Violence in the Administrative State

Drawing on an original, interview-based case study of Immigration and Customs Enforcement (ICE) and a synthesis of six decades of social science literature, this Article offers a theory of physical violence in the administrative state that challenges foundational assumptions about administrative law. Approximately one-fifth of federal employees work for administrative agencies that police, fight wars, enforce immigration law, or incarcerate people—in other words, agencies that use force to execute the laws. These agencies are saturated with administrative law that obligates front-line administrators to confer due process, give notice, behave non-arbitrarily, and comply with law. Yet this law often fails to constrain administrators, with unauthorized violence as the result.

Conventional administrative law fails in these agencies because it has developed out of a model of the administrative state to which they do not conform. Most administrative law is designed around the organizational form and norms of bureaucracy. The characteristic role of bureaucracy is rational information processing using standardized rules and procedures. But the lower levels of these agencies belong to the domain of violence. They empower administrators to use physical violence in response to exigent circumstances––a different role, carried out within a different organizational form that has different norms and decision-making techniques. Ordinarily designed to safeguard core rule-of-law values, when bureaucratic administrative law is applied to agencies in the domain of violence, it often masks, and at worst accelerates, unauthorized violence.

For administrative law to potentially address unauthorized violence, it would have to override individuals’ responses and alter agency culture, requiring it to take different forms than bureaucratic administrative law usually does. Recognizing the domain of violence has practical and conceptual payoffs. It can help us understand how and whether law can check violence in the administrative state. And it opens up a universe of questions about the exceptional legal norms applied to agencies in this domain.

Table of Contents Show

Introduction

In May 2023, the Biden Administration began applying a long-planned, widely publicized regulation that allowed migrants to request asylum at the border, albeit with limited legal process.[1] Days later, a reporter followed a group of migrants to the San Diego port of entry. There, armed U.S. Customs and Border Protection (CBP) agents first misdirected the migrants, then turned them away without letting them apply for asylum[2] in a pattern also observed elsewhere on the border.[3] It seemed possible that front-line agency officials were deliberately violating the regulation. The report followed months of open hostility between the workforce and Biden Administration officials, with one agent telling Department of Homeland Security (DHS) Secretary Alejandro Mayorkas in a leaked recording that the Secretary had turned his back on agents the day he was appointed.[4]

The agents’ actions continued a pattern familiar from the Trump Administration, when CBP agents unlawfully turned away thousands of migrants at the border.[5] Across hundreds of incidents, monitors documented agents turning migrants away, threatening them with guns, and “throwing” and “knocking” them “to the ground,” even “stepping on [the] neck” of one asylum seeker as she lay prone on the floor.[6]

Reports like this emerge routinely across administrative agencies that use force. In May 2023, the same month that CBP agents were caught flouting the new asylum policies, reporters revealed that corrections staff across New York state prisons had systematically abused inmates. They then lied to investigators to protect each other, punished the inmates after the abuse was discovered, and evaded sanctions in cases where inmates died or suffered permanent injury—all in violation of administrative regulations.[7] This type of violence extends to the armed forces as well. The U.S. Navy reported that in a breach of agency policy and medical ethics, Navy medics had refused care to injured recruits in the Navy Sea, Air, and Land Teams (SEAL) basic training and even “participated in the[ir] abuse,” leading to hospitalizations and a death but no significant disciplinary action.[8] Similar incidents of apparently unauthorized force have occurred within Immigration and Customs Enforcement (ICE), the subject of this Article’s case study.[9]

Administrative agencies that police, fight wars, enforce immigration law, or incarcerate people create at a massive scale the “field of pain and death” that the law occupies.[10] As many as seven hundred and fifty thousand federal employees[11] work as uniformed personnel or sworn officers for military combatant commands, CBP, ICE, the Bureau of Prisons (BOP), and federal law enforcement agencies like the Federal Bureau of Intelligence (FBI); the Drug Enforcement Agency (DEA); and the Bureau of Alcohol, Tobacco, and Firearms (ATF).[12] In 2019, the most recent year for which comprehensive data is available, these agencies detained nearly one hundred and seventy-five thousand federal criminal prisoners[13] and over five hundred thousand immigrants,[14] engaged in six armed conflicts,[15] and carried out two hundred thousand domestic arrests.[16] These agencies are saturated with administrative law from top to toe, from their authorizing statutes, to regulations governing everything from use of force to procurement, to their internal guidance and organizational charts.[17] Much of this law obligates administrators to follow rule-of-law norms like due process, non-arbitrariness, or—most basically—legality or compliance with law.[18] Yet as the examples above illustrate, administrative law often does not perform this basic function and thus fails to stop arbitrary violence[19] and the misuse of force within these agencies.

Drawing on classic social theory, organizational sociology, a synthesis of six decades of qualitative social science literature, and a novel case study of ICE incorporating original qualitative research, this Article argues that administrative law fails in these agencies because it has developed out of a model of the administrative state to which these agencies do not conform. Most of our administrative law is designed to regulate bureaucracy.[20] Bureaucracy is an organizational form that applies standardized rules and procedures to process evidence and make reasoned decisions.[21] Its goal is to implement government programs.

But the front lines[22] of agencies that do policing, immigration enforcement, war making, and incarceration belong to a fundamentally distinct domain of the state—what this Article calls the domain of violence. Agencies in the domain of violence, which this Article calls force agencies,[23] have different organizational forms, norms, and roles. They underwrite law, social order, and the integrity of the state’s borders,[24] executing the law when the use of physical force or violence becomes necessary.[25] Many parts of government rely on these agencies, turning to them when enforcement without coercion fails. But these administrators are distinctive because they are directly authorized to use physical force against other people.[26]

When ordinary administrative law designed for bureaucracy—what this Article will call bureaucratic administrative law—is applied to these agencies,[27] the results can be pathological. Normally designed to safeguard core rule-of-law values, bureaucratic administrative law within force agencies masks, and at its worst accelerates, the use of arbitrary violence. In noting this lack of fit between administrative law and the organizational form and norms of force agencies, this Article identifies a fundamental irony at the heart of the administrative state. Current administrative law struggles to constrain agencies whose fundamental role is to underwrite and backstop all administration.[28]

Scholars have long known that some street-level bureaucrats do not follow the rules, but the failure of bureaucratic rules in the domain of violence has different causes than the failure of these rules in the domain of bureaucracy. At their front lines, force agencies are not bureaucracies but have a different organizational form and different norms. Two features of this ideal type generate norms and patterns of behavior that thwart bureaucratic administrative law. First, administrators in these agencies respond to perceived dangers or risks to self or others in situations that are dynamic and contextual. Second, the workforce is allowed to use physical violence against other people, which is normally a forbidden activity and may confer on these administrators a sense of exceptionalism.[29] These two features of front-line administrators’ work have been observed to give rise to individual behavior, workforce-wide cultural dynamics, and even gradual, bottom-up changes to organizational form that cut against the top-down imposition of standardized, bureaucratic administrative legal rules.[30] And non-compliance here also has distinctive costs: because of the powers we grant these administrators, non-compliance with rules often manifests as unauthorized violence.

To potentially address unauthorized violence, we would need a different administrative law for the domain of violence.[31] Bureaucratic administrative law provides standards and criteria for assessing evidence and making decisions through regulations, guidance, and other legal forms. Administrative law for the domain of violence would look different. It would use tools like training, discipline, and limits on unionization to routinize the behavior of administrators and change how they are acculturated within agencies.[32] Certain force agencies, like military agencies and some police agencies, explicitly have tried to develop administrative law for this purpose that might help address similar problems at other agencies in the domain.

Along with its implications for our understanding of immigration law and the law of the police, discussed in Parts III and IV,[33] this Article primarily makes several contributions to our understanding of administrative law: conceptual, doctrinal, and descriptive.[34] It asks us to reconsider how we define “administrative law,” suggesting that training, acculturation, and other personnel management tools are core administrative law that deeply shapes agencies and their application of law and policy.[35] It broadens the lens of the administrative state to include agencies that use physical force, a conceptual shift the author explores further in future work.[36]

But perhaps most important, at the highest conceptual level, this Article advances a different way of understanding substantive concepts, like accountability and legitimacy, that preoccupy the current Supreme Court[37] and administrative law scholars.[38] It suggests that as an alternative to assessing whether the administrative state is accountable or legitimate by looking at whether it conforms to top-down, formal legal institutions, we must also assess concepts like accountability and legitimacy from the bottom up.

Our law and scholarship alike tend to assess the quality of governance with reference to top-down, formal institutions like the presidency and the separation of powers. This Article lays bare significant limits to this approach.[39] Picking up an insight of legal realism,[40] it suggests that the way administrative agencies execute the laws involves them to some extent in making the law.[41] If this is so, then concepts like accountability and legitimacy must also be assessed as the law is constructed and applied from the bottom up in day-to-day interactions.[42] Is there legality, accountability, legitimacy, and so on at the level where individual members of the public interact with administrators? How do these individuals perceive the state’s accountability to them and its legitimacy?

Asking these questions from the bottom up gives us a very different perspective in two ways. First, the social science literature suggests that in the domain of violence, the constraints placed on administrators by formal legal institutions can disintegrate.[43] Some informal norms seem to replace them, but these norms are not what we would call accountable or legitimate. This Article does not focus on measuring accountability, legitimacy, legality, or other values that the Supreme Court, legal scholars, and Congress have suggested administrative law is meant to serve. But it does strongly suggest that our assessment of whether our administrative state is accountable, legitimate, and so on will look very different at the front lines of the agencies that actually execute the laws. While recent years’ scholarship on the rule of law[44] and democratic decline has focused on top-down formal institutions—like the presidency, constitutions, and elections[45]—a bottom-up perspective suggests an alarming degree of continuity and perhaps transformation rather than decline in democracy, in the United States at least.

Second, the Article starts to bring into focus the ways that top-down formal institutions and an agency’s workforce can manipulate legal norms in tandem.[46] Both presidents and administrators are meant to operate under the constraints placed on them by Congress’s statutes and the Constitution. But if both the President and his administrators are intent on stretching the law—say, to deploy federal agents to suppress domestic racial justice protests—they will stretch these constraints more effectively than either could alone.[47] Our system arguably offers few constraints on this dynamic, and policymakers are already using this insight to alter how our legal norms are articulated and applied. For instance, the Heritage Foundation has proposed that the next Republican President transform the administrative state by replacing a significant number of civil servants with specially trained political personnel, an ostensibly non-legal change that could reshape the way agencies wield vast bodies of law.[48]

Finally, it is worth asking whose perspective is relevant when we assess administrative action from the top down versus the bottom up and what we lose by excluding some people from the frame. The people who tend to bear the brunt of law’s execution are poor people of color and non-citizens inside and outside U.S. borders[49] who come into contact with a different part of the administrative state than legal theorists typically examine. Moreover, they may have a very different perspective on the administrative state’s accountability, legitimacy, and so on, as a number of political scientists have begun to observe.[50] Yet our political system and administrative state tends to marginalize these same people from decision-making and debates about law and government institutions, often through the operation of law itself. It may be that by marginalizing these voices, we do not just inflict a moral injury and perpetuate hierarchies of race, national origin, and class. We deprive ourselves of the chance to understand what these groups of people know about our administrative state, which may become relevant to all of our lives as political subjects as administration grows and our politics change.

This Article proceeds as follows: Part I briefly illustrates that contemporary administrative law and its study have coalesced around bureaucracy as the ideal organizational type of the administrative state, with bureaucratic administrative law as the means to constrain and shape it. Part II offers a different ideal type for force agencies that can help us understand why bureaucratic administrative law fails in the domain of violence. Applying a methodology with antecedents and current uses in administrative, constitutional law,[51] and public administration scholarship,[52] it synthesizes six decades of social science literature on policing, prisons, the military, and immigration enforcement to build an inductive theory of this ideal type and its features. These studies already suggest a mismatch between this ideal type and bureaucratic administrative law. They identify an important set of dynamics originating within force agencies that encourage their administrators to bend or evade internal policies, legal rules, and accountability mechanisms in their agencies.[53]

Through a granular original case study of ICE, Part III explores the organizational consequences of this mismatch for force agencies in greater detail.[54] While Part II’s ideal type is a synthesis, drawing on many prior studies, Part III illustrates through the concrete lens of a single force agency how the distinctive norms that arise in such an agency accompany—and in some cases, evidently seem to condition—the misapplication or manipulation of bureaucratic administrative law. The case study draws on a mixed-method archive of sources including a group of fourteen semi-structured interviews, some with former ICE and DHS officials, as well as case filings, public statements by ICE’s union, government oversight studies, Freedom of Information Act (FOIA) documents, and investigative journalism reports. This level of detail on ICE’s workforce and organizational structure makes the case study an independent, descriptive contribution to immigration scholarship.[55]

Drawing on both the ideal type elaborated in Part II and the mismatch illustrated in detail in Part III, Part IV proposes that an administrative law made for the domain of violence would regulate an agency’s workforce. The law would acculture and hardwire its workforce with forms of accountability tailored to organizational dynamics within these agencies. It considers training, discipline, and unionization as examples of administrative law tools that can shape a workforce’s routines and culture. Part IV also points out that the Article’s theory of the organizational mismatch between administrative law and force agencies contributes to policing scholarship. The theory helps distinguish administrative law tools that may help reduce unauthorized police violence, like scenario-based de-escalation training, from bureaucratic tools that cannot, like police rulemaking.[56]

I. Background: Bureaucratic Administrative Law

Imagine two people, both seeking asylum in the United States from abuse by a paramilitary group. The first person, A, applies for refugee status from outside the country. A State Department staffer looks at the facts in A’s application and compares them with the Department’s refugee status regulations, perhaps with the Department’s handbook at their side. The official determines that yes, A qualifies as a refugee, and A receives the visa that A has been waiting for. Because decision-making in bureaucracies is inconsistent, a different administrator might deny A’s application, even after applying the regulation to the very same evidence.[57] And other problems, like delay, might hamper A’s path to the United States. But either way, the decision-making process would entail a certain engagement with standardized rules, evidence, and reason-giving.

Now imagine a second asylum-seeker, B, who comes to the southern border and encounters immigration enforcement agents. ICE has criteria determining which asylum seekers to detain and which to release on parole.[58] On a fair reading of the regulation, it favors release for most people—including B.[59] But B, like thousands of other asylum seekers, is detained, because the ICE administrator who receives B’s application does not implement the regulation.[60] Instead, parts of ICE’s workforce obscure the existence of the parole process, decline to accept applications, and do not review the ones they receive.[61] B spends over a year in detention, their mental health deteriorating due to post-traumatic stress,[62] before an immigration judge in the Department of Justice grants them asylum. Only then is B released.

These narratives exemplify how administrative law applies differently in the domain of bureaucracy and the domain of violence. Most of what we think of as “administrative law,” including the regulations that apply to both A and B, is actually bureaucratic administrative law. It is tailored to a particular type of organization, bureaucracy, with particular norms. Bureaucratic administrative law consists of standardized rules and procedures according to which administrators assess evidence. The law imposes weak and sporadic forms of contemporaneous and ex post oversight.

The difference has to do with who applies this law and its institutional context. The State Department staffer who reads A’s visa application is part of the domain of bureaucracy. In an ideal bureaucratic organization, an administrator applies standardized process or criteria and deliberately analyzes information to produce a reasoned decision—however it comes out. But the ICE administrator who receives B’s parole application is within a distinctive domain of the state. A different organization and different norms lead to a very different process (or lack of process) for B. This Part describes bureaucratic administrative law and the assumptions that underwrite it.

Until the 1960s, policymakers and scholars understood administrative law as substance-specific law, tailored to particular agencies with particular roles.[63] Their vision of the administrative state was also broader, encompassing military, prison, and police agencies.[64] Since that time, however, the field of administrative law largely has coalesced around a streamlined model of the administrative state, and a core of trans-substantive law, allowing courts and legal scholars to see across the entire administrative state by simplifying it.[65]

This view assumes that the administrative state, especially at its lower levels, is a bureaucracy defined above all by its rationality.[66] It neutrally analyzes evidence to adjudicate applications and develops regulations based on data and technical expertise.[67] The archetypal administrator in this influential, Weberian model of bureaucracy[68] might be a Social Security Administration (SSA) staffer processing an application for disability relief[69] or a group of Environmental Protection Agency scientists setting a threshold for particulate matter. In an influential work drawing on his deep immersion in SSA and on the sociologist (and trained attorney) Max Weber’s theory, Professor Jerry Mashaw defined the ideal type of bureaucracy[70] by setting out its value, goal, organizational structure, and key decision-making technique.[71] The ideal bureaucracy seeks values of accuracy and efficiency, implements programs, has a hierarchical organization, and employs a decision-making technique that Professor Mashaw calls “information processing.”[72] Generations of scholars have embraced or responded to this concept of bureaucracy as they develop a picture of the administrative state and the actors within it. [73]

Ideal Type	Legitimating Values	Primary Goal	Organizational Structure	Decision-Making Technique
Bureaucracy	Accuracy and Efficiency	Program Implementation	Hierarchical	Information Processing

Though recent judicial attacks on the administrative state have cast bureaucracy in a harsher light,[75] administrative law doctrine reflects similar assumptions about bureaucracy and bureaucratic administrators. The notion of administrative expertise as a justification for deference[76]—which has been a premise of Chevron, Auer, and some other deference doctrines—is one example.[77] Similarly, the presumption of regularity evolved out of assumptions that administrators are rational bureaucrats.[78] And Professor Blake Emerson has suggested that even the ascendant major questions doctrine rests on this idea of bureaucracy, because it assumes technocratic administrators cannot resolve questions of political value.[79]

Much of contemporary administrative law is congruent with the idea that administrators are bureaucrats. It equips them with standardized policies, rules, and procedures to channel their behavior and provides relatively weak and sporadic ex post oversight. This is intuitive if we assume that administrators’ occupational role is to rationally process evidence. Standardized terms support them in performing this role, and they ought to comply without much oversight or sanction.[80] The core of trans-substantive administrative law—executive-branch and congressional controls, judicial review under the Administrative Procedure Act (APA), and internal administrative law—reflects this view.

External control through the political branches. In the conventional picture of the administrative state, Congress and the White House control agencies in two primary (but certainly not exclusive[81]) ways: by writing or influencing the law and policy that applies to them, and by appointing, removing, and otherwise overseeing senior officials. Congress provides standardized rules, policies, and procedures for agencies through statutes,[82] and the White House does so through executive orders,[83] OIRA review,[84] and informal contacts, among other tools.[85] Another major lever in the conventional view is confirmation, appointment, and removal of senior officials.[86]

External control through judicial review. The APA is the primary vehicle for judicial review of agency action. In theory, buttressed by a presumption in favor of judicial review,[87] many actions by administrative agencies can be challenged on substantive or procedural grounds.[88] But in practice, judicial review is circumscribed.[89] Few agency actions make it to court,[90] some that do are expressly committed to agency discretion by law,[91] and judicial review rarely reaches the highly informal work of front-line administrators.[92]

The remedies courts apply are also limited. This constraint is consistent with an image of the administrative state where the public can usually rely on bureaucrats to apply standardized rules.[93] Generally speaking, if an agency action is unlawful, courts will set it aside without ordering additional relief.[94] As Professor Nicholas Parrillo has documented, even in the rare cases where courts enter contempt orders, they almost never sanction agencies, and sanctions district courts do levy are often overturned on appeal.[95]

Agency process and internal administrative law. The policies that agencies make connect external forms of law and front-line administrators.[96] At the highest level of generality, agencies do notice-and-comment rulemaking to create regulations that administrators often implement.[97] But statutes, rules, and other directives often need to be fleshed out before administrators can use them, regardless of whether they are made by the agency, Congress, or the President. This is where “internal administrative law” comes in. Defined by Professors Gillian Metzger and Kevin Stack, “internal administrative law” is “the internal directives, guidance, and organizational forms through which agencies structure the discretion of their employees.”[98] Examples include the State Department handbook and ICE parole policy described at the beginning of this Part. In practice, agencies must operationalize many forms of administrative law this Part discusses through internal administrative law before administrators can apply them.

Professors Metzger, Stack, Mashaw, and Elizabeth Magill have documented key tools of internal administrative law that Parts II and III will discuss and examine.[99] Among the essential forms of internal administrative law is agency structure.[100] Even relatively decentralized agencies need a hierarchical organizational chart that designates officials who make rules and oversee administrators.[101] Another principal tool is guidance,[102] a category of informal document that instructs administrators in procedure, standards, and priorities for their work.[103] Quality assurance and management systems are a third tool.[104] Metrics for case completion and work rate, performance review and discipline systems, and random review of administrative dispositions all let officials oversee administrators.[105]

Mostly, the tools at the core of this simplified picture of administrative law—from congressional authorization and presidential supervision to judicial review, notice-and-comment rules, informal guidance, and quality assurance—are adapted to bureaucratic rationality. Very often, they produce policies and procedures for administrators to implement. In other cases, they designate the officials who have authority to make those policies and procedures. Generating standardized policies, bureaucratic administrative law evidences a particular vision of the work administrators do and the forms of law that will direct them in carrying it out.

Moreover, to a remarkable degree, bureaucratic administrative law relies on administrators to carry out the law with few real checks on their behaviors. Oversight of front-line administrators’ implementation is relatively weak in practice, with judicial review sporadic and sanctions limited. Internal quality control mechanisms and disciplinary systems can lead to action against an employee, but firing is the limit at which sanctions in the civil service stop.[106] The President and Congress may exert control over agency leadership and even an agency’s budget, but most agencies have an entrenched professional civil service that outlasts administrations.[107]

Of course, the stylized picture of bureaucracy and its law that this Part presents is the subject of many caveats. The bureaucratic ideal type of the administrative state is just that: an idealized model from which real bureaucracies diverge. Bureaucrats are corrupt or stubborn, offices are mismanaged, courts fail to step in, and Congress is gridlocked or captured. Scholars have long known this. Within the field we call “administrative law,” Professors Emily Bremer, Daniel Ho, and Anne O’Connell, among others, have complicated stylized assumptions about the dominance of trans-substantive procedure,[108] the utility of internal administrative law for directing street-level bureaucrats,[109] and the boundaries of the administrative state.[110] Indeed, the problem of constraining bureaucratic discretion has been a subject for administrative law scholars at least since it was raised by Professor Kenneth Culp Davis.[111]

There is also arguably much literature sitting outside the formal bounds of administrative law scholarship that calls into question the stylized picture of bureaucracy. For instance, scholars studying bureaucratic agencies that deal with child welfare,[112] patents,[113] and asylum adjudication[114] have noted pathologies within these agencies, including specifically problems related to administrators’ discretion. Among the most famous works on public administration is a classic study of the use of discretion by street-level bureaucrats.[115]

In other words, real-world bureaucracies are beset by fraud, waste, abuse, variable decision-making, and bad governance, including at their front lines.[116] Most law that deals with these problems will necessarily be administrative. A solution that fixes systemic problems rather than simply providing individual relief will have to reach into the agency to change how it works.

Rather than explain how administrative law allows for gaps between the real and the ideal in the domain of bureaucracy, this Article explores a different kind of gap: between bureaucratic administrative law and a distinctive domain of the state where this law is nonetheless applied.

II. The Domain of Violence

Agencies in the domain of violence are inside the administrative state, even if they may not appear that way.[117] Though in some cases they have their own bespoke administrative law,[118] they are also awash in bureaucratic administrative law. For instance, these agencies are mostly covered by the judicial review provisions of the APA,[119] avail themselves of APA procedures like notice and comment rulemaking,[120] and get taken to court under the statute, albeit with mixed results.[121] At the federal level,[122] these agencies derive initial authority to act from statutes[123] and receive ongoing direction from Congress and the President.[124] Moreover, these agencies are largely internally self-regulated, using guidance and other forms of internal administrative law to direct their front-line officers. In other words, bureaucratic administrative law is everywhere in the domain of violence. But in this very different domain of the state,[125] with a different ideal type of organization that has different modal features, bureaucratic administrative law has the opposite of its ordinary, benign effect. Applied to force agencies, bureaucratic administrative law does not promote rule of law values but instead masks and even accelerates the imposition of arbitrary violence. It helps create a feedback loop of organizational pathologies—like rule non-compliance, the overuse of discretion, and tension between line employees and agency leadership—that the rest of this Article will explore.

This Part builds an ideal type for force agencies that helps theorize why bureaucratic administrative law fails within them. To generate the ideal type, I synthesized six decades of largely underutilized[126] qualitative research on organizations that police, incarcerate, and fight in combat. I used an inductive approach to identify a set of internal mechanisms (such as cultural norms and workforce-level organizational structures) that force agencies share and condition non-compliance within them. This model is not meant to be a definitive statement; through further qualitative research on force agencies, it could be tested and improved. Nor is it meant to be an exhaustive depiction of the reasons administrators in force agencies violate rules. However, it does identify a set of internal organizational dynamics that are particular to force agencies and incompatible with bureaucratic administrative law.

