Vol. 31 No 1 (January) pp. 12-47

LAW AND LEVIATHAN: REDEEMING THE ADMINISTRATIVE STATE, by Cass R. Sunstein and Adrian Vermeule. Cambridge, Harvard University Press/Belknap Press, 2020. 208pp. Cloth $25.95. ISBN 9780674247536.

Reviewed by Paul Gowder, Northwestern University Pritzker School of Law. Email:

Sunstein and Vermeule offer, in LAW AND LEVIATHAN, a rule of law defense of American administrative law. Their interlocutors are a collection of scholarly and judicial critics of the American administrative state, such as Philip Hamburger, Gary Lawson, Richard Epstein, Judge Janice Rogers Brown, and Justice Neil Gorsuch. As that list suggests, Sunstein and Vermeule appear to be primarily concerned with defending the administrative state against libertarian/conservative (broadly speaking) critics. Sunstein and Vermeule label that school of administrative law critics “the New Coke” in reference to Edward Coke and the common lawyers’ resistance to Stuart power and to the tendency of some of those critics to equate the American administrative state to monarchical abuses like the Star Chamber. In this respect, it’s quite obvious that Hamburger (2014) looms large among those whom Sunstein and Vermeule are motivated to answer, as it is Hamburger who has focused most vigorously on the lessons to be gleaned from that period of English history.

The New Coke critics believe that the American framers had a vision of separation of powers drawn from Seventeenth Century English history, as well as from Montesquieu, and that the administrative state violates it. The touchstone of the New Coke, to the extent there is one, may be Madison’s statement in Federalist 47 that:

The accumulation of all powers, legislative, executive, and judiciary, in the same hands, whether of one, a few, or many, and whether hereditary, selfappointed, or elective, may justly be pronounced the very definition of tyranny. Were the federal Constitution, therefore, really chargeable with the accumulation of power, or with a mixture of powers, having a dangerous tendency to such an accumulation, no further arguments would be necessary to inspire a universal reprobation of the system.

For Hamburger, Lawson, and others, American administrative law is in fact tyrannical—it is tyrannical because Congress is permitted to delegate legislative power to agencies through the rulemaking process, and judicial power through the power of agencies to make binding determinations of citizens’ legal rights—thus indeed unifying the three powers of government in the single hand of the President in his role as the ultimate manager of the administrative apparatus.

Sunstein and Vermeule directly address the New Coke only in Chapter 1, and then only to very briefly articulate the New Cokes’ fear of unconstrained executive power and sketch a few points toward the notion that those fears do not accurately represent the views of the constitutional framers. They do not directly engage with the arguments of any particular one of the critics of the administrative state. Instead, the bulk of the volume is devoted to a presentation of Sunstein and Vermeule’s positive case for the lawful character of administrative law. [*13]

That positive case begins with Lon Fuller. In THE MORALITY OF LAW, Fuller articulated 8 requirements which he took to represent the “inner morality of law”—morally advantageous features of governance that were, to Fuller, also requirements for a form of governance to be law-like at all. These requirements—which we can broadly summarize as prospective rather than retroactive general rules which are stable, clear and public, not internally contradictory (and hence capable of being obeyed), and actually corresponding to what officials enforce—have often been pressed into service by later scholars as a conception of the moral ideal of the rule of law.

Sunstein and Vermeule, over several chapters, argue that existing administrative law does a reasonably good job of corresponding to the Fullerian conception of the rule of law. A variety of judicially created doctrines—doctrines which, they argue, are at best only loosely rooted in positive legal enactments such as the Administrative Procedure Act’s (APA) prohibition against arbitrary agency action or the Due Process Clause, and more rooted in judges’ internal sense of the morality of law—have been layered on top of the textual sources of administrative law in order to promote its conformity with Fullerian standards.

For example, nothing in the APA prohibits agencies from making rules that apply retroactively, but such retroactivity raises obvious concerns with respect to the basic normative criteria underlying the rule of law: people cannot obey retroactive legislation, and retroactive legislation can unsettle expectations about the legal system on which private actors had previously relied. Accordingly, the Supreme Court, in BOWEN V. GEORGETOWN UNIVERSITY HOSPITAL, 488 U.S. 204 (1988) announced a new rule of legislative construction: administrative agencies would not be allowed to engage in retroactive rulemaking unless Congress clearly and specifically authorized retroactivity by statute. Notably, the Court did not announce this rule as an interpretation of the APA, because, Sunstein and Vermeule argue, it isn’t one. Nor did the Court rely on the Due Process Clause. Rather, the Court relied on the bald declaration that “[r]etroactivity is not favored in the law”—a fairly direct appeal to something like Fullerian principles.

Sunstein and Vermeule make similar points about a number of other judicially-created administrative law doctrines: the requirement that agencies obey their own rules, the resistance to “telephone justice” (direct executive interference in agency adjudication), at least one decision prohibiting an agency (Amtrak) from regulating its competitors, and others.

Having defended that broad mode of thinking about judicially-created doctrines of administrative law, they sketch out some fairly uncontroversial limitations to it. Not everything that administrative agencies do really looks like legal regulation of private conduct, or the sort of thing that Fuller’s King Rex is trying to do; accordingly, not everything agencies do ought to be subject to the internal morality of law. The FCC handing out radio frequencies is the classic example. Moreover, even in the appropriate domain for Fullerian constraint, considerations of judicial competence and the practical limits of agency procedural capacity might impose external bounds on the reach of legal morality. Moreover, essentially welfarist considerations might make it prudent for agencies to sometimes regulate with more open-ended standard-like provisions rather than rule-like provisions; more abstractly, agencies must also serve morally important ends that go beyond Fullerian legal morality and may sometimes conflict with it. [*14]

They conclude by interpreting some recent decisions of the Roberts court as a kind of system of concessions to the New Coke critics of the administrative state, but concessions that maintain the broader structure of administrative law intact and simply reemphasize the Fullerian doctrines that hold the regime within the bounds of the rule of law. Thus, interpreting GUNDY V. UNITED STATES, 588 U.S. __(2019), a case concerning the retroactive administrative application of a sex offender registration statute to prior convictions, they argue that the Court took a textually open-ended delegation of power to the Attorney General and interpreted it to actually be a fairly narrow and purposive command to come up with a way to apply its registration statute to pre-convicted offenders posthaste. In doing so, Sunstein and Vermeule say, the Court largely defanged those who would resurrect the infamously moribund “nondelegation doctrine” to prevent Congress from granting lawmaking authority to agencies—but it did so using Fullerian principles. They also offer a defense of the controversial AUER and CHEVRON doctrines, under which the courts are required to defer to agency interpretations of, respectively, their own regulations and the statutes they are asked to administer, on the basis of the idea that the Supreme Court, in several recent cases, has indicated a willingness to maintain judicial scrutiny over administrative interpretations by policing those interpretations for, inter alia, consistency, good faith rather than pretext, litigation-independent origins, and relative modesty of subject matter (AUER V. ROBBINS, 519 U.S. 452 (1997) and CHEVRON U.S.A., INC. V. NATURAL RESOURCES DEFENCE COUNCIL, INC., 467 U.S. 837 (1984)).

