Interview Comment No. 1116 (speaking of using “an industry dictionary . . . that is an important interpretive source for us . . . because . . . it’s what’s commonly known in the . . . industry”); Interview Comment No. 1678 (“I remember when we got [a draft rule produced by another agency with a related but not coterminous purview], we looked at it and said it looks like it’s been written by someone who doesn’t know the [relevant] industry at all.”). Part of the interagency process, of course, involves ensuring that experts in the relevant industry have input into a rule. ↑
172. . Interview Comment No. 284; see also Interview Comment No. 35 (noting an agency’s designation of a private organization to give input on what metrics the agency should use for industry members). ↑
173. . See Sant’Ambrogio & Staszewski, supra note 16, at 819 (reviewing the informal means used by agencies to engage in public consultation); Cary Coglianese & Daniel E. Walters, Agenda-Setting in the Regulatory State: Theory and Evidence, 68 Admin. L. Rev. 93, 99 (2016) (“[M]any agency rules get started because of informal interactions between regulators and regulated entities.”). ↑
174. . Interview Comment No. 139. ↑
175. . Interview Comment No. 272; see also Interview Comment No. 633 (noting use of “census data” to illuminate relevant social trends). ↑
176. . Interview Comment No. 742. ↑
177. . Interview Comment No. 1853; see also Interview Comment No. 335 (“You would have stakeholder meetings. You would have many rounds of back and forth.”); Interview Comment No. 900 (explaining that most policy processes involved taking “a really, really good look at the complaints we were getting from [the public], inputs from [the agency’s economic analysts], from our [leadership], from partners in law enforcement . . . , observations from all the many organizations that we deal with on a regular basis” as well as “[w]orkshops that we convened on various topics to explore and learn about them,” and that such research was “pretty similar from year to year”); Interview Comment No. 1255 (describing a policy process arising out of “listening sessions or round tables” where agency leadership “had always heard of the challenges of applying for [a benefits] program” and asking staff, “‘Look, how do we simplify, how do we streamline?’”); Interview Comment No. 1365 (describing an agency having “round tables [to] try to get industry input”). ↑
178. . Interview Comment No. 466 (describing a rulemaking process having to do with incarceration conditions, in which the agency consulted with “people who operate[d] correctional facilities at the federal and the state level[,] . . . a lot of sheriffs and stuff like that[,] . . . civil rights groups[, as well as] . . . survivors of [the harm addressed by the rule]”). One interviewee explained that particular agency guidance was catalyzed when a political appointee attended a conference on the subject area in the 1960s. “[H]e was asked a question and found himself saying, ‘It would be very good if there were guidelines so that people could more easily decide’” what actions conformed with legal requirements. Interview Comment No. 972. “A year later, he was up there [again] and he was asked, ‘So, where are your guidelines?’ . . . And . . . sure enough, roughly the day before he left office, he issued . . . guidelines. He could sort of see the clock ticking and he finally just said, ‘I gotta get these out or it’ll be embarrassing.’” Id. ↑
179. . Interview Comment No. 1755. ↑
180. . Id. ↑
181. . See Sant’Ambrogio & Staszewski, supra note 16, at 801 (noting that although agencies use a wide variety of agenda-setting public consultation approaches, these “tend to be relatively unstructured, unsystematic, and ad hoc”). ↑
182. . See supra note 141 (collecting citations); West & Raso, supra note 156, at 496 (analyzing the influence of business and professional groups); Wendy E. Wagner, Administrative Law, Filter Failure, and Information Capture, 59 Duke L.J. 1321, 1386–87 (2010) (describing the “information capture” of agencies). It is also worth noting and exploring further how the influence of particular groups grows or diminishes with changes in presidential administration. For instance, whereas the agencies with responsibility for immigration regulation during the (first) Trump years paid no heed to immigrant rights advocacy groups, during the early Biden years officials informally sought and listened to their perspectives. ↑
183. . See Sant’Ambrogio & Staszewski, supra note 16, at 831–35 (proposing best practices for agenda-setting public consultation). ↑
184. . See, e.g., Interview Comment No. 1347 (“[A]fter the fact you will hear from industry groups . . . ‘Oh I didn’t know that was gonna happen, I think we should change it . . . because if you don’t, we won’t get [the agency-managed benefit].’ . . . So it’s just a constant adjustment.”); see also Wendy Wagner, William West, Thomas McGarity & Lisa Peters, Dynamic Rulemaking, 92 N.Y.U. L. Rev. 183, 189 (2017) (showing that new rules often adjust existing rules and that agencies “engage in a great deal of informal review and modification of existing rules, often beginning before their effective dates”). Scholars have shown that industry groups also lobby the Office of Management and Budget (OMB) to pressure agencies to change rules, leading to another path for dynamic rulemaking. See Rena Steinzor, Michael Patoka & James Goodwin, Ctr. for Progressive Reform, Behind Closed Doors at the White House: How Politics Trumps Protection of Public Health, Worker Safety, and the Environment 15–17 (2011), http://www.progressivereform.net/articles/OIRA_Meetings_1111.pdf [https://perma.cc/QNN9-UESY] (quantifying the frequency of such meetings and the attendance of diverse civil-society constituencies); Simon F. Haeder & Susan Webb Yackee, Influence and the Administrative Process: Lobbying the U.S. President’s Office of Management and Budget, 109 Am. Pol. Sci. Rev. 507, 518 (2015) (finding that industry lobbying of OMB tended to produce regulatory change, whereas public-interest lobbying had no analogous effect). ↑
185. . Interview Comment No. 742. The responses we got in our interviews diverge from a widespread scholarly assessment of the notice-and-comment process as a kind of “Kabuki theater” that has little effect on policy development. See, e.g., E. Donald Elliott, Re-Inventing Rulemaking, 41 Duke L.J. 1490, 1492 (1992) (“No administrator in Washington turns to full-scale notice-and-comment rulemaking when she is genuinely interested in obtaining input from interested parties. Notice-and-comment rulemaking is to public participation as Japanese Kabuki theater is to human passions . . . .”). Our material does not allow us to determine, however, whether this difference is due to differing assessment of what constitutes a relevant or significant effect. ↑
186. . Interview Comment No. 172. ↑
187. . Id. (“[P]ublic comments, certainly, influenced the agency to adopt some significant payment reforms in a much more measured and careful way.”); see also Interview Comment No. 1579 (discussing how comments influenced the agency to waive enforcement of a regulation based on the particulars of how the relevant industry worked). One agency official told the story of one such circumstance:There was one comment that was, it was just someone who wrote in the Regulations.gov, like the little text box, it wasn’t a letter or anything, it was just like two paragraphs. And it was from a lawyer who pointed out a question that he thought we got wrong. And there were detailed letters from the [industry representative group] and all kinds of other groups who had lawyers writing letters. And these people, this team of people who was working on that rule, mostly lawyers, people had reviewed this and we got it wrong. If that one person had not made that comment, we would not have changed that.Interview Comment No. 1764. ↑
188. . Interview Comment No. 92. ↑
189. . Id.; see also Interview Comment No. 1835 (“We would use the comments as our sort of source of, ‘How can we make this sound more reasonable? Well, what did people say about this?’ And then we would respond to what they said, like, ‘Wow, this sounds a lot more reasonable.’”). ↑
190. . See, e.g., Interview Comment No. 1576 (“[W]e had spreadsheets basically where we categorized the comments by issue, [with] keywords about which stakeholders it was, which provisions of the rule they addressed, which issue they were responding to, [and at] twice-weekly meetings . . . [staff would share] a synopsis of the comments . . . , and they proposed recommendation[s] for my review.”). Comment assessment systems like this likely limit the impact of mass-produced comments, since agencies respond to the comments’ contents, not their quantity. See Nina A. Mendelson, Rulemaking, Democracy, and Torrents of E-Mail, 79 Geo. Wash. L. Rev. 1343, 1363 (2011) (“Agencies have frequently treated . . . multiple postings briefly and with little real engagement.”); Cary Coglianese, Citizen Participation in Rulemaking: Past, Present, and Future, 55 Duke L.J. 943, 959 (2006) (“According to one recent study of about 500,000 comments submitted on an especially controversial EPA rule, less than 1 percent of these comments reportedly had anything original to say.”); see also Interview Comment No. 1853 (“We got a huge number of comments in that rulemaking, and I think I read every one. That’s actually worse than grading exams . . . because at some point they all are the same.”). ↑
191. . For our extended account of the forms of accountability reflected in the administrative state based on this study, see generally Bernstein & Rodríguez, supra note 15. ↑
192. . William F. West, Inside the Black Box: The Development of Proposed Rules and the Limits of Procedural Controls, 41 Admin. & Soc’y 576, 583 (2009) (noting that scholarship has “practically ignored” the “identification and framing of issues,” one of the “most important decisions” in policymaking). ↑