The goals of this Part are two-fold. First, this Part grounds the Article’s contention that approaching force agencies as a group has analytic benefits beyond either a purely trans-substantive or agency-by-agency approach. This Part illustrates that force agencies share a common value, goal, and front-line organizational structure, and that their administrators deploy a particular decision-making technique. In other words, this Part suggests that these agencies experience what the organizational theorists Paul DiMaggio and Walter Powell have called “normative isomorphism,” or convergence in organizational structures due to norms shared among their workforces.[127] These norms make force agencies similar to each other and different from bureaucracies, especially at the lower levels of their organization.[128] Second, this Part suggests that bureaucratic administrative law can fail when applied to force agencies because of two distinctive aspects of these agencies: first, the perception of dynamic and contextual dangers or risks to self and others, and second, the capacity to use violence.

A.     An Ideal Type for Force Agencies

1.  Legitimating Value: Order[129]

In force agencies, administrators exert control through the use or threat of physical violence so that they can enforce and protect order. For administrators, fulfilling this role legitimates their work.

The state designates these agencies to carry out its basic police functions: to underwrite law, social order, and the integrity of the state’s borders.[130] To facilitate this role, the state delegates force agencies its monopoly of legitimate force.[131] The value of order is embedded in concepts such as the “thin blue line”—these administrators hold back the disorder that otherwise would burst in upon society.[132]

2. Primary Goal: Control

If the overall value in force agencies is order, this value is enacted by exerting control over physical spaces, situations, and people. This makes control the goal that force agencies and their administrators pursue.

For instance, social scientists and theorists have described police as emergency problem solvers,[133] as crime fighters,[134] and as agents of the normative social order.[135] Their tasks may include handling a noise complaint,[136] arresting or chastising a “disorderly” person on the street,[137] catching a criminal perpetrator,[138] or maintaining the dominant social hierarchy of race and class through a program like stop and frisk.[139] Similarly, the work of immigration agents involves “control [of] situations they perceive as threatening,” like encounters with undocumented migrants at the border.[140] And the role of corrections staff implicates the physical control of incarcerated people whom they come to see as dangerous.[141]

Synthesizing the literature, the goal in force agencies is to use the agency’s authority, backstopped by the capacity to use violence,[142] to assert control over spaces (like streets, detention centers, and prisons), situations (encounters with undocumented migrants, instances of disruptive noise), and people (detained people, members of subordinated groups) to avert disturbances to order. The goal of control, the circumstances under which administrators pursue it, and the tools they deploy shape numerous features of these organizations and the people within them. These features include everything from who joins the workforce and what attitudes these employees hold,[143] to how they perceive the legitimate uses of their powers,[144] to how the public perceives the agency.[145]

3.  Organizational Structure: In-Group/Out-Group

One important feature of many force agencies is an in-group/out-group dynamic that develops at lower levels of the organization. At upper levels, force agencies are hierarchical. Indeed, their upper levels tend to be bureaucratic—perhaps one reason they make bureaucratic law for front-line administrators.[146] But an in-group/out-group dynamic can arise at these agencies’ front lines, leading to tension between these two groups, and between front-line administrators and the public.

Of course, there are cliques or staff-management tensions in all kinds of agencies.[147] But in force agencies, administrators operate under distinctive conditions of perceived physical danger and vigilance, as they patrol, surveil, or walk the beat.[148] Relying on each other for physical protection, material support, and understanding, workgroups can form strong internal ties while becoming isolated from senior officials[149] and outsiders,[150] including the people they are meant to control[151] and the broader public.[152] Researchers have found that this dynamic can enclose the workgroup in its own normative world, one where behavior that is generally prohibited—like bending rules—becomes justified.[153]

4.  Decision-Making Technique: Identify Risks in a Dynamic Context

At a granular level, force agencies and their administrators understand their day-to-day work as responding to and establishing control of perceived vectors of danger or risk.[154] This means that identifying these risks, in contexts that are fast-moving and dynamic, is the core decision-making technique. From the moment they enter a force agency, administrators are taught that their role is to identify possible dangers to public order or their own safety.[155]

Oriented in this way, some of these administrators tend to overestimate threats. Recent research has found that police and Border Patrol agents tend to overestimate the dangers they face in their work, a tangible illustration of how perceptions of risk by administrators in force agencies can decouple from the facts.[156]

Interestingly, some theorists of force agencies, like the administrators they study, have come to the conclusion that administrators must have broad discretion to identify and respond to threats in dynamic contexts.[157] Theorists like James Q. Wilson, who are sympathetic to the order maintenance goals of these agencies, have contended that administrators must have latitude to decide which threats most require their attention and how to address them in unpredictable and risky situations.[158] Work by Wilson and similar theorists not only reflect attitudes among administrators in this domain. It has long helped to shape a broader perception in society and among policymakers that force agencies require discretion, likely reinforcing the beliefs and practices this article critiques.

Ideal Type	Legitimating Values	Primary Goal	Organizational Structure	Decision-Making Technique
Bureaucracy	Accuracy and Efficiency	Program Implementation	Hierarchical	Information Processing
Force Agency	Order	Control	In-Group/Out-Group	Identify Risks in Dynamic Context

B. Why Bureaucratic Administrative Law Fails in the Domain of Violence

As Part I discusses at length, bureaucratic administrative law is made for bureaucracies whose goal is to implement programs, and for bureaucrats who analyze evidence in accordance with standardized policies.[159] But force agencies are not bureaucracies, and many of their administrators are not bureaucrats. The culture and behavior of the workforce in these agencies are critically shaped by both perceived danger or risk and the capacity to use violence, in ways that give standardized rules designed for bureaucracy limited purchase on these administrators’ work.

Empirically, we know that administrative rules often fail in force agencies. The decades of studies cited in this Part have found that some administrators bend or break standardized rules routinely, and even express the idea that they are justified in doing so.[160] Studies of police agencies, for instance, have long observed that some agencies are shot through with complex patterns of excessive force,[161] secrecy,[162] lying,[163] and covering for colleagues.[164]

One reason for these patterns of non-compliance may be the perceived mismatch between standardized rules and situations of perceived danger or risk. Part of this mismatch may be about decision-making technique. As this Part details, force agencies do not statically present agents with a series of applications or cases to rationally determine. Rather, they ask administrators to respond to what they perceive as dangers or risks in dynamic conditions.[165] Bureaucratic regulations may be a poor match for this domain because they do not fit the dynamic and contextual nature of this work. This mismatch also may be one reason that force agencies often allow administrators to use discretion, and that administrators sometimes seek additional discretion, justifying it in terms of their work.

It is worth noting, though, that other administrators who respond to dynamic and contextual risks do not believe their work requires vast discretion, nor do their agencies display the same systematic non-compliance with administrative rules.[166] Consider firefighters. Like administrators in the domain of violence, firefighters face unpredictability and risk, and some scholars have found that the job attracts employees with similar attitudes.[167] Firefighters also break rules: while embedded with the U.S. Forest Service, the sociologist Matthew Desmond saw wildland firefighters break organizational safety codes in the course of their work.[168] Yet unlike in force agencies, here firefighters throughout the organization expressed very positive attitudes to the codes, suggesting the breaches were accidental, not the result of conscious attitudes to the rules.[169] Rather than suggest non-compliance could be justified, firefighters voiced the view that compliance was essential, going so far as to attribute deaths on the line to failure to follow the rules.[170] Other emergency responders, like paramedics, have been found to closely hew to standardized rules, even if the rules cause frustration or do not perfectly fit their work.[171]

These examples suggest another reason bureaucratic administrative law may be incompatible with the risk identification aspect of this domain. Administrators may perceive bureaucratic administrative law as being meant to protect the targets of force, rather than the administrators themselves, even as administrators believe that their role implicates great risk.

The capacity to do violence itself may also be an important part of the story. Professor Jerome Skolnick has suggested that violence—both the capacity to do it and the possibility of suffering it—leads “procedural requirements” to “take on a ‘frilly’ character” or “a secondary position in the face of circumstances seen as threatening.”[172] Alternately, because violence is only authorized in limited circumstances, it may add to administrators’ sense that their work is important and special—which may in turn seem to justify procedurally irregular means.[173] And perhaps most fundamentally, Professor Robert Cover long ago implied that there is an inherent tension between violence, whose imposition requires administrators to overcome the social and legal conventions that ordinarily prohibit it, and the law that we expect these administrators to continue to apply.[174] Compounding this problem, most important work in force agencies involves informal, discretionary enforcement actions, like arrests,[175] that often leave little or no paper trail for oversight mechanisms to pick up.

To summarize, we can expect bureaucratic administrative law to poorly suit force agencies—indeed, decades of observation by researchers tell us as much. Both decision-making under dynamic contexts of danger and risk, and the authority to use violence itself, may breed attitudes and institutional dynamics that encourage non-compliance with bureaucratic law. The case study in Part III explores these dynamics in detail as they play out within a single agency.

III. Administrative Law in the Domain of Violence: A Case Study

How does administrative law function differently in the domain of violence than in the domain of bureaucracy? This Part adds illustrative granularity to the theory of organizational mismatch elaborated in Part II[176] through an original case study of ICE’s Enforcement and Removal Operations (ERO) division, a force agency that does interior immigration enforcement. The case study shows how parts of ICE’s front-line workforce—often alone, and sometimes in collaboration with sympathetic agency leadership—have defeated, manipulated, or thwarted oversight using ICE’s bureaucratic administrative law, in ways consistent with the theory of organizational mismatch developed in Part II.[177]

ICE ERO is an ideal case study to illustrate how bureaucratic administrative law functions in force agencies. Unlike its predecessor the Immigration and Naturalization Service (INS), which combined benefits and enforcement functions in a single agency, ICE ERO is largely focused on enforcement. The majority of its front-line administrators have the power to use force, or directly assist those who do.[178] Many details of its organizational structure and internal rules are public; it is larger than most;[179] and its eight thousand agents frequently do a mix of arrest, detention, and deportation work and more traditional desk work.[180] Perhaps most important of all, as an immigration enforcement agency, ICE is often framed as “exceptional,” even though conventional constitutional law and bureaucratic administrative law ostensibly apply to many facets of its operations.[181] These features, which in some contexts would disqualify ICE as a case study, make it paradigmatic within the domain of violence.

This Part relies on mixed-methods sources selected for their ability to paint a rich picture of ICE’s organizational environment, making it a freestanding contribution to our understanding of ICE as an organization in immigration law scholarship. Federal law enforcement agencies can be notoriously hard to research with qualitative methods, and immigration enforcement is no exception.[182] As the sociologist Irene Vega, who produced a pathbreaking qualitative study of field-level officers in the U.S. Border Patrol, noted in 2019, “DHS deploys both formal and informal gatekeeping mechanisms that keep researchers out. While there are real challenges in gaining access to DHS employees, researchers must not turn away from scrutinizing the state and its agents because of them.”[183] By incorporating a set of fourteen semi-structured interviews with well-located informants; corroborating the interviews with an archive of primary and secondary sources, including two thousand pages of primary source guidance documents; carrying out the study against the backdrop of the author’s field experience;[184] and indexing it against the large body of existing literature on similar organizations discussed in Part II, the case study contributes to our still-limited understanding of ICE’s institutional design and the culture and behavior of parts of its workforce.[185]

A.     ICE’s Bureaucratic Design on Paper

As Parts I and II observe, bureaucratic administrative law often applies in the domain of violence. ICE is no exception, sharing in many features of bureaucratic administrative law on paper. This Part focuses on four features:

Hierarchy and centralization. As Part I explains, hierarchy and centralization are core features of bureaucratic administrative law. They help identify the officials who get to make policy for front-line administrators.[186] On paper, ICE as a whole is designed to be hierarchical and at least somewhat centralized. It has a headquarters that sets a policy agenda, led by a Senate-confirmed Director.[187] ICE ERO in particular is also so structured, with a headquarters overseeing 25 unionized, geographically bounded, and hierarchically organized local field offices.[188]

Guidance. Guidance is also an essential tool to channel administrators’ discretion in bureaucratic administrative law. A broad variety of internal policies, manuals, and handbooks give administrators standardized, granular procedures and rules for carrying out their duties.[189]

Given ICE’s size and the complexity of its operations, it is no surprise that its headquarters has developed an extensive body of guidance for its administrators. ICE’s responsibilities are diverse. They of course include arrest, detention, and deportation, all tasks ultimately backed by the ability to threaten or use force.[190] But at any given time, ICE is also monitoring millions of immigrants to see if they attend immigration court and check-ins.[191] This is a classically bureaucratic task, even if ultimately ICE can directly enforce monitoring requirements while other bureaucratic agencies must delegate this work to a force agency. Similarly, Congress has given ICE the obligation to manage a large carceral system[192] in a way that ensures immigrants are physically safe,[193] can complete their immigration cases with access to counsel,[194] and can access disability accommodations and medical care.[195] And Congress has required ICE to report on its compliance with detention standards.[196]

Guidance is important to translate these high-level directives into more refined policies for administrators. The guidance ICE has issued for its workforce often takes the form of memoranda, bulletins, and handbooks, and governs arrests,[197] detention policies, and removals.[198] At least some of this guidance asks officers to apply standardized, bureaucratic regulations to individual cases. For instance, ICE gives its administrators criteria to decide who will be released from detention on their own recognizance, who will be electronically monitored, and who will be detained.[199]

Management and quality assurance. As Part I describes, bureaucratic administrative law assumes that most administrators will implement guidance from agency leadership with relatively limited oversight and sanctions. As a result, management, training, discipline, and quality assurance processes within these agencies are suited to civil servants who operate with autonomy.[200] ICE has long had at least some management and quality assurance structures, with the Biden Administration reportedly making an effort to implement more.[201] ICE agents attend a centralized basic training tailored to law enforcement.[202] Subsequently, though, ICE relies on bureaucratic management and quality assurance tools, with policy and procedural updates primarily communicated by oral and written guidance.[203] ICE also has a system for employee evaluation and discipline that its headquarters oversees using quality assurance metrics.[204]

Judicial review. Part I notes that in practice, judicial review in bureaucratic administrative law is sporadically applied to administrative agencies, captures little informal activity, and typically does not reach deep into the minutiae of front-line administrators’ work. Doctrines like finality and the presumption of regularity are meant to shield the day-to-day activity of administrators from scrutiny, with judicial review seen as both disruptive and unnecessary.[205]

ICE is taken to court relatively often, and at a high level of generality, courts review the agency along the lines described in Part I. Many suits against the agency concern written policies, rather than informal agency action, and judges have tended to strike down the policy as a remedy.[206] With limited exceptions, which I discuss in other work, courts have also tended to apply finality and the presumption of regularity to the actions of ICE’s front-line administrators.[207]

In other words, the administrative law that applies to ICE shares many features of traditional, bureaucratic administrative law. However, while the law that applies to ICE is law for bureaucracy, ICE is a force agency and has a different organizational form, as Part II sets out. The physically violent aspects of ICE’s role deeply shape the agency, warping how this law is applied.

B.     ICE’s Internal Environment: The Perception of Risk

Part II suggests that in the domain of violence, the twin characteristics of perceived danger or risk and capacity to use violence shape agency culture and workforce behavior. Moreover, this is true independent of how the workforce spends its time or the objective risks it faces. This case study indicates that the same is true within ICE. Though ICE’s statutory mandates envision that officers will do desk work and help manage carceral spaces that meet detained people’s multifaceted needs,[208] the idea that ICE’s work involves risk, danger, and responding to public safety threats resonated throughout the interviews, primary sources, and secondary sources collected for this case study. The context of ICE’s work and relevant data call these ideas into question. Unauthorized presence is a civil, not criminal, offense, and detention is lawful because it does not serve criminal purposes like incapacitation.[209] Immigrants commit crimes at lower rates than U.S. citizens,[210] the majority of people in ICE detention have no criminal record, and only one in ten have what ICE calls a “serious” criminal record—which is not the same as a violent criminal record.[211] In the past ten years, none of ICE’s eight thousand staff were killed by an immigrant in the line of duty.[212] However, in its organizational focus on violence prevention and danger despite incompatibility with observed occupational risks, ICE resembles the Border Patrol and police agencies discussed in Part II.[213]

This organizational focus starts with hiring and training. ICE is hiring, at least to some extent, out of a law-enforcement oriented pool. A significant number of new hires come to it from other agencies in the domain,[214] and a few interviewees suggested many ICE ERO administrators would have preferred careers with agencies like the FBI or DHS’s Homeland Security Investigations (HSI) that have a stronger law enforcement mandate and fewer carceral responsibilities.[215] Prior FOIA work reveals that once administrators enter the agency, ICE’s trainings reinforce a particular vision of the agency’s mandate, preparing the workforce for risky and unpredictable tasks that may require physical force. These risky and unpredictable tasks include enforcing criminal warrants,[216] arresting “fugitive” immigrants to meet quotas,[217] and pretending to be local police to gain entry to immigrants’ homes.[218] As one interviewee put it, administrators enter the agency “being told that they’re the blue line.”[219] These trainings also incorporate explicit messages about danger. ICE’s basic training conveys that “Law Enforcement work is inherently dangerous,” “Officer safety is paramount,” and immigrants being arrested are “threat[s].”[220]

Interviews and primary and secondary sources suggested that ICE’s workforce sees itself as police-like, and that its union positions the street police role of walking the beat and identifying threats as an important expression of its work.[221] Interviewees reported that ICE officers, in their experience, saw themselves as “police” or “cops,”[222] or that the workforce tended to be “enforcement minded.”[223] A former government interviewee stated that ICE officers “wanted to be seen as going out there and making arrests, preventing crime,” and that the agency also treats detention as “crime prevention. It’s like a Minority Report kind of thinking: well, if I take them off the streets, they can’t hurt anybody.”[224] ICE’s extremely powerful[225] union, which represents nearly all front-line officers, has advocated expanding ICE administrators’ discretionary enforcement work, urging that the agency “have our officers do less paperwork,” “expan[d] . . . the law enforcement duties our officers currently perform in the field,” and give officers power to conduct street arrests, as police have.[226] These observations of third parties and statements of ICE’s workforce speak to a particular vision of ICE’s front-line work that centers risk and danger, where—like street police—ICE administrators identify and respond to the greatest threats.

Messages about immigrant dangerousness came up over and over again in interviews and sources collected for this case study. ICE’s union president has stated that “countless lives will be saved” with increased enforcement,[227] and that politicians who criticize ICE put “the safety and lives of our officers at risk” by “empower[ing]” “criminals” to “become resistant and aggressive.”[228] Interviewees with extensive experience monitoring or working within ICE facilities recounted stories of ICE officers treating visibly disabled detained people as dangerous[229] or justifying their work on community safety grounds that clash with ICE’s statutory purposes.[230] In an emblematic image, an interviewee recalled being “shocked” to discover that ICE “transport[s] people for civil detention . . . in chains, to their belly, to their feet.”[231] Interviewees also suggested that the notion that ICE’s role is immigrant crime prevention goes hand in hand with workforce reluctance to get involved in detention management, seen as a less prestigious form of law enforcement work in the professional hierarchy of this domain, and one imbricated with caretaking responsibilities.[232] One former government interviewee stated: “There’s no appetite in the workforce to actually do the detention piece,” where “you deal with food, you deal with hygiene.”[233] Two interviewees suggested that these preferences have played a role in notorious problems with detention conditions in some ICE facilities.[234]

C.    Bureaucratic Administrative Law in Action at ICE

1.     Hierarchy and Centralization

On paper, ICE looks like a hierarchical, centralized agency that filters principals’ preferences down to the front line.[235] But as Part II describes, features of the domain of violence cut against bureaucratic hierarchy and centralization. Agents may resist management as an out-of-touch “out group,”[236] and perceive a tension between centrally promulgated, standardized rules and exigencies on the ground.[237]

A similar dynamic prevails within ICE. Numerous interviewees painted a picture of a “terrifically decentralized” agency[238] where management has struggled to exert control over parts of the workforce.[239] Interviewees characterized ICE field office directors as “little kings and queens”[240] “in charge of their own shop,”[241] and suggested that in practice they have great discretion as to whether to implement policies promulgated from above.[242] One former employee explicitly linked this structure to ICE’s “law enforcement role,” suggesting that such agencies tend to practice “a fairly delegated form of management.”[243] But this is more than a deliberate design choice by political principals—it is the product of workforce preferences and action. For instance, two interviewees reported that the Biden Administration had introduced more discretion into a revised draft of recent enforcement guidelines partly in response to contentious interactions with the ICE and CBP workforces.[244] In an earlier case, detailed below, the workforce refused to implement new enforcement guidelines entirely.[245]

Devolved authority at ICE may not be by Congress’s or the President’s design,[246] but as interviewees indicated, ICE’s union has sought to devolve power to the workforce and weaken agency leadership. The union frequently publicly criticizes agency management,[247] has successfully intervened to derail ICE director nominations in collaboration with members of Congress,[248] has sued the agency for having too lenient an enforcement policy,[249] and has interfered in internal policymaking in ways described below.[250] All of these interventions recall the in-group/out-group organizational structure discussed in Part II, with ICE’s union taking hostile positions against the agency’s management.

2.     Guidance

Like many bureaucracies, ICE is designed so that the agency often makes policy by guidance. But guidance is ill-tailored to narrow discretion in the domain of violence. The granular, standardized policies and procedures it provides are suited to a bureaucratic decision-making technique. However, they are in tension with the perception discussed in Part II that extensive discretion is key to these administrators’ role.

This incompatibility is reflected within ICE, where the workforce has sometimes expressed opposition to guidance issued by agency leadership. In line with the decision-making technique described in Part II, ICE’s union connects its opposition to guidance to the danger it confronts, suggesting that constraints on its discretion thwart its ability to respond to situations that implicate both self-risk and risk to the public.[251] The ICE union’s President, Christopher Crane, has stated that enforcement guidelines “thro[w]” the safety of the U.S. public “out the window,”[252] and the union has opposed detention guidelines on grounds of “officer safety.”[253]

The union has at times exerted unusual power over the process of making guidance, often to weaken or derail constraints on discretion or to delay its development or implementation. ICE’s union has delayed or vetoed even humanitarian policies requiring ICE officers to ask detainees “if they need medication” or to check for “signs of trauma, abuse, or trafficking.”[254]

And where guidance does exist, ICE field offices have developed strategies to avoid implementing it, paralleling the findings about rule-bending in the literature from Part II.[255] Damus v. Nielsen is illustrative.[256] In 2018, detained asylum seekers sued ICE over its implementation of a policy regarding parole from detention. The Damus complaint describes how some field offices had applied (or rather, not applied) the written parole policy:

Asylum seekers routinely are not notified of the availability of parole, in violation of the Parole Directive. . . Some asylum seekers are denied parole even before they are notified of their ability to seek it, again in violation of the Parole Directive. Others are not given interviews before their parole requests are denied, likewise in violation of the Parole Directive. . . Indeed, sometimes the ICE Field Offices simply do not respond to asylum seekers’ parole requests at all. Even where the ICE Field Offices ostensibly consider parole requests, they no longer provide individualized determinations of flight risk and danger, in violation of the Parole Directive. . . [but rather] form denials. . .[257]

As a result of these practices, five of the country’s largest ICE field offices had gone from granting over 90 percent of parole applications by asylum seekers to denying between 92 and 100 percent of them.[258]

Earlier research by Professors Adam Cox and Cristina Rodríguez similarly found that ICE’s leadership lost a conflict with the workforce over enforcement discretion during the Obama Administration, when ICE’s workforce refused to implement new enforcement guidelines.[259] And in other cases, the workforce has not evaded internal law, but has gradually manipulated it to enhance its discretion to respond to ostensible risks posed by immigrants. Professors Robert Koulish and Kate Evans have documented through FOIA work how front-line administrators were able to increase their discretion to detain immigrants by manipulating ICE’s risk assessment algorithm until the algorithm found that most immigrants ought to be detained.[260]

3.     Management and Quality Control

Management and quality control practices within ICE have emphasized the law enforcement nature of the agency’s work and the danger of the role while providing minimal, bureaucratic checks on officer discretion. This combination reinforces the idea that officers have broad latitude to enforce the law as they see fit in response to exigent situations. As discussed above, ICE’s initial training materials extensively discuss danger and risk.[261] Yet, in a 2017 report, the Inspector General characterized ICE ERO’s post-hiring training as “primarily on-the-job and informal.”[262] It described the agency passively communicating guidance and criteria for decision-making in a disorganized way, noting that ICE sent guidance to “field-office personnel orally or by email,” and posted written policies “on websites where they are poorly organized and difficult for field office staff to find.”[263] In-person training was ad hoc and one-off, with attendance not required in practice.[264]

On the discipline end, in recent years ICE had weakly monitored procedures that were set by negotiation with its employee union.[265] ICE’s policy placed discipline for most administrators at the field office level, so that their close colleagues managed the process.[266] For quality assurance, ICE headquarters monitored the timeliness of misconduct cases, rather than any substantive metric.[267] This system did not provide powerful oversight. The Government Accountability Office found in 2018 that immediate supervisors responsible for discipline had not reviewed half of all cases in the system.[268] Approximately 70 percent of misconduct cases resulted in no documented action of any kind.[269] And investigative reporting has found that ICE’s (and DHS’s) disciplinary processes have failed repeatedly to redress on-the-job sexual harassment, sexual abuse, and other actionable offenses.[270] As noted above, two interviewees described interventions by the Biden Administration to improve management and quality assurance within the agency.[271] Time will tell whether these bear fruit.