Based on that evidence, Sunstein and Vermeule ultimately contend that, contra the New Coke, administrative law is not tyrannical. Appropriately guided by Fullerian judges, it can be the object of “celebration” as well as the impetus for “reform.” In short, they claim to have offered an interpretation (even, I dare say, in a kind of Dworkinian sense) of existing administrative law, drawn from serious observation of its doctrinal structure, which largely describes that doctrine in its best light, while retaining the capacity to generate an internal critique of particular ways in which that system might fall short of its own aspirations.

Such an interpretive project is always in some danger of being upset by reality. To be sure, an interpretative theory is robust to some deviation from the idealized system to which it is read to aspire, for no social institution always acts in accordance with its best self. But too frequent deviations from the ideal undermine the theory’s claim to be genuinely interpretive. In other words, Sunstein and Vermeule are entitled to look at the administrative state in its best light, and to say that some shortfalls from full-fledged rule of law morality are errors rather than evidence that the whole isn’t really guided by the ideal. But we can only seriously attribute to the American administrative state a kind of Fullerian aspiration (constructively speaking) if, when given the opportunity, the system fairly regularly acts in accordance with those aspirations. Otherwise, we are merely imposing external legal standards on a system that bears no serious resemblance to it—and the door to interpreting it not as law but as tyranny is reopened.

Moreover, the interpretive approach is subject to a second constraint in the form of a kind of conceptual partitioning. By this I mean that we might discover that the interpretation only really applies to a part of the thing interpreted. That is, suppose there is some domain within administrative law that generally does comport with Fullerian internal morality, but there is another discrete domain that systematically [*15] fails to do so—and suppose that both domains share the defining characteristics of administrative law. Those facts, if true, suggest that perhaps the internal morality isn’t a property of administrative law as such, but a property of the privileged subdomain of administrative law—and that the less privileged subdomain might just be left open to tyranny.

I intend to argue that Sunstein and Vermeule’s argument founders on both constraints. With respect to the first, their account of administrative law almost completely ignores adjudication, even though the adjudicative process is at the core of the Anglo-American conception of the rule of law—and of the rule of law worries about administration. With respect to the second, their interpretation of administrative law really is just an interpretation of post-New-Deal industrial regulation. By contrast, in those areas of administrative law where executive power touches on the wellbeing of discrete individuals and particularly where the federal government directly wields its violent power—with the core of that dark side of the administrative state largely centered on those domains encompassed by the Departments of Homeland Security and Justice—Stuart-style tyranny is permitted to run rampant.

Before we get there, however, a few brief remarks on the project as a whole. I must confess a certain degree of pre-reading skepticism for any rule of law account of administrative law by these two authors, whose previous intellectual record on the subject did not offer much cause for optimism. One of the authors is known for, among other things, the robust advocacy of an admittedly paternalistic (e.g. Sunstein and Thaler 2003) use of bureaucratic-style interventions to shape individual choices. Such a project is arguably quite anti-Fullerian, at least to the extent one accepts an interpretation of the moral basis of Fuller’s theory of legality that focuses on individual dignitary interests in autonomously applying the law to oneself (e.g. Waldron 2012). I have previously expressed some (not unqualified) reservations about the potential undemocratic and autonomy-undermining character of the “nudge” project (Gowder 2016a). The other author has become known, in recent years, for two positions which I confess to finding deeply troubling: an integralist rejection of liberalism as an organizing principle of the American political order (e.g. Vermeule 2017)—this despite the fact that the rule of law is unquestionably a liberal principle—and an abiding skepticism about the necessity or even possibility of legal constraints on executive power (Posner and Vermeule 2010). Vermeule has even argued that administrative law is unavoidably “Schmittian” and hence incapable of being brought within the bounds of ordinary legality, particularly in emergencies (Vermeule 2009). In light of the well-known difficulty with bounding even the determination of what constitutes an emergency by law (a difficulty we’re presently experiencing in the form of Donald Trump’s decision to pretend that there’s an emergency at the border entitling him to spend funds that Congress refused to appropriate to build his border wall), the Schmittian position seems to be entirely irreconcilable with the notion that there is an administrative internal legal morality at all.

So when I learned of a book by Sunstein and Vermeule which was to defend the proposition that the administrative state may be interpreted as within the rule of law, at least in a weak Fullerian sense, I confess to mixed feelings of dismay and hope—the former due to the previous work of both authors, and the latter due to the potential promise of a retreat from it. At least in part, the hope side won out. The core of Sunstein and Vermeule’s argument elucidating the way in which the [*16] courts have imposed a Fullerian morality on (some parts of) administrative law has a certain capacity to persuade—and maybe even to lead the reader to think that the scholars who wrote it might be brought back into the rule of law “liberal legalism” fold.

In some ways, the authors could take their argument even further than they do. While Sunstein and Vermeule explicitly disclaim the aspiration to do jurisprudence or to say anything useful about what “law” is (p. 42), I don’t think we can fully understand what the doctrines they describe are doing or really get the force of their overall argument without being more robustly Fullerian than Sunstein and Vermeule are willing to get. At least some parts of the body of administrative law they describe make the most sense not merely as requirements for good law but as requirements for a legal system to operate as such at all. And if that’s true, it may strengthen their overall case for a Fullerian administrative law, insofar as it suggests that judges seeing themselves as operating a legal system will necessarily apply such constraints.