193. . See West & Raso, supra note 156, at 502, 504 (finding, based on a sample of 276 rules published in the Unified Regulatory Agenda in 2007, that 34% were catalyzed by Congress through some sort of statutory requirement, 3% by the President, 3% by court order, and 60% by the agency’s own discretion, as authorized in the relevant statute). ↑
194. . See, e.g., Interview Comment No. 1345 (“We want to [fulfill our mission] in a way that[] adheres to the statutory language but is as efficient and least burdensome as can be based on the language of the statute.”); Interview Comment No. 846 (“It needs to be practical . . . and effective.”). ↑
195. . See, e.g., Interview Comment No. 388 (“Agencies . . . publish policy documents about . . . their top priorities, what are the top things they’re going to focus their resources on. . . . [S]o there are a lot of foundational documents which establish the official position of various parts of the government as to what that particular agency thinks is important.”). ↑
196. . Interview Comment No. 28. ↑
197. . Interview Comment No. 1922. ↑
198. . Interview Comment No. 599 (“[W]ith regulations when Congress tells us to write something in a year, we get started immediately.”); Interview Comment No. 1054 (“[Y]ou’re just kind of off and running, and it’s almost a reactive kind of thing because you’re trying to meet these deadlines.”); Interview Comment No. 1331 (“I mean [Congress] often give[s] us deadlines . . . like 200 days to do a rule, which is not possible. . . . And if you do that, you might get a rule that you don’t really like because you’re going to rush and not get enough input. It might defeat the whole purpose.”). ↑
199. . Although federal statutes are usually enacted by Congress plus the President, our interviewees consistently treated statutes as primarily congressional texts. They were clearly aware of presentment requirements, but it was the contents of statutes, as drafted through the congressional process (often with agency participation) that were central. ↑
200. . See Freeman & Spence, supra note 19, at 72 (“[W]hen courts review the consistency of agency policy choices with the underlying enabling legislation, they must consider two congresses—the Congress that passed the enabling legislation in question and the current Congress, which may or may not be moved to pass legislation.”); Mashaw, supra note 19, at 508 (“Statutes persist while presidents and congresses change. In this context, the agency becomes the guardian or custodian of the legislative scheme as enacted.”). ↑
201. . Interview Comment No. 279. For discussion of scholarly work that explores the relationship between the enacting and existing Congresses, see infra note 321 and accompanying text. ↑
202. . Interview Comment No. 1269. ↑
203. . Id. ↑
204. . See Kenneth Lowande & Rachel Augustine Potter, Congressional Oversight Revisited: Politics and Procedure in Agency Rulemaking, 83 J. Pol. 401, 402–03 (2021) (explaining that Congressional committees can obtain “policy concessions” from agencies through procedural maneuvers); Charles R. Shipan, Regulatory Regimes, Agency Actions, and the Conditional Nature of Congressional Influence, 98 Am. Pol. Sci. Rev. 467, 478 (2004) (finding that agency actors must be sensitive to the preferences of oversight committees in making policy); Jason A. MacDonald & Robert J. McGrath, Retrospective Congressional Oversight and the Dynamics of Legislative Influence over the Bureaucracy, 41 Legis. Stud. Q. 899, 906 (2016) (“[C]ommittees can directly address [agency discretion] using oversight, rather than through the more burdensome process of legislation.”); Kenneth Lowande, Who Polices the Administrative State?, 112 AM. Pol. Sci. Rev. 874, 888 (2018) (finding that “congressional oversight is far more diffuse and ubiquitous than previously thought” in part because “oversight is often informal and conducted by individual legislators”). ↑
205. . By holding that Congress may act legislatively only as a full body, through the constitutional process of bicameralism and presentment, Supreme Court case law has made congressional power more ephemeral still. See INS v. Chadha, 462 U.S. 919, 956 (1983) (holding that legislative acts are subject to the bicameralism and presentment requirements except for in “narrow, explicit” circumstances). ↑
206. . See Walker, supra note 16, at 9 (discussing the congressional practice of requesting technical drafting assistance from agencies); Shobe, supra note 7, at 474 (demonstrating agencies’ central role in statutory drafting); see also Interview Comment No. 196 (“[T]o be the primary communication source with Congress, with the individual offices, individual members, committee offices. That’s one function. Two is to provide technical assistance . . . to ensure the drafting is correct and that the agency could actually implement what was being intended.”). ↑
207. . Cf. Russell W. Mills, Nicole Kalaf-Hughes & Jason A. MacDonald, Agency Policy Preferences, Congressional Letter-Marking and the Allocation of Distributive Policy Benefits, 36 J. Pub. Pol’y 547, 549 (2016) (finding that congressional communications with agency officials did not meaningfully alter the distribution of policy benefits). ↑
208. . As one interviewee described:Oftentimes . . . pressure would come from Capitol Hill. Legislators . . . would . . . realize . . . that they wrote a bad law. It just didn’t make sense . . . [, and they] would come to us and really push [us] . . . , and we would have to fully evaluate that and either say, ‘Yes, we can do this,’ or ‘No, we can’t do this.’ . . . They claim certain parentage to a law, and they want to see the law implemented the way they thought it should be.Interview Comment No. 268. Congress can also change its mind. See Interview Comment No. 1841 (describing a regulatory avenue explicitly foreclosed by legislation in the wake of agency action under the statute as “Congress directing us to do it and then beating us up for doing it, which is not an infrequent eventuality”). ↑
209. . Interview Comment No. 279. ↑
210. . Interview Comment No. 1868. ↑
211. . Id. ↑
212. . Interview Comment No. 514. ↑
213. . Id. ↑
214. . Interview Comment No. 371. ↑
215. . Id. ↑
216. . Id. (emphasis added). ↑
217. . E.g., Interview Comment No. 375. ↑
218. . Id. ↑
219. . Id. ↑
220. . See id. (discussing the commercial implications). ↑
221. . Id. ↑
222. . Id. ↑
223. . See, e.g., Martin Gilens, Affluence and Influence 234 (2012) (concluding that the U.S. government “correspond[s] more closely to a plutocracy than to a democracy” because of the outsized influence of the wealthy on Congress); Benjamin I. Page & Martin Gilens, Democracy in America? 135–37 (2020) (finding that interest group preferences have a “substantial impact” on policy outcomes and that “business-oriented groups” have “nearly twice as much influence” as non-business “mass-membership groups”). ↑
224. . Interview Comment No. 126; see also Interview Comment No. 1307 (“[O]ftentimes Congress will get upset that it’s not being implemented the way they intended it to be implemented, but their language was so vague to begin with.”). ↑
225. . Interview Comment No. 129. This interviewee explained that they had been involved in “technical assistance,” in which the “agency itself is not taking a position on the legislation. It’s just helping the drafters do what the drafters intend.”^⁠ Interview Comment No. 130; see also supra note 206 (citing sources documenting technical assistance). ↑
226. . Interview Comment No. 129. ↑
227. . Id. ↑
228. . Id.; see also id. (describing an experience in which members of Congress “wrote letters [to the agency] . . . saying ‘Well, this is what we meant, this is how you should implement it,’ but the agency can’t do that,” since “[w]hat two Senators say or what any Senators say, even if it was 75 Senators, can’t be used in place of what’s in the statutory language,” because otherwise, it “would mean that Congress could enact legislation, the Congressional Budget Office would give it a score,” but some group of Members “could just change their mind and say, ‘We don’t want to implement in the way we drafted it [and] [w]e’re just going to tell the agency to implement it a different way”). Moreover, there are “a lot of rules about who can talk to Congress, who can talk to staff. And it would be completely inappropriate for anybody to just pick up the phone and call someone and say, ‘What did you mean by this?’ or, ‘What’s your view?’ No. That would never happen.” Interview Comment No. 1785. ↑
229. . Brian D. Feinstein, Congress in the Administrative State, 95 Wash. U. L. Rev. 1187, 1192 (2018); see Douglas L. Kriner & Eric Schickler, Investigating the President: Congressional Checks on Presidential Power 8 (2016) (“Even without spurring legislation, however, investigations can lead to direct policy changes by prompting the president to change course on his own initiative.”). ↑
230. . See J.R. DeShazo & Jody Freeman, The Congressional Competition to Control Delegated Power, 81 Texas L. Rev. 1443, 1490 (2003) (“Members of appropriations committees also exert considerable influence over agencies. They enjoy all the tools described above, but because of their gatekeeping role over the appropriations process, they also have the power to control, and occasionally withhold, the flow of agency funds.”); Harold H. Bruff & Ernest Gellhorn, Congressional Control of Administrative Regulation: A Study of Legislative Vetoes, 90 Harv. L. Rev. 1369, 1421–22 (1977) (describing moments where appropriations committees have “directed [an] agency not to spend its funds to establish regulation . . . until legislation settled the issue”); Charles Tiefer, Controlling Federal Agencies by Claims on Their Appropriations? The Takings Bill and the Power of the Purse, 13 Yale J. on Regul. 501, 531 (1996) (explaining that agencies that wish to roll over unspent funds must submit proposals for approval to appropriations committees). ↑
231. . Lowande & Potter, supra note 204, at 403. Oversight hearings may often be conducted to “serve an electoral purpose in the form of position taking” visible to key constituencies and donors like industry representatives. Id. at 402–03, 406 (finding that “the marginal probability of engaging in oversight . . . generally increas[es]” with a “legislator’s [ideological] distance from the policy proposal,” because conducting oversight allows legislators to demonstrate opposition to a policy proposal for specific audiences). ↑