4.     Agency Manipulation of Judicial Review

ICE manipulates bureaucratic administrative law to mask enforcement practices at the front line when it comes to judicial review. Consider the presumption of regularity, a background norm of administrative law that assumes bureaucrats carry out their duties as they describe.[272] In recent years, ICE has taken advantage of the presumption to mask front-line administrators’ policy violations.

In the run-up to Damus v. Nielsen, the lawsuit about ICE’s parole policy mentioned above,[273] Trump Administration officials had initially intended to eliminate the parole policy.[274] But the agency realized there was a drawback. In a Supreme Court case about immigration bond hearings, Jennings v. Rodriguez, the government had taken the position that bond hearings for detained immigrants were unnecessary because parole was available.[275] Rather than eliminate the policy and jeopardize its position in Jennings, ICE leadership left the policy on the books and communicated to the workforce that it could use its discretion to implement it.[276] It continued to claim in Jennings that the policy remained in place.

The strategy proved effective, as some courts took ICE administrators’ presumed compliance with the parole policy at face value. The government won Jennings and did not have to grant bond hearings.[277] Parole rates plummeted, because many of ICE’s front-line administrators exercised discretion to detain.[278] And attorneys for the agency could claim in Damus that the parole policy remained in place. The asylum seekers ultimately prevailed on their claim after producing extensive quantitative evidence that its implementation had changed. But the process of gathering evidence sufficiently robust to build a successful lawsuit and overcome the agency’s claims took over two years, during which many people remained detained.[279] A similar manipulation of guidance to mask conditions on the ground occurred in the COVID-19 pandemic. Attorneys for the agency often successfully argued that ICE’s detention facilities were following public health guidance, even as investigative reporting revealed mass violations and few efforts by ICE administrators to correct them.[280]

Damus also illustrates how parts of ICE’s workforce have learned to manipulate the APA’s finality requirement, sometimes in collaboration with ICE leadership.[281] Finality ensures that review of agency action happens at the appropriate time, so that courts do not waste resources or invade agencies’ decision-making processes.[282] By pocketing applications or not announcing policies, parts of ICE’s workforce have learned to obscure final actions. In Damus and similar cases,[283] the workforce developed tactics to avoid denying parole applications and generating a final agency action.[284] In other cases, ICE leadership increased enforcement by pushing discretion to the workforce, and then—when sued—claimed that plaintiffs had attempted to manufacture a policy from a set of diffuse and unrelated enforcement actions.[285] In some cases, ICE succeeded in concealing from courts what were effectively final actions by administrators.[286]

IV. Administrative Law for the Domain of Violence

This Article develops a theory of violence in the administrative state that explains why bureaucratic administrative law has persistently failed as a check against the unauthorized use of physical force. While most of the Article highlights the descriptive uses of this theory of organizational mismatch, Part IV discusses its policy implications.[287] As this Article describes, bureaucratic administrative law often applies to force agencies. But bureaucratic law fails here. Force agencies have a different value and goal from bureaucracies, demand different decision-making techniques, and evidence different norms and workforce behavior. To potentially address unauthorized violence, administrative law would have to account for the domain of violence as a separate domain of the state. It would take form around the ideal type of a force agency (not a bureaucracy) and seek to realize different goals than it conventionally does. But how should administrative law for this domain be different? Taking current public policy on law enforcement, incarceration, and other aspects of state coercion as a constant,[288] this Part draws on the theory of organizational mismatch in Part II to develop a tentative answer.

One important note before the normative argument proceeds. The answer is “tentative” because of tension between the conventions of legal scholarship in making normative arguments and some of the theoretical assumptions of this Article. Legal scholarship builds normative arguments around the assumption of a high level of agency[289] in controlling and steering state institutions, often through top-down legal mechanisms.[290] Whether it is reformist, calling for training or new rules of legal liability, or radical, calling for abolition of government agencies, the scholarship often assumes that legal change and contestation can profoundly alter the behavior and composition of complex, path-dependent state institutions.[291]

But this Article assumes a more complicated relationship between agency and structure, or the aspects of social organization that create regularity in organizational and individual action.[292] Specifically, this Article suggests that the quintessential state role of enforcing order through force gives rise to organizational forms and norms that help produce arbitrary violence. Because the targeted behavior relates to a fundamental role of the state, this argument implies limits on agency to change this behavior.[293]

This Article foregrounds ways that structure at various levels shapes the action of front-line administrators. But this focus does not necessarily imply that legal actors cannot take any effective, autonomous top-down action to reduce administrative violence against individuals. Accordingly, and because the relationship between structure and agency in a social theory of law exceeds the bounds of this Article,[294] the Article adopts the basic, agentive assumptions of most legal scholarship as it makes normative recommendations. The Article assumes that top-down legal interventions can have an effect and that administrative law can achieve policy goals. At the same time, taking the Article’s argument seriously means acknowledging that even domain-specific administrative law will not be able to eliminate arbitrary state violence.[295] Domain-specific administrative law will run up against many other social structures, including (other aspects of) law itself, that will limit its utility.[296] Hence my reasons for calling these proposals “tentative.”

Returning to the main argument, based on the theory of organizational mismatch developed in Part II, an administrative law for force agencies might foster automatic routines, and eliminate institutional factors that reinforce a distorted workforce culture. As the policing literature observes, one way to ensure that distortions in workforce culture in force agencies do not harm the public is to disaggregate agencies’ functions, shifting services to more task-relevant agencies with fewer cultural problems.[297] Since force agencies often ask administrators to fulfill functions well beyond execution of the laws, this idea is a natural fit for the domain of violence. For instance, were ICE to hire social workers to review parole applications, rather than law enforcement officers, the agency would likely implement its parole policy in a different way.

Still, significant parts of both ICE and the other agencies in this domain would continue to execute laws using physical coercion. This means that a domain-specific law would still be necessary. To explore this idea, this Part briefly considers three domain-specific administrative law tools that some force agencies—most notably the military and some police departments—have developed in an attempt to alter routines and culture: training, discipline, and unionization. Though the effect of these tools is difficult to know and measure, they are at least the product of agency-specific thinking about the particular challenges inherent in constraining the use of force.

Routines are a worthwhile focus because they can create automatic responses, even in situations of perceived emergency.[298] As Professor Stephen Holmes noted in the counterterrorism context, many professionals who confront emergencies, like emergency room staff and flight crew, are not encouraged to use discretion in essential aspects of their work.[299] Instead, they are repetitively put through routines until they become ingrained and automatic, so that instead of responding to risky situations with discretion, they use specific protocols.[300]

Culture is also a worthwhile focus because, as Parts II and III suggest, it can reinforce counterproductive dynamics that arise through administrators’ work.[301] Foundational work in the 1980s by Professor Jerry Mashaw noted the role training and acculturation play in influencing agencies’ implementation of the law,[302] but acculturation has been peripheral to our understanding of administrative law.

A.     Training

Even as ICE’s training creates a culture focused on danger, it provides limited routine. As the case study discusses, from the moment ICE agents enter the agency, they are told they will do law enforcement work and risk life and limb.[303] But ICE’s training has been tailored by and for bureaucracy rather than for a force agency, especially after agents go through basic training.[304] As described in Part III, ICE’s ongoing training for employees primarily uses written documents that have been disorganized and difficult to access.[305] Fundamentally, ICE’s approach relies on the notion that communicating policy in writing will be enough to get administrators to apply it.

At the opposite end of the spectrum, military agencies center the “[p]rogressive repetition of tasks” in much of their training, with the explicit goal of making reactions automatic.[306] Military training is structured this way partly to moderate the use of force.[307] As two senior officials in the Air Force Judge Advocate General Corps put it, the goal of military training and discipline is to prepare officers to “overcome self-risk” and to “control the use of lethal force” in high-pressure situations.[308] Accordingly, military agencies have long emphasized rigid, repetitive, and highly structured training that drills routines.[309] For instance, the U.S. Army puts all staff through boot camp. Enlisted staff who complete boot camp are then funneled to specialized training schools that last four weeks to one year.[310] Officers go through pathways that last twelve weeks to four years.[311] And after staff join their units, regular training in operational routines is considered essential to keep the Army ready, consuming significant time and resources.[312]

While the military has a particularly totalizing approach to routines, other agencies similarly train for routine. Some police de-escalation programs incorporate scenario-based training that cuts against instinctive reactions to threats.[313] Limited evidence suggests that these trainings may reduce both use-of-force incidents and injuries to officers.[314]

B.    Discipline

Discipline processes are the other side of training. While training can help establish automatic routines, discipline can de-establish responses and mold agency culture—for instance, by promoting certain norms of behavior.[315] ICE’s discipline systems are so limited that in practice, administrators enjoy a relatively high level of independence, even compared with their civil service peers. Consider that administrative judges, agency adjudicators whose independence implicates due process, “are subject to performance appraisals” with impact on pay, in some cases regularly, and “are not entitled to any particular protection for removal.”[316] Though this system may function differently in practice across agencies, ICE’s discipline system is not more stringent, given how weakly its procedures are enforced.[317]

Again, the administrative law in military agencies sits at a different extreme. As with training, the military uses discipline to encourage its “administrators” to take on risk and to constrain their responses.[318] On the formal side, employees are subject to the Uniform Code of Military Justice (UCMJ), a discipline system and legal code Congress passed.[319] While the military has filled in many aspects of the UCMJ through regulations and guidance,[320] the statute provides the essential substantive and procedural framework for courts-martial.[321] The UCMJ creates an entirely separate system of military justice, one where penalties for crimes are often higher than in the civilian justice system. Heightened penalties exist partly on the recognition that firing—the ultimate penalty for administrators in bureaucratic administrative law—is too weak a sanction to serve the military’s disciplinary purposes.[322] Drawing the parallel between police and military, police reformers have proposed creating a version of the UCMJ for police so that police officers who fail to carry out their legal and job responsibilities will face similarly heightened liability.[323]

Alongside this formal disciplinary system, the military also allows informal discipline meant to “improv[e] performance” and allow senior officials to perform their “command responsibilities.”[324] The Army, for instance, allows senior officials to order “corrective measures” by administrators that include everything from “extra training,” counseling, and “administrative reprimands and admonitions” to “[b]rief physical exercises” and “denial of privileges.”[325] These strategies create a close connection between administrators’ behavior (including instinctive behavior) and discipline, to condition administrators’ responses and promote compliance with rules.

C.     Unionization

A final factor, this one critical to agency culture, is unionization. Unions within law enforcement agencies can be constructive. By bargaining for higher wages and increasing job satisfaction, unions may make jobs in these agencies more desirable, so that they attract better staff.[326] But these unions can also reinforce negative aspects of workforce culture by blocking beneficial policies or loosening rules for accountability and discipline.[327] The example of ICE demonstrates the impact of unions. ICE’s union exerts a high level of control over ICE’s policies,[328] using its power to thwart efforts at humanitarian policymaking,[329] loosen both ex-ante and ex-post constraints on front-line discretion,[330] and weaken the agency’s leadership.[331] It has also helped set ICE’s disciplinary process.[332] In all these cases, ICE’s union has helped accelerate pathologies in the agency.

To some extent, any public-sector union can reinforce bad bureaucratic behavior, but this Article suggests that unions in force agencies are distinctive from other public-sector unions in critical ways. The mechanisms that lead to rule-bending behavior in these agencies suggest there may be a higher need for control here than in other parts of the administrative state.[333] Moreover, the costs of this behavior are in some senses higher than ordinary bureaucratic malfeasance: the state’s misuse of force can harm individuals in uniquely grave ways.[334] To the extent union bargaining weakens constraints on administrators, the resulting costs alone might be enough to justify treating unionization differently here.

There is at least one other reason to think that unionization is uniquely problematic in violent agencies. The foundational role of violence in our administrative state supports stricter public control over employees in the domain of violence. As Weber observed long ago and as is still true, agencies in the domain of violence underwrite the political community.[335] The military and CBP maintain its formal borders; ICE, prison agencies, and law enforcement agencies help to maintain internal order. Since the scope and boundaries of political community arguably ought to be an essential object of democratic control,[336] political officials should have stricter control over agencies that carry out this work. If unionization weakens political control here, there is an additional case for limiting it.

In line with these arguments, policymakers and scholars have long expressed discomfort with unions in force agencies. On voting out a bill to ban unionization among uniformed personnel in the military, the House Committee on Armed Services observed that unions there would undercut “good order and discipline,” and saw the “threat” as sufficiently grave that “legislative, rather than administrative regulation” was warranted.[337] Similarly, labor law scholars have noted that police unions can thwart accountability for misconduct, even as they point out that they are not an unalloyed ill.[338] Accordingly, administrative law tailored for these contexts limits unionization. In the case of the military, Congress banned unionization by uniformed personnel in 1978, criminalizing some organizing efforts.[339]

A less extreme alternative, proposed for police agencies, would limit the range of subjects for union bargaining in force agencies.[340] Allowing only collective bargaining over wages and benefits, as Professor Benjamin Sachs proposed, would limit unions in force agencies to issues where their incentives align with the goal of restricting unauthorized violence.[341] By contrast, bargaining over job responsibilities has allowed ICE’s union to block humanitarian policies.[342]

While the possibility that unions might reinforce negative aspects of workforce culture is intuitive, the evidence on this issue is far from clear. Recent empirical research has pointed in opposite directions,[343] with one large study showing no statistically significant difference in misconduct between unionized and non-unionized police departments.[344]

***

The normative insights this Part derives resonate beyond the field of administrative law. This Part indicates that the Article’s theory of organizational mismatch can help evaluate the increasing number of administrative law proposals for reforming and reimagining police departments, distinguishing proposals that may port to the policing context from those that cannot. To give a concrete example, proposals by scholars such as Professors Barry Friedman, Maria Ponomarenko, and Christopher Slobogin to adopt police rulemaking,[345] or community participation in making administrative rules for use of force, may serve values like transparency. However, they are unlikely to resolve the problem of unauthorized violence against community members, at least without being operationalized through other tools. Police rulemaking is paradigmatic bureaucratic administrative law. It operates by making new policies to implement, without acting on the mechanisms that can thwart implementation in force agencies. On the other hand, legal changes that impose or promote stronger training for police are tailored to this domain.[346] Training, discussed in this Part, can address routines and responses directly, by attempting to alter them.

This Part will close by reiterating the caveat it opens with. In light of the ideal type this Article builds, administrative law that creates routines and changes culture may help to reduce unauthorized violence in this domain. But as with the gap between ideal and real bureaucracies discussed in Part I, there is still a trade-off between the model and reality. Our autonomous capacity to either reform or radically reimagine these agencies is not straightforward. It is mediated in complex ways by a variety of other laws, institutions, and social structures.

Ultimately, accepting the Article’s argument means acknowledging that even administrative law for the domain of violence would be significantly limited in its ability to constrain state violence. And if some level of unauthorized violence is endemic, there is a case to be made that the ideals of liberal legality sit in fundamental and unresolvable tension with the American coercion and control state.

Conclusion

This Article has argued that bureaucratic administrative law does not adequately regulate force agencies, and that this lack of fit between law and ideal type has fostered unauthorized violence in the administrative state. Yet the Article has not just made an argument about how law works in a group of similar agencies. It has also gestured to a broader argument about a large domain of the administrative state. By gesturing to this macro argument, this Article lays the foundation to consider questions that lie outside its scope. This Conclusion points to an especially large one. How should we understand the exceptional norms that seem to characterize agencies in the domain of violence? More specifically, if the domain of violence represents (by one metric) a fifth of the administrative state,[347] then what does the exceptionalism of this substantial part of the state tell us about the operation of law across the whole?

Whether you favor an administrative law that is relaxed or strict in its control of agencies, you likely believe it should serve a similar set of values. A basic list might include “due process, fairness, consistency, transparency, and public accountability.”[348] Some people might add equality or anti-subordination.[349] But applied to the domain of violence, the same bureaucratic administrative law serves different values. Rather than promote fairness and consistency, it supports arbitrariness. Rather than support accountability, it breeds discretion. And rather than promote equality, it reinforces inequalities of race, national origin, and class.

In administrative law, agencies in the domain of violence operate under exceptional norms,[350] with a legal and political order ordinarily designed to safeguard core rule of law values here doing the opposite by masking or even promoting arbitrary violence.[351] The same is true if we look at constitutional doctrine. Agencies in the domain of violence possess powers that are “inherent,”[352] “plenary,”[353] exercised by the federal government despite arguably being meant for the states,[354] or political in nature, with few counter-majoritarian checks.[355] This Article suggests one reason these agencies are exceptional: because of their shared function. They respond to situations of perceived danger or risk, with the authority to use physical violence if necessary, for the purpose of maintaining order and the boundaries of the state. These ideas will not sound unfamiliar to scholars who study immigration, prisons, national security, and police. But drawing the contours of the domain of violence drives home how much of the administrative state is not just exceptional, but exceptional in the same way, and for the same reasons.

Because we have not had the categorical nature of this exceptionalism front of mind, we may not have a sense of how deep and wide claims of exceptionalism in our public law are, nor of how relatively coherent they are. A recent Supreme Court case in an area of law far outside this Article’s subject matter helps illustrate. In Students for Fair Admissions v. Harvard, arguably the most sweeping case of the 2023 Term, the majority indicated that two specific types of institution might be exempt from its rollback of affirmative action: prisons and military academies.[356] Justice Clarence Thomas suggested in his concurrence that “only those measures the state must take to provide a bulwark against anarchy, or to prevent violence” could justify exempting an institution from the court’s new anti-classification norm.[357]

Constitutional equality doctrine, and litigation over affirmative action specifically, is a paradigmatic area of public law, taught routinely in first-year constitutional law class. If any norms in public law have been “ordinary,” the ones governing affirmative action fall into this category. In this context, the repeated references to prisons and the military seem like odd carve-outs for two types of agencies that the Court shows special solicitude.[358]

But when we understand a fifth of the state by personnel as “a bulwark against anarchy” that “prevent[s] violence,” Justice Thomas’s statement takes on a different cast, one not contradicted by other aspects of the Court’s recent equal protection jurisprudence.[359] Even the most “normal” public law is cross-cut, perhaps increasingly so, by exceptions for a group of large agencies that arguably wield more power than the rest of government or most private organizations. Exceptionalism in the domain of violence stretches wide and deep, from these agencies’ day-to-day operational rules to the high-level equality norms at the heart of constitutional law.

By surfacing the breadth and unity of this exceptionalism, which may encompass one-fifth of the state, this Article starts to complicate the common assumption that a meaningful distinction between ordinary and exceptional public law norms can be sustained. In doing so, it suggests just how much the study of administrative law, by surfacing the depth of these contradictions and their roots in agency function, can contribute to our understanding of public law.

Appendix A: Interview Methodology

The case study for this Article is based in part on fourteen semi-structured interviews I carried out with former government officials and members of civil society between August 2022 and January 2023.[360] I relied on my field experience and on primary and secondary source research to prepare two open-ended interview agendas, one for government employees and one for civil society interviewees. Both agendas covered topics related to front-line policy implementation at ICE, but the government agenda included additional topics related to the internal structure of the agency and of DHS. Because of the personalized nature of the interviews and the possible impacts of identification on participants, the study protocol involved keeping the interviews and transcripts confidential.

A.     Interview Recruitment

After receiving approval from Columbia University’s Institutional Review Board (IRB), I initially used convenience sampling, relying on personal contacts from my immigration defense work to solicit names. Convenience sampling was an appropriate choice given the indefinite pool of potential interviewees and the sensitive subject of the study. I then reached out to potential participants with a recruitment email approved by the IRB. After conducting the initial interviews, I used snowball sampling to identify additional participants, soliciting names from interviewees. Some of my interviewees ultimately were at three or four degrees of separation from my initial contacts. In total, I received forty-nine names, secured contact information for and reached out to thirty-one people, and interviewed fourteen.

B.     Sample Characteristics

The interviewees for this project included six former government officials and eight members of civil society. Of the former government officials, five had worked in roles at ICE or DHS involving direct exposure to field offices or other relevant information. One had worked in a congressional role with extensive exposure to the agency. One of the officials held their role during the Bush Administration, four during the Obama Administration, two during the Trump Administration, and three during the Biden Administration, with some interviewees serving across multiple Administrations.[361]

Of the eight participants from civil society, five were in primarily litigation roles, two were in primarily policy roles, and one was in a primarily monitoring role. I recruited interviewees with extensive exposure to field-level conditions or evidence of field-level conditions through their work, well beyond what a single direct-representation attorney would have; and who either had insight into conditions nationwide or represented geographically diverse perspectives. Four of the interviewees were engaged in nationwide work; three had experience with conditions in the West, one with conditions in the South, one with the Southwest, and one with the Northeast. Most had worked on immigration issues since at least the Obama Administration.

C.    Interview Procedure and Analysis

Interviews mostly lasted sixty minutes, and all were conducted on Zoom. To best capture interviewees’ impressions and experiences, I used the interview agendas to ask open-ended questions.[362] To establish a connection with interviewees and assess how best to target the interview, I began with broad prompts about their roles or former roles and how they came to have knowledge of the environment inside ICE. After getting a sense of which topics each interviewee might be well placed to address, I began asking personalized, open-ended questions about how they had observed policies communicated or enforced at the field-office level. I generally kept my questions open-ended in order to elicit interviewees’ observations and views of the factors shaping policy implementation, rather than asking questions that presumed a causal link. In many cases, these questions organically led interviewees to comment on ICE’s field-level occupational culture, a line of inquiry that greatly shaped the overall conclusions of the case study.

I gave interviewees a choice to be digitally recorded or, at their preference, to have me take written notes only. All interviewees consented to digital recording, though a few asked to go “off the record,” or to switch to written notes, at certain points.

I took notes immediately after each interview, so that I could retain my initial impressions of the interviewees’ demeanors and responses. I used an iterative approach to analyze the interviews. After transcribing an initial set of interviews, I began coding the transcripts for high-level themes, and coded additional interviews as I conducted them. Coding as I worked allowed me to begin developing a sense of the factors influencing front-line policy implementation, which made it easier for me to explore these factors with participants in later interviews. Once I finished coding, I compared the transcripts to see which themes tracked across the interviews. The themes I derived informed the findings in the case study, helping to contextualize other primary or secondary sources; in other cases, the interviews provided important factual detail and color that corroborated existing sources.[363] While I did not attempt to establish a theory of ICE’s front-line institutional culture purely through the interviews, coding the transcripts was valuable. Coding helped me observe patterns in the interviews that I might not otherwise have seen and would not have been able to predict based on my field experience.[364] Because the coding process made it easy to compare interviews with former government and civil society participants, it also increased my confidence in the study’s core insights into institutional context, helping to triangulate the different positions, experiences, and views of these two groups.

Copyright © 2024 Emily R. Chertoff, Associate Professor of Law, Georgetown University Law Center. In 2019 and 2020, I was a litigator and removal defense attorney for immigrants and asylum seekers detained in Southern California. For helpful comments and conversations at different stages of the drafting process, I would like to thank Ashraf Ahmed, Kate Andrias, José Argueta Funes, Seyla Benhabib, Adam Bradlow, Jessica Bulman-Pozen, Jennifer Chacón, David Chen, Adam Cox, William Eskridge, Maeve Glass, Bernard Harcourt, Oona Hathaway, Michael Heller, James Hicks, John Inazu, Vicki Jackson, Olatunde Johnson, Emma Kaufman, Jeremy Kessler, Madhav Khosla, Monika Leszczynska, Benjamin Levin, Lev Menand, Gillian Metzger, Nicholas Parrillo, Eloise Pasachoff, Robert Post, David Pozen, K. Sabeel Rahman, Shalini Bhargava Ray, Kate Redburn, Daniel Richman, Noah Rosenblum, Charles Sabel, Miriam Seifter, Karen Tani, Shoba Sivaprasad Wadhia, Wendy Wagner, Christopher Walker, Matthew Waxman, and participants in the 2023 AALS New Voices in Administrative Law Program, the 2021 Administrative Law New Scholarship Roundtable, and the Columbia Law School Academic Fellows Workshop. Many thanks also to Jung Kim and the editors of the California Law Review for their outstanding work editing and preparing this Article for publication.