A case for such a broader ambition is perhaps most obvious when we consider the requirement that an agency obey its own rules. Imagine that were not the case, but that two other basic propositions of administrative law were true: (1) agencies, when so authorized by Congress, have the power to make rules that have the force of law, and (2) courts are obliged to exercise the power of judicial review to overturn agency action “otherwise not in accordance with law” (5 U.S.C. § 706(2)(a)). What would have to be the case for those two principles to exist simultaneously with (3) agencies may disobey their own rules?

Consider a more concrete example: suppose the EPA enacts a regulation establishing a maximum amount of sulfur dioxide that a factory may spew into the air without violating the Clean Air Act. And suppose some factory pollutes less than that amount, but the EPA nonetheless imposes a penalty on the factory owner, and the owner turns to the courts seeking judicial review of that penalty. Sustaining the three claims that

(1) the EPA regulation has the force of law,

(2) a court may review agency action for its accordance with law, and

(3) the EPA may punish those whom its regulation classified as not punishable

is only sustainable if we make, at best, highly contestable claims about what it might mean for a regulation to be “law.” After all, the EPA regulation cannot be the thing that authorizes the agency to punish people who spew too much sulfur dioxide if the agency has the right to punish people for spewing sulfur dioxide regardless of what the regulation says. We’d have to understand the notion of “the force of law” as a kind of purely communicative threat rather than a speech act that changes an agency’s legal powers and the legal rights of those whom they regulate: an agency merely announcing to the public “right now, we’re inclined to punish you if you pollute more than this, but we might also punish you if you pollute less.” And, to be sure, we could interpret “the force of law” that way—but doing so would put unavoidable pressure on our conception of law itself, or at least on the kinds of law that agencies create.

In addition to this potential strengthening of their core claim, the book benefits from a kind of helpful commonsense perspective about the whole administrative [*17] law debate. The New Coke critics have unquestionably overplayed their hand in many respects. It is ludicrous to suppose, as libertarian critics of economic regulation so often do, that the Obama Administration’s administrative regulations smoothing out the bumps in the Affordable Care Act were somehow analogous to some kind of witches’ brew of John I, Charles I, and George III disposing of their enemies in Star Chambers and exacting ship-money from inland towns. As Sunstein and Vermeule (pp. 34–35) very aptly point out, such critiques, in addition to massively overblowing the tyrannical threats, also suffer from an inability to balance costs and benefits, risks and rewards: administration brings welfare benefits that also matter, morally speaking, and ought not to be just ignored.

Consider a recent example of the irresponsibility of some of the critics of the administrative state. In July 2020, 26 bar owners—represented by Arizona State University administrative law professor Ilan Wurman—filed suit against their governor contending that his order to close bars in order to prevent the spread of COVID-19 violated the state constitution because the statute permitting the governor to issue such emergency orders violated the Arizona equivalent of the nondelegation doctrine. Sunstein and Vermeule don’t comment on this lawsuit, but I take it that were they to do so, their comments would, quite rightly, begin with words to the effect of “oh come on!” Gubernatorial emergency orders to close bars, on the advice of health experts, in order to protect the public against a rapidly-spreading deadly respiratory virus, are hardly the appropriate venue for law professors to work out the chips on their shoulders about the nondelegation doctrine.

And yet, the administrative state does not merely look like the post-LOCHNER world of sensible industrial regulation and public health orders. Rather, the administrative state often directly disposes of individual rights in ways which are profoundly troubling from the rule of law perspective.

The Problem of Adjudication

In this respect, it is particularly striking that Sunstein and Vermeule say very little about adjudication. With the exception of a few words about the tension between notice and comment rulemaking and rulemaking by adjudication, and a few other words about “telephone justice,” or the direct interference by the President or his political officials in ongoing adjudication, the disposal, in the administrative state, of copious important individual interests by executive branch bureaucrats rather than by independent and neutral judges goes unexamined.

This inattention to adjudication is not just a weakness of Sunstein and Vermeule, but also of the broader Fullerian conception of the rule of law they adopt. While Fuller’s theory of legality hardly ignores adjudication—and some elements of his theory, like the requirement of congruence, absolutely entail constraints on adjudicative process—the core of his account at least in the MORALITY OF LAW is fundamentally legislation-forward. Fuller’s primary analytic device is the articulation of the constraints that a rulemaker must follow if the rules are to actually be effective at bringing about the behavior that the rulemaker intends. Such a focus puts the legislative process, not adjudication, in the driver’s seat. Unfortunately, in Sunstein and Vermeule’s hands, that framing takes the adjudicative side of administrative law almost entirely out of focus. [*18]

Where Sunstein and Vermeule do take up adjudication, they minimize it in a way that is both implausible and troubling. In the course of their discussion of “telephone justice” (pp. 82–83), they suggest, in part based on dictum in a footnote in a Supreme Court case involving the placement of cellphone antennas, that when administrative agencies engage in adjudication, they carry out “a core executive task,” namely, “the (preliminary) application of statutes to facts” (pp. 82-83). Later, they characterize administrative adjudication’s genuinely executive nature as “the traditional and mainstream understanding in American public law” (p. 128).

The problem is that this “(preliminary)” does an immense amount of work in insulating executive adjudication from serious analysis, by implying that even if that adjudication is grossly unfair, someone more responsible might take up the problem later. And it is simply false to suggest that administrative adjudications, as opposed to ordinary executive branch applications of law to facts in the classical sense, are preliminary. In normal (non-administrative) executive contexts, like criminal prosecutions, it is certainly true that the executive conducts a genuinely preliminary kind of application of law to fact, and, more importantly, preliminary finding of those facts in the first place. But the preliminary character amounts to the fact that it does so in the course of coming to its own conclusions about what the government’s position is, conclusions which must be defended in an ordinary adversary judicial proceeding with no thumb on the scale in favor of what the government happens to believe.

For example, when an alleged criminal is brought to trial, presumably the executive branch, embodied in its police and prosecutor’s offices, genuinely believes that the person charged is guilty of the crime. In that sense, the executive has preliminarily decided “that’s the person who did the deed” and decided that the deed as described meets the elements of the crime. Because the determination is actually preliminary, in order for it to have any legal effect on the individual thus condemned, it must be confirmed by an independent judge. Put differently, under any genuinely traditional conception of the difference between executive and legislative power, the executive’s application of law to facts is done only for the purpose of deciding who to bring before a court for judgment. (This, to be sure, is only traditional on American timescales—moving to pre-Constitutional times there has been longstanding theoretical confusion about the difference between execution and adjudication, and, after all, the English courts were long an annex of the Crown. Indeed, that history is precisely what Edward Coke helped overthrow.)