232. . See Bernstein & Rodríguez, supra note 15, at 1614–37 (explaining key features of the political control of administrative rulemaking). ↑
233. . See Interview Comment No. 2 (“[I]t’ll blow your mind, how strict and detailed that [ACA] oversight was.”); Interview Comment No. 4 (describing the “massive chart” and multiyear scheduling effort to implement the ACA); Interview Comment No. 651 (stating that any regulation that would affect the ACA was carefully vetted at the White House level); see also Bernstein & Rodríguez, supra note 15, at 1622–24 (describing White House oversight of ACA rulemaking). ↑
234. . See Bernstein & Rodríguez, supra note 15, at 1620, 1622–23 (“White House involvement [in policymaking] ultimately ranged from regularized, sometimes daily, contact to episodic, events-driven, and context-dependent interaction.”). ↑
235. . See Interview Comment No. 7 (recalling that the interviewee “had to go to meetings at the White House to answer questions about very small individual elements” about a proposed rule over a thousand pages long). Interactions with the EOP’s Office of Management and Budget (OMB), one typical comment explained, “would definitely influence the way we interpreted [the] statute. . . . [W]e would be thinking about what do we think is the best policy option, and then what do we think is the best policy option that we think we can get through the bureaucracy, including OMB.” Interview Comment No. 157. ↑
236. . See Bernstein & Rodríguez, supra note 15, at 1670 (“[W]e found that a President’s influence tends to be diffuse and atmospheric . . . . [P]residential priorities do not have [a] particular or concrete form, and the political influence of an administration arises from the sharing of high-level values and policy preferences by officials . . . .”). Our conclusions accord with recent scholarship finding that presidential “neglect—rather than proactive building or deconstructing of capacity—is the norm for most agencies.” Nicholas R. Bednar & David E. Lewis, Presidential Investment in the Administrative State, 118 Am. Poli. Sci. Rev. 442, 443 (2024). ↑
237. . See, e.g., Interview Comment Nos. 61–62 (explaining that the President’s influence was “more like the values that he would bring to conversations with his staff, and then his staff were in the meetings with [agency officials] and were sort of carrying those values” and that the interviewee could not “think of an instance where it was like, ‘The President wants [X] in terms of policy,’ but it was much more how he was communicating values to people and how those values were informing the broader policymaking process”). ↑
238. . See Interview Comment No. 781 (“Every administration should have an agenda, otherwise they’re just sorta tacking with the wind.”). Political appointees extend, but do not simply instantiate, presidential directives. See, e.g., Bernstein & Rodríguez, supra note 15, at 1628 (explaining how political connections, expertise, and experience affect policymakers’ evaluation of policy choices). Rather, while they imbue agencies with general orientations or values that resonate with their appointing President, they often pursue their own particular agendas. Their policy initiatives generally operate at a level of specificity impossible to attribute to the President or even the White House; yet they still work on the electoral cycle, precipitated by a new election and limited by the course of a political regime. See Bernstein & Rodríguez, supra note 15, at 1615–16, 1625–26 (explaining how in the absence of specific presidential priorities, appointees and career civil servants create policy by deriving meaning from “general [presidential] administration philosophy”). ↑
239. . Interview Comment Nos. 119–20 (discussing career staff’s expertise with respect to the ACA); Interview Comment No. 1548 (explaining subject-matter experts’ role in creating a particular rule); see also Interview Comment No. 290 (“There would be different times where the general counsel or some career staff or others [would] say, ‘We’ve been look[ing] at this this way . . . , but we think we could actually do this with this and make these additional changes. . . . Should we look at some of these new different options?’”); Interview Comment No. 395 (noting that “agencies are always briefing up”). ↑
240. . Bernstein & Rodríguez, supra note 15, at 1628. ↑
241. . Theories of “presidential administration” claim that agencies follow presidential wishes and act on presidential priorities, while “unitary executive theory” places the President at the center of administrative action in a normative vision of the constitutional order. See Kagan, supra note 31, at 2246 (adhering to the “presidential administration” theory); Calabresi, supra note 31, at 86 (“The best we can do is to let the spokesman for a quadrennial, energized national majority administer the laws.”). While our research suggests that a presidential administration can act as an important catalyst and input for agency action, the President’s role was usually not as direct, specific, or central as either theory would have it. See also Bednar & Lewis, supra note 236, at 454 (suggesting that while presidents have the authority to build agency capacity, they rarely invest time in ensuring that the administrative state runs smoothly); West & Raso, supra note 156, at 511 (“Although presidents have the ability to influence agency agendas when they so choose, . . . and although they do so on occasion, this does not appear to be an important or at least a systematic means of shaping policy.”). ↑
242. . See Bednar & Lewis, supra note 236, at 443 (“When presidents do build capacity, they tend to focus on agencies that (1) implement policies central to the president’s agenda, (2) share the president’s ideological leanings, or (3) face a high risk of experiencing a publicly salient failure.”); Freeman & Spence, supra note 19, at 65 (“While mobilizing agencies can be costly and time consuming for the President, when the political benefits of doing so are substantial enough and the legal means are readily available, it can be done by a motivated White House.”); Cristina M. Rodríguez, Complexity as Constraint, 115 Colum. L. Rev. Sidebar 179, 192 (2015) (“An energized President can certainly champion particular initiatives or take control through his advisors of arms of the government to advance elements of his political platform. . . . But even a well-run White House cannot hope to touch but a small fraction of the administrative state.”). At the same time, a President intent on not executing the law can probably have a substantial effect by curbing the federal workforce, bringing it under his control, or preventing it from implementing statutory directives. ↑
243. . See, e.g., Nicholas Bagley, The Procedure Fetish, 118 Mich. L. Rev. 345, 363 (2019) (“[P]roceduralism drains agency resources, introduces delay, and thwarts agency action.”); Cristina M. Rodríguez, The Supreme Court, 2020 Term—Foreword: Regime Change, 135 Harv. L. Rev. 1, 97 (2021) (“[T]he rule of law can be mobilized as an excuse to insert the judiciary in the policymaking process to thwart outcomes judges regard with suspicion, with limited discernable benefit flowing from the insistence on reason-giving.”); Neil Komesar & Wendy Wagner, The Administrative Process from the Bottom Up: Reflections on the Role, if Any, for Judicial Review, 69 Admin. L. Rev. 891, 926 (2017) (“[T]he adjudicative process and the dynamics of litigation create distinct advantages for concentrated interests. These forces mean that rather than alleviating the minoritarian bias in the administrative process, judicial review can aggravate it.”). ↑
244. . See Interview Comment No. 84 (“I think everyone’s risk tolerance is influenced by [questions like:] . . . does anyone have standing?”); Interview Comment No. 101 (noting that agency officials determined their litigation risk by applying the logical outgrowth doctrine and asking, “Do we have logical outgrowth on this issue?”); Interview Comment No. 400 (“I think the political question doctrine was the biggest judicial question in our minds when we were doing . . . rulemaking.”); Interview Comment No. 1208 (“We’re always aware of Chevron and we’re always, you know, will almost always make a plain meaning or argument first if we can and then, fall back on ambiguity arguments and deference argument.”); Interview Comment No. 1120 (noting that the Supreme Court’s decision in Martin v. Occupational Safety & Health Review Commission, 499 U.S. 144 (1991), is “an incredible tool for agency policy-making”). Some agencies took a “common law” approach to their own regulations, developing an overarching framework through long-term litigation efforts. Interview Comment No. 992 (“[I]t’s an ongoing common law.”); Interview Comment No. 1171 (“It was really a common law type of development.”). ↑
245. . Interview Comment No. 232; see also Interview Comment No. 106 (“[I]t was . . . intuitive . . . . The people would be asked to give their opinion about sort of how severe the litigation risks of a particular course of action was, and they would try to say high, medium, low in a way that was accurate as possible.”); Interview Comment No. 258 (“Then if they become high risk, the question is, ‘What’s the risk of being sued on this? Is it 50–50, is it 80–20? . . . [H]ow much of a risk taker do we want to be on this?’”); Interview Comment No. 296 (“Sometimes you just say, ‘We know whatever we’re going to do here, we’re going to get sued. So the question is, what’s our risk on this one versus that one of winning the case in court?’”); Interview Comment No. 175 (“[W]e felt like the policy . . . had high litigation risk in that there would be litigation, but . . . we felt like it was the option that was required by the statute, so the risk of losing is low[,] . . . provided we articulated very well our statutory interpretation and our implementation of the provision.”); Interview Comment No. 1079 (“[Y]ou never know what it is that’s going to end up in court. There are always surprises[,] . . . always things that don’t seem as important when you’re going through it as it turns out to be.”). ↑