           [1].     See 88 FR 31314, Circumvention of Lawful Pathways (May 16, 2023); see also 8 C.F.R. 208, Procedures for Asylum and Withholding of Removal.

           [2].     Kate Morrissey, U.S. Border Officials Have Been Turning Asylum Seekers Away at Ports of Entry Despite New Rules, San Diego Union-Tribune (May 20, 2023), https://www.sandiegouniontribune.com/news/immigration/story/2023-05-20/border-asylum-seekers-turn-backs-ports [https://perma.cc/YV26-GQVB].

           [3].     See Haitian Bridge Alliance, American Civil Liberties Union, Amnesty International, Human Rights First, National Immigration Law Center, National Immigrant Justice Center, Tahirih Justice Center, UndocuBlack Network & #WelcomeWithDignity, Lives at Risk: Barriers and Harms as Biden Asylum Ban Takes Effect 4 (May 2023), https://humanrightsfirst.org/wp-content/uploads/2023/05/Barriers-and-Harms-As-Biden-Asylum-Ban-Takes-Effect31.pdf [https://perma.cc/36XB-9L6U] (observing turnbacks after the end of Title 42 at ports of entry in Texas).

           [4].     Geneva Sands, Leaked Audio and Video Show Border Patrol Agents Confronting Homeland Security Secretary at Meetings, CNN (Feb. 1, 2022), https://www.cnn.com/2022/02/01/politics/border-patrol-agents-mayorkas/index.html [https://perma.cc/4PJ3-7RUJ]. Leaked audio from listening sessions in 2022 similarly depicts an agent telling Border Patrol Chief Raul Ortiz that “[f]or evil to triumph is for good men to do nothing. That’s exactly what’s happening here. Good men are doing nothing. You’re allowing illegal aliens to be dropped off in communities.” Ali Bradley, New: Tense Exchange Between U.S. Border Patrol Agents and USBP Chief Ortiz Today in Laredo, Facebook, at 3:01 (Jan. 28, 2022), https://fb.watch/qHq2QeV9lV/ [https://perma.cc/V67F-3WDU].

           [5].     See Complaint, Al Otro Lado v. Nielsen, No. 3:17-cv-02366-BAS-KSC, 2018 WL 10613416, No. 189 at 27–28 (S.D. Cal. Nov. 13, 2018).

           [6].     See id. at 17, 44–45. Advocates ultimately won several of their legal claims on summary judgment.

           [7].     Joseph Neff, Alysia Santo & Tom Meagher, How a ‘Blue Wall’ Inside N.Y. State Prisons Protects Abusive Guards, N.Y. Times (May 22, 2023), https://www.nytimes.com/2023/05/22/nyregion/ny-state-prison-guards-abuse.html [https://perma.cc/Q52V-6FZW]. For an overview of literature on violence within prisons, see Sharon Dolovich, The Failed Regulation and Oversight of American Prisons, 5 Ann. Rev. Criminology 153, 157–58 (2022).

           [8].     Dave Phillips, Leaders Let Problems Mount at Brutal SEAL Course, Navy Finds, N.Y. Times (May 25, 2023), https://www.nytimes.com/2023/05/25/us/navy-report-seal-course-kyle-mullen.html [https://perma.cc/LM86-ES34]; Andrew Dyer, Navy SEAL Leaders Reprimanded Over Death of Trainee During “Hell Week” in February, L.A. Times (Oct. 14, 2022), https://www.latimes.com/california/story/2022-10-14/navy-seal-leaders-reprimanded-over-trainee-death-hell-week [https://perma.cc/W42Y-K9K7].

           [9].     See, e.g., Hamed Aleaziz, There’s Been a Major Increase in the Use of Force Against Immigrants at ICE Detention Centers During the Pandemic, Buzzfeed News (Aug. 5, 2020), https://www.buzzfeednews.com/article/hamedaleaziz/use-of-force-immigrants-ice-pandemic [https://perma.cc/VJT6-8JM2] (describing use of force incidents involving pepper spray and pepper balls in immigration detention during the pandemic).

         [10].     Robert M. Cover, Violence and the Word, 95 Yale L.J. 1601, 1601 (1986). In Violence and the Word, Professor Cover advances a famous observation that violence underwrites public law in the execution of a court’s judgments. Id. This Article, by contrast, makes the related but distinct observation that the state entities carrying out violence are not constrained by such judgments (or by democratically authorized laws) in meaningful ways. This argument is about legality and the legitimacy of public law as much as it is about the inherent violence of law. I return to the deeper issue of the inherent violence of administrative law in work in progress. Professor Cover’s work is an important point of reference for this Article for a second reason: his insistence on centering the relationships between culture, law in action, and the individuals implementing the law prefigures this Article and the work of many other scholars.

         [11].     At the federal level, as of 2020, the agencies in this group collectively employed approximately one hundred and thirty thousand sworn law enforcement officers, with a few thousand employed by a range of agencies primarily engaged in non-coercive, bureaucratic work. See Connor Brooks, U.S. Dep’t of Just., Off. of Just. Programs, Bureau of Just. Stats., NCJ 304752, Federal Law Enforcement Officers, 2020 – Statistical Tables 4 (Sept. 29, 2023), https://bjs.ojp.gov/document/fleo20st.pdf [https://perma.cc/A5NR-GMEL]; for a more exhaustive disaggregation of these agencies and components, see Emily R. Chertoff & Jessica Bulman-Pozen, The Two Faces of the Administrative State (manuscript on file with author). Nearly half are employed by DHS, particularly ICE and CBP; and nearly one-third by the Department of Justice (DOJ), particularly the BOP and the FBI. See id. at 1, 4. The military’s combatant commands—i.e., those that have operational responsibility for fighting wars—collectively command close to six hundred thousand personnel. Mark Nevitt, The Operational and Administrative Militaries, 53 Ga. L. Rev. 905, 953 (2019). And the U.S. Coast Guard has about forty thousand active-duty personnel. See 14 U.S.C. § 4904; U.S. Dep’t of Def., 2022 Demographics Profile of the Military Community, at iii (2022), https://download.militaryonesource.mil/12038/MOS/Reports/2022-demographics-report.pdf [https://perma.cc/FS5R-XDNG] (reporting 39,485 active-duty Coast Guard members in 2022). This is as compared with 4.2 million federal employees in total, meaning that agency components that stress occupational roles involving physical force may employ close to a fifth of all federal personnel. See Cong. Rsch. Serv., R43590, Federal Workforce Statistics Sources: OPM and OMB 6 (June 28, 2022), https://sgp.fas.org/crs/misc/R43590.pdf [https://perma.cc/CL78-4YTZ].

         [12].     In Part II, the Article develops an ideal type of the domain of violence that is functional. A component of any agency could fall into the domain if it serves the function and displays the characteristics associated with it. That said, the Article is structured around a core set of agencies discussed above, and the organization-level effects of this function are at the heart of the Article’s argument.

         [13].     See Todd D. Minton, Lauren G. Beatty & Zhen Zeng, U.S. Dep’t of Just., Off. of Just. Programs, Bureau of Just. Stats., NCJ 300655, Correctional Populations in the United States, 2019 – Statistical Tables, 11 (July 2021), https://bjs.ojp.gov/sites/g/files/xyckuh236/files/media/document/cpus19st.pdf [https://perma.cc/4JHM-DLJ2].

         [14].     See U.S. Immigr. & Customs Enf’t, U.S. Immigration and Customs Enforcement Fiscal Year 2019 Enforcement and Removal Operations Report 5 (2019), https://www.ice.gov/sites/default/files/documents/Document/2019/eroReportFY2019.pdf [https://perma.cc/8ZUJ-QVRQ].

         [15].     In 2019, the United States was engaged in overseas conflicts in Afghanistan, Iraq, Yemen, Somalia, Syria, and Libya. See Scott R. Anderson & Benjamin Wittes, Trump Administration Releases Overdue War Powers Report in Response to Lawsuit, Lawfare (Oct. 20, 2020), https://www.lawfareblog.com/trump-administration-releases-overdue-war-powers-report-response-lawsuit [https://perma.cc/H96F-2LFL].

         [16].     See Mark Motivans, U.S. Dep’t of Just., Off. of Just. Programs, Bureau of Just. Stats., NCJ 301158, Federal Justice Statistics, 2019, at 3 (Oct. 2021), https://bjs.ojp.gov/content/pub/pdf/fjs19.pdf [https://perma.cc/MP7R-T9LF].

         [17].     I follow Gillian Metzger in broadly defining “administrative law” to include “not simply measures that are typically identified as part of federal administrative law, such as the Administrative Procedure Act (APA) and judicial review doctrines,” but also “internal executive-branch practices and structures that overlap with public administration, as well as administrative law scholarship.” Gillian E. Metzger, Through the Looking Glass to a Shared Reflection: The Evolving Relationship between Administrative Law and Financial Regulation, 78 J.L. & Contemp. Probs. 129, 129 n.4 (2015). Yet ultimately, my view of “administrative law” sweeps wider, to include a variety of personnel- and culture-focused measures rarely studied by administrative law scholars even as they profoundly shape the law in action. See infra Part IV.

         [18].     See infra Parts I, III.

         [19].     By “violence,” I here mean the excessive use of physical force to the point of causing injury. This physical force could involve direct contact, like an aggressive shove, or “indirect” forms of contact that are still physically coercive, like incarceration. The use of force becomes violent when it leads to either a physical or a psychological injury. Under this definition, an unauthorized chokehold or excessive time in solitary confinement would be examples of violence. An agency’s failure to treat an injury sustained in custody, thereby aggravating it, might also be a form of violence. My choice to focus on physical violence is not meant to deny that other forms of state action can be considered “violent” in a structural sense but to offer due attention to important and often hidden infringements of bodily liberty and autonomy that are the result of law execution in the administrative state.

         [20].     This Article uses the word “bureaucracy” primarily to describe two things: an organizational form that many state agencies take and a type of law—“bureaucratic administrative law”—that codeveloped with this organizational form in order to regulate it. See Max Weber, Economy and Society: A New Translation 350, 352 (Keith Tribe ed., 2019) [hereinafter Weber, Economy and Society] (articulating a vision of bureaucracy as organizational form defined by rationality and the domination of knowledge). That being said, when the Article alludes to “domains” of bureaucracy—and of violence—it broaches a more abstract question: the possible connection between these mid-range organizational forms and the macro dynamics of our system of government. By operating differently across these two different organizational forms, does administrative law facilitate the state’s use of physical force? In what ways, and how? These broad questions are a subject for future work.

         [21].     See infra Part I.B.

         [22].     At their higher echelons these organizations may be bureaucratic, and social scientists have suggested that conflict arises between management and the workforce within these agencies because they hew to different norms at different levels. See infra note 127–128.

         [23].     Distinct from “violence” discussed supra note 19, “force” refers to any direct or indirect application of physical contact (e.g., detention or pushing), whether or not it is excessive. I call these agencies “force agencies” since not every application of coercive force they undertake is violent, though they are defined by their authority to use both force and violence when necessary. In writing about the domain of the state that has this authority, German sociologist Max Weber—whose state theory this Article draws on—uses a word that can be translated as either “force” or “violence.” This Article accordingly refers to the “domain of violence” when it discusses this domain of the state. See Dieter Grimm, The State Monopoly of Force, in International Handbook of Violence Research 1043–56 (Wilhelm Heitmeyer & John Hagen eds., 2003) (discussing the plurality of uses of the German term as well as Weber’s use and theory of the monopoly of violence).

         [24].     See Weber, Economy and Society, supra note 20, at 136 (suggesting that violence is a core capacity of a political community and that it uses violence or force to underwrite “order” in the community and “secure” “rule over a definite territory”); Max Weber, Politics as a Vocation, in From Max Weber: Essays in Sociology, at 78–79 (H.H. Gerth & C. Wright Mills eds., 1946) [hereinafter Weber, Politics] (describing “legality,” or “the belief in the validity of legal statute and functional ‘competence’ based on rationally created rules,” as one of three means to legitimate domination generally and violence specifically).

         [25].     “Violence” appears in Weber’s state theory as a core function of the state. See supra note 24; see also Weber, Politics, supra note 24, at 78 (“Today the relation between the state and violence is an especially intimate one. . . . [A] state is a human community that (successfully) claims the monopoly of the legitimate use of physical force within a given territory.”).

         [26].     Because this Article focuses on the ways that perceived risk or danger and the capacity to do violence shape these administrators and their immediate institutional environment, its argument does not implicate agencies that can call on these powers without exercising them directly. See infra Part II. Nor does it necessarily implicate the significant number of agencies that have a small cadre of armed officials, representing only a fraction of total staff—though as a functional rather than a formal category, it could apply to these components. See Brooks, supra note 11, at 4. Rather, the three Departments at the heart of my argument—DOJ, DHS, and the Department of Defense—each proportionately employ at least twenty percent of their staff who are authorized to carry arms and use physical force. These staff are often localized in particular components where they make up a balance of all employees. See Chertoff & Bulman-Pozen, supra note 11. Though this Article does not specify a precise cutoff, from the perspective of its argument, it makes sense to distinguish agencies where violence is prominent in shaping the organization from those where violence is a de minimis part of the work.

         [27].     This is true whether we apply the label “administrative law” narrowly, i.e., only to agency action that involves the APA, or more broadly. As Parts II and III discuss, many of the activities of agencies in this domain are covered by administrative law narrowly construed: they fall within the judicial review provisions of the APA and are reviewed by courts under the statute. See infra notes 119–124, Part III.A; 5 U.S.C. §§ 701–706. At the same time, these agencies’ internal guidance documents, hierarchies, and procedures correspond to what Professors Gillian Metzger and Kevin Stack have influentially called “internal administrative law.” See Gillian E. Metzger & Kevin M. Stack, Internal Administrative Law, 115 Mich. L. Rev. 1239, 1239 (2017). And ultimately, if we construe “administrative law” broadly the way Professors Metzger and Stack, and others have done, as a lens that describes a set of homologous legal and organizational structures within agencies, these agencies parallel conventional administrative agencies in many features of their design. Of course, this is not to say that there is no agency or domain-specific administrative law in force agencies. Jessica Bulman-Pozen and I survey this domain-specific law in work in progress. See Chertoff & Bulman-Pozen, supra note 11.

         [28].     This Article just begins to note the deeper interdependence of physically coercive and ordinary agencies. Though there are meaningful distinctions between the two domains—some of which this Article details—they are also inextricably linked. I intend to return to this aspect of the argument in future work. There are many important antecedents of this project in social and sociological theory. See, e.g., Charles Tilly, War Making and State Making as Organized Crime, in Bringing the State Back In 169­–86 (Peter Evans, Dietrich Rueschemeyer, & Theda Skocpol eds., 1985) (arguing that administrative and violent functions of the state supported each other during the process of state formation and state-building in Europe).

         [29].     Cover, supra note 10, at 1615 (“Persons who act within social organizations that exercise authority act violently without experiencing the normal inhibitions or the normal degree of inhibition which regulates the behavior of those who act autonomously.”).

         [30].     My argument here suggests that force agencies experience “normative isomorphism”—that their organizational structures tend to converge because of pressures coming from the agency’s workforce. See Paul J. DiMaggio & Walter W. Powell, The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields, 48 Am. Soc. Rev. 147, 150 (1983). DiMaggio and Powell, in their classic article, identified selection of personnel, the circulation of personnel among organizations, and professionalization as mechanisms of normative isomorphism. See id. at 152–53. This Article suggests that the content of administrators’ roles may also be a mechanism of normative isomorphism.

         [31].     I discuss in Part IV how the social theory of law this Article draws on problematizes the common practice of making normative law reform recommendations, and the reason the Article concludes with such recommendations nonetheless.

         [32].     Attention to these mechanisms is relatively rare in administrative law scholarship, but see Jerry L. Mashaw, Bureaucratic Justice 155–66 (1983); Daniel E. Ho & Sam Sherman, Managing Street-Level Arbitrariness: The Evidence Base for Public Sector Quality Improvement, 13 Ann. Rev. Soc. Sci. 251, 253–54 (2017) (developing a typology of internal management tools to improve decisional consistency in agencies, including training, but noting that across the social sciences, “the literature turns out to be remarkably thin”).

         [33].     Part III will highlight implications for the study of immigration law, and Part IV will note the relevance of this Article to the law of the police.

         [34].     Among other things listed above, it illustrates how scholarship could use organizational theory to understand and model how agencies use administrative law. Interdisciplinarity has long been a feature of administrative law scholarship partly because work from other disciplines helps legal scholars conceptualize how law will actually be applied. See Cary Coglianese, Empirical Analysis and Administrative Law, 2002 U. Ill. L. Rev. 1111, 1113 (2002) (“Administrative law is constructed and reconstructed on the basis of assumptions about how particular procedural arrangements will affect the behavior and performance of government officials and organizations. As a result, the insights and methods of other disciplines, such as political science, economics, and organizational behavior have contributed greatly to the development of administrative law scholarship.”). Moreover, the Article offers another reason to be skeptical of fully trans-substantive accounts of administrative law, suggesting that a domain-specific approach to the administrative state best accounts for parallels between some agencies and discontinuities with others. By arguing for a domain-specific approach, this Article joins in a critique of trans-substantivity explicit or implicit in recent work by Professors Brian Feinstein, Lev Menand, Joy Milligan, and Karen Tani, among others. See Brian D. Feinstein, Legitimizing Agencies, 91 U. Chi. L. Rev. 920, 996–97 (2023) (suggesting based on empirical research that an agency-specific, substance-specific approach to agency design would support agencies’ sociological legitimacy); Lev Menand, Why Supervise Banks? The Foundations of the American Monetary Settlement, 74 Vand. L. Rev. 951, 963 n.49 (2021); Joy Milligan & Karen Tani, Seeing Race in Administrative Law: An Interdisciplinary Perspective, J. Reg. Notice & Comment (Sept. 16, 2020), https://www.yalejreg.com/nc/seeing-race-in-administrative-law-an-interdisciplinary-perspective-by-joy-milligan-and-karen-tani/.

         [35].     See also Elizabeth Fisher & Sidney Shapiro, Disagreement About Chevron: Is Administrative Law the “Law of Public Administration”?, 70 Duke L.J. Online 111, 112 (2021) (suggesting that administrative law be seen as a body of law that facilitates administrative competence, rather than one that more narrowly constrains agencies).

         [36].     See Chertoff & Bulman-Pozen, supra note 11, at 5–7.

         [37].     Regulation-skeptic justices have increasingly voiced concerns about anti-democratic aspects of (parts of) the administrative state as they transform the doctrine that applies to (some) agencies. See, e.g., West Virginia v. EPA, 142 S. Ct. 2587, 2617 (2022) (Gorsuch, J., concurring) (connecting major questions doctrine to democratic accountability concerns); see also Gillian E. Metzger, 1930s Redux: The Administrative State Under Siege, 131 Harv. L. Rev. 1, 33–42 (2017) (discussing this trend). By exposing the deep democratic concerns raised by a large group of agencies left out of “the administrative state” by courts as well as scholars, the Article brings into stark relief an inconsistency in the norms applied to different parts of the administrative state. See infra Conclusion.

         [38].     For a few recent examples of arguments by scholars addressing theoretical questions of democratic legitimacy and accountability, see, e.g., Blake Emerson, The Public’s Law: Origins and Architecture of Progressive Democracy (2019) (describing how American thinkers, and later Progressive-era state builders, drew on the philosophy of G.W.F. Hegel and other German state theorists to incorporate democratic values into administration); Daniel E. Walters, The Administrative Agon: A Democratic Theory for a Conflictual Regulatory State, 132 Yale. L.J. 1, 5–10 (2022); K. Sabeel Rahman, Reconstructing the Administrative State in an Era of Economic and Democratic Crisis, 131 Harv. L. Rev. 1671, 1675–76 (2018).

         [39].     It is notable and I suspect not accidental that the law of the police is one prominent area of public law where this bottom-up approach to accountability and legitimacy is more of the norm. For examples of important recent works adopting such an approach, see, e.g., Jocelyn Simonson, Police Reform Through a Power Lens, 130 Yale L.J. 778, 788–89 (2021) (bringing perspectives of social movements—a view “from the bottom”—to analyze and understand police accountability); Monica C. Bell, Police Reform and the Dismantling of Legal Estrangement, 126 Yale L.J. 2054, 2061, 2126 (2017). Many parts of what we call the “administrative state” developed originally out of local police to exercise the police power, later differentiating into coercive and non-coercive arms of the administrative state. See Mark Neocleous, A Critical Theory of Police Power: The Fabrication of Social Order 19 (2d ed. 2021). Perhaps the continuing relevance of the view from the bottom up in policing has to do with policing’s local nature.

         [40].     Cf. Kate Andrias, The President’s Enforcement Power, 88 N.Y.U. L. Rev. 1031, 1045 (“[E]nforcement vel non matters: The law on the books is different from the law in action, and enforcement is a vital part of law’s identity as law.”).

         [41].     “Execution” here refers to a specific aspect of the broader enforcement activities agencies undertake: the point where those activities bottom out in physical enforcement, often carried out by the agencies in the domain of violence. While enforcement in general receives less attention than adjudication, and certainly less than rulemaking, there is a literature on the enforcement-related powers of the President and agency leadership. See generally Gillian E. Metzger, The Constitutional Duty to Supervise, 124 Yale L.J. 1836, 1844 (2015) [hereinafter Metzger, Constitutional Duty] (discussing President’s “duty to supervise” or superintend agency implementation of statutes, including through enforcement); Zachary S. Price, Enforcement Discretion and Executive Duty, 67 Vand. L. Rev. 671, 673–75 (2014); Andrias, supra note 40, at 1034–36. But the execution of laws is not traditionally a focus of administrative law scholarship even though it is at the heart of the study of policing. See generally Maria Ponomarenko, Rethinking Police Rulemaking, 114 Nw. L. Rev. 1, 7 (2019); John Rappaport, How Private Insurers Regulate Public Police, 130 Harv. L. Rev. 1539, 1546 (2017); Andrew Manuel Crespo, Systemic Facts: Towards Institutional Awareness in Criminal Courts, 129 Harv. L. Rev. 2049, 2059 (2016).

         [42].     Cf. Vesla M. Weaver & Gwen Prowse, Racial Authoritarianism in U.S. Democracy, 369 Sci. 1176 (2020) (raising similar questions vis-à-vis the understanding of “authoritarianism” as a concept in political science); Joe Soss & Vesla M. Weaver, Police Are Our Government: Politics, Political Science, and the Policing of Race-Class Subjugated Communities, 20 Ann. Rev. Poli. Sci. 565 (2017) (arguing that democracy and other concepts in political theory must be assessed from the perspectives of subordinated groups, not just that of formal institutions).

         [43].     James Q. Wilson, Bureaucracy: What Government Agencies Do and Why They Do It 11–12 (1989) [hereinafter Wilson, Bureaucracy] (noting drawbacks of a top-down approach to understanding organizations).

         [44].     My concern here is not with how we define “rule of law,” “accountability,” or “legitimacy,” or precisely what any of these things require from the perspective of normative theory. Indeed, my assumption would be that as a descriptive matter, governing elites, and perhaps also the general public, already have some coherent concept of how to define these terms, or at least of what qualities government should have to satisfy them. My methodological question is whether we measure adherence to these concepts however defined by looking at top-down, formal institutions and metrics or at bottom up, on-the-ground practices and experience.

         [45].     See generally Rosalind Dixon & David Landau, Abusive Constitutional Borrowing: Legal Globalization and the Subversion of Liberal Democracy 11–20 (2021) (positing that liberal constitutional provisions have been repurposed to advance political authoritarianism); Tom Ginsburg & Aziz Z. Huq, How to Save a Constitutional Democracy 2 (2018).

         [46].     While clearly in dialogue with a recent literature on bureaucratic resistance to presidential directives, this Article points to a different dynamic, one that is endemic to certain agencies rather than situational. See generally, e.g., Bijal Shah, Civil Servant Alarm, 94 Chi.-Kent L. Rev. 627 (2020) (describing how civil servants resisted norm-breaking behavior and maintained fidelity to core agency missions during the Trump Administration); Jennifer Nou, Civil Service Disobedience, 94 Chi.-Kent L. Rev. 349 (2019) [hereinafter Nou, Civil Service Disobedience]; Jennifer Nou, Bureaucratic Resistance from Below, 36 Yale J. on Reg. Notice & Comment (Nov. 16, 2016), https://www.yalejreg.com/nc/bureaucratic-resistance-from-below-by-jennifer-nou/ [https://perma.cc/MB6X-YHCC].