Yet in the administrative context often no actual court is to be found. The administrative process yields a judgment that, in the absence of an appeal, at least sometimes conclusively settles individual legal rights. And this obvious fact has been recognized in numerous Supreme Court opinions over many decades. (E.g. UNITED STATES V. UTAH CONSTRUCTION & MINING CO., 384 U.S. 394, 422 (1966); ASTORIA FEDERAL SAVINGS AND LOAN ASSOCIATION V. SOLIMINO, 501 U.S. 104 (1991)). No real court never needs to be involved, and the executive determination is not preliminary at all.

Moreover, in at least some administrative domains, even if an individual does make some kind of appeal to a court, they can be subject to quasi-judicial executive determinations which may barely be examined under a deferential standard by a court, if at all. This is particularly (but not exclusively) true in the case of factual [*19] findings. For example, fact-finding in Social Security adjudications is reviewed under a substantial evidence standard (42 USC § 405(g)).

All of this is wholly unlike genuinely preliminary and genuinely executive fact finding and legal-conclusion-reaching as it appears in traditional executive contexts like the criminal law. On the traditional model, where the judiciary comes to its own conclusions about the facts and their legal interpretation. Moreover, traditionally, the status quo is in the individual’s favor: the executive is obliged to incur the difficulty and expense of bringing a person before the court before doing him any harm. By contrast, on the administrative model, the executive can directly make rulings on a person’s legal interests, and those rulings become the status quo unless the individual affirmatively assumes the burden and expense of appealing it to some court, after already paying to defend themselves before the executive. And except in the cases where the Article III courts are given the right to review the executive decision de novo, by the time the poor individual reaches a court there’s already a built-in bias against them.

There are even some areas where Congress has entirely deprived the independent courts of even appellate jurisdiction over some executive determinations. For example, there are a number of determinations that “immigration judges” (i.e., DOJ lawyers without even the minimal independence protections enjoyed by the administrative law judges attached to other agencies) may make without any judicial review whatsoever. If a person shows up at the border without a visa, they can be subjected to a non-judicial “expedited removal” process, and the main way to avoid expedited removal is to claim asylum. If they claim asylum, an immigration official has to determine if they have a “credible fear” of one of the legally relevant kinds of oppression in their home country. Should that official think their fear is not credible, they may appeal to an immigration judge, and then… that’s it. No independent judge ever gets to rule on the question. Just this year in DHS V. THURAISSIGIAM, 591 U.S. __ (2020), the Supreme Court upheld this lack of access to the courts.

I have spilled so much ink on that single word in its pregnant parentheses “(preliminary)” because I suspect that the misconception according to which such proceedings are dismissed as “(preliminary)” and in that way assimilated to the genuinely preliminary applications of law to fact that police and prosecutors and the like do explains why defenders of administrative adjudication such as Sunstein and Vermeule have been so sanguine about the procedural irregularities that often attach to administrative adjudication. For example, political actors fairly routinely interfere with adjudication in immigration law. Consider an incident from 2002: annoyed with a backlog amounting to some 56,000 cases at the Board of Immigration Appeals, the Attorney General at the time ordered a speedup which caused adjudicators to dispose of appeals at a breakneck rate, in some cases, one every ten minutes. Unsurprisingly, given that highly-abbreviated review, this also led to an increase in the percentage of immigration judge cases that the board affirmed from 59% to 86% (Getter and Peterson 2003). Were those executive adjudications merely preliminary, that speedup wouldn’t be so bad: the affected individuals could go to the District Court and get a trial de novo and have the full protections of the judicial process. But that is not the system we have.

In the United States, there is a grim history to the dismissal as merely preliminary of kangaroo-court adjudications of fundamental individual legal interests. One argument offered in defense of the commissioner system of the Fugitive Slave Act [*20] of 1850 was that the proceedings before those commissioners—special quasi-judicial officers who were paid on an infamous bounty system, five dollars if they ruled against the claim of a putative slaveowner to a person who had been seized, and ten dollars if they ruled in favor of enslavement—were merely preliminary, and that if an alleged fugitive slave was actually free, they would have a subsequent remedy in the courts of the state to which they were kidnapped (Campbell 1970, 35-44). In reality, of course, there was a good argument that the proceedings of the commissioners were final de jure, and they were manifestly final de facto: once a person had been sold down the river they were highly unlikely to be able to have an effective judicial remedy, if for no other reason than because they’d have to escape from a plantation to get to the court in the first place. (Compare the current state of affairs in immigration law, in which those who appeal a removal order to a Circuit Court do not receive an automatic stay of removal while the appeal is pending; if they don’t realize they need to apply for one or aren’t granted one, they can be dragged off to another country, far away from the American courts and their lawyers and evidence, while their case is still in progress.)

Unsurprisingly, the commissioner system occasioned a constitutional debate similar to the contemporary debate about administrative adjudication. One of the challenges that abolitionists raised against the Fugitive Slave Act was that if commissioner decisions were final, they were really exercising the federal judicial power, and hence could only be appointed pursuant to Article III of the Constitution. In 1921, the Yale Law Journal published an article that flipped the whole debate on its head: according to the author, the progress of administrative adjudication up to that point was such that it had become clear that non-judicial officers were perfectly free to dispose of vital individual interests within the executive branch; ergo, viewed from the perspective of the administrative law developments of the period, the Fugitive Slave Act should retrospectively be viewed as constitutional (Johnson 1921, 181-182).

In what must be the ultimate example of the philosopher’s adage that “one person’s modus ponens is another’s modus tollens,” Johnson even trotted out the immigration adjudication process under the Chinese Exclusion Act as an exemplar of administrative adjudication which made the Fugitive Slave Act look perfectly normal:

To boards of immigrant inspectors, subject to appeal to the Secretary of Labor, belongs the decision of cases involving no less a matter than American citizenship. Chinese, born in the United States and subject to its jurisdiction, are citizens by the terms of the Fourteenth Amendment. Yet should such a citizen have the temerity to leave the country and then seek re-entrance, his constitutional rights would be determined by the immigration board without trial by jury. He becomes subject to the exclusive provisions of our immigration laws, according to the ruling of the Supreme Court.