246. . See Bernstein & Rodríguez, supra note 15, at 1636–37 (discussing the differences in risk tolerance across lawyers, offices, and agency roles). ↑
247. . Interview Comment No. 1439. ↑
248. . Id. ↑
249. . Interview Comment No. 1628. ↑
250. . Interview Comment No. 875; see also Interview Comment No. 1374 (“[T]he rule that comes out should be defensible by anybody . . . . [B]ecause of all that oversight by OMB and the other agencies . . . , there’s always that kind of balance to not go too far . . . .”); Interview Comment No. 1632 (“I think it is a reasonable position for an administration to say that they are going to advance defensible, reasonable, justifiable interpretations that not everybody will agree with.”). ↑
251. . Interview Comment No. 174 (“[O]ur expectation [in a particular rule-making] was that there was going to be a significant litigation risk no matter what the agency did . . . , but if we wrote the policy to be as defensible as possible, then we would be in a good position to . . . defend whatever policy we adopted . . . .”); see also Interview Comment No. 176 (“In situations . . . where there’s a lot of litigation risk . . . the attorney would say, ‘Well, you can adopt this policy, but . . . it’s going to be . . . hard . . . to defend.’ It would be up to the decisionmaker . . . whether . . . to . . . take the risk, or . . . adopt a policy that was more defensible.”); Interview Comment No. 265 (noting that there were times when the Office of General Counsel “just said, ‘It would just be too costly for the federal government to defend this’”); Interview Comment No. 741 (noting that addressing public comments is “important for legal defensibility”); Interview Comment No. 523 (“[T]he more likely we were to get legally challenged, the deeper [our] research would need to be.”). ↑
252. . Interview Comment No. 747; see also Interview Comment No. 876 (“[W]hen you do this before a court, you have to prove that you’re being reasonable and you’re being consistent with the statute, and you are abiding [by] the Administrative Procedure Act and you’re doing all the right things.”). One interviewee described how litigation affected the composition of a rule:So like in [a particular] rule, if you look at the final rule, it’s like 85 pages . . . in the Federal Register. The regulatory text at the end is like five pages. It has an 80-page explanation. And there’s a whole section [of] . . . legal policy justification. Every section of that rule is written with the idea that somebody is going to say we didn’t do our job.Interview Comment No. 1781. ↑
253. . Interview Comment No. 407 (“You’re thinking about almost not necessarily even the litigation risk of this particular rule or action ending up in court, as the effects it’s going to have more broadly on your ability to regulate and litigate.”). ↑
254. . Interview Comment No. 691. ↑
255. . Id. ↑
256. . See Interview Comment No. 1067 (“[Y]ou have the additional consideration of just the reputation of the office.”); Interview Comment No. 1719 (noting that a principal part of the agency lawyer’s job is maintaining agency credibility by not “doing stupid shit”). ↑
257. . Interview Comment No. 995 (“[A]fter the [litigation loss on a particular rule] debacle that almost got the [agency] put out of business, [the statutory term at issue in the litigation] became something that was sort of radioactive. It made people very nervous.”). One interviewee described “litigat[ing] a case for a rule” the interviewee “had been involved in,” which implemented a statutory provision that the interviewee, in assisting legislative drafters, “had written and . . . wrote all the legislative history to.” Interview Comment No. 757. See Shobe, supra note 7 (discussing agencies’ extensive participation in drafting legislation). In litigation about whether the rule the interviewee had worked on properly interpreted the statutory provision the interviewee had drafted, “the D.C. Circuit . . . ended up saying this rule fails on Chevron 1.” Interview Comment No. 757. This “[m]ean[t] that I had somehow interpreted my own [statutory] provision in a way that couldn’t meet Chevron 1. I knew damn well what I intended but it definitely wasn’t what the D.C. Circuit thought it meant.” Id. ↑
258. . Interview Comment No. 1202; Interview Comment No. 457 (“But yeah, the general mess of statutory interpretation that [is] in the courts is reflected at the agency level as well. Without any specific sense of what is necessarily going to be smiled upon by any particular court, you just look for whatever stuff you think helps.”). One agency official provided a more thorough description:[T]hey decide where they wanna come out, and then they pick the rule of statutory construction that’ll get them there. You can always find a rule of statutory construction that will get you to where you wanna go . . . because, there’s always an exception, or there’s always a contrary rule, or whatever . . . . Yeah, we’re very much aware of [rules of statutory construction], and we always use them but I think we use them as inconsistently and as results-orientedly as the courts do . . . .Interview Comment Nos. 1202–05. ↑
259. . Interview Comment No. 661. ↑
260. . Interview Comment No. 668; see, e.g., Interview Comment No. 173 (“I would say the General Counsel’s office was very influential. They would never say, ‘Yes, you could do something,’ or, ‘No, you couldn’t do something.’ They would always couch it in terms of litigation risk.”); Interview Comment No. 265 (“You would really be pretty reliant on your subject matter experts and Office of the General Counsel.”). Depending on the agency, subject matter and policy experts might work separately from attorneys evaluating litigation risk. See Interview Comment No. 925 (“[N]obody [on the subject-matter expert team is] going out and reading the case law or anything like that. That’s not our job . . . . They have lawyers whose job it is to read the case law . . . and determine how that would affect our litigation posture, were we to proceed to court.”); Interview Comment No. 830 (“[The] Office of General Counsel . . . may be mindful of certain risks that wouldn’t be necessarily evident to people who are kind of so mired in the substance of the issues and the risks to [the protected population].”); Interview Comment No. 1633 (explaining that as a high-ranking career subject matter expert, “[o]bviously I worked very, very closely with the Office of General Counsel, but I think to the extent that it is the agency’s role to evaluate litigation risk and counsel policymakers not to take the positions that will create litigation risk, that wasn’t my job”). ↑
261. . Interview Comment No. 107; see also Interview Comment No. 1044 (“Meetings. Well, I’d be listening to them. They’d be listening to me, and we hopefully come to an agreement one way or another.”). ↑
262. . See, e.g., Interview Comment No. 1208 (“We’re always aware of Chevron and we’re always, you know, will almost always make a plain meaning or argument first if we can and then, fall back on ambiguity arguments and deference arguments.”); Interview Comment No. 809 (“[C]ertainly in dealing with the Justice Department who defends these things we were always speaking the same language, Chevron 1, Chevron 2. Are you so far outside the bounds of statutory ends that you can’t even defend it under Chevron 2?”). ↑
263. . Interview Comment Nos. 181–82 (suggesting awareness of Chevron was not evenly distributed); Interview Comment No. 1719 (noting that “every interpretive question takes place in the context of deference”). ↑
264. . Interview Comment No. 1209; see also Interview Comment No. 444 (“I think if there was like a policy goal you’re really driving at . . . , you have, of course, the Chevron doctrine in your back pocket . . . .”); Interview Comment No. 1103 (“[A]s long as they’re within reason, I view my job as . . . say[ing] . . . , ‘Well, if this is the policy concern . . . , and you have a poorly written or unclearly written standard . . . , let’s see if there’s a way to interpret that given . . . Chevron, see if we can get to the right result.’”); Interview Comment No. 1543 (“[W]e look for ways to characterize whatever has happened in a way that gets Chevron.”). ↑
265. . Interview Comment No. 755. Another official sounded a similar theme, noting that Chevron did not so much serve the aggressive purpose of “put[ting] a thumb on the scale” in favor of the agency, but rather helped officials identify the “questions that Congress punted to the agency.” Interview Comment No. 1131; see also Interview Comment No. 754 (describing Chevron as helping them “[p]reserve the maximum range of legal actions for the decision-makers, who were the political people”); Interview Comment No. 179 (“I guess Congress[, in considering curbing Chevron,] feels that it would be happier with judicial . . . rather than agency interpretation. It doesn’t trust the agency, it does trust the court[s]. Again, . . . careful what you ask for. You’re not necessarily going to get an outcome that’s . . . better for the people that you’re trying to help.”). ↑
266. . Interview Comment No. 1509 (“It’s utilized to remind the circuit court of its role in an appeal from an administrative action and to frame how the appellate court should take on its judicial review . . . .”). ↑
267. . Interview Comment No. 1831; see Interview Comment No. 773 (citing interpretation as being “antithetical to the goals of the Act if it ignores critical facts”); Interview Comment No. 810 (noting that if the agency looks “exclusively to cost and . . . ignore[s] any environmental implications . . . it’s just way too outside the bounds of what the statute is getting at,” Chevron notwithstanding). ↑
268. . Interview Comment No. 1142; see also Interview Comment No. 536 (“[B]ut 98% of what we did you could like or dislike, but there were no Chevron questions raised at all . . . .”). ↑