         [47].     The federal government’s response to the George Floyd protests in 2020 embodies this dynamic. It had a weak statutory justification but enthusiasm from both the White House and some rank-and-file federal agents, with alarming results: “[The protestors] had barely made it half a block when an unmarked minivan pulled up in front of them. ‘I see guys in camo,’ O’Shea said. ‘Four or five of them pop out, open the door and it was just like, ‘Oh s***. I don’t know who you are or what you want with us.’ . . . ‘I am basically tossed into the van,’ Pettibone said. ‘And I had my beanie pulled over my face so I couldn’t see, and they held my hands over my head.’” See Jonathan Levinson, Conrad Wilson, James Doubek & Suzanne Nuyen, Federal Officers Use Unmarked Vehicles to Grab People in Portland, DHS Confirms, NPR (July 17, 2020), https://www.npr.org/2020/07/17/892277592/federal-officers-use-unmarked-vehicles-to-grab-protesters-in-portland [https://perma.cc/RRB2-VL3A]. For the legal ambiguity around the deployment of federal agents to suppress the George Floyd protests, see Jonathan Blitzer, Is It Time to Defund the Department of Homeland Security?, New Yorker (July 24, 2020), https://www.newyorker.com/news/daily-comment/is-it-time-to-defund-the-department-of-homeland-security [https://perma.cc/TJ3U-E529].

         [48].     See Jonathan Swan & Maggie Haberman, Heritage Foundation Makes Plans to Staff Next G.O.P. Administration, N.Y. Times (Apr. 20, 2023), https://www.nytimes.com/2023/04/20/us/politics/republican-president-2024-heritage-foundation.html [https://perma.cc/4XY5-QBTB]. Successful applicants to this personnel program join a “conservative LinkedIn,” and can participate in trainings and earn certificates in management and procurement. Ultimately, Heritage hopes that five times more of these political appointees than normal will join the next Republican Administration, replacing many civil servants. See id. This approach stands to transform how administrative law is applied, yet the tools Heritage is using—i.e. cultural change mechanisms including extensive training—are mostly absent from the administrative law toolkit scholars tend to rely on. This Article addresses these tools. See Presidential Administration Academy, Project 2025: Presidential Transition Project, https://www.project2025.org/training/presidential-administration-academy/ [https://perma.cc/48EN-RH76].

         [49].     As Professor Monica Bell has noted, legal research on policing has often looked to a thin concept of legitimacy, assessing it in terms of the public’s acceptance of and cooperation with the police. See Bell, supra note 39, at 2126. Administrative law scholars, on the other hand, have often looked to thicker conceptions of legitimacy that ask whether administrative law serves substantive democratic values and promotes robust public participation in decision-making. Applying these “thick” conceptions of legitimacy to the domain of violence gives us different frameworks within which to answer the question of how and whether these agencies serve democracy.

         [50].     See Michael McCann & Filiz Kahraman, On the Interdependence of Liberal and Illiberal/Authoritarian Legal Forms in Racial Capitalist Regimes . . . The Case of the United States, 17 Ann. Rev. L. & Soc. Sci. 483 (2021); Soss & Weaver, supra note 42 (arguing that the perspectives of race-class subordinated communities on democratic quality must be a focus of political science literature).

         [51].     Theorists of both constitutional and administrative law have deployed the concept of “domains,” which we could essentially characterize as social functions, to explain why law does or should operate differently in some areas of social life than in others. See generally Robert Post, Constitutional Domains: Democracy, Community, Management 2–18 (1995) (offering a theory of the First Amendment built on three domains of communication that serve three different social functions); Mashaw, supra note 32, at 23–34 (1983) (building a theory of bureaucratic self-regulation by internal administrative law on a functional account of three ideal types of bureaucracy). While administrative law scholarship in recent years has been thoroughly trans-substantive, a domain-specific view has increasingly colored our understanding of constitutional law. See Emma Kaufman, The New Legal Liberalism, 86 U. Chi. L. Rev. 187, 205–14 (2019).

         [52].     For recent examples developing ideal types for organizations, see, e.g., Pernille S. Strobaeke & Marek Korczynski, Client Abuse to Public Welfare Workers, 52 Sociology 762, 764 (2018); Yuen Yuen Ang, Beyond Weber: Conceptualizing an Alternative Ideal-Type of Bureaucracy in Developing Contexts, 11 Reg. & Gov. 282, 283 (2017); see also Danielle S. Allen, Justice By Means of Democracy 13–15 (2023) (work of political theory ideal typing models of justice and subsidiary ideals across the domains of politics, economy, and society).

         [53].     Other attitudes, norms, and organizational constraints—like racial and xenophobic bias, social-dominance orientation, or resource scarcity—also influence the implementation of the law. This Article does not deny that these forces exist or condition some rule-breaking by administrators in force agencies. Rather, it seeks to identify a set of organizational dynamics that arise within these agencies, that are unique to them, and that contribute to rule-breaking (in some cases, possibly by amplifying or helping spread attitudes and norms that some administrators bring with them).

         [54].     ICE is a representative agency within the domain of violence, making it an appropriate subject for the case study. See infra notes 176–179 and accompanying text.

         [55].     Qualitative research on immigration tends to focus on adjudication rather than enforcement, and though scholars have discussed the problem of front-line non-compliance within ICE, the field has few granular accounts of internal dynamics within the agency, outside of work by Professors Adam Cox and Cristina Rodríguez. Adam B. Cox & Cristina M. Rodríguez, The President and Immigration Law, ch. 6 (2020) The case study surfaces specific forms of non-compliance or manipulation of internal administrative law that this earlier work has not discussed and offers detail on ICE’s workforce culture that helps explain dynamics within the agency. It suggests an alternative explanation for problems within ICE that is grounded in workforce culture and organizational dynamics, rather than in individual bias or the exceptionalism often attributed to immigration law. Cf. Jennifer Lee Koh, Executive Defiance and the Deportation State, 130 Yale L.J. 948, 953, 973–82 (2021) (framing wrongful deportations by immigration enforcement officials as a question of the “deportation bureaucracy’s competence and respect for the rule of law,” and connecting this “defiance” to the “structural dynamics at play in immigration law—a field in which the normal rules of constitutional law do not consistently apply”); Peter L. Markowitz, Abolish ICE . . . And Then What?, 129 Yale L.J. F. 130, 131 (2019) (citing ICE’s “brutality” as it enforces immigration law).

         [56].     See, e.g., Christopher Slobogin, Policing as Administration, 165 U. Pa. L. Rev. 91, 134–51 (2016) (suggesting that the APA should apply to police agencies, and specifically detailing how the APA rulemaking process might apply in practice); Barry Friedman & Maria Ponomarenko, Democratic Policing, 90 N.Y.U. L. Rev. 1827, 1832 (2015).

         [57].     See generally Jaya Ramji-Nogales, Andrew I. Schoenholtz, & Philip G. Schrag, Refugee Roulette: Disparities in Asylum Adjudication and Proposals for Reform (2009) (explaining a classic study of disparities in adjudication at the Asylum Office—a bureaucratic component of the U.S. Citizenship and Immigration Services that makes similar asylum determinations using bureaucratic administrative law).

         [58].     U.S. Immigr. & Customs Enf’t, Directive 11002.1, Parole of Arriving Aliens Found to Have a Credible Fear of Persecution or Torture (Dec. 8, 2009), https://www.ice.gov/doclib/dro/pdf/11002.1-hd-parole_of_arriving_aliens_found_credible_fear.pdf [https://perma.cc/4TLX-CKNA].

         [59].     See infra note 258 and accompanying text (describing high rate of grants under parole policy from its implementation in 2009 until 2014); see also Damus v. Nielsen, Case No. 18-578 (JEB), 2018 WL 3232515, No. 2, Class Complaint for Injunctive and Declaratory Relief, at 12–14 (D.D.C. Mar. 15, 2018) (noting that the “overwhelming majority” of asylum seekers meet the criteria for parole under ICE’s parole regulation).

         [60].     See Damus v. Nielsen, 2018 WL 3232515, No. 2, Class Complaint for Injunctive and Declaratory Relief, at 14.

         [61].     Id. at 15–16.

         [62].     See generally M. von Werthern, K. Robjant, Z. Chui, R. Schon, L. Ottisova, C. Mason & C. Katona, The Impact of Immigration Detention on Mental Health: A Systematic Review, 18 BMC Psychiatry 382 (2018) (reviewing studies showing immigration detention’s negative effects on mental health, including post-traumatic stress disorder).

         [63].     This was true from the founding through (and beyond) the passage of the APA. See Jerry L. Mashaw, Creating the Administrative Constitution: The Lost One Hundred Years of American Administrative Law 13, 32–35 (2012) (describing early American administrative law, including substance-specific law made for military agencies); see also David Marcus, Trans-Substantivity and the Processes of American Law, 2013 BYU L. Rev. 1191, 1210 (2014) (describing factors in the rise of trans-substantivity in administrative law, including a shift to trans-substantive legal process theories in the academy and greater acceptance of the APA’s model among New Dealers who had once opposed it).

         [64].     For early examples, see Frank J. Goodnow, The Principles of the Administrative Law of the United States 19, 120–21, 126, 150, 155 (1905); and Bruce Wyman, The Principles of the Administrative Law Governing the Relations of Public Officers 9, 10–11, 16–17, 58–61, 91–92, 104–05, 113–14, 195, 197–98, 234–35 (1903); for a late example, see generally Kenneth Culp Davis, Discretionary Justice 81–89 (1970) (using police as core example of administrative agencies exercising discretion).

         [65].     This Section does not capture the nuances of doctrinal controversies or scholarship, but briefly sets out a group of conventional heuristics for administrative law that may look familiar to readers. My goal here is to offer a broad overview, similar to Jeremy Kessler & Charles Sabel, The Uncertain Future of Administrative Law, 150 Daedalus 188, 196 (2021) (discussing the “progressive synthesis” in administrative law). This Article is also not the first to note that some aspects of the administrative state are not well integrated into core debates in administrative law. See Adam B. Cox & Emma Kaufman, The Adjudicative State, 132 Yale L.J. 1769, 1817–18 (2023) (noting the relative neglect of adjudication by scholarship). However, the neglect of the domain of violence and related issues of law execution may run especially deep.

         [66].     See Weber, Economy and Society, supra note 20, at 343–54 (discussing this defining orientation of bureaucracy, as well as the organizational aspects of bureaucracy that allow this orientation to flourish).

         [67].     See Mashaw, supra note 32, at 25­–26.

         [68].     On the general dominance of Weberian state theory in U.S. scholarship on the state, legal and otherwise, see William J. Novak, The Myth of the “Weak” American State, 113 Am. Hist. Rev. 752, 761 (2008); but see Wilson, Bureaucracy, supra note 43, at xviii–xix, 11–13 (1989) (rejecting aspects of Weber’s approach, including his theory building and focus on top-down institutions rather than bottom-up processes and norms as drivers of behavior, in a classic work on the sociology of bureaucracy).

         [69].     See Mashaw, supra note 32, at 34–40 (suggesting that bureaucratic rationality is the dominant ideal type within the Social Security Administration, supplemented by the models of professional treatment and moral judgment).

         [70].     See id. at 26 n.13 (citing Weber as he defines the “bureaucratic administration” ideal type of organization).

         [71].     See id. at 26, 31.

         [72].     See id. at 26–27, 31.

         [73].     See Blake Emerson, Administrative Answers to Major Questions: On the Democratic Legitimacy of Agency Statutory Interpretation, 102 Minn. L. Rev. 2019, 2053–54 (2018) (discussing the Weberian view of bureaucracy as it appears in the work of Landis, Freund, and Mashaw, as well as the work of Edward Rubin, Jacob Gersen, Jeannie Suk, and the scholars of the Legal Process school). For other examples of recent scholarship in dialogue with Weber, see, e.g., Nicholas R. Parrillo, Against the Profit Motive: The Salary Revolution in American Government, 1780-1940, at 1–8 (2013) (framing his empirical contribution partly as a modification and refinement to Weber’s thesis about the connection between salarization and bureaucratization); Anya Bernstein & Cristina Rodríguez, The Accountable Bureaucrat, 132 Yale L.J. 1600, 1635 (2023) (connecting interview statements about bureaucratic technical expertise to Weber’s observations on expertise in bureaucracy).

         [74].     This chart is adapted from Mashaw, supra note 32, at 31.

         [75].     As Gillian Metzger has charted, years of opinions by several members of the current Supreme Court have criticized “bureaucrats” as unaccountable and focused on arrogating power. See, e.g., City of Arlington v. FCC, 569 U.S. 290, 305 (2013) (invoking “unelected federal bureaucrats”); Free Enter. Fund v. Pub. Co. Acct. Oversight Bd., 561 U.S. 477, 499–500 (2010); Metzger, Constitutional Duty, supra note 41, at 35–36.

         [76].     See James M. Landis, The Administrative Process 22–24 (1938); see also Louis Jaffe, The Illusion of the Ideal Administration, 86 Harv. L. Rev. 1183, 1186–87 (1973) (noting similarities between Landis’s view of bureaucracy and Weber’s).

         [77].     See, e.g., Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 865 (1984) (invoking expertise as a core reason to defer to agencies); Kisor v. Wilkie, 139 S. Ct. 2400, 2413, 2417 (2019) (invoking as a justification for Auer deference to agency interpretation of regulatory language the “unique expertise” of agencies, “often of a scientific or technical nature”).

         [78].     See, e.g., Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402, 415 (1971) (noting that agency action receives “a presumption of regularity”); United States v. Morgan (Morgan IV), 313 U.S. 409, 421–22 (1941).

         [79].     See Emerson, supra note 73, at 2025. As personnel change on the Supreme Court has led to greater scrutiny of the “undemocratic” administrative state, Professors Emerson, Rodríguez, and Bernstein, among others, have connected bureaucratic administration to democratic values. They have suggested that a bureaucratic organization populated by professional civil servants can channel public input in relatively direct ways, see Emerson, supra note 73, at 2073–76; and have demonstrated that democratic values infuse bureaucrats’ attitudes to their work, see Bernstein & Rodriguez, supra note 73, at 1613–14. These accounts also tend to treat bureaucracy as a neutral entity—at least in the sense that it serves as a conduit for democratic political preferences rather than imposing its own, assessing evidence and giving reasoned decisions in service of public control and accountability.

         [80].     The basic point here, implicit in sociologically minded work by scholars like Robert Post, is that any body of law governing an occupational role embeds assumptions about that role that are either more or less congruent with the role in practice. See, e.g., Robert C. Post, Democracy, Expertise, and Academic Freedom 55–59 (2012) (arguing that the emergence of a “constitutional sociology of knowledge” privileging certain occupations’ “disciplinary practices” is inevitable in attempts to regulate professional speech with law).

         [81].     This picture is, of course, simplified. See generally Jack M. Beermann, Congressional Administration, 43 San Diego L. Rev. 61 (2006) (detailing numerous levers of congressional control over agencies, including appropriations and reporting requirements).

         [82].     For an important empirical account, see Lisa Schultz Bressman & Abbe Gluck, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part II, 66 Stan. L. Rev. 725, 765–77 (2014) (discussing statutory drafters’ beliefs and attitudes about delegation to agencies).

         [83].     See Cristina M. Rodríguez, The Supreme Court 2020 Term: Foreword: Regime Change, 135 Harv. L. Rev. 1, 49 (2021) (noting that executive orders can “launch a slew of policy processes that reach deep into the administrative state,” but also that “executive orders must be followed with concrete actions by lower-level officials who have the express authority to bring about change”).

         [84].     See Cass R. Sunstein & Adrian Vermeule, The Unitary Executive: Past, Present, Future, 2021 Sup. Ct. Rev. 83, 111 (2021) (naming OIRA review as a lever of presidential control).

         [85].     See generally Paul R. Verkuil, Jawboning Administrative Agencies: Ex Parte Contacts by the White House, 80 Colum. L. Rev. 943 (1980) (describing practice of “jawboning,” or backchannel White House advocacy with administrative agencies).

         [86].     See Elena M. Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2277, 2328 (2001) (noting appointment and removal powers as key presidential management tools).

         [87].     See Abbott Labs. v. Gardner, 387 U.S. 136, 140 (1967).

         [88].     See generally 5 U.S.C. § 706 (setting out grounds of judicial review).

         [89].     See, e.g., Christopher J. Walker, Constraining Bureaucracy Beyond Judicial Review, 150 Daedalus 155, 159–63 (2021) (noting numerous ways that judicial review fails to capture the bulk of agency actions); William H. Simon, The Organizational Premises of Administrative Law, 78 L. & Contemp. Probs. 61, 72–74 (2015) (noting vast areas of “virtually unreviewable discretion” in administrative law due in part to formality requirements); Edward Rubin, It’s Time to Make the Administrative Procedure Act Administrative, 89 Cornell L. Rev. 95, 97–98, 106–09 (2003) (criticizing the APA for failing to “recognize” informal agency action “such as priority setting, resource allocation, research, planning, targeting, guidance, and strategic enforcement” and setting out the ways the statute fails to provide standards for these activities).

         [90].     Walker, supra note 89, at 160–61.

         [91].     Id. at 160.

         [92].     Rubin, supra note 89, at 97–98, 106–09; Simon, supra note 89, at 72–74.

         [93].     Again, this idea is reflected in the presumption of regularity. See supra note 78.

         [94].     5 U.S.C. § 706.

         [95].     See Nicholas Parrillo, The Endgame of Administrative Law: Governmental Disobedience and the Judicial Contempt Power, 131 Harv. L. Rev. 685, 697 (2018).

         [96].     See Kessler & Sabel, supra note 65, at 197.

         [97].     See 5 U.S.C. § 553.

         [98].     Metzger & Stack, supra note 27, at 1239.

         [99].     See id.; Elizabeth Magill, Agency Self-Regulation, 77 Geo. Wash. L. Rev. 2855 (2009); see also Mashaw, supra note 32, at 213–17.

      [100].     See Metzger & Stack, supra note 27, at 1253; Magill, supra note 99, at 884–86.

      [101].     Cf. H.L.A. Hart, The Concept of Law 94–99 (3d ed. 2012) (discussing rules of recognition, or the second-order conventions in any legal system that stipulate the authoritative sources of law).

      [102].     See Nicholas R. Parrillo, Federal Agency Guidance and the Power to Bind: An Empirical Study of Agencies and Industries, 36 Yale J. on Reg. 165, 167–68 (2019).

      [103].     See Metzger & Stack, supra note 27, at 1251–53; Magill, supra note 99, at 866–69.

      [104].     Professor Mashaw’s scholarship on quality assurance processes in adjudication has for many years been the primary work on this issue. See Mashaw, supra note 32, at 145–68; Jerry L. Mashaw, Management Side of Due Process: Some Theoretical and Litigation Notes on the Assurance of Accuracy Fairness and Timeliness in the Adjudication of Social Welfare Claims, 59 Cornell L. Rev. 772, 791 (1974); see also Kent Barnett, Against Administrative Judges, 49 U.C. Davis L. Rev. 1643, 1655–56, 1661–62 (2016) (briefly discussing quality assurance processes applied to administrative judges and administrative law judges).

      [105].     See Mashaw, supra note 32, at 146–63.

      [106].     Even recent efforts to more strictly discipline agency employees have remained within this ambit. See Nou, Civil Service Disobedience, supra note 46, at 380, 380 n.173 (noting efforts by the Trump Administration to strengthen sanctions against and presidential control of the civil service by expediting the timeline on which an employee can be fired or otherwise disciplined).

      [107].     See supra note 46 (gathering citations to recent legal scholarship on autonomous behavior by this entrenched civil service). An extensive literature on this phenomenon began to develop during the Trump Administration.

      [108].     See Emily S. Bremer, Reckoning with Adjudication’s Exceptionalism Norm, 69 Duke L.J. 1749, 1752, 1753–54 (2020) (noting that most agencies that adjudicate do not do so under APA procedures and that “[a]djudication’s exceptionalism norm turns . . . on its head” conventional beliefs that underpin “a unified descriptive and normative account of how administrative law operates” drawn from the realm of rulemaking).

      [109].     See Daniel E. Ho, Does Peer Review Work? An Experiment of Experimentalism, 69 Stan. L. Rev. 1, 5–10 (2017) (gathering empirical studies of widely variable administrative decision-making by bureaucrats and noting limited utility of administrative law to solve these problems).

      [110].     See, e.g., Anne Joseph O’Connell, Bureaucracy at the Boundary, 162 U. Pa. L. Rev. 841 (2014) (describing the existence of agencies on the margins of the conventional view of the federal administrative state, including those that cross public-private, federal-state, and interbranch boundaries).

      [111].     See Davis, supra note 64, at 233.

      [112].     See Dorothy Roberts, Shattered Bonds: The Color of Child Welfare 55–61 (2002) (describing extensive discretion, limited training, and vague guidelines provided to child welfare workers, all of which encourage them to rely on racialized stereotypes about Black mothers).

      [113].     See Michael D. Frakes & Melissa F. Wasserman, Is the Time Allocated to Review Patent Applications Inducing Examiners to Grant Invalid Patents?: Evidence from Micro-Level Application Data, 99 Rev. Econ. & Stat. 550, 550–51 (2017).

      [114].     See Ramji-Nogales et al., supra note 57, at 2–3.

      [115].     See generally Michael Lipsky, Street-Level Bureaucracy: Dilemmas of the Individual in Public Service 3–4 (1983).

      [116].     Organizational sociologists have devoted significant attention to the causes of mistaken or pathological behavior, including rule-bending, within bureaucratic organizations. See generally Diane Vaughan, The Dark Side of Organizations: Mistake, Misconduct, and Disaster, 25 Ann. Rev. Soc. 271 (1999) (surveying classic literature on organizational pathologies and typologizing these pathologies).

      [117].     See, e.g., Nevitt, supra note 11, at 953 (analyzing the military as an agency).

      [118].     We discuss this bespoke administrative law, which is also often shared (or at least mirrored) across the domain of violence, at length in Emily R. Chertoff & Jessica Bulman-Pozen, The Two Faces of the Administrative State 16–30 (manuscript on file with author).

      [119].     As Professor Kati Kovacs has noted, Congress exempted only “military authority exercised in the field in time of war or in occupied territory” from APA judicial review. See Kathryn E. Kovacs, A History of the Military Authority Exception in the Administrative Procedure Act, 62 Admin. L. Rev. 673, 677, 679–708 (2010). This was a deliberate choice against military protests, grounded in fears of totalitarianism that the war in Europe had thrown into stark relief. See id. at 710.

      [120].     For instance, though military and foreign affairs functions are exempt from notice-and-comment rulemaking, the Department of Defense has published over four thousand final rules in the Federal Register, and the Department of Homeland Security nearly ten thousand, over twenty years. See Document Search, Fed. Reg., https://www.federalregister.gov/documents/search?conditions%5Bagencies%5D%5B%5D=defense-department&conditions%5Btype%5D%5B%5D=RULE [https://perma.cc/HB5M-CTAX]; Document Search, Fed. Reg., https://www.federalregister.gov/documents/search?conditions%5Bagencies%5 D%5B%5D=homeland-security-department&conditions%5Btype%5D%5B%5D=RULE [https://perma.cc/LQ8A-JYHN].

      [121].     As discussed infra, the APA was largely meant to cover these agencies, but courts often have interpreted its exceptions broadly, deferred to these agencies, or otherwise refrained from reviewing them. See Elena R. Chachko, Toward Regulatory Isolationism? The International Elements of Agency Power, 57 U.C. Davis L. Rev 57, 75–90 (forthcoming 2023) (detailing “permissive” framework applied to agencies that invoke international considerations, including many force agencies); Oona A. Hathaway, National Security Lawyering in the Post-War Era: Can Law Constrain Power?, 68 UCLA L. Rev. 2, 15–20 (2021) (noting broad array of deference doctrines courts use in national security contexts); Justin Driver & Emma Kaufman, The Incoherence of Prison Law, 135 Harv. L. Rev. 515, 536–37 (2021) (noting deference in prison contexts). Other hurdles to suit include information asymmetries and the limited power that affected parties often have. See, e.g., David Pozen, Deep Secrecy, 62 Stan. L. Rev. 257, 267–68 (2010) (offering criteria to assess the “depth” of state secrets as a way of assessing how they contribute to “knowledge asymmetries” between government actors and the public).

      [122].     Local and state control of police and prison authorities sometimes follows similar contours, though some states exempt prison authorities from their Administrative Procedure Acts. See Giovanna Shay, Ad Law Incarcerated, 14 Berkeley J. Crim. L. 329, 346–47 (2009). While these state agencies fall within the domain of violence, a dissection of the ways these agencies differ from or are similar to other agencies of state and local government is beyond the scope of this Article.