The people of the United States have travelled far on the road to government by administrative commissions since the middle of the last century. So long as they acquiesce in the extraordinary determinations reached by administrative boards regarding rights of citizenship, they are likely to look with less and less passion upon the controversy which culminated in the assault upon the constitutionality of the Fugitive [*21] Slave Acts. Measured by the developments of a half-century and interpreted in the light of reason, these acts must be declared constitutional in every particular. (Johnson 1921, 181-182)

Immigration serves as a particularly important challenge to administrative adjudication because of its legally liminal character, and I shall be discussing it at egregious length in this review. In most immigration law, the United States maintains the fiction that immigrants are mere petitioners seeking a discretionary boon from the executive which may be granted or denied at whim—thus the lack of procedural protections attached to those decisions. In effect, under the plenary power doctrine, immigrants are outside the law—in the Hobbesian terms that the title of LAW AND LEVIATHAN invites us to use, immigrants are in a state of nature with respect to the United States. But immigration law’s claims to not really be law at all are somewhat dishonest: in particular, asylum claimants unquestionably have at least a skeleton of legal rights, since the United States has signed several treaties providing for same and then codified those treaties in statute. Hence, it serves as a kind of test case for the Fullerian morality of administrative law: if it breaks down there (and I contend that it does), we have some reason to suspect that administrative modes of governance in general might be more susceptible to the failure of legality than would otherwise seem to be the case.

Moreover, as Gabriel Chin (2002) has cogently argued, immigration law is closer to the center of administrative law as a whole than might be supposed from the limited attention paid to it by mainline administrative law scholars. Chin shows that the very idea that administrative adjudications can be final rather than “(preliminary)” is traceable to immigration innovations from the Chinese Exclusion era. Chin focuses on UNITED STATES V. JU TOY (198 U.S. 253 (1905)), a particularly horrifying case in which a U.S. citizen was deported on the basis of a final administrative adjudication which the dissent (foreshadowing Hamburger) characterized as “a star chamber proceeding of the most stringent sort.” As Chin demonstrates, JU TOY was cited in a number of early administrative law cases involving agencies like the NLRB and the FCC, and hence should be understood as a foundational precedent establishing the constitutional permissibility of final administrative adjudications.

However, immigration is not the only area where administrative adjudication falls drastically far from even minimal due process propriety. Richard Epstein (2019, 783–5, 787–8) describes openly stacked administrative adjudications in both securities and patent law. On Epstein’s account the SEC apparently gets to pick its own administrative law judges on a case-by-case basis when it charges people, and the politically appointed chief judge of the Patent Trial and Appeal Board (PTAB) apparently gets to add more judges to a case midstream if they don’t like the result toward which a case is proceeding. When I read Epstein’s article, I found both of these procedures so blatantly lawless that I—a person who knows nothing about either securities or patent law—simply couldn’t believe that he describes them accurately. But, as far as I can tell, he’s more or less correct. As to the PTAB, the panel-stuffing process is recounted in a reported federal court opinion (NIDEC MOTOR CO. V. ZHONGSHAN BROAD OCEAN MOTOR CO., 868 F.3d 1013 (Fed. Cir. 2017)). While the SEC’s regulations say that the Chief Administrative Law Judge is the one who picks the ALJs (17 CFR § 200.30-10), the SEC seems to reserve the authority to pick a non-ALJ adjudicator, or even adjudicate cases [*22] itself (17 CFR § 201.110); as well as to take away the authority of the chief ALJ to select an ALJ.

What About the Old Coke?

Sunstein and Vermeule’s inattention to adjudication partly explains the disjuncture between administrative law as they see it and administrative law as many of its critics see it. Hamburger, in particular, draws heavily on the English rule of law tradition, and that tradition has always been somewhat judge-centered. Consider that the Old Coke himself (I struggle mightily to refrain from calling him Coke Classic) was a judge before he became a parliamentary leader of the opposition to the Stuarts. And the rulings that got Coke fired from his judicial posts included a number of cases asserting the independence and in some cases superiority of the judiciary over its nominal royal master.

For example, Coke directly spoke to what Sunstein and Vermeule would call “telephone justice” in the Case of Prohibitions. James I had purported to take over the judicial role in a case relating to disputes over ecclesiastical authority. The Crown had asserted that “the Judges are but the delegates of the King, and that the King may take what causes he shall please to determine, from the determination of the Judges, and may determine them himself.” Au contraire, Coke said, “the King cannot take any cause out of any of his Courts, and give judgment upon it himself.” Similarly, in Peacham’s Case, Coke insisted that the King not consult his judges individually for their opinions on a case before them (in effect tampering with them), but speak to the court as a whole in its collective, properly judicial, i.e., independent, capacity.

Were the Old Coke brought forward through time to consider the American administrative state, his first complaint would doubtless be that we seem to have made little progress in protecting the independence of the judges. And matters may actually be getting worse. If we believe Justice Breyer’s hybrid concurrence-dissent in 2018’s LUCIA V. SEC, 585 U.S. __ (2018), the Supreme Court’s recent appointments clause jurisprudence has endangered the APA’s overall statutory program of insulating ALJs from political pressure—the next case may take away their civil service protections altogether.

A couple hundred years after Coke, the next classical representative of the English rule of law tradition was the constitutional scholar A.V. Dicey. One of the core features of the English rule of law, as Dicey saw it, was that both the punishment of ordinary citizens and the remedy for official misconduct was handled under ordinary law, and in ordinary courts (Dicey [1885] 1982, 110, 114, 120–1).

Another century down the road, a third great English articulation of the rule of law tradition, appeared in the work of historian E.P. Thompson, who hinted at the rationale for Dicey’s emphasis on the ordinary courts. Defending, against his fellow Marxists, the proposition that the law is more than merely a cover for class domination, Thompson observed that “[i]t is inherent in the especial character of law, as a body of rules and procedures, that it shall apply logical criteria with reference to standards of universality and equity… [i]n the case of an ancient historical formulation like the law, a discipline which requires years of exacting study to master, there will always be some men who actively believe in their own procedures and in the logic of justice.” (Thompson 1975, 262–3). Put differently, the process of becoming and serving as a generalist judge in a formalist legal [*23] system has a socializing effect: it leads judges to adopt the values of the legal system, where those values include the aspiration to deliver equal justice to all.