269. . Interview Comment No. 1014; cf. Interview Comment No. 1878 (“There’s some macro out there that every agency uses, but if you look at the briefs that we filed, we don’t really hang our hat on Chevron.”). ↑
270. . See Interview Comment No. 1202 (“You can always find a rule of statutory construction that will get you to where you wanna go . . . .”). ↑
271. . Interview Comment No. 1078 (“The reality is in litigation, you have so much more time to think about whatever the particular narrow issue is that you might not have thought about or done very well with when you were going through [a] rule that had 100 issues like that.”). ↑
272. . Interview Comment No. 1090. ↑
273. . Interview Comment No. 1628; see also Interview Comment No. 183 (“[An agency] promised the court that it would go back and . . . restore [a formula it had changed] . . . . I don’t know that the agency necessarily changed its interpretation, but based on the litigation, I think it concluded it didn’t have a defensible case and just decided to concede [the point].”). ↑
274. . See, e.g., Cary Coglianese, Heather Kilmartin & Evan Mendelson, Transparency and Public Participation in the Federal Rulemaking Process: Recommendations for the New Administration, 77 Geo. Wash. L. Rev. 924, 928 (2009) (noting within the context of judicial review of agency action that “[w]hen the public monitors agency behavior, regulators are inclined to choose policies that best advance the overall public welfare and the agency’s statutory mandate; while under the microscope of public scrutiny, regulators are reluctant to choose policies that are sloppy or expedient”); Alan B. Morrison, The Administrative Procedure Act: A Living and Responsive Law, 72 Va. L. Rev. 253, 263 (1986) (arguing that because litigants have “easier access to courts to challenge agency decisions” under the APA, “public participation has deterred the agencies from straying too far from their assigned missions”); Harold J. Krent & Nicholas S. Zeppos, Monitoring Governmental Disposition of Assets: Fashioning Regulatory Substitutes for Market Controls, 52 Vand. L. Rev. 1705, 1772 (1999) (“Given the potential for interest group influence in both Congress and agencies, judicial review provides another forum to minimize the chance that agency policy initiatives serve private interests at the expense of the general public.”); Colin A. Olivers, Has the Federal Courts’ Successive Undermining of the APA’s Presumption of Reviewability Turned the Doctrine into Fool’s Gold?, 38 Env’t L. 243, 259 (2008) (arguing that federal courts should reassert the APA’s presumption in favor of judicial review in part because “judicial review goes to . . . the right of the public to monitor and interact in its government”). ↑
275. . Jonathan H. Adler & Christopher J. Walker, Delegation and Time, 105 Iowa L. Rev. 1931, 1945 (2020) (discussing the dead-hand problem in the context of delegations to agencies); Freeman & Spence, supra note 19, at 3 (concluding that “when adapting old statutes to new problems, agencies are surprisingly accountable, not just to the President, but also to Congress, the courts, and the public”). ↑
276. . This focus on intertemporal mediation resonates with theories of dynamic statutory interpretation. See William N. Eskridge, Jr., Dynamic Statutory Interpretation 5–6 (1994) (arguing that statutory interpreters should not be bound by what “the original legislature would have endorsed” when circumstances have changed); William N. Eskridge, Jr., Dynamic Statutory Interpretation, 135 U. Pa. L. Rev. 1479, 1479 (1987) (“Statutes, however, should—like the Constitution and the common law—be interpreted ‘dynamically,’ that is, in light of their present societal, political, and legal context.”); Edward Rubin, Dynamic Statutory Interpretation in the Administrative State, Issues Legal Scholarship, 2002, at 1, 1 (observing that while “Eskridge’s article is addressed to judges, . . . [d]ynamic statutory interpretation is the only sensible approach to statutory interpretation in a modern administrative state for both agencies . . . and courts”). Current Supreme Court doctrine increasingly ignores and even rejects this aspect of agency work, a crucial democratic need served by no other branch. ↑
277. . See Mashaw, supra note 19, at 498–99 (remarking that “forests have been laid waste to accommodate the outpouring of legal commentary” on judicial review of agency statutory interpretation, but “virtually no one has even asked . . . some simple questions [such as,] ‘how do agencies interpret statutes?’”); see also Peter L. Strauss, When the Judge Is Not the Primary Official with Responsibility to Read: Agency Interpretation and the Problem
of Legislative History, 66 Chi.-Kent L. Rev. 321, 321 (1990) (discussing the shortfalls of the scholarly “paradigm” of statutory interpretation). While scholars differ on what precisely constitutes agency statutory interpretation, the question requires knowing how agencies actually work with statutes. Compare Richard J. Pierce, Jr., How Agencies Should Give Meaning to the Statutes They Administer: A Response to Mashaw and Strauss, 59 Admin. L. Rev. 197, 199, 202–05 (2007) (limiting agency statutory interpretation to determining whether a statute is ambiguous in the same way as judicial interpretation), with Jerry L. Mashaw, Agency-Centered or Court-Centered Administrative Law? A Dialogue with Richard Pierce on Agency Statutory Interpretation, 59 Admin. L. Rev. 889, 893 (2007) (distinguishing agency from court statutory interpretation). ↑
278. . In contrast, significant work has explored interested parties’ role in shaping regulations. See supra notes 141, 156 and accompanying text. ↑
279. . See, e.g., Herz, supra note 11, at 96–98, 102, 106 (emphasizing the institutional dimension of purposivism and evaluating purposivism from an agency’s, rather than a court’s, perspective); Stack, supra note 11, at 875 (developing a purposivist theory of congressional delegation); Walker, supra note 12, at 2020, 1023 (surveying regulation drafters on attitudes toward canons of interpretation, legislative history, and administrative review doctrines); Semet, supra note 12, at 2258–59 (evaluating interpretive patterns in NLRB decisions with reference to interpretive patterns in judicial opinions). ↑
280. . We thus contribute to a smaller body of scholarship exploring interpretation from the agency perspective—a shift we think essential to keeping the literature on statutory interpretation relevant. See, e.g., West, supra note 192, at 583 (noting scholarly neglect of many issues of rulemaking); Shobe, supra note 7, at 454 (stating that his 2017 article is “the first extensive empirical study into the role of agencies in the legislative process”). ↑
281. . See, e.g., Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1745 (2020) (rejecting the argument that the “consequences that might follow a ruling for the [appellants]” entitles judges to “ignore the law as it is”). This theoretical rejection does not stop textualist adherents from taking consequences into account in practice. Schacter, supra note 38, at 1009 (“[W]hile textualism on the books conspicuously eschews the legitimacy of consequentialism in statutory interpretation, textualism in action often uses strikingly consequentialist methods.”); Victoria Nourse, The Paradoxes of a Unified Judicial Philosophy, An Empirical Study of the New Supreme Court: 2020–2022, 38 Const. Comment. 1, 2 (2024); Krishnakumar, supra note 24, at 586. ↑
282. . Dynamic statutory interpretation, which allows statutory interpreters to go beyond what “the original legislature would have endorsed,” comes closer. Eskridge, supra note 276, at 5; see supra note 276 and accompanying text. ↑
283. . See infra notes 317 & 319 and accompanying text. ↑
284. . See Bressman & Gluck, supra note 7, at 767 (“[Legislative drafters] rejected the idea of courts as partners.”). ↑
285. . Cf. Neil K. Komesar, Imperfect Alternatives: Choosing Institutions in Law, Economics, and Public Policy 5 (1994) (“The choice is always a choice among highly imperfect alternatives.”). ↑
286. . See, e.g., Sant’Ambrogio & Staszewski, supra note 16, at 801 (showing that agencies vary widely in the extent to which they involve affected publics in agenda setting). ↑
287. . See supra note 141 and accompanying text. ↑
288. . See, e.g., Bijal Shah, Administrative Subordination, 91 U. Chi. L. Rev. 1603, 1612 (2024) (arguing that agencies prioritize institutional interests over the interests of marginalized people); Karen M. Tani, Compensation, Commodification, and Disablement: How Law Has Dehumanized Laboring Bodies and Excluded Nonlaboring Humans, 119 Mich. L. Rev. 1269, 1272 (2021) (book review) (“[T]he U.S. legal system embraced a paradigm that treated working people impersonally—almost fungibly—when it came to the harms they experienced on the job.”). ↑
289. . The wealthy have significantly more influence over Congress than do middle-class and lower-class Americans. See, e.g., Gilens, supra note 223, at 1 (“The American government does respond to the public’s preferences, but that responsiveness is strongly tilted toward the most affluent citizens. Indeed, under most circumstances, the preferences of the vast majority of Americans appear to have essentially no impact on which policies the government does or doesn’t adopt.”); Larry M. Bartels, Unequal Democracy 6 (2d ed. 2016) (“[I]nsofar as constituents’ views do matter [to the roll call votes cast by Senators and members of the House], political influence seems to be limited mostly to affluent and middle-class people. In the case of senators, the opinions of millions of ordinary citizens in the bottom one-third of the income distribution have no discernible impact on the behavior of their elected representatives.”); Robert S. Erikson, Income Inequality and Policy Responsiveness, 18 Ann. Rev. Pol. Sci. 11, 23 (2015) (estimating that “the richest Americans enjoy at least 2.5 times the political influence of the poorest Americans”). ↑