      [123].     See, e.g., Homeland Security Act of 2002, Pub. L. 107-296, 116 Stat. 2135 (creating Customs and Border Protection and Immigration and Customs Enforcement); An Act to Reorganize the Administration of Federal Prisons, Pub. L. 71-218, 46 Stat. 325 (1930) (creating the federal Bureau of Prisons); National Security Act of 1947, Pub. L. 80-253, 61 Stat. 495) (1947) (creating the modern Department of Defense, originally named the National Military Establishment).

      [124].     For statutes regulating the military, see, e.g., Pub. L. 81-506, 64 Stat. 107 (1950) (creating the Uniform Code of Military Justice, and with it a system of military discipline, later to be filled in by military regulation); Pub. L. 107-40, 115 Stat. 224 (2001) (authorizing use of military force against parties involved in the September 11, 2001, terrorist attacks); see also Zachary S. Price, Congress’s Power over Military Offices, 99 Tex. L. Rev. 491, 495 (2021) (noting dense web of statutes governing appointment and removal of military officers).

      [125].     See Weber, Economy and Society, supra note 20, at 112 (“An order will be called a . . . Law, where its validity is externally underwritten by the Chance that physical or mental coercion will be applied by a specialised staff of people whose task is to enforce conformity or punish contravention.”); id. at 135–36 (“An institutionally organised political enterprise . . . will be called a state if, and to the extent that, its administrative staff can lay claim to a monopoly of legitimate physical force in the execution of its orders.”).

      [126].     “The last thirty years, legal scholars have largely lost sight of overall police function, goals, or police crime control metrics”—the subjects the policing literature that I cite addresses. Shima Baradaran Baughman, Crime and the Mythology of the Police, 99 Wash. U. L. Rev. 65, 73 (2021). It is largely within the past two years that legal scholars have begun to return to this earlier social science literature. See generally id. (using historical and social scientific literature to investigate the “function and purpose” of police); Friedman & Ponomarenko, supra note 56, at 147.

      [127].     See DiMaggio & Powell, supra note 30, at 150.

      [128].     Compare Tom Tyler, Patrick E. Callahan & Jeffrey Frost, Armed, and Dangerous(?): Motivating Rule Adherence Among Agents of Social Control, 41 Law & Soc’y Rev. 457, 458 (2007) (characterizing military and law enforcement employees as “order maintenance” agents, and writing, after the Abu Ghraib prisoner abuse scandal, that “[a]n important reason for the persistence of problems in preventing misconduct among those responsible for order maintenance is the nature of the situation in which social control agents work, i.e., the nature of the tasks they perform and the institutional structure and dynamics that surround [them]”).

      [129].     See also Weber, Economy and Society, supra note 20, at 136; cf. Mashaw, supra note 32, at 25–34 (using this four-part framework to distill and define three “ideal types” of bureaucracy from similar qualitative evidence of organizational culture and institutional environment in the claims processing context).

      [130].     See supra note 24.

      [131].     See Weber, Politics, supra note 24, at 78 (suggesting that the core characteristic of a state is whether it can “clai[m] the monopoly of the legitimate use of physical force within a given territory”).

      [132].     For examples from policing, see, e.g., Didier Fassin, Enforcing Order: An Ethnography of Urban Policing 176–78 (2013) (documenting the self-perception of police as promoting “force and order”); William A. Westley, Violence and the Police 192 (1970) (describing police in their role conception as “agents of order”); and Ronald Weitzer, Can the Police Be Reformed?, 4 Contexts 21, 22 (2005) (“Officers see themselves as a ‘thin blue line’ between order and chaos.”); see also Jean-Paul Brodeur, The Policing Web 39 (2010). For examples from immigration enforcement, see, e.g., Irene Vega, Empathy, Morality, and Criminality: the Legitimation Narratives of U.S. Border Patrol Agents, 44 J. Ethnic & Migration Studs. 2544, 2550–54 (2018) (describing narratives that border crossers are dangerous criminals and that border crossers pose uncertain risks as core scripts that Border Patrol agents use to legitimate their work); see also infra, e.g., notes 227–229 (documenting perception that ICE administrators are fighting lawbreaking or taking dangerous people off the streets).

      [133].     See, e.g., Egon Bittner, Florence Nightingale in Pursuit of Willie Sutton: A Theory of the Police, in The Potential for Reform of Criminal Justice 32–35 (Herbert Jacob ed., 1974) [hereinafter Bittner, Willie Sutton] (characterizing the police role in terms of the capacity to respond to emergencies that may require the use of force).

      [134].     Though the view that police primarily fight crime has been challenged by empirical literature showing that 70–80 percent of their time is downtime, see Christine N. Famega, Variation in Officer Downtime: A Review of the Research, 28 Policing 388, 391 (2005), the crime-fighting view plays a critical role in structuring public and officer expectations. See Michael Sierra-Arévalo, Reward and “Real” Police Work, in The Ethics of Policing: New Perspectives on Law Enforcement 67–70 (Ben Jones & Eduardo Mendieta eds., 2021); see also Nat’l Acad. of Scis., Fairness and Effectiveness in Policing: The Evidence 58–60 (Wesley G. Skogan & Kathleen Frydl eds., 2004) (“Patrol officers also spend a significant part of their unassigned time conducting what is known as

‘preventive patrol[’] . . . the core police activity, designed to prevent crime by deterring potential offenders through a visible police presence.”).

      [135].     See, e.g., Fassin, supra note 132, at xv; Brodeur, supra note 132, at 18, 22, 35; James Q. Wilson, Varieties of Police Behavior: The Management of Law and Order in Eight Communities 16–17 (1978) [hereinafter Wilson, Police Behavior].

      [136].     Bittner, Willie Sutton, supra note 133, at 32.

      [137].     Wilson, Police Behavior, supra note 135, at 16–17.

      [138].     See supra note 134.

      [139].     See, e.g., Dorothy E. Roberts, Foreword: Race, Vagueness, and the Social Meaning of Order-Maintenance Policing, 89 J. Crim. L. & Criminology 775, 779–80 (1999) (noting that order-maintenance policing “entrenches the racialized division of Americans into the presumptively lawless whose liberties deserve little protection and the presumptively law-abiding who are entitled to rule over them”); Bernard E. Harcourt, Reflecting on the Subject: A Critique of the Social Influence Conception of Deterrence, the Broken Windows Theory, and Order-Maintenance Policing New York Style, 97 Mich. L. Rev. 291, 352–56 (1998).

      [140].     See Irene I. Vega, “Reasonable” Force at the U.S.-Mexico Border, 69 Social Probs. 1154, 1161 (2021).

      [141].     See Wilson, Bureaucracy, supra note 43, at 38–39; Ethan M. Higgins, Justin Smith, & Kristin Swartz, “We Keep the Nightmares in Their Cages”: Correctional Culture, Identity, and the Warped Badge of Honor, 60 Criminology 429, 434 (2022); John R. Hepburn, The Exercise of Power in Coercive Organizations: A Study of Prison Guards, 23 Criminology 145, 145, 160 (1985).

      [142].     Accord Fassin, supra note 132 at 125–26, 125 n.7 (2013) (“[V]iolence is in fact constitutive of the very role of law enforcement.”); Jean-Paul Brodeur, The Policing Web 103–38 (2010) (suggesting that distinguishing feature of police is their legal powers, which are not shared by the general public, but which are underwritten by the potential to use force); see, e.g., Albert J. Reiss, Jr., The Police and the Public 17 (1972) (“Though superficially [the police’s] role is to preserve the peace by coping with any individuals in the large aggregation who violate laws, they must be prepared to restore order [through force] as well.”); Bittner, Willie Sutton, supra note 133, at 35 (“The policeman, and the policeman alone, is equipped, entitled, and required to deal with every exigency in which force may have to be used”); see also Trevor Jones & Tim Newburn, Private Security and Public Policing 258 (1998) (noting that “sociologists have not improved upon [Bittner’s] definition [of the police] in the past twenty-five years”).

      [143].     See Richard H. McAdams, Dhammika Dharmapala & Nuno Garoupa, The Law of Police, 82 U. Chi. L. Rev. 135, 143–47 (2015) (reviewing evidence of police self-selection, particularly evidence suggesting that people who become police tend to have more punitive attitudes).

      [144].     See, e.g., Westley, supra note 132, at 34–41 (describing rationalizations that police deploy to justify the illegal use of violence in service of crime control goals); see also infra note 160–164.

      [145].     See, e.g., P.A.J. Waddington, Policing Citizens: Authority and Rights 22–23 (1999) (suggesting that public ambivalence towards the police stems from “the fact that police may use violence against ‘us’” in service of their goals “and therefore they present a threat”). As a consequence of this dynamic, there is an extensive literature on conditions under which community members see the police as trustworthy or legitimate. See generally Bell, supra note 39; Tom R. Tyler & Yuen J. Huo, Trust in the Law: Encouraging Public Cooperation with the Police and Courts, at xiii–xvi (2002).

      [146].     See, e.g., Elizabeth Reuss-Ianni, Two Cultures of Policing: Street Cops and Management Cops 2–3 (1983) (describing a “management cop culture” that “is positively oriented towards public administration and looks to scientific management,” paralleling Mashaw’s description of bureaucratic rationality).

      [147].     The recent literature on civil servant disobedience, mostly focused on very different agencies, exemplifies this. See Nou, Civil Service Disobedience, supra note 46, at 357, 371 (discussing the Environmental Protection Agency, Department of Housing and Urban Development, and Department of Labor alongside ICE).

      [148].     For police, see, e.g., John E. Van Maanen, Working the Street: A Developmental View of Police Behavior, in The Potential for Reform of Criminal Justice 221 (Herbert Jacobs ed., 1974) (describing how the separation between police and the public, and the risky nature of police work, leads to a subculture that binds police together and alienates them from society at large). For corrections officers, see, e.g., Higgins, Smith, & Swartz, supra note 141, at 434 (“[T]he code of correctional staff established via correctional training and informal enculturation emphasizes solidarity in the face of isolation . . . where officers prioritize the safety of coworkers over the incarcerated at any cost.”) (citations omitted).

      [149].     For police, see, e.g., Peter Moskos, Cop in the Hood 145–55 (2008) (describing conflict between management and street cops over minimum arrest quotas). For immigration enforcement, see, e.g., Vega, supra note 140, at 1164 (noting theme of tensions between Border Patrol line officers and management over use of force in interviews). For corrections officers, see, e.g., Higgins, Smith, & Swartz, supra note 141, at 434 (collecting literature finding “mistrust towards administrators” among front-line corrections officers). See also Elisabeth Jean Wood, Armed Groups and Sexual Violence: When Is Wartime Rape Rare?, 37 Pol. & Soc. 131, 136–42 (2009) (synthesizing organizational theories of unauthorized violence by armed groups, especially theories revolving around the relationships and relative influence of a group’s leadership versus workforce in norms around the use of violence).

      [150].     For police, see, e.g., Fassin, supra note 132, at 100; Reuss-Ianni, supra note 146, at 1–7, 75 (1983) (theorizing, based on a study of the New York City Police Department, that street police within modern police organizations have developed an in-group/out-group perspective). For prisons, see, e.g., Higgins, Smith, & Swartz, supra note 141, at 434 (documenting “us-versus-them ideologies” among corrections officers vis-à-vis incarcerated people). For a comparison of prisons and police, see, e.g., Mary Ann Farkas & Peter K. Manning, The Occupational Culture of Corrections and Police Officers, 20 J. Crime & Just. 51, 55–56 (1997) (noting in-group solidarity at the front-line level of both organizations, as well as differences in norms and orientations among organizational levels that lead to mistrust). For a comparison of military and police, see Joseph Soeters, Culture in “Uniformed” Organizations, in Handbook of Organizational Culture and Climate 473–74 (Neal M. Ashkanasy, Celeste P.M. Wilderom & Mark F. Peterson eds., 1st ed. 2000) (noting, in a work synthesizing literature on military and police, the presence of a common us/them perception).

      [151].     For police, see, e.g., Moskos, supra note 149, at 83–85 (describing how police officers develop intuitions about who is involved in drug activity in a highly policed Black neighborhood); John E. Van Maanen, The Asshole, in Policing: A View from the Streets 221–237 (Peter K. Manning & John Van Maanen eds., 1978) (surveying a literature on “the various ‘outgroups’ perceived by the police” in the course of patrol work). For immigration enforcement, albeit in a different (and arguably less punitive) national context, see, e.g., Mary Bosworth, Inside Immigration Detention 150 (2014) (writing of immigration detention in the United Kingdom that antagonistic or stereotyping behavior by detention staff, while observed relatively infrequently, evidenced a “desire [of staff] to distance themselves” emotionally and interpersonally from detainees).

      [152].     See, e.g., Reuss-Ianni, supra note 146, at 53–57 (documenting street cops’ suspicion of management, media, and politicians during a police brutality trial).

      [153].     See infra notes 161–167.

      [154].     For police, see, e.g., Bittner, Willie Sutton, supra note 133, at 28, 31–32, 35 (arguing that police are “permanently and specifically attuned” to risks in their vicinity whose resolution may require the use of force) (emphasis in original); Michael Sierra-Arévalo, American Policing and the Danger Imperative, 55 Law & Soc. Rev. 70, 70–71 (2021) (hereinafter Sierra-Arévalo, The Danger Imperative) (reviewing literature on connection between the notion that police officers are under constant physical threat and non-compliance with procedural rules); Jason R. Ingram, William Terrill & Eugene A. Paoline III, Police Culture and Officer Behavior: Application of a Multilevel Framework, 56 Criminology 780, 796–801 (2019) (discerning a relationship in most cases between the perceived danger of the environment officers work in and their use of coercion); Jerome H. Skolnick, Justice Without Trial: Law Enforcement in a Democratic Society 42–62 (4th ed. 2011) [hereinafter Skolnick, Justice Without Trial] (placing persistent exposure to or perception of danger at the core of what shapes police occupational culture). For immigration enforcement, see Vega, supra note 140, at 1159 (finding that Border Patrol agents characterized their work as “uniquely dangerous among the law enforcement profession,” and that this perception was a function of both the border environment and “bureaucratic messages about high assault risks”).

      [155].     See, e.g., Soeters, supra note 150, at 473–74 (noting that military and police share an occupational culture based around danger or risk, inculcated partly through training); Seth Stoughton, Law Enforcement’s “Warrior” Problem, 128 Harv. L. Rev. F. 225, 228 (2015) (explaining that police training and culture promote “hypervigilance” and a perception of threat).

      [156].     See Sierra-Arévalo, The Danger Imperative, supra note 154, at 70–71; Vega, supra note 140, at 1160–61.

      [157].     As Egon Bittner put it, “no human problem exists, or is imaginable, about which it could be said with finality that this certainly could not become the proper business of the police.” Bittner, Willie Sutton, supra note 133, at 30; Wilson, Police Behavior, supra note 135, at 64–66 (describing centrality of discretion to police work); see also Harcourt, supra note 139, at 345.

      [158].     Bittner, Willie Sutton, supra note 133, at 29–30.

      [159].     Mashaw, supra note 32, at 25–26.

      [160].     See, e.g., Moskos, supra note 149, at 9–10 (“Police violate departmental regulations all the time.”); Jerome H. Skolnick & James J. Fyfe, Above the Law: Police and the Excessive Use of Force 89–171 (1993) (identifying features particular to police agencies that encourage a lack of accountability for the use of excessive force); Jennifer Hunt & Peter K. Manning, The Social Context of Police Lying, 14 Symbolic Interaction 51, 52–53, 56–65 (1991) (collecting literature documenting police lying, and typologizing contexts in which such lying is considered normal); Jennifer Hunt, Police Accounts of Normal Force, 13 Urb. Life 315, 316–17 (1985) (describing, based on observational study, that police perceptions of “normal” and permissible force, versus “immoral” “[b]rutality,” permit more violent behavior than do legal rules for the use of force).

      [161].     See Marie Ouellet, Sadaf Hashimi, Jason Gravel & Andrew V. Papachristos, Network Exposure and Excessive Use of Force: Investigating the Social Transmission of Police Misconduct, 18 Criminology & Pub. Pol’y 675, 686–89 (2019) (finding that network effects influenced the use of excessive force).

      [162].     See, e.g., Jerome H. Skolnick, Corruption and the Blue Code of Silence, 3 Police Practice & Rsch. 7, 7 (2002) (detailing evidence of a “blue code of secrecy” among police); Egon Bittner, The Functions of the Police in Modern Society 63 (1980) (connecting the concept of police “esprit de corps” to a “code of secrecy” by police).

      [163].     See Christopher Slobogin, Testilying: Police Perjury and What to Do About It, 67 U. Colo. L. Rev. 1037, 1041–48 (1996). For literature suggesting that aspects of the police role may create the conditions for lying, see, e.g., Hunt & Manning, supra note 160, at 56–65 (1991); Peter Manning, Police Lying, 3 Urb. Life 283, 286–89 (1974).

      [164].     See, e.g., Bethan Loftus, Police Occupational Culture: Classic Themes, Altered Times, 20 Policing & Soc’y 1, 13–14 (2010) (discussing observed police perception of danger and its role in promoting covering for colleagues); Westley, supra note 132.

      [165].     See supra notes 154–157.

      [166].     It is more challenging to make the case that violence alone does not explain the incongruence between this domain and administrative law. One exception, and a complicated one, is the imposition of the death penalty, which prison agencies carry out at the direction of the courts and which is not implemented in situations of perceived risk or danger, but in contexts where an individual is highly restrained. See Corinna Barrett Lain, Death Penalty Exceptionalism and Administrative Law, 8 Belmont L. Rev. 552, 553–54 (2021). The death penalty is a complicated example because its imposition is highly proceduralized. Yet, the administrative rules governing how it is carried out tend to be weak, delegating extensive discretion to Department of Corrections staff. See id. at 554–72 (detailing widespread failures in the administration of lethal injection by corrections departments).

      [167].     See Matthew Desmond, Making Firefighters Deployable, 34 Qual. Socio. 59, 60–61 (2011) (discussing firefighters in the context of the literature on high-risk occupations, including policing and soldiering, and canvassing associated theories of masculine role as a driver of agents’ behavior in accepting risk).

      [168].     See Matthew Desmond, On the Fireline: Living and Dying with Wildland Firefighters 145–69 (2007) [hereinafter Desmond, On the Fireline] (describing training in, socialization to, application of, and ultimate impossibility of full compliance with two sets of canonical organizational safety rules among firefighters).

      [169].     See id. at 152, 180–85, 190; see also Desmond, Making Firefighters Deployable, supra note 167, at 63–67 (noting that since “it is impossible to fight a wildfire without violating” at least one of these rules, “we reasonably would expect firefighters to look on the[m] with suspicion,” and yet “every member of the Forest Service who I met valued these mandates”).

      [170].     See Desmond, On the Fireline, supra note 168, at 180–85.

      [171].     See, e.g., Leo McCann, Edward Granter, Paula Hyde & John Hassard, Still Blue-Collar After All These Years? An Ethnography of the Professionalization of Emergency Ambulance Work, 50 J. Mgmt. Stud. 699, 762–64 (2013) (noting reluctance by studied paramedics to deviate from protocols and use discretion, and pointing to distress participants voiced in situations where they were forced to deviate).

      [172].     Skolnick, Justice Without Trial, supra note 154, at 62; see also Sierra-Arévalo, The Danger Imperative, supra note 154, at 75.

      [173].     See, e.g., Michael A. Caldero & John P. Crank, Police Ethics: The Corruption of Noble Cause 2 (3rd ed. 2010) (describing “noble cause corruption” among police as corruption “committed in the name of good ends”); Carl B. Klockars, The Dirty Harry Problem, 452 Annals Am. Acad. Pol. Soc. Sci. 33, 38 (1980) (suggesting that police officers routinely confront situations that require them to trade off rule-bending and their order maintenance mandate, with the importance of the latter leading them to violate rules).

      [174].     See Cover, supra note 10, at 1613–15 (canvassing a set of explanations for how officials overcome inhibitions against violence, and concluding, “[C]ommon to all of these theories is a behavioral observation in need of explanation. Persons who act within social organizations that exercise authority act violently without experiencing the normal inhibitions . . . which regulat[e] the behavior of those who act autonomously”).

      [175].     See Eisha Jain, Arrests as Regulation, 67 Stan. L. Rev. 809, 820–25 (2015) (reviewing the significant consequences that flow from arrests).

      [176].     While this case study helped inform the analysis in Part II and was, in turn, informed by it, Part II does not rely primarily on Part III’s case study. Part III focuses on organization-level dynamics, while Part II synthesizes literature on both organizational dynamics and administrators’ orientations.

      [177].     An observer might here see administrative law working as intended to channel public sentiment against immigrants. But this view is too simple. Even as it built our contemporary immigration system, Congress also granted immigrants new substantive and procedural rights, an incongruity that the conventional pictures of democratic engagement that populate administrative law do not explain. See, e.g., Pub. L. 82–414, 66 Stat. 163, Sec. 311 (1952) (amending Immigration and Naturalization Act to both strip procedural protections in deportation hearings and end restrictions on immigration “because of race or sex”); see also William Funk, The Rise and Purported Demise of Wong Yang Sun, 58 Admin. L. Rev. 881, 883–85 (2006). Moreover, Congress presumably did not intend for ICE to act unconstitutionally. For instance, the Fifth Amendment requires that ICE maintain adequate conditions in its facilities at a standard higher than the prison context. Yet, some of ICE’s actions have violated this standard. See Jones v. Blanas, 393 F.3d 918, 933–34 (9th Cir. 2004), cert. denied, 126 S. Ct. 351 (2005).

      [178].     See 8 U.S.C. § 1357(a).

      [179].     For an overview and explanation of ICE’s detention capacity and its evolution, see generally Jennifer M. Chacón, Immigration Detention: No Turning Back?, 113 S. Atl. Q. 621 (2014).

      [180].     ICE ERO employs over eight thousand people directly. See Career Frequently Asked Questions (FAQs), U.S. Immigr. & Customs Enf’t (Sept. 1, 2022), https://www.ice.gov/careers/faqs [https://perma.cc/YP8V-2QKM]. ICE also manages the country’s largest carceral system, spread across as many as 637 different ICE facilities, some of which are overseas. See New Data on 637 Detention Facilities Used by ICE in FY 2015, TRAC Immigr. (Apr. 12, 2016), https://trac.syr.edu/immigration/reports/422/ [https://perma.cc/RVL2-2BL2]; see also infra notes 190–196 and accompanying text.

      [181].     A reader might wonder whether conditions at ICE reflect the reality of plenary power over immigration. This view greatly overstates immigration exceptionalism. Plenary power applies to the executive’s ability to decide which non-citizens can enter and remain in the country. See, e.g., Trump v. Hawaii, 138 S. Ct. 2392, 2402 (2018) (defining “the admission and exclusion of foreign nationals” as “a fundamental sovereign attribute . . . largely immune from judicial control”); Kerry v. Din, 576 U.S. 86, 101–05 (2015) (Kennedy, J., concurring) (largely upholding the doctrine of consular non-reviewability as a concomitant of the executive’s plenary power over immigration). However, plenary power does not apply to how ICE manages its detention facilities or treats immigrants passing through the immigration system, nor does it excuse ICE from complying with administrative rules or guidance it adopts. See, e.g., Wong Wing v. U.S., 163 U.S. 228, 231, 235–36 (1896) (finding that deportation is not a punishment, and establishing limitations on conditions in which detained immigrants can be held); Jones v. Blanas, 393 F.3d 918, 933–34 (9th Cir. 2004), cert. denied, 126 S. Ct. 351 (2005) (suggesting that constitutional conditions minima in the immigration context exceed those in criminal detention).

      [182].     ICE has a long history of obscuring information from the public. See, e.g., Lucas Guttentag, Trump Administration Designates ICE a “Security/Sensitive” Agency, Immigr. Pol’y Tracking Project (June 26, 2020), https://immpolicytracking.org/policies/reported-trump-administration-designates-ice-as-a-security-agency [https://perma.cc/PWG2-4J26]; Use of ICE Detainers: Obama vs. Trump, TRAC Immigr. (Aug. 30, 2017), https://trac.syr.edu/immigration/reports/479/ [https://perma.cc/J9XS-4KFA] (noting that agency repeatedly withheld mandated disclosures on its detainer program).

      [183].     See Irene I. Vega, Toward a Cultural Sociology of Immigration Control Work: A Call for Research, 69 Am. Behav. Sci. 1172 (2019), https://journals.sagepub.com/doi/epub/10.1177/0002764219835251; see also Emily Ryo, Understanding Immigration Detention: Causes, Conditions, and Consequences, 15 Ann. Rev. L. & Soc. Sci. 97, 105–07 (2019) (noting dearth of qualitative research on immigration detention in the United States, especially on detention officers and “less-visible” groups of detained people).