If Thompson is right about the socialization of real judges, then defenders of administrative adjudication need to ask themselves whether the bureaucratic organizations in which the adjudicators are embedded—oriented as they are toward policy implementation rather than neutral adjudication—can nonetheless manage to also orient their adjudicators toward legal justice. If they cannot, the classical English rule of law school stands behind the objections of those contemporary scholars, such as Hamburger, who draw on it to decry executive branch adjudication.

I’m not sure that Sunstein and Vermeule would agree with Thompson about the effective socialization of real judges. In criticizing Pojanowski’s “neoclassical administrative law,” they suggest that, at least in the context of CHEVRON deference, the legal realism genie cannot be stuffed back into the bottle; judges will never be fooled into thinking that there is a clean distinction between “law” and “policy” (pp. 112–14). But Thompson’s hypothesis rests on a kind of presumptive minimal judicial formalism. Insofar as ordinary judges are just thoroughgoing legal realists, then they will not perceive law as an autonomous domain of universal rules with its own moral standpoint, and hence cannot be the subject of a distinctive socialization process.

But I think that Sunstein and Vermeule might agree more with Thompson than they might want to recognize. For their entire argument seems to rest on the supposition that judges are not thoroughgoing legal realists. Time and again, they argue that some Fullerian feature of administrative law is not found in standard positive legal materials—not an implication of the Due Process Clause, for example, or the APA’s prohibition on “arbitrary action”—but instead imposed by judges because of their own senses of the law’s inner morality (e.g. pp. 50–56, 58–61, 67–70, 76–77, 80–81, 84–85, 86–87, 125). That’s just what Thompson and Dicey would expect from officials, and what the Old Coke exemplified. Article III judges are imposing legal morality on a deficient legal framework because they have been brought up and embedded in systems that reward them for acting like law is an autonomous domain with a distinctive morality. (And Sunstein, one of the leading expositors of behavioral science and bounded rationality in the law, surely can accept Thompson’s idea that officials who are rewarded for giving lip service to the autonomous morality of law are likely to come to believe it themselves.)

But if that’s so, then a system that systematically channels adjudication away from real judges—that disposes of individual legal interests in “(preliminary)” administrative proceedings by bureaucrats subject to interference from political officials and where those proceedings aren’t actually preliminary at all—puts them at serious danger of losing access to the benefits of legal morality on a day-to-day basis, even if Article III judges do have the capacity to sometimes impose that morality on the general processes of agencies from time to time in the ways that Sunstein and Vermeule recount. And this is a potentially fatal threat to any effort to claim some kind of legal morality for the administrative state.

I want to be careful not to overstate the point here. Our history also provides examples of administrative adjudication making up for judicial injustice. The most important example is the system of tribunals set up by the Freedmen’s Bureau at [*24] the beginning of Reconstruction as a substitute for state courts which refused to accept Black testimony. That example illustrates a point that Sunstein and Vermeule doubtless would happily accept, namely, that any superiority of judicial over administrative socialization is likely to be a contingent fact of the legal culture in which they are embedded. But Thompson’s argument still holds as to ordinary Article III judges and their state-level counterparts for the more normal case where an entire legal system isn’t corrupted, as it was in the former Confederacy. In general, Thompson’s argument suggests that ordinary judges are likely to be embedded in cultures of legal procedural justice. Whereas if Alexander Hamilton in Federalist 70 is to be believed, executive bureaucracies, which may value procedural justice, are also likely to value “[d]ecision, activity, secrecy, and despatch.” Such values, and the overall orientation of the executive toward policy articulation and implementation rather than neutral dispute resolution, are in tension with procedural justice. Examples of this tension include Epstein’s (2019) descriptions of rigged hearings in the SEC and PTAB and Jain’s (2019) interviews with immigration judges who said that they experienced their workplace as more like a bureaucracy than a court.

The Connection Between Realistic Forms of Possible Tyranny and Adjudication

As Sunstein and Vermeule point out, and as I have been emphasizing, the foundational fear of critics of the administrative state is a kind of executive “absolutism” or “tyranny” (p. 19). But if we were looking for tyranny in the administrative state, the quasi-legislative regulatory process would be a terrible place to find it—for exactly the reasons Lon Fuller (1964) laid out in THE MORALITY OF LAW. If agencies actually want to make regulations that bring it about that people do the things they’re commanded to do, they will only be effective if they do stuff like announce the rule in advance, publish it somewhere that the people whom it wants to do the obeying can find, more or less enforce it kind of as written, not change it every day, and so forth. This was, after all, precisely Fuller’s point: he didn’t claim to be giving a theory of some external constraint on lawful governance called “the rule of law,” but rather to be elucidating the moral character of the practice of regulating human behavior by rules. Assuming that our administrative agencies are more or less motivated to actually achieve their policy goals, Fuller would predict that they’d achieve a fair simulacrum of his morality in the regulatory process regardless of what the judges supervising them made them do.

To suppose that the Fullerian character of a regulatory process is any kind of remedy to tyranny, or to the kinds of affronts to the rule of law that are useful to tyrants, is to misconceive the kind of tyranny that a functional administrative state might threaten. Nobody in their right mind thinks that anything resembling the modern American administrative state has the immediate potential to threaten a sort of rule-less tyranny, where officials just operate by shouting arbitrary commands at people. The American administrative state just doesn’t work that way—it couldn’t—it’s so heavily bureaucratized and formalized and has so many complex moving parts and interrelationships between public and private actors that to layer such a model of tyranny on it would require totally disassembling the whole system and reassembling it on a different institutional foundation altogether.

Rather, the model of tyranny that the U.S. administrative state might be most plausibly said to threaten probably resembles Ernst Fraenkel’s (1941) famous [*25] analysis of Nazi Germany’s “dual state.” According to Fraenkel, the Nazi legal system was structured as both a “normative state,” which resembled an ordinary rule of law order and primarily operated in the economic domain, and a “prerogative state,” in which unbounded power could be exercised over the political domain. In other words, Germany could achieve the economic benefits of rule-governed behavior on the Fullerian model, while still keeping the capacity to have individuals shot for disagreeing with the dictator. Another classic example of the dual state is Pinochet’s Chile, and I have suggested elsewhere that China seems to have something like a dual state as well (Gowder 2020, 145–148).