290. . Komesar, supra note 285, at 22 (“The crucial question concerns the relative merits of . . . institutions, when compared to each other.”); see also Cass R. Sunstein & Adrian Vermeule, Interpretation and Institutions, 101 Mich. L. Rev. 885, 886 (2003) (“[D]ebates over legal interpretation cannot be sensibly resolved without attention to [institutional] capacities. The central question is not ‘how, in principle, should a text be interpreted?’ The question instead is ‘how should certain institutions, with their distinctive abilities and limitations, interpret certain texts?’”). ↑
291. . See Komesar, supra note 285, at 3 (defining the term “comparative institutional analysis”). ↑
292. . Our research thus gives empirical support to Jerry Mashaw’s suggestions that agencies are “the primary official interpreters of federal statutes” and that their interpretations “[b]end[] with the political winds.” Mashaw, supra note 9, at 499, 516. ↑
293. . Mashaw, supra note 140, at 17. ↑
294. . For a popular version of this form of argument, see David French, Overturning Chevron Can Help Rebalance the Constitutional Order, N.Y. Times (Jan. 21, 2024), https://www.nytimes
     .com/2024/01/21/opinion/supreme-court-chevron.html [https://perma.cc/7C6B-L5RC]. ↑
295. . See supra note 204 and accompanying text. ↑
296. . See INS v. Chadha, 462 U.S. 919, 958 (1983) (holding that Congress only acts legislatively through bicameralism and presentment). ↑
297. . For discussions of the limits of Congress’s capacity, see William G. Howell & Terry M. Moe, Relic: How Our Constitution Undermines Effective Government And Why We Need a More Powerful Presidency 62–66 (2016) and Eric A. Posner & Adrian Vermeule, The Executive Unbound 28–29 (2010), detailing the vast personnel and resource differential between congressional staffs and the administrative state. ↑
298. . See, e.g., Jesse M. Cross, The Staffer’s Error Doctrine, 56 Harv. J. on Legis. 83, 85 (2019) (“[M]embers of Congress . . . no longer participate in the drafting of statutory text . . . . Instead, . . . their . . . knowledge about the content of bills typically comes from summaries that outline the [statute’s] purpose . . . provided either via memoranda or in-person staffer briefings. . . . [They] provide direct review only of the purpose of statutory provisions . . . .”). ↑
299. . See Shobe, supra note 7, at 498–99 (discussing agencies’ superior subject matter expertise as compared with Congress). ↑
300. . See Cross, supra note 298, at 112–17 (describing the incentives and challenges of individual members of Congress). ↑
301. . See Gundy v. United States, 139 S. Ct. 2116, 2132–33, 2143 (2019) (Gorsuch, J., dissenting) (“These unbounded policy choices have profound consequences for the people they affect.”). Judge Neomi Rao, former OIRA Administrator, has echoed this view, arguing that “[w]ith significant opportunities for regulatory action, a single bureaucrat can at times exercise an authority that exceeds that of a member of Congress.” Neomi Rao, The Hedgehog & the Fox in Administrative Law, Daedalus, Summer 2021, at 220, 228. The Supreme Court has ruled that a single member of Congress has no legislative authority. Chadha, 462 U.S. at 951 (“[The requirement that legislation must be passed by both Houses of Congress] represents the Framers’ decision that the legislative power of the Federal Government be exercised in accord with a single, finely wrought and exhaustively considered, procedure.”). ↑
302. . See Gundy, 139 S. Ct. at 2131 (Gorsuch, J., dissenting) (“The Constitution promises that only the people’s elected representatives may adopt new federal laws restricting liberty. Yet the statute before us scrambles that design.”); Philip Hamburger, Is Administrative Law Unlawful? 372 (2014) (arguing that “disdain for popular legislative power” goes hand in hand with “advocates of administrative power[’s] . . . admiration for expert control”). See generally Gillian E. Metzger, The Supreme Court, 2016 Term—Foreword: 1930s Redux: The Administrative State Under Siege, 131 Harv. L. Rev. 1 (2017) (documenting historical roots and contemporary political features of anti-administrativism). ↑
303. . See Bernstein & Rodríguez, supra note 15, at 1639, 1655 (describing diverse participants in policymaking processes). Some interviewees did note that a single person in the presidentially created OIRA could obstruct the implementation of a statutory mandate enacted by Congress. Cf. Ashraf Ahmed, Lev Menand & Noah A. Rosenblum, The Making of Presidential Administration, 137 Harv. L. Rev. 2131, 2153 (2024) (arguing that the creation of OMB shocked contemporaries as an Executive power grab). ↑
304. . Such delegation has a long history. Julian Davis Mortenson & Nicholas Bagley, Delegation at the Founding, 121 Colum. L. Rev. 277, 281 (2021) (“[E]arly Congresses adopted dozens of laws that broadly empowered executive and judicial actors to adopt binding rules of conduct[ for private parties].”); Nicholas R. Parrillo, A Critical Assessment of the Originalist Case Against Administrative Regulatory Power: New Evidence from the Federal Tax on Private Real Estate in the 1790s, 130 Yale L.J. 1288, 1302 (2021) (“[O]riginalist skeptics of rulemaking are mistaken to say that no early congressional grant of rulemaking power was coercive and domestic.”). ↑
305. . Bernstein & Rodríguez, supra note 31, at 371 (arguing that the Constitution and the government it authorizes “should be understood as making room for—perhaps even requiring—a workable government”). A formalistic interpretation of the Constitution might support an alternative conclusion aimed at constraining unelected bureaucrats. That approach, however, ignores the fact that our current government itself emerged through constitutional processes over time. Claims about returning government to a design conceptualized in the past thus present a normative plea, not a constitutional command. See id. at 369 (advocating for an interpretation of the Constitution that is both “workable and normatively attractive”). ↑
306. . See Bernstein & Rodríguez, supra note 15, at 1606 (identifying three mutually supporting aspects of agency accountability: “(1) diffuse, rather than concentrated, forms of political control; (2) nonhierarchical organizational structures of negotiation and deliberation among numerous actors and groups; and (3) practices that keep agencies attuned to affected publics and events in the regulated world”). ↑
307. . This problem underscores how efforts to revive the nondelegation doctrine represent a formal fantasy, motivated either by indifference to the need for a working government or by the ideological preference to have less federal regulation, relying instead on the private sector or the states. See Neil Gorsuch & Janie Nitze, Over Ruled: The Human Toll of Too Much Law 4–5 (2024) (arguing for reducing federal law and regulation on policy and consequentialist grounds); see also Eric A. Posner & Adrian Vermeule, Interring the Nondelegation Doctrine, 69 U. Chi. L. Rev. 1721, 1724 (2002) (denying that theories of nondelegation add “any constitutionally meritorious content to the relevant debates”); Keith E. Whittington & Jason Iuliano, The Myth of the Nondelegation Doctrine, 165 U. Pa. L. Rev. 379, 429 (2017) (concluding that historical justifications for nondelegation are “more mythical than historical”). ↑
308. . See, e.g., Nat’l Fed’n of Indep. Bus. v. OSHA, 142 S. Ct. 661, 669 (2022) (Gorsuch, J., concurring) (explaining that a “major questions doctrine guards against” the possibility that an agency “may seek to exploit some gap, ambiguity, or doubtful expression in Congress’s statutes to assume responsibilities far beyond its initial assignment”). ↑
309. . See, e.g., Adam B. Cox & Cristina M. Rodríguez, The President and Immigration Law 201–08 (2020) (exploring limits of executive action to update immigration laws direly in need of reform). ↑
310. . See Shah, supra note 288, at 1606 (discussing the impact of administration on marginalized communities); Jodi L. Short, In Search of the Public Interest, 40 Yale J. on Regul. 759, 762 (2023) (“[R]egulation in the public interest opens the door to the arbitrary exercise of tyrannical state power.”); see also Matthew Desmond, Poverty, by America 120–22 (2023) (showing how the tax code and other statutes advantage those with more resources and disadvantage those with less). ↑
311. . See David L. Noll, Administrative Sabotage, 120 Mich. L. Rev. 753, 762 (2022) (discussing the problem of officials acting to deliberately undermine their superiors’ policy goals); Jody Freeman & Sharon Jacobs, Structural Deregulation, 135 Harv. L. Rev. 585, 587 (2021) (explaining some of the ways in which a president can undermine agency efficacy); Bagley, supra note 243, at 391 (arguing that procedural safeguards are insufficient to prevent agency capture). ↑
312. . See Sant’Ambrogio & Staszewski, supra note 16, at 801 (reviewing a range of ways agencies involve affected publics in developing policy ideas); K. Sabeel Rahman, Policymaking as Power-Building, 27 S. Cal. Interdisc. L.J. 315, 355, 357 (2018) (calling for agencies to be redesigned to redress disparities of power); Blake Emerson, The Values of the Administrative State: A Reply to Seidenfeld, 119 Mich. L. Rev. Online 81, 91 (2021) (arguing that agency action should be required to openly consider normative values); Short, supra note 310, at 834 (calling for affected parties to frame their appeals to agencies “on a clearly articulated, substantive vision of what the public interest requires”). The Biden Administration took steps to acknowledge and address these distributional considerations in policymaking. See Off. of Mgmt. & Budget, Exec. Off. of the President, Circular No. A-4, Regulatory Analysis, 5 (2023) (establishing that a materially complete benefit–cost analysis requires a “summary of the effects on social welfare or of the evidence intended to inform determination of the most net-beneficial alternative”). ↑