      [184].     Between 2019 and 2020, I worked as a litigator and removal defense attorney for immigrants held at two large detention facilities in Southern California. This experience was essential to prepare and carry out this study. See Mario Luis Small & Jessica McCrory Calarco, Qualitative Literacy: A Guide to Evaluating Ethnographic and Interview Research 18–19 (2022) (discussing the importance of “exposure,” or “direct contact” with “the social world or its people,” for qualitative research).

      [185].     Importantly, readers should not take me as saying that rank-and-file ICE officers are homogeneous in their perceptions and approach to their jobs. A number of interviewees of all persuasions emphasized this, and indeed no group of eight thousand people could be this way.

      [186].     See supra notes 100–101 and accompanying text.

      [187].     See ICE Leadership, U.S. Immigr. & Customs Enf’t (Oct. 14, 2022), https://www.ice.gov/leadership [https://perma.cc/2S63-BAXR].

      [188].     See Organizational Structure, U.S. Immigr. & Customs Enf’t (Nov. 22, 2021), https://www.ice.gov/leadership/organizational-structure [https://perma.cc/73Q2-XH8K]; Enforcement and Removal Operations, U.S. Immigr. & Customs Enf’t (Oct. 6, 2022), https://www.ice.gov/about-ice/ero [https://perma.cc/CCM6-CMXQ]; U.S. Immigration and Customs Enforcement: 2021 ERO Field Offices, U.S. Immigr. & Customs Enf’t (accessed Nov. 4, 2022), https://www.ice.gov/doclib/about/offices/ero/pdf/eroFieldOffices.pdf [https://perma.cc/J3VS-XS2A].

      [189].     See supra notes 102–103 and accompanying text.

      [190].     See Identify and Arrest, U.S. Immigr. & Customs Enf’t (Oct. 27, 2022), https://www.ice.gov/identify-and-arrest [https://perma.cc/9XP7-EHC2].

      [191].     By the end of fiscal year 2019, ICE had nearly 3.3 million people on its non-detained removal docket. See U.S. Immigr. & Customs Enf’t, U.S. Immigration and Customs Enforcement Fiscal Year 2019 Enforcement and Removal Operations Report 10 (2019), https://www.ice.gov/sites/default/files/documents/Document/2019/eroReportFY2019.pdf [https://perma.cc/T9ME-KNW5].

      [192].     See, e.g., 8 U.S.C. § 1226(c)(1) (establishing mandatory detention of certain immigrants during the pendency of their proceedings).

      [193].     See, e.g., 34 U.S.C. § 30307(c) (applying the Prison Rape Elimination Act to immigration facilities).

      [194].     See 8 U.S.C. § 1362 (establishing “right to counsel” of immigrant’s choosing “[i]n any removal proceedings before an immigration judge and in any appeal proceedings,” so long as representation comes “at no expense to the Government”).

      [195].     Title II of the Americans with Disabilities Act, 42 U.S.C. § 12132, and Section 504 of the Rehabilitation Act of 1973, 29 U.S.C. § 794, impose obligations related to detainees’ ability to participate in their cases and receive disability accommodations.

      [196].     See H.R.J. Res. 31, 116th Cong. (2019) (requiring ICE to report on its compliance with detention standards). The main documents setting out detention standards are, for ICE-only facilities, U.S. Immigr. & Customs Enf’t, Performance-Based National Detention Standards 2011 (Dec. 2016), https://www.ice.gov/doclib/detention-standards/2011/pbnds2011r2016.pdf [https://perma.cc/HG7Y-SVKM]; for facilities that house both ICE and non-ICE detainees, U.S. Immigr. & Customs Enf’t, National Detention Standards for Non-Dedicated Facilities (2019), https://www.ice.gov/doclib/detention-standards/2019/nds2019.pdf [https://perma.cc/DUC9-RPGP]; and for immigrant family detention, U.S. Immigr. & Customs Enf’t, Family Residential Standards (FRS): Program Philosophy, Goals, and Expected Outcomes (2020), https://www.ice.gov/doclib/frs/2020/2020family-residential-standards.pdf [https://perma.cc/LC3N-PHCK].

      [197].     See, e.g., Memorandum from Alejandro Mayorkas, Sec’y of Dep’t of Homeland Security, to Tae D. Johnson, Acting Dir. of Immigr. & Customs Enf’t, et al., Worksite Enforcement: The Strategy to Protect the American Labor Market, the Conditions of the American Worksite, and the Dignity of the Individual, Dep’t of Homeland Sec. (Oct. 12, 2021), https://www.dhs.gov/sites/default/files/publications/memo_from_secretary_mayorkas_on_worksite_enforcement.pdf [https://perma.cc/64W6-S5F6] (directing agencies to end worksite raids); Memorandum from John Morton, Dir. of Immigr. & Customs Enf’t, to All ICE Employees, Civil Immigration Enforcement: Priorities for the Apprehension, Detention, and Removal of Aliens, U.S. Immigr. & Customs Enf’t (Mar. 2, 2011), https://www.ice.gov/doclib/news/releases/2011/ 110302washingtondc.pdf [https://perma.cc/8ZC2-GY8M] (earlier attempt to set enforcement priorities).

      [198].     See, e.g., ICE Detention and Deportation Officer’s Field Manual, Am. Immigr. Lawyers’ Ass’n (Oct. 5, 2009), https://www.aila.org/library/ice-detention-deportation-officers-field-manual [https://perma.cc/W88N-FPW9] (offering guidelines on how to execute removals).

      [199].     For instance, after arrest, some asylum seekers are eligible for release (“parole”) from detention to finish their cases. U.S. Immigr. & Customs Enf’t, Directive 11002.1, Parole of Arriving Aliens Found to Have a Credible Fear of Persecution or Torture (Dec. 8, 2009), https://www.ice.gov/doclib/dro/pdf/11002.1-hd-parole_of_arriving_aliens_found_credible_fear.pdf [https://perma.cc/4TLX-CKNA]. The Parole Directive asks ICE agents to weigh each asylum seeker’s facts and make an assessment of “flight risk” and “danger to the community.” See id. at 6–8.

      [200].     See supra notes 104–105 and accompanying text (discussing management and quality assurance mechanisms in conventional administrative law).

      [201].     See Government Interview #1 at 6–7, 8–10 (describing internal appeals process and dashboard giving the Secretary of DHS the ability to monitor enforcement actions on the ground).

      [202].     This centralized training has been criticized in the past for its lack of standardization. See Off. of Inspector Gen., U.S. Dep’t of Homeland Sec., OIG-16-19, DHS’ Oversight of its Workforce Training Needs Improvement (Jan. 20, 2016), https://www.oversight.gov/sites/default/files/oig-reports/OIG-16-19-Jan16.pdf; Off. of Inspector Gen., U.S. Dep’t of Homeland Sec., OIG, 18-42, Management Alert — ICE’s Training Model Needs Further Evaluation (Jan. 25, 2018), https://www.oig.dhs.gov/sites/default/files/assets/Mga/2018/oig-18-42-jan18.pdf [https://perma.cc/EE8A-NE7Q] [hereinafter Off. of Inspector Gen., Management Alert].

      [203].     See infra notes 262–264 and accompanying text.

      [204].     See infra notes 267–270 and accompanying text.

      [205].     For the presumption of regularity, see supra note 78. For finality, see, e.g., FTC v. Standard Oil Co., 449 U.S. 232, 242–43 (1980) (articulating classic argument for finality, on grounds that it prevents judicial review from “interfer[ing] with the proper functioning of the agency” or causing “delay”); Reliable Automatic Sprinkler Co. v. Consumer Prod. Safety Comm’n, 324 F.3d 726, 731 (D.C. Cir. 2003).

      [206].     See, e.g., infra, notes 273–280.

      [207].     In work in progress, I document recent litigation where courts have varied from this trend. See Emily R. Chertoff, Administrative Civil Liberties Law (manuscript on file with author).

      [208].     See supra notes 192–195.

      [209].     See César Cuauhtémoc García Hernández, Immigration Detention as Punishment, 61 UCLA L. Rev. 1346, 1351–55 (2014); see also Wong Wing v. United States, 163 U.S. 228, 237 (1896).

      [210].     See, e.g., Michael T. Light, Jingying He & Jason P. Robey, Comparing Crime Rates Between Undocumented Immigrants, Legal Immigrants, and Native-Born U.S. Citizens in Texas, 117 Proc. Nat’l Acad. Sci. 32340, 32340 (2020), https://www.pnas.org/doi/epdf/10.1073/pnas.2014704117 [https://perma.cc/JS98-GQCX]; see also Alex Nowrasteh, Criminal Immigrants in Texas in 2019: Illegal Immigrant Conviction Rates and Arrest Rates for Homicide, Sex Crimes, Larceny, and Other Crimes, Cato Inst. (May 11, 2021), https://www.cato.org/immigration-research-policy-brief/criminal-immigrants-texas-2019?fbclid=IwAR24Xdm6VfbCHau9GcIa8j0KvbR0A7aoBU4rPJMAXBB259ojrTZrgvsRNzA [https://perma.cc/N5UL-37VY] (collecting citations to earlier studies).

      [211].     See Decline in ICE Detainees with Criminal Records Could Shape Agency’s Response to COVID-19 Pandemic, TRAC Immigr. (Apr. 3, 2020), https://trac.syr.edu/immigration/reports/601 [https://perma.cc/8WQZ-GCM9].

      [212].     See Wall of Honor, U.S. Immigr. & Customs Enf’t (May 11, 2022), https://www.ice.gov/topics/eow [https://perma.cc/79ZV-39Q9].

      [213].     See supra note 154. The case study is also broadly congruent with a legal literature on danger narratives in immigration enforcement, often drawing particular attention to the link between narratives of danger or security threat and race. See, e.g., Jennifer M. Chacón & Susan Bibler Coutin, Racialization Through Enforcement, in Race, Criminal Justice, and Migration Control: Enforcing the Boundaries of Belonging 159–75 (Mary Bosworth, Alpa Parmar, & Yolanda Vázquez eds., 2018).

      [214].     See Cox & Rodríguez, supra note 55, at 173 n.54 (alluding to circulation of personnel between ICE and CBP).

      [215].     Government Interview #3 at 6–7 (describing gradations between different federal law enforcement roles and suggesting that ICE ERO agents had an “inferiority complex . . . ICE is seen as this low-rung player in federal law enforcement”); Government Interview #4 at 8 (stating that “working with the [immigrant] population, in many scenarios, is not seen as prestigious” within ICE); Civil Society Interview #3 at 7 (sharing interviewee’s understanding from speaking to former officials “that ERO is not the sexy side of ICE”); see also infra note 232.

      [216].     See, e.g., ERO Physical Evidence Lesson Plan: Field Operations Training Program – Prosecutions, U.S. Immigr. & Customs Enf’t Acad., Enf’t & Removal Operations Training Div., Jan. 2013, at 4–5, 7, 10, https://s3.documentcloud.org/documents/3317930/ICE-Academy-Field-Operations-Training-Program.pdf [https://perma.cc/WT22-RX8R];
(“You are probably already obtaining and executing criminal warrants . . . [s]o why not have a complete understanding of the proper [f]ederal procedures as well as ICE police?”).

      [217].     See, e.g., National Fugitive Operations Program, U.S. Immigr. & Customs Enf’t Acad., Enf’t & Removal Operations Training Div., at 9, https://s3.documentcloud.org/documents/3317933/ICE-Academy-National-Fugitive-Operations-Program.pdf [https://perma.cc/M3W9-JQZT] (“[E]ach Field Office should seek to achieve a goal of at least 1,000 arrests per team.”).

      [218].     See, e.g., 4th Amendment Training (Refresher for Enforcement & Removal Operations Fugitive Operations), Off. of the Principal Legal Advisor, U.S. Immigr. & Customs Enf’t (Aug. 2017), https://mijente.net/wp-content/uploads/2019/07/ICE-4th-Amendement-Training.pdf [https://perma.cc/PVN7-XPTA] (“Ruses are permitted but must not be coercive.”).

      [219].     Civil Society Interview #2 at 2 (describing this as part of an “entrenched and ingrained workforce culture” among ICE line officers).

      [220].     See, e.g., Basic Tactics, U.S. Immigr. & Customs Enf’t Acad., Enf’t & Removal Operations Training Div., at 4, 5, 6, 9, 28, https://s3.documentcloud.org/documents/3317926/ICE-Academy-Basic-Tactics-Training.pdf [https://perma.cc/N8JX-PEFK] (referring to targets of field operations enforcement actions as “threats”).

      [221].     See, e.g., supra notes 133, 142, 154, 157 and accompanying text. ICE’s union, which has played a significant role in governance problems at the agency, will be discussed in more detail later in the Part.

      [222].     Civil Society Interview #2 at 4 (“ERO deport officers see themselves as cops. . . So how does that show up? I mean, the way it shows up is that if you’re a cop, you think your job is to arrest as many people as possible.”); Civil Society Interview #5 at 13 (“I don’t think they think of it as a civil agency. That’s the problem. I think they think of themselves as police.”); see also infra note 224.

      [223].     Government Interview #1 at 10–11 (“If you imagine any enforcement agency . . . like the FBI, the people who work there are going to be enforcement minded. . . It’s colored by their experiences, and their experiences are heavily influenced by where they work.”); Government Interview #3 at 2, 7 (describing ICE has having a “macho, machismo, bravado culture”); see also Civil Society Interview #8 at 12 (recounting that the prior assistant field office director “in charge” at a major Western detention facility where the interviewee works “would call himself the ‘alien hunter’”); Cox & Rodríguez, supra note 55, at 172 n.47, 173 n.54 (citing interviews with Obama Administration officials describing ICE field officers’ “mentality” as “paramilitary,” and noting that “many came from the Border Patrol, which had a ‘zero tolerance’ approach to enforcement”).

      [224].     Government Interview #3 at 3, 6 (“ICE’s zeitgeist, when it comes to detention, is they’re doing crime prevention. It’s like a Minority Report kind of thinking. . . [ICE officers] hated being likened to correctional guards. . . . [they] wanted to be seen as going out there and making arrests, preventing crime.”); see also Civil Society Interview #2 at 4 (“ERO deport officers see themselves as cops. . . So how does that show up? I mean, the way it shows up is that if you’re a cop, you think your job is to arrest as many people as possible.”).

      [225].     See, e.g., Government Interview #2 at 17 (relating that ICE union was able to block the Obama Administration from implementing new detention standards for a year and adding, “that’s the extent of the power they have”); Government Interview #3 at 5 (describing union role in blocking or delaying an array of new internal policies).

      [226].     Statement of Chris Crane, President, National Immigration & Customs Enforcement Council 118 of the American Federation of Government Employees, Before the U.S. Senate Committee on Homeland Security & Government Affairs, Comm. on Homeland Sec. & Governmental Affs. 3 (Mar. 22, 2017), https://www.hsgac.senate.gov/imo/media/doc/Testimony-Crane-2017-03-221.pdf [https://perma.cc/8GDW-D26Y] [hereinafter Mar. 22, 2017, Statement of Chris Crane]; Senator Grassley’s Questions for Chris Crane, ICE Union, Senate Judiciary Comm. 1–2 (Feb. 13, 2013), https://www.judiciary.senate.gov/imo/media/doc/021313QFRs-Crane.pdf [https://perma.cc/JLG3-5LCC].

      [227].     Mar. 22, 2017, Statement of Chris Crane, supra note 226, at 2.

      [228].     Id. at 5.

      [229].     Civil Society Interview #3 at 3–4 (recounting giving a pencil to a “nearly blind” detained asylum seeker at “the lowest risk classification” so the asylum seeker could sign a document, at which point “the AFOD [assistant field office director] screamed at me . . . in his words, that can cause a security problem, because those can be turned into shanks”); Civil Society Interview #5 at 11–12 (recounting that interviewee’s client, a 65-year-old woman, had been “chained to the bed” when she was taken to a hospital to be treated for a pulmonary embolism).

      [230].     Government Interview #3 at 3 (describing “belief” inside the agency that “I’m taking people off the streets, it’s helping the American public”); Civil Society Interview #4 at 7, 9 (noting that “on paper, more cleanly [ICE] say[s] compliance” with removal orders and immigration court is the purpose of detention and monitoring, but recounting a closed meeting where an ICE official justified it on “safety” grounds); Civil Society Interview #7 at 13 (recounting conversations with ICE officers “trying to justify to me that . . . they’re keeping their communities safe by locking up these bad people”).

      [231].     Civil Society Interview #5 at 13. The interviewee noted that the practice continues to this day.

      [232].     See Government Interview #3 at 11 (“The job nobody wanted to be in charge of was custody management. Because you dealt with all of that. You dealt with healthcare, you dealt with food, you dealt with hygiene, you dealt with all this stuff.”); Government Interview #4 at 8 (“If you’re at ICE, and you’re working for HSI, those are the folks who are doing investigations. They have some stature within the organization. The detention work – well, detention work, I mean, literally is relegated, you know, to the private providers or, or the county jails, wherever the population lands.”); Civil Society Interview #3 at 7 (“ERO also does not actually know how to run carceral spaces . . . in addition to not really being that interested in running detention”); see also Civil Society Interview #2 at 5 (characterizing ICE officers as having a “cop mentality” that “exclude[s] . . . anything having to do with social services [or] humanitarian protections”).

      [233].     Government Interview #3 at 6, 11.

      [234].     Government Interview #3 at 11; Civil Society Interview #3 at 7.

      [235].     See supra notes 100–101, 186–188 and accompanying text.

      [236].     See supra notes 146–149 and accompanying text.

      [237].     See supra notes 154–155, 157–158, 160, and accompanying text.

      [238].     Government Interview #4 at 4; see also Civil Society Interview #1 at 9–10 (describing ICE detention system as “such a decentralized system, where so much authority is delegated to the local level”); Civil Society Interview #3 at 5 (calling ICE’s “org chart” “chaos”).

      [239].     See Government Interview #1 at 7 (suggesting that more “centralized” decision-making was warranted at ICE); Government Interview #2 at 8 (stating that the latitude given field offices “created problems in terms of implementing headquarters policy, from my perspective. Policy only stuck as much as a field office director ensured that it happened”).

      [240].     See Government Interview #1 at 5 (“The field office directors traditionally are like little kings and queens. They rule their kingdom.”).

      [241].     Government Interview #2 at 6.

      [242].     “[I]t’s up to them to implement the guidance that’s given to them. It’s not really up to the ICE Director, per se, or even lots of the senior headquarters management.” Government Interview #1 at 5; see also Civil Society Interview #1 at 1–3 (describing massive variations in field offices as to whether they implemented parole policy and saying that “local leadership played a significant role” in differences); Government Interview #2 at 5, 7 (“A recalcitrant field office director, or one who didn’t want to make a change, wouldn’t let it happen. . . . Field offices really have a lot of authority around the country instead of headquarters.”); Government Interview #3 at 9 (describing struggles to secure effective implementation of policies at the field office level); Government Interview #4 at 8 (“[I]f lots of things are getting pushed to the field, then you’re not getting that policy direction.”); Government Interview #5 at 11 (suggesting that ICE would benefit from more standardization of policy across detention).

      [243].     See Government Interview #1 at 4.

      [244].     See Civil Society Interview #2 at 5–7 (reporting that the Biden Administration had shown “a real deference to the workforce” at ICE, that the current Secretary was “very swayed” by listening sessions, and that senior DHS leadership subsequently intervened to draft guidelines that “really allowed for anyone to justify any enforcement action”); Civil Society Government Interview #6 at 2 (confirming that listening sessions and workforce preferences “absolutely” affected revisions to enforcement guidelines).

      [245].     See infra note 259 and accompanying text.

      [246].     The same interviewee who suggested that ICE’s structure reflected its law enforcement role, and who spoke positively about the agency and its enforcement role, also noted that too much authority had been delegated to field offices. See Government Interview #1 at 7.

      [247].     See, e.g., Mar. 22, 2017, Statement of Chris Crane, supra note 226, at 4 (avowing that “our worst employees are the ones promoted to supervisory and leadership positions,” and noting that “employees have zero trust” in leadership); see also Government Interview #4 at 3 (suggesting that at ICE “labor” and “administration” “hardly ever otherwise agree on things,” except that “the positive outcome [in an immigration case] is removal”).

      [248].     See, e.g., National ICE Council, National Immigration and Customs Enforcement Council Applauds President Trump’s Withdrawal of ICE Director Nominee, PR Newswire (Apr. 8, 2019), https://www.prnewswire.com/news-releases/national-immigration-and-customs-enforcement-council-applauds-president-trumps-withdrawal-of-ice-director-nominee-300826487.html [https://perma.cc/3AYB-VFZT]; Rebecca Rainey, Vitiello Confirmation Vote Delayed over ICE Union Concerns, Politico (Nov. 28, 2018), https://www.politico.com/story/2018/11/28/vitiello-confirmation-vote-delayed-ice-1024583 [https://perma.cc/5UYC-NSZ7]; Brandon Darby & Ildefonso Ortiz, Official Who Released Criminal Aliens into U.S. for Obama Now Directing ICE, Breitbart (July 1, 2017), https://www.breitbart.com/border/2017/07/01/official-released-criminal-aliens-u-s-obama-now-directing-ice [https://perma.cc/G8RP-A7U9].

      [249].     See Crane v. Johnson, 783 F.3d 244 (5th Cir. 2015).

      [250].     See infra notes 254–258 and accompanying text.

      [251].     See supra notes 151–153 and accompanying text.

      [252].     See Julia Preston, Agents Sue over Deportation Suspensions, N.Y. Times (Aug. 23, 2012), https://www.nytimes.com/2012/08/24/us/agents-sue-over-deportation-suspensions.html [https://perma.cc/35GF-N8QN] (reporting that union President said constraints on enforcement discretion meant that “safety has just been thrown out the window”).

      [253].     Government Interview #2 at 17–18.

      [254].     Government Interview #3 at 5; see also Government Interview #2 at 17–18 (recounting that the union delayed adoption of new detention guidelines for a year); see also Cox & Rodríguez, supra note 55 at 173 n.55 (citing interview with former Obama Administration official for proposition that DHS leadership did not consult with field on enforcement priorities “in part due to concern that ICE union would be very frustrating to deal with”).

      [255].     See supra notes 160–164 and accompanying text.

      [256].     Damus v. Nielsen, 313 F. Supp. 3d 317 (D.D.C. 2018).

      [257].     Damus v. Nielsen, Case No. 18-578 (JEB), 2018 WL 3232515, No. 2, Class Complaint for Injunctive and Declaratory Relief, at 15–16 (D.D.C. Mar. 15, 2018).

      [258].     Id. at 14.

      [259].     See Cox & Rodríguez, supra note 55, at 166, 172–73.

      [260].     See Kate Evans & Robert Koulish, Punishing with Impunity: The Legacy of Risk Classification Assessment in Immigration Detention, 34 Geo. Immigr. L.J. 1, 57–62 (2021); Kate Evans & Robert Koulish, Manipulating Risk: Immigration Detention Through Automation, 24 Lewis & Clark L. Rev. 789, 817, 833–34, 838 (2020); See also Government Interview #4 at 6, 7 (describing evidence of tampering with risk assessment algorithm, ending at the point at which “there was only one recommendation, and that was detain. . . . ICE overmanages everybody, regardless of the risk assessment. You know, I mean, we just drive them crazy”).

      [261].     See supra notes 216–220 and accompanying text.

      [262].     Off. of Inspector Gen., Dep’t of Homeland Sec., OIG-17-51, ICE Deportation Operations 8 (Apr. 13, 2017), https://www.oig.dhs.gov/sites/default/files/assets/2017/OIG-17-51-Apr17.pdf [https://perma.cc/SJ8F-HK78] [hereinafter Off. of Inspector Gen., ICE Deportation Operations]; see also Off. of Inspector Gen., Management Alert, supra note 202.

      [263].     Off. of Inspector Gen., ICE Deportation Operations, supra note 262, at 7.

      [264].     Id. at 8.

      [265].     See Professional Employees Agreement: U.S. Immigration and Customs Enforcement and AFGE Local 511, Immigr. Pol’y Tracking Proj. (2019), https://immpolicytracking.org/media/documents/afge-ice-511-cba_05sep19---final.pdf [https://perma.cc/T4MR-L3J9]; see also Government Interview #3 at 5 (noting that large number of policy changes had to be run by the ICE union).