Of course, no rational critic of American administrative law thinks that the rule of law threats it poses are on the level of Nazi Germany or Xi Jinping’s China or Pinochet’s Chile. But the word “prerogative” captures the heart of the potential threat even in its milder American-style version: the threat that notwithstanding a system of more-or-less lawful rules, the executive will have a kind of reserve power to exercise lawlessly in individual cases of the application of those rules. This is a key part of the prerogative that the Stuarts exercised: the King maintained a reserve capacity to barge into a courtroom of his choice and declare that in this one case, for reasons of state, the Crown would give the judgment, and that judgment would be motivated by the policy or personal interests of the person wearing the crown.

Thus, the Five Knights Case is typically taken as the pinnacle of Stuart tyranny. That’s a case that originated in Charles I’s efforts to procure “forced loans” from the populace. He demanded those loans from Thomas Darnell and four other knights. The knights refused, and the King ordered them thrown into a cell. In response to their habeas petitions, the King said that they were imprisoned as a bare act of his royal will. Scandalously, the court accepted that “justification” and refused to release the knights—leading fairly directly to the Petition of Right 1628, and thereafter serving as a major contributing factor to the escalation of the conflict between Crown and Parliament that ultimately resulted in the civil war and in Charles losing his tyrannical head.

Five-Knights-style prerogative tyranny doesn’t tend to show up in the regulatory process. It’s not the sort of thing we’re likely to see in administrative agencies using notice and comment rulemaking to change their interpretation of statutes, hence forcing courts to change their own interpretations under the CHEVRON doctrine. Rather, it’s something we’re likely to see in the application of law to individuals, i.e., in adjudication. When the King can show up in court cases to warp the application of law, take away the right of habeas corpus, dictate outcomes to the judges, and use that power to reach results unpermitted by the law or even to lash out against his political enemies—there is where we might find tyranny. And there is where critics of administrative adjudication like Epstein claim to have actually found some tyranny.

Moreover, administrative tyranny in adjudication could come in an even milder form than the Stuart kind. What we could call Full Stuart Tyranny would be the case where an agency just flat-out announces that it’s acting for reasons of state. For example, the EPA has a rule saying that the maximum amount of sulfur dioxide that a factory can spew into the air is X, some factory spews out some amount Y>X nonetheless prosecutes the owner of that factory, and, when the factory owner complains, simply says “actually, we have discretion to prosecute you regardless of what we said earlier.” Sunstein and Vermeule offer a pretty [*26] convincing argument that our Fullerian judges have built barriers protecting us against Full Stuart Tyranny, notably in their discussion of the requirement that agencies follow their own rules and not take advantage of Auer deference to willy-nilly change those rules and upset reliance interests in doing so (pp. 64-77).

But there’s also Weak Stuart Tyranny, which we can see as a less blatant version of the strong kind: a kind of tyranny in which agencies administer rules in a grossly biased way via things like policy statements and rigged adjudications—in which they don’t outright say “we’re going to violate our rules now,” but in which their interpretation of those rules in the spaces of unavoidable legal ambiguity or the application of relatively open standards subtly shifts on a case-by-case basis to accommodate the desires of the executive to strike out against individuals or to quietly massage the law to achieve novel policy outcomes on the backs of individual interests. This is what Richard Epstein (2019, 785–6) complained about with respect to the PTAB: it’s not that they violate their own regulations, but that they can quietly shift the outcomes of individual cases by mucking around with the judges—and in that way deal tyrannically with the individuals who come before them, who have their individual cases adjudicated not according to law but according to the whim of some unaccountable official.

As Epstein hints, mucking about with individual adjudications also can entail mucking about with the law—if a chief judge who really dislikes technology industry patents happens to be in charge, that judge can gently nudge the law in their preferred direction by making aggregate use of procedurally unfair adjudication to tinker with the results in individual cases. In other words, the agency can shift the legal landscape without ever openly violating its own rules, but simply by strategically deploying the capacity for biased judges to make motivated decisions within the domain of legal ambiguity. But the core wrong, from the rule of law standpoint, is not the policy-manipulating end (that’s more of a democracy complaint than a rule of law complaint) but the adjudication-twisting means—the subjection of individuals to biased and arbitrary decision-making on a case by case basis.

I like to think about the problem is in Kantian terms: when the executive can engage in policy-based adjudication, individuals and their legal rights are used as mere means for the executive to achieve external policy goals in an adjudicative process that is supposed to treat them as autonomous legal agents. But the same point can be made in the Fullerian morality that Sunstein and Vermeule invoke: Fuller’s entire conception of legal morality rests on the notion of a relationship of reciprocity between the lawmaker and the individual, represented as an autonomous rational agent, called upon to obey the law (Murphy 2005). Such a relationship is shattered if the executive can show up in the adjudication of an individual’s rights to make decisions (which may even retroactively alter those rights) based not on good-faith reasoning about the arguments offered by the individual in their case, but on an externally predetermined policy position to be achieved on the back of the individual. This is why when Fuller did turn to adjudication he insisted that, at least in the domain in which legal adjudication is appropriate, the essence of the process lies in its responsiveness to rational arguments by individual litigants based on claims of right (Fuller 1978, p. 364-8).

The reader may worry that Weak Stuart Tyranny is an insipid sort of tyranny. But before leaping to that conclusion, recall John Ashcroft’s 2002 BIA speedup. That directive led immigration adjudicators to dispose of thousands of appeals in [*27] summary fashion, and led, recall, to a marked increase in the proportion of immigration judge decisions that were affirmed. We don’t know whether those appeals affirmed favorable or unfavorable immigration judge decisions, but we can make a good guess: according to the Department of Justice Executive Office for Immigration Review’s Statistical Yearbook for 2001 (p. 12), immigration judges had ordered deportations in almost 75% of cases before them. So it’s likely that the overwhelming majority of the cases the BIA rubberstamped were deportations. Probably some of those people managed to get appeals in to the federal courts, but doubtless some did not—there’s always some dropoff at a new level of appeal.

In other words, hundreds—maybe thousands—of people who would not have been deported probably got deported because of Ashcroft’s speedup. In view of the fact that many immigration cases involve claims to asylum on the basis of the premise that a person is in danger of life and limb if they’re sent back to their oppressive home country, it is entirely reasonable to conclude that some unknown number of people were imprisoned, tortured, or killed when they were deported.

So the Attorney General’s decision to sacrifice fair appeals with his speedup likely killed an unknown number of actual human beings who will forever remain nameless. That sounds pretty tyrannical to me. Admittedly, the chain of inference I just took to get there is fairly long, and it could have gone astray—maybe the BIA mostly rubberstamped decisions that were favorable to the immigrants, for example—but if it did, that’s merely a contingent fact about the world; the capacity for such tyranny still visibly exists within the system.

At this point, we can sketch out something like a definition of the kind of tyranny that the American administrative state might realistically threaten. Define Weak Stuart Tyranny as the conjunction of:

1. the capacity for arbitrary executive action; and

2. social and political conditions that permit and encourage this arbitrary power to be used to systematically do unjustified harm to individuals (to “oppress”).

That conception of Weak Stuart tyranny is consciously in conversation with Vermeule’s prior work with Eric Posner (2010), which seems to me to provide the background conceptual framework underlying the notion of LAW AND LEVIATHAN that however the administrative state might fall short of the rule of law, it does not rise to the level of tyranny. In that earlier book, Posner and Vermeule argue that a Schmittian administrative state does not lead to tyranny because of the political constraints on the president. But Posner and Vermeule base that rosy conclusion on an unreasonably strict conception of tyranny—in their chapter on “tyrannophobia,” they equate tyranny to outright dictatorship. That’s far too narrow—executives can behave tyrannically in some respects even while being susceptible to elective and legal control in others. The obvious example is Jim Crow: Southern political leaders were electively accountable, and were obliged in most cases to follow the law, however, those legal and political constraints were utterly shattered when it came to Black citizens, who were subject to unrestrained tyranny by any reasonable definition of the word.

Posner and Vermeule were right, however, that the capacity for tyranny is subject to political control—the problem in the Jim Crow South was that Black Americans had been violently deprived of real political power. That point suggests that if [*28] we’re looking for tyranny, we should go looking first at the executive’s treatment of those who lack political power, for only there can arbitrary capacity be used oppressively without political consequences. It is no coincidence that Ashcroft’s tyrannical BIA speedup systematically did unjustified harm to immigrants, one of the most powerless groups of people in American society.

Actually-Existing Tyranny

Now that we have more of an idea what “tyranny” might actually be, it’s easier to notice the vivid impedance mismatch between the sorts of legal regimes (environmental regulation, labor regulation, patent law, securities regulation, etc.) on which the New Coke tends to focus and Stuart-style tyranny. It’s a little bizarre to imagine that there’s any kind of serious relationship between the administrative capacity to adjust the number of permissible parts per billion of sulfur dioxide that some factory can spew into the air, or even the power of the Arizona governor to shut down some bars for a few months to prevent people dying in a pandemic, and Charles I deciding to clap Darnell and the other four knights into prison for not handing over a forced loan. This kind of gratuitous overclaiming seems to me to be one of the key reasons that the New Coke hasn’t really caught on beyond the most committed libertarians.

But it is impossible to write this review, particularly in 2020, and not confront the fact that our government contains a substantial amount of what I shall call actually-existing tyranny to oppose it to the hypothetical, and largely theoretical, tyranny imagined by the libertarian critiques of industrial regulation to whom LAW AND LEVIATHAN is primarily addressed.

Consider an event that happened last summer. After the police killing of George Floyd, Black Lives Matter protests erupted across the country, and those protests became particularly conflictual in Portland, Oregon. If one believes the law enforcement narrative (I don’t, but we can assume it arguendo), the Portland protests went off the rails, lasting for an extended period of time and spilling over into opportunistic property destruction by anarchists who were there to create chaos rather than to protest injustice. Unfortunately, those alleged opportunistic anarchists did some damage to a federal courthouse, opening the door for Donald Trump to engage in an egregious display of force.

Against the wishes of the Mayor of Portland and the Governor of Oregon, the Department of Homeland Security deployed a SWAT/special forces team from the Border Patrol to Portland, and that team (“BORTAC”) (Pilkington 2020), maybe together with other Homeland Security agents, engaged in wildly lawless uses of force against protestors—for example, grabbing people off the streets, stuffing them into vans, covering their eyes so they could not see where they were being taken—effectively using military-style tactics to kidnap Americans off the streets of Portland, before releasing them without charge—no charges being available—as they lacked probable cause to believe that the specific people they had grabbed had committed any offense (Olmos, Baker, and Kanno-Youngs 2020). Their most notorious infamy, doubtless because of the sympathetic nature of their victim, is this: a 53-year old Navy veteran named Christopher David went to the protests for the specific purpose of speaking with the federal agents. He walked up to a police line and peacefully asked them to reflect on their compliance with the oath they took to defend the Constitution. In response, he was pepper-sprayed and savagely beaten, his hand broken, in an act of unrestrained aggression by federal agents [*29] against a peaceful citizen who was merely acting as a democratic citizen ought—remonstrating with the officials who broke the law in his name, and asking them to conform their behavior to that law (Ismay 2020).

This may seem somewhat orthogonal to the worries about administrative law, about Montesquieu and Coke, about separation of powers, that Sunstein and Vermeule are concerned to answer. But it’s actually deeply entwined with the administrative state for at least three reasons.

First, as a report by the Center for American Progress (Jawetz, Wolgin, and Flores 2020) describes, the administrative structure of the Department of Homeland Security directly facilitated these abuses: because the DHS, after 9/11, consolidated a variety of agencies with disparate missions into one organizational structure, it was easy for the lackeys of an authoritarian President interested in a show of force in a liberal city to have the BORTAC team cross-designated as members of the “Federal Protective Service”—the ordinarily milquetoast agency responsible for guarding government buildings—and then use damage to the federal courthouse as an excuse to send those troopers in to roam the streets looking for people to kidnap and beat.

Second, the ethical and legal dispositions of government officials are not separate from their organizational structure, and the organizational structure of America’s immigration and border agencies is notoriously lawless and brutal (Cohen 2020, ch. 1; Graff 2014). The isolation of CBP and its sister agencies within DHS from meaningful legal scrutiny and from any kind of culture of lawfulness or respect for individual rights, in a context in which those whom they encounter are represented as people without rights and subject to copious officially sanctioned brutality—from the large-scale, such as family separation policies, to the small-scale, such as their participation in the official destruction of lifesaving water supplies in the desert (Carroll 2018)—has, it’s impossible not to conclude, warped their characters and predisposed them to treating those with whom they interact in the interior as likewise also without rights. As Jacob Levy (2018) has explained, lawlessness at