313. . Our conception of democratic governance might also be criticized as overly statist. Indeed, much of the critique of the administrative state is animated by the contention that it cannot work well and that regulation imposes a drag on the economy and innovation. Metzger, supra note 302, at 6 (identifying the discursive forebears of the anti-administrative movement as “conservative opponents of an expanding national bureaucracy in the 1930s,” fueled by “business and legal interests deeply opposed to pro-labor regulation and economic planning”). But some of these critics would be no happier with an active, regulatory Congress; a comparative institutional analysis need not be sidelined by bigger picture debates over the validity of regulation in principle. ↑
314. . See, e.g., Guido Calabresi, A Common Law for the Age of Statutes 164 (1982) (arguing that courts should either “updat[e]” statutes that have become obsolete themselves or prompt legislatures and agencies to do so); Eskridge, supra note 276, at 161, 164 (arguing that deference to agencies’ dynamic interpretation under the Chevron framework “threatens to unbalance government,” whereas judges are better equipped to “[f]igur[e] out statutory purpose and harmoniz[e] applications of statutes with legal and constitutional principles”). ↑
315. . See Anya Bernstein, Judicial Accountability, 113 Geo. L.J. (forthcoming 2025) (manuscript at 62), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4962653 [https://perma.cc
     /3JVX-G45Q] (arguing that courts’ capacity for democratic accountability is lower than agencies’). ↑
316. . See, e.g., Sant’Ambrogio & Staszewski, supra note 16, at 801 (detailing the efforts by federal agencies to engage the public). ↑
317. . See, e.g., R. Shep Melnick, Regulation and the Courts 13–14 (1983) (“Far removed from the daily operation of administrative agencies, judges may fail to appreciate the complexity of the issues before them . . . .”). In this sense, scholars’ focus on courts’ countermajoritarian nature may be somewhat misplaced. See Barry Friedman, The History of the Countermajoritarian Difficulty, Part One: The Road to Judicial Supremacy, 73 N.Y.U. L. Rev. 333, 334 (1998) (“The ‘countermajoritarian difficulty’ has been the central obsession of modern constitutional scholarship.”); Alexander M. Bickel, The Least Dangerous Branch 16 (1962) (“The root difficulty is that judicial review is a counter-majoritarian force in our system.”). It is instead courts’ inability to convene pluralistic deliberation and evaluate policy effects that should limit their governance role. ↑
318. . See, e.g., Bernstein & Rodríguez, supra note 15, at 1670–71 (discussing the interdependent nature of political appointees and civil servants, as well as different categories of each). ↑
319. . As scholars have documented, for instance, prominent judicial theories of statutory interpretation portray legislative drafting in highly inaccurate, seemingly uninformed ways. See Gluck & Bressman, supra note 71, at 906 (“Interpretive doctrines designed to reflect how members actually participate in the drafting process would look very different, and certainly less text oriented, than the ones we currently have.”); Jarrod Shobe, Congressional Rules of Interpretation, 63 Wm. & Mary L. Rev. 1997, 2003–04 (2022) (showing that the use of judicial canons is in tension with the congressional enactment of interpretive rules); Shobe, supra note 7, at 468 (explaining how Congress often relies heavily on agencies’ expertise when drafting statutes); Cross & Gluck, supra note 136, at 1543 (arguing that textualist scholarship mistakenly focuses on elected officials to the exclusion of congressional staff’s role in legislative drafting); Christopher J. Walker, Legislating in the Shadows, 165 U. Pa. L. Rev. 1377, 1390 (2017) (revealing the ways in which federal agencies help draft statutes through various forms of technical assistance). ↑
320. . See Nicholas R. Parrillo, Leviathan and Interpretive Revolution: The Administrative State, the Judiciary, and the Rise of Legislative History, 1890-1950, 123 Yale L.J. 266, 328–42 (2013) (tracing the federal government’s unique role in briefing and encouraging courts to refer to legislative history). These connections make agencies’ use of legislative history raise fewer concerns about reliability and propriety. ↑
321. . See supra Parts I–II. For scholarly considerations of how courts reconcile the interests and influences of different Congresses over statutory interpretation, see Jarrod Shobe, Intertemporal Statutory Interpretation and the Evolution of Legislative Drafting, 114 Colum. L. Rev. 807, 855 (2014), arguing that statutory interpretation theory should be sensitive to differences in interpretive practices and expectations across different Congresses. There is scholarly debate over which Congress—the enacting or the current Congress—should take primacy in statutory interpretation. Compare Einer Elhauge, Preference-Estimating Statutory Default Rules, 102 Colum. L. Rev. 2027, 2037 (2002) (“Contrary to ordinary supposition, . . . the default rule that overall best maximizes the political preferences of the enacting government will track the preferences of the current government where they can be reliably ascertained from official action.”), with Elizabeth Garrett, Preferences, Laws, and Default Rules, 122 Harv. L. Rev. 2104, 2106 (2009) (book review) (“It is unlikely that legislators would support a system that allows ‘current enactable preferences’ to trump the views of the enactors . . . .”). See also Eric S. Fish, Choosing Constitutional Remedies, 63 UCLA L. Rev. 322, 374 (2016) (“[T]he enacting legislature is far more likely to have coherent preferences [for a particular] statute . . . . [S]ubstituting the views of the current legislature undermines . . . intertemporal democracy—it harms a past electoral majority by refusing to preserve its legislative accomplishments, and grants an unfair windfall to the current electoral majority . . . .”). ↑
322. . Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 845 (1984); see Daniel E. Walters, Four Futures of Chevron Deference, 31 Geo. Mason L. Rev. 635, 652 (2024) (“It is no surprise . . . that the Court’s most conservative Justices began changing their tune on Chevron in recent years—deference on average could be plausibly forecasted to favor a pro-regulatory agenda offered by Democratic Presidents.”); Elinson & Gould, supra note 23 at 478, 481–82 (tracing the political history of deference and showing that what was once a doctrine favored by conservatives has come to be essential to progressive conceptions of government); Rodríguez, supra note 243, at 110–15 (describing the demise of deference doctrines in the hands of “Justices whose conception of law fits awkwardly with Chevron’s basic premise that a law can reasonably be read to yield more than one conclusion or possibility”). ↑
323. . See Christopher J. Walker, Attacking Auer and Chevron Deference: A Literature Review, 16 Geo. J.L. & Pub. Pol’y 103, 104, 112 (2018) (noting calls from the federal courts to rethink deference doctrines and concerns that Chevron deference specifically diminishes Congress). ↑
324. . Bernstein & Staszewski, supra note 5, at 1766. ↑
325. . See Mila Sohoni, The Major Questions Quartet, 136 Harv. L. Rev. 262, 263 (2022) (“While ostensibly applying existing major questions case law, the quartet in actuality altered the doctrine of judicial review of agency action in its method and content, in ways that will have momentous consequences.”); Todd Phillips & Beau J. Baumann, The Major Questions Doctrine’s Domain, 89 Brook. L. Rev. 747, 753 (2024) (noting how recent Roberts Court decisions have created doctrinal instability, including by leaving discretion to judges as the result of the vagueness of the major questions doctrine’s factors); Ronald M. Levin, The Major Questions Doctrine: Unfounded, Unbounded, and Confounded, 112 Calif. L. Rev. 101, 105 (2024) (arguing that the major questions doctrine “cannot be defended”). ↑
326. . The major questions doctrine “displaces the ordinary meaning of statutory text in the name of normative values.” Kevin Tobia, Daniel E. Walters & Brian Slocum, Major Questions, Common Sense?, 97 S. Cal. L. Rev. 1153, 1157 (2024); see also Daniel E. Walters, The Major Questions Doctrine at the Boundaries of Interpretive Law, 109 Iowa L. Rev. 465, 472 (2024) (stating the major questions doctrine “in effect allows systemic departure from plausible readings of statutes on the basis of judicial values and preferences that are at best weakly tethered to higher sources of law”); Chad Squitieri, Who Determines Majorness?, 44 Harv. J.L. & Pub. Pol’y 463, 506 (2021) (describing the risk that the Court favors its political views when applying the major questions doctrine); West Virginia v. EPA, 142 S. Ct. 2587, 2641 (2022) (Kagan, J., dissenting) (“Some years ago, I remarked that ‘[w]e’re all textualists now.’ . . . It seems I was wrong. The current Court is textualist only when being so suits it.” (quoting Harvard Law School, The 2015 Scalia Lecture | A Dialogue with Justice Elena Kagan on the Reading of Statutes, YouTube (Nov. 25, 2015), https://www.youtube.com/watch?v=dpEtszFT0Tg)); Biden v. Nebraska, 143 S. Ct. 2355, 2400 (2023) (Kagan, J., dissenting) (“Small wonder the majority invokes the [major questions] doctrine. The majority’s ‘normal’ statutory interpretation cannot sustain its decision.”). For instances of the lower courts applying the doctrine, see North Carolina Coastal Fisheries Reform Group v. Capt. Gaston LLC, 76 F.4th 291, 297 (4th Cir. 2023), holding that “whether returning bycatch qualifies as a ‘discharge’ of a ‘pollutant’ under the [Clean Water] Act is a major question,” and Georgia v. President of the United States, 46 F.4th 1283, 1296 (11th Cir. 2022), holding that power to require that government contractors be vaccinated against COVID-19 is major and requires clear statutory authorization. ↑
327. . See, e.g., Deacon & Litman, supra note 5, at 1015 (arguing that the doctrine “supplies an additional means for minority rule in a constitutional system that already skews toward minority rule”); Sohoni, supra note 325, at 266–67 (critiquing the doctrine’s theoretical underpinnings); Aditya Bamzai, Judicial Deference and Doctrinal Clarity, 82 Ohio St. L.J. 585, 602 (2021) (criticizing the doctrine’s underdeterminancy); Parrillo, supra note 304, at 1313 (“Vesting power in administrators to make sweeping discretionary decisions with high political stakes was not alien to the federal lawmakers who first put the Constitution into practice.”); Mortenson & Bagley, supra note 304, at 280 (“[T]he Constitution at the Founding contained no discernable legalized prohibition on delegations of legislative power, at least so long as the exercise of that power remained subject to congressional oversight and control.”); Bernstein & Staszewski, supra note 5, at 1766 (criticizing the doctrine as one of a number of recent doctrinal moves disempowering political branches); Rachel Rothschild, The Origins of the Major Questions Doctrine, 100 Ind. L.J. (forthcoming 2025) (manuscript at 8–9), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4602927 [https://perma
     .cc/B8SM-RG3K] (finding that key precedent cited to justify the major questions doctrine involved “exaggeration of economic effects, ignoring the statute’s constraints on agency discretion, and deference to unqualified experts,” allowing judicial claims to “concerns about delegation and ‘major questions’ to put a thumb on the scale for corporate interests over those of the public”). ↑
328. . West Virginia, 142 S. Ct. at 2643 (Kagan, J., dissenting); see supra note 5 and accompanying text. ↑
329. . As we have noted, courts lack agencies’ capacity for convening a broad array of participants to inform policy decisions; they similarly lack a historically evolving sense of statutory mission. Their decisions about the application of regulatory statutes are thus likely less informed, less pluralistic, and less politically accountable than those of agencies. See Rodríguez, supra note 243, at 58–91 (defending many elements of presidential administration as not only legitimate, but oftentimes democratically necessary); Bernstein & Rodríguez, supra note 15, at 1678 (arguing that courts’ interpretations should be shaped by “a realistic understanding” of “agencies’ robust accountability practices”). ↑
330. . By preventing statutory effects from adapting to changing circumstance, the doctrine renders its target statutes effectively obsolete. This obsolescence, moreover, is sporadic: courts employ the major questions doctrine only at the instance of litigants, and only in cases they choose to construe as sufficiently major. Courts thus get to decide which statutes should not be effectuated. ↑
331. . See Buffington v. McDonough, 143 S. Ct. 14, 18 (2022) (Gorsuch, J., dissenting from the denial of certiorari) (“Rather than provide individuals with the best understanding of their rights and duties under law a neutral magistrate can muster, we outsource our interpretive responsibilities.”); Brett M. Kavanaugh, Fixing Statutory Interpretation, 129 Harv. L. Rev. 2118, 2154 (2016) (book review) (noting that, unlike agency policymakers, “[j]udges are trained” to determine the “best reading of the statutory text . . . in a neutral and impartial manner”). ↑
332. . See Bressman & Gluck, supra note 7, at 767 (“[Chevron] seems to assume that Congress is talking to the courts when Congress signals an intent to delegate. In fact, our respondents told us that Congress is communicating with agencies about delegation, . . . utilizing internal and structural cues . . . .”). ↑
333. . Skidmore v. Swift & Co., 323 U.S. 134 (1944). ↑
334. . Interview Comment No. 110 (describing agency responsibility as “we had to build [a] thing”); see also Foote, supra note 14, at 697–702 (arguing that the Chevron doctrine misconstrues agency statutory work as similar to judicial statutory interpretation and calling on courts to recognize agencies’ duty to—and institutional competence at—carrying out a statute’s mandates). ↑
335. . See Whitman v. Am. Trucking Ass’ns, Inc., 531 U.S. 457, 468 (2001) (“Congress, we have held, does not alter the fundamental details of a regulatory scheme in vague terms or ancillary provisions—it does not, one might say, hide elephants in mouseholes.”); MCI Telecomms. Corp. v. Am. Tel. & Tel. Co., 512 U.S. 218, 231 (1994) (“It is highly unlikely that Congress would leave the determination of whether an industry will be entirely, or even substantially, rate-regulated to agency discretion—and even more unlikely that it would achieve that through such a subtle device . . . .”). ↑
336. . See 5 U.S.C. § 706 (establishing the “arbitrary, capricious, [or] an abuse of discretion” standard); Daniel T. Deacon, Responding to Alternatives, 122 Mich. L. Rev. 671, 673–74 (2024) (analyzing agencies’ responsibility to respond to policy options). ↑
337. . See Bernstein & Rodríguez, supra note 15, at 1676–77 (explaining the valuable role of judicial review in enhancing accountability). ↑
338. . See supra notes 322–327 and accompanying text. ↑
339. . See, e.g., Bonnie Honig, Political Theory and the Displacement of Politics 205 (1993) (explaining that politics necessarily consists of conflict and negotiation between subjects); Chantal Mouffe, Deliberative Democracy or Agonistic Pluralism?, 66 Soc. Rsch. 745, 752 (1999) (describing the “dimension of power and antagonism” in democratic politics); Robert Post, Theorizing Disagreement: Reconceiving the Relationship Between Law and Politics, 98 Calif. L. Rev. 1319, 1338 (2010) (“[P]olitics is the practice we use when we agree to continue to disagree.”). ↑
340. . See Metzger, supra note 302, at 2–3, 6, 9–11 (describing recent “political attack[s]” by the Executive and Congress on the administrative state). President Trump’s Schedule F—which purports to transform swaths of career civil servants into at-will presidential employees—exemplifies an effort to dismantle the civil service and hobble entire domains of government. Exec. Order No. 14171, 90 Fed. Reg. 8625 (Jan. 20, 2025) (citing Exec. Order No. 13957, 85 Fed. Reg. 67631 (Oct. 21, 2020)); Donald P. Moynihan, Public Management for Populists: Trump’s Schedule F Executive Order and the Future of the Civil Service, 82 Pub. Admin. Rev. 174, 174 (2022) (describing Schedule F as “invit[ing] a politicization” of the independent civil service); Harold Meyerson, The Blueprint, Am. Prospect (Nov. 27, 2023), https://prospect.org/politics
     /2023-11-27-far-right-blueprint-america/ [https://perma.cc/CW5G-LA5X] (describing the Heritage Foundation’s plans to assist the Trump administration in remaking American government in 2025). ↑
341. . For example, agencies can systematically exacerbate the subordination of the politically powerless. Shah, supra note 288, at 1619–20. They can overvalue financial considerations at the expense of less quantifiable, but no less important, values, leading to a similar subordination of a broad public in favor of particular economic interests. Short, supra note 310, at 766. And they can be over-influenced by highly resourced parties that skew decisionmaking in their own favor. See supra note 141 and accompanying text. ↑
342. . This Methods Appendix is a slightly modified version of the one that appears in our previous work drawing on this data set in the Yale Law Journal. Anya Bernstein & Cristina Rodríguez, The Accountable Bureaucrat, 132 Yale L.J. 1600 (2023). ↑
343. . Although of course any interview study depends on the people who agree to be interviewed, we sought to mitigate the danger of self-serving descriptions by keeping our questions largely practice-oriented, rather than value-oriented, clearly marking off areas where we sought the interviewee’s personal opinion. Our discussion in this Article rests on general trends we saw prevalent in the answers of different people in different positions. Of course, participants could still give self-serving accounts in many ways, and we assume that our subjects, like many people, might often like to portray themselves in a favorable light. This does not, we think, diminish the reliability of the mutually reinforcing descriptions subjects gave of the practices they engaged in, nor does it undermine the information they provided about their own insider views on their workplace. ↑
344. . We list the primary agency by which subjects were employed. For instance, CMS is a component agency of HHS. If a subject was employed within CMS, we list CMS. If a subject worked in the HHS central department, we list HHS. To help protect subjects’ confidentiality, we do not list the number of interviewees in each agency. ↑
345. . One employee contacted the ACLU, arguing that this instruction impinged on employees’ speech rights. After some negotiation, the agency, months later, issued a letter to the ACLU modifying its instruction so that it only prohibited employees from participating in the project during work time. This letter, however, did not go out to the whole department the way the initial email prohibiting participation had. ↑
346. .  We have omitted the portion of the study’s disclosure with our contact information. ↑
347. . The project was approved by the Institutional Review Board (IRB) at both of our home institutions. All research assistants received IRB-provided confidentiality training. We used rev.com for interview transcription, with a nondisclosure agreement. ↑
348. . See Stanton Wortham, Katherine Mortimer, Kathy Lee, Elaine Allard & Kimberly Daniel White, Interviews as Interactional Data, 40 Language Soc’y 39, 42–43 (2011) (noting that interviews “often include positioning by the interviewee that reveals habitual social actions—sometimes actions that characterize the individual interviewee and sometimes actions typical of an interviewee’s social group” and that interview narratives are “particularly rich vehicles for communicating social positions and enacting characteristic actions,” in part because “[n]arrators always ‘voice’ narrated characters as having some recognizable social role, and they always evaluate those characters, taking their own position with respect to narrated characters and events”). ↑
349. . See Kathy Charmaz, Constructing Grounded Theory 3 (2d ed. 2014) (discussing the methods of the “grounded theory” approach in social science, in which “data form the foundation of our theory and our analysis of these data generates the concepts we construct,” and “[g]rounded theorists collect data to develop theoretical analyses from the beginning of a project”). ↑