      [266].     See Gov’t Accountability Off., GAO-18-405, Department of Homeland Security: Components Could Improve Monitoring of the Employee Misconduct Process 10 (July 2018), https://www.gao.gov/assets/gao-18-405.pdf [https://perma.cc/3VR4-UDLZ] (“For cases involving misconduct outside of [the Office of Professional Responsibility’s] jurisdiction, generally including non-senior and bargaining unit employees—which represent most cases—local management proposes and decides discipline.”).

      [267].     Id. at i.

      [268].     See id.; see also Off. of Inspector Gen, Dep’t of Homeland Sec., OIG-06-57, A Review of Immigration and Customs Enforcement Discipline Procedures 8–9 (Aug. 2006), https://www.oig.dhs.gov/sites/default/files/assets/Mgmt/OIG_06-57_Aug06.pdf [https://perma.cc/SF82-CL25] (noting flaws in ICE’s discipline tracking procedures and management record-keeping); Off. of Inspector Gen, Dep’t of Homeland Sec., OIG-20-54, Special Report - ICE Should Document Its Process for Adjudicating Disciplinary Matters Involving Senior Executive Service Employees 7 (July 20, 2020), https://www.oig.dhs.gov/sites/default/files/assets/2020-07/OIG-20-54-Jul20.pdf [https://perma.cc/KC8G-2R95] (documenting allegations of unfair disciplinary practices for senior ICE officials).

      [269].     See Gov’t Accountability Off., supra note 266, at 26.

      [270].     See Adam Zagorin & Nick Schwellenbach, Protecting the Predators at DHS, Proj. on Gov’t Oversight (Apr. 7, 2022), https://www.pogo.org/investigation/2022/04/protecting-the-predators-at-dhs [https://perma.cc/EXP9-XRVM].

      [271].     See supra note 201.

      [272].     See supra note 278 and accompanying text.

      [273].     See supra notes 256–258 and accompanying text.

      [274].     See Memorandum from John Kelly, Sec’y of Dep’t of Homeland Sec., to Kevin McAleenan, Acting Comm’r, U.S. Customs & Border Prot. et al., Implementing the President’s Border Security and Immigration Enforcement Improvements Policies 1–2 (Jan. 25, 2017), https://knightcolumbia.org/documents/e4ece6adb2 [https://perma.cc/JBM3-TVNX].

      [275].     See also Civil Society Interview #1 at 1–2 (suggesting DHS was “in a bind” because it was relying on the Directive in Jennings v. Rodriguez to argue that immigrants did not need bond hearings).

      [276].     See Memorandum from John Kelly, Sec’y of Dep’t of Homeland Sec., to Six Senior Dep’t of Homeland Sec. Officials, Implementing the President’s Border Security and Immigration Enforcement Improvements Policies (Feb. 20, 2017), https://www.dhs.gov/sites/default/files/publications/17_0220_S1_Implementing-the-Presidents-Border-Security-Immigration-Enforcement-Improvement-Policies.pdf [https://perma.cc/5BGB-A73F] (suggesting that “the practice of granting parole . . . has contributed to a border security crisis” but not directly eliminating it).

      [277].     Jennings v. Rodriguez, 138 S. Ct. 830, 836 (2018).

      [278].     See supra note 258 and accompanying text.

      [279].     See Damus v. Nielsen, 313 F. Supp. 3d 317, 339, 343 (D.D.C. 2018).

      [280].     Compare, e.g., Order, A.S.M. v. Donahue, No. 7:20-cv-00062-CDL-MSH (M.D. Ga. June 3, 2020), ECF No. 94 at 17 (finding that ICE had “taken extensive measures to prevent the spread of coronavirus” and had “substantially complied with the CDC’s recommendations”) with Seth Freed Wessler, Fear, Illness and Death in ICE Detention: How a Protest Grew on the Inside, N.Y. Times (June 4, 2020), https://www.nytimes.com/2020/06/04/magazine/covid-ice.html [https://perma.cc/B8EW-2NVQ] (revealing sweeping violations of COVID-19 guidance inside the same Georgia facility the day after the judge’s opinion issued).

      [281].     Cf. Note, Stephen Hylas, Final Agency Action in the Administrative Procedure Act, 92 N.Y.U. L. Rev. 1644, 1666 (2017) (noting that in the context of policymaking, “the finality requirement creates incentives for agencies to strategically abuse the prongs of the Bennett test”).

      [282].     See FTC v. Standard Oil Co., 449 U.S. 232, 242 (1980).

      [283].     See Aracely R. v. Nielsen, 319 F. Supp. 3d 110, 122–23 (D.D.C. 2018) (noting that two of three named plaintiffs never received a decision on their parole applications but continued to be detained).

      [284].     See supra note 257 and accompanying text.

      [285].     See Ramirez v. ICE, 310 F. Supp. 3d 7, 20 (D.D.C. 2018).

      [286].     See, e.g., CGB v. Wolf, 464 F. Supp. 3d 174, 225 (D.D.C. 2020) (holding that suit over ICE transgender detainee policy, as a challenge to “continuing (and thus constantly changing) operations,” did not involve a final agency action) (quotations omitted); De Dandrade v. DHS, 367 F. Supp. 3d 174, 187 (S.D.N.Y. 2019) (finding that several plaintiffs had disavowed that they were seeking review of final agency action as to their individual waiver determinations).

      [287].     As Professor Emma Kaufman has noted in the constitutional law context, there are pitfalls to a normative argument for a domain-specific approach to administrative law. See Kaufman, supra note 51, at 214. As the examples of bespoke administrative law in Part III show, the people who are targets of violence are often not sympathetic to Congress and the courts (and sometimes not to the President, either). But the domain-neutral approach to administrative law is so counterproductive in these agencies that a domain-specific approach may still make sense here.

      [288].     Of course, one way to reduce or eliminate unauthorized state violence would be to shrink or eliminate the state agencies that carry out that violence. Partly for the reasons discussed infra, this Article assumes current policy arrangements that favor some use of state coercion, and makes proposals that would reduce harm within that context.

      [289].     Not agencies, in the sense this Article has generally talked about; but rather agency in the sense of autonomy and ability to take independent actions.

      [290].     For new theoretical work that engages with the structure/agency relationship from the perspective of a historical materialist social theory and that notes legal scholarship’s bias towards assuming agency, see Jeremy Kessler, Law and Historical Materialism, 74 Duke L.J. _, *49–51 (2024), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4745408 [https://perma.cc/U6AQ-6KDX].

      [291].     Cf. id. (noting both material and political reasons many critical scholars rely on a social theory of law that privileges agency, and arguing that these same factors cut against the widespread adoption of historical materialism, which privileges structure, and specifically the relations of production, as a causal explanation).

      [292].     Specifically, Weberian and post-Weberian theories of the state provide the scaffolding for this Article’s model of the relationships between law, state, and society (as when the Article argues that important components of the contemporary state are not bureaucratic in organization, or suggests that front-line administrators generate legal norms). Perhaps most salient of the Weberian assumptions here are that the state is just one form of social organization; that it can to some extent operate autonomously from the rest of society and the economy (and perhaps more in the domain of violence than elsewhere); that the monopoly of violence is one of its core capacities; and that state action displays certain structural regularities that mediate the possibilities for agency and contingency. For a helpful description (and critique) of some basic premises of Weberian state theory in twentieth-century American intellectual life, the social and political scientists who adopted them, see William J. Novak, Beyond Max Weber: The Need for a Democratic (Not Aristocratic) Theory of the Modern State, 36 Tocqueville Rev. 43, 63–80 (2015).

      [293].     I follow some sociological theorists in assuming that culture is part of social structure. See Sharon Hays, Structure and Agency and the Sticky Problem of Culture, 12 Socio. Theory 57, 65–66 (1994).

      [294].     And in fact, are core questions for any social theory of law to resolve. Like all the immediately above, this is a subject for future work.

      [295].     Indeed, the examples in this Part empirically demonstrate that there are significant constraints on what a domain-specific administrative law can accomplish. This Part draws heavily on the military as a comparator, suggesting obvious limits to the Article’s argument. Military administrative law clearly has a mixed record at preventing unauthorized violence, given the prevalence of sexual assault within these agencies and reports of prisoner abuse and unauthorized airstrikes in recent conflicts. See, e.g., Melinda Wenner Moyer, “A Poison in the System”: The Epidemic of Military Sexual Assault, N.Y. Times (Aug, 3, 2021), https://www.nytimes.com/2021/08/03/magazine/military-sexual-assault.html [https://perma.cc/UHX5-8QMS] (sexual assault); Dana Priest & Barton Gellman, U.S. Decries Abuse but Defends Interrogations, Wash. Post (Dec. 26, 2002), https://www.washingtonpost.com/archive/politics/2002/12/26/us-decries-abuse-but-defends-interrogations/737a4096-2cf0-40b9-8a9f-7b22099d733d/ [https://perma.cc/YQN3-KMR5] (reporting detainee abuse during detentions); Dave Philipps, Azmat Khan, & Eric Schmitt, A Dam in Syria Was on a “No-Strike” List. The U.S. Bombed It Anyway, N.Y. Times (Jan. 20, 2022), https://www.nytimes.com/2022/01/20/us/airstrike-us-isis-dam.html [https://perma.cc/96Y8-JG9Y] (unauthorized airstrikes). Many agencies in the domain of violence enjoy relaxed transparency requirements, but the military in particular is isolated from the rest of society, frequently exempt from the Freedom of Information Act, and the subject of highly deferential judicial review. See Chertoff & Bulman-Pozen, supra note 11, at 24–26, 35–37; Hathaway, supra note 121, at 15–20 (noting use of deference, political question, state secrets, and other doctrines to limit judicial inquiry into military and national security issues). This isolation limits what we can know about how well administrative law functions here. And military training and discipline also impose very high costs on administrators, granting the agency control of many aspects of their lives and behavior and allowing it to restrict their sexual, gender, and religious expression in ways that would be unconstitutional in other settings. See, e.g., Cook v. Gates, 528 F.3d 42, 65 (1st Cir. 2008) (finding Don’t Ask, Don’t Tell policy for LGBTQ servicepeople constitutional); Goldman v. Weinberger, 475 U.S. 503, 506–08 (1986) (finding that Free Exercise Clause did not prevent Air Force from barring Orthodox Jew from wearing a yarmulke while on duty).

      [296].     While this observation might seem obvious to a social scientist, it has deeper implications for the normative questions animating much administrative law scholarship. The structures in this domain, including many other aspects of legal doctrine applied to these agencies, tend to work against the standard, liberal norms of American administrative and public law. See infra Conclusion; see also Chertoff & Bulman-Pozen, supra note 11.

      [297].     See Tracey L. Meares & Tom R. Tyler, The First Step Is Figuring Out What Police Are For, Atlantic (June 8, 2020), https://www.theatlantic.com/ideas/archive/2020/06/first-step-figuring-out-what-police-are/612793/ [https://perma.cc/X4JB-EFXR] (suggesting reassignment of functions).

      [298].     See generally Stephen Holmes, In Case of Emergency: Misunderstanding Tradeoffs in the War on Terror, 97 Calif. L. Rev. 301, 309 (2009) (noting that through training to routinize reactions, “urgent threats that have appeared repeatedly in the past can be managed according to protocols that have become automatic”).

      [299].     Id. at 308–09.

      [300].     Id. at 315.

      [301].     See supra, e.g., notes 153, 155–156 and accompanying text.

      [302].     See Mashaw, supra note 32, at 155–66.

      [303].     See supra notes 213–224 and accompanying text.

      [304].     See supra notes 261–264 and accompanying text.

      [305].     See id.

      [306].     Marine Corps Order (MCO) 1553.1B, The Marine Corps Training and Education System 1 (1991), https://www.marines.mil/Portals/1/Publications/MCO%201553.1B.pdf [https://perma.cc/EHY4-AAA6].

      [307].     See Amelia Hoover Green, The Commander’s Dilemma: Creating and Controlling Armed Group Violence, 53 J. Peace Rsch. 619, 619–20 (2016) (noting that in armed conflict, the “Commander’s Dilemma” involves how to cultivate “restraint, rather than violence,” and noting that “institutions to control combatant behavior” in militaries and other armed groups “include recruitment processes, military training, political training, formal and informal socialization experiences, and disciplinary regimes”).

      [308].     See Colonel Kenneth M. Theurer & James W. Russell III, Why Military Justice Matters, 37 Air Force Rep. 7–99 (2010).

      [309].     See Howard Brotz & Everett Wilson, Characteristics of Military Society, 51 Am. J. Socio. 371, 371 (1946) (describing chain of command as one of the key social values communicated in basic training).

      [310].     See Advanced Individual Training, U.S. Dep’t of the Army, https://www.goarmy.com/careers-and-jobs/job-training/advanced-individual-training.html [https://perma.cc/GH2F-QMAS].

      [311].     See Path for Army Officers, U.S. Dep’t of the Army, https://www.goarmy.com/careers-and-jobs/find-your-path/army-officers.html [https://perma.cc/SL6T-HTDL].

      [312].     See G. James Herrera, Cong. Rsch. Serv., R46559, The Fundamentals of Military Readiness 1 (Oct. 2, 2020) (describing “military readiness,” including training, as a component of budget requests and staff development).

      [313].     See Robin S. Engel, Nicholas Corsaro, Gabrielle T. Isaza & Hannah D. McManus, Assessing the Impact of De-Escalation Training on Police Behavior: Reducing Police Use of Force in the Louisville, KY Metro Police Department, 21 Criminology & Pub. Pol’y 199, 207–08 (2022).

      [314].     Id. at 199.

      [315].     See Theurer & Russell, supra note 308, at 9 (“In the simplest sense, punishment is necessary to correct or modify the undesirable behavior of a person. From an organizational perspective . . . punishment serves a number of important purposes, including reinforcing behavioral standards, making an example of the violator, and maintaining the perception that the organization is a just place.”).

      [316].     See Barnett, supra note 101, at 1661.

      [317].     See supra section III.A.3.c.

      [318].     See Theurer & Russell, supra note 308, at 8.

      [319].     See 10 U.S.C. §§ 801–946 (the Uniform Code of Military Justice).

      [320].     See generally, e.g., U.S. Joint Serv. Comm. on Mil. Just., Manual for Courts-Martial United States (2019), https://jsc.defense.gov/Portals/99/Documents/2019%20MCM %20(Final)%20(20190108).pdf?ver=2019-01-11-115724-610 [https://perma.cc/E3WE-4PUR] (setting out substantive charges, procedures, and rules of evidence for courts-martial); U.S. Dep’t of the Army, Army Pamphlet 27-9, Military Judge’s Benchbook para. 1–1 (2020) (noting that the benchbook is designed to “assist military judges in the preparation of trial instructions” and outline their responsibilities); U.S. Dep’t of the Army, Army Pamphlet 27-7, Guide for Summary Court-Martial Trial Procedure para. 1–1 (stating that pamphlet is “a practical guide for officers detailed as a Summary Courts-Martial officer”).

      [321].     See, e.g., 10 U.S.C. §§ 801–946.

      [322].     See Theurer & Russell, supra note 308, at 10.

      [323].     See generally Monu Bedi, Toward a Uniform Code of Police Justice, 2016 U. Chi. L. F. 13, 13–16 (2016).

      [324].     See, e.g., U.S. Dep’t of the Army, Reg. 600-20, Army Command Policy 28 paras. 4–6 (distinguishing “nonpunitive corrective measures,” “nonjudicial punishment,” and [t]rial by court-martial”) [hereinafter U.S. Dep’t of the Army, Reg. 600-20]; see also U.S. Dep’t of the Army, Reg. 27-10 ch. 3 (setting out in detail Army procedures for nonjudicial punishment, penalties, suspension of punishment, and appeals).

      [325].     See U.S. Dep’t of the Army, Reg. 600–20, supra note 324, at 28 paras. 4–6.

      [326].     See Dhammika Dharmapala, Richard H. McAdams & John Rappaport, Collective Bargaining Rights and Police Misconduct: Evidence from Florida, 38 J.L. Econ. & Org. 1, 2–3 (2022). Compare also Bell, supra note 39, at 2131–36 (discussing how issues around police compensation translate into lower-quality policing in low-income areas).

      [327].     See, e.g., Max Schanzenbach, Policing the Police: Personnel Management and Police Misconduct, 75 Vand. L. Rev. 1523, 1525–26 (2022) (surveying evidence that police unionization imposes significant constraints on management); Anthony O’Rourke, Rick Su & Guyora Binder, Disbanding Police Agencies, 121 Colum. L. Rev. 1327, 1347 (2021) (detailing empirical evidence and past scholarship indicating that collective bargaining agreements are associated with higher levels of police misconduct and hamper accountability through investigative and disciplinary protections); Catherine L. Fisk & L. Song Richardson, Police Unions, 85 Geo. Wash. L. Rev. 712, 749–55 (2017) (exploring limits collective bargaining agreements place on police discipline); Stephen Rushin, Police Union Contracts, 66 Duke L.J. 1191, 1201–07 (2017) (describing sources of employment protection for police, including collective bargaining agreements, as a problem for police discipline).

      [328].     See supra note 225 and accompanying text.

      [329].     See supra note 254 and accompanying text.

      [330].     See supra notes 226, 249, 265–270 and accompanying text.

      [331].     See supra note 248 and accompanying text.

      [332].     See supra note 265 and accompanying text.

      [333].     See supra Section II.B.

      [334].     See supra, e.g., notes 2–89, 257–258, and many more examples.

      [335].     Compare Weber, Economy and Society, supra note 20, at 135–36.

      [336].     At a higher conceptual level, this is one of the most fraught questions in democratic theory, which must deal with an antinomy. No democratic theory can itself specify the group that should draw the boundaries of the political community without slipping into a regressus ad infinitum. But at the same time, a democratic theory requires that the boundary be drawn democratically. See Arash Abizadeh, Democratic Theory and Border Coercion: No Right to Unilaterally Control Your Own Borders, 36 Pol. Theory 37, 45–46 (2008). As a matter of non-ideal theory, however, it seems uncontroversial that individual administrators in this domain should not be freer to make unilateral decisions about community scope than bureaucratic administrators are in their (less democratically fraught) domain.

      [337].     See, e.g., H.R. Rep. No. 95-894(I), at 7 (1978).

      [338].     See, e.g., Fisk & Richardson, supra note 327, at 717–21; Benjamin I. Sachs, Police Unions: It’s Time to Change the Law and End the Abuse, On Labor (June 4, 2020), https://onlabor.org/police-unions-its-time-to-change-the-law [https://perma.cc/NJT4-82Q4].

      [339].     10 U.S.C. § 976.

      [340].     See Sachs, supra note 338.

      [341].     Id.

      [342].     See supra note 254 and accompanying text.

      [343].     Compare Dharmapala, McAdams & Rappaport, supra note 326, at 5–6 (suggesting that violent misconduct incidents increased within police agencies where the workforce gained a right to collective bargaining) with Felipe Goncalves, Do Police Unions Increase Misconduct? (Mar. 2021) (working paper), https://static1.squarespace.com/static/58d9a8d71e5b6c72dc2a90f1/t/60622724b 6a902732b636324/1617045285669/Goncalves_Unions.pdf [https://perma.cc/FNR3-MWR8] (finding no statistically significant increase in police misconduct, both fatal and nonviolent, with unionization).

      [344].     See Goncalves, supra note 343, at *3–4.

      [345].     See, e.g., Slobogin, supra note 56, at 134–51 (suggesting that APA should apply to police agencies, and specifically detailing how APA rulemaking process might apply in practice); Friedman & Ponomarenko, supra note 56, at 1832 (calling for rulemaking to govern police uses of force). For a reevaluation of these proposals, see Maria Ponomarenko, Rethinking Police Rulemaking, 114 Nw. U. L. Rev. 1, 5–6 (2019) (detailing a lack of fit between “policing’s various ills” and rulemaking as an administrative law tool, with particular attention to the incongruity between public-facing rulemaking and the internal character of the rules governing police).

      [346].     See, e.g., The President’s Task Force on 21st Century Policing, Implementation Guide: Moving from Recommendations to Action 16–17 (2015) (recognizing the structure and content of police training as core tools for police reform).

      [347].     See supra note 11.

      [348].     Walker, supra note 89, at 157; see also Kevin M. Stack, An Administrative Jurisprudence: The Rule of Law in the Administrative State, 115 Colum. L. Rev. 1985, 1987–89 (2015) (identifying and describing five rule of law values to be served by administrative law, namely authorization, notice, justification, coherence, and procedural fairness).

      [349].     See generally Bijal Shah, Administrative Subordination, 91 U. Chi. L. Rev. 1, 57–90 (2024) (exploring how the redesign of administrative agencies might promote anti-subordination).

      [350].     See, e.g., Driver & Kaufman, supra note 121, at 570–76 (criticizing existing prison law exceptionalism and suggesting that prison law “should be exceptional . . . because punishment is an exceptional act of state power”); Curtis Bradley, A New American Foreign Affairs Law?, 70 U. Colo. L. Rev. 1089, 1091–97 (1999) (enumerating elements of foreign affairs exceptionalism); Peter H. Schuck, The Transformation of Immigration Law, 84 Colum. L. Rev. 1, 1 (1984) (“Probably no other area of American law has been so radically insulated and divergent from those fundamental norms of constitutional right, administrative procedure, and judicial role that animate the rest of our legal system [as immigration law].”); Aziz Z. Huq, Against National Security Exceptionalism, 2009 Sup. Ct. Rev. 225, 225–26 (2009) (noting widespread belief that national security emergencies merit the application of exceptional legal norms).

      [351].     The relationship between rule of law and arbitrary state violence is important to an emerging literature on law and authoritarianism. My argument here suggests a distinctive perspective on the relationship between the rule of law and authoritarianism as a political phenomenon, an issue I plan to take up in future work. Cf. Weaver & Prowse, supra note 42, at 1176–77 (suggesting, against the focus of mainstream political science on authoritarianism in formal institutions, that the discipline devote more attention to bottom-up experiences of “authoritarian practices”).

      [352].     See Nishimura Ekiu v. U.S., 142 U.S. 651, 659 (1892) (finding power “inherent in sovereignty . . . to forbid the entrance of foreigners”).

      [353].     See U.S. v. Curtiss-Wright Export Corp., 299 U.S. 304, 320 (1936) (foreign affairs); Chae Chan Ping v. United States, 130 U.S. 581, 609 (1889) (immigration); Sarah H. Cleveland, Powers Inherent in Sovereignty: Indians, Aliens, Territories, and the Nineteenth Century Origins of Plenary Power over Foreign Affairs, 81 Tex. L. Rev. 1, 7 (2002) (grounding inherent powers jurisprudence in nineteenth-century case law).

      [354].     See Santiago Legarre, The Historical Background of the Police Power, 9 U. Pa. J. Const. L. 745, 777–79 (2007); Kathleen F. Brickey, Criminal Mischief: The Federalization of American Criminal Law, 46 Hastings L.J. 1135, 1138 (1995).

      [355].     See Curtiss-Wright, 299 U.S. at 316–17; Chae Chan Ping, 130 U.S. at 609–10.

      [356].     See Students for Fair Admissions v. Harvard, 600 U.S. __, *15, *22, *23–24 (2023).

      [357].     Id. (opinion of Thomas, J., concurring) at *26 (citing Grutter v. Bollinger, 539 U.S. 306, 326 (2003) (op. Thomas, J.)).

      [358].     See, e.g., Driver & Kaufman, supra note 121, at 536–37; Hathaway, supra note 121, at 15–20.

      [359].     See Trump v. Hawaii, 138 S. Ct. 2392, 2420 (2018) (subjecting religion-based regulation of immigration to rational basis review after finding that plenary power doctrine applied).

      [360].     The methodology I used here is relatively common across the study of immigration agencies and the administrative state. See, e.g., Cox & Rodríguez, supra note 55, ch. 6 (using interviews with ten DHS and Department of Justice sources along with other materials to explain how an enforcement discretion policy was implemented across both agencies over time); Shoba Sivaprasad Wadhia, Banned 133 (2019) (incorporating twenty-one semi-structured interviews, including six with former INS and DHS officials); Nicholas R. Parrillo, Federal Agency Guidance: An Institutional Perspective, Admin. Conf. U.S. (Oct. 12, 2017), https://www.acus.gov/sites/default/files/documents/parrillo-agency-guidance-final-report.pdf [https://perma.cc/4TS9-ANCM] (relying on unstructured interviews to understand how agency officials saw and used guidance).

      [361].     In addition to the interviews, I had brief, informal conversations with current and recent government employees who declined to participate in the study but shared information on background. I do not cite these conversations, but they broadly corroborated some of the themes in the on-the-record interviews.

      [362].     See Small & McCrory Calarco, supra note 184, at 102–04.

      [363].     See Kathy Charmaz, Constructing Grounded Theory: A Practical Guide Through Qualitative Analysis 42, 45–46 (2006).

      [364].     See id. at 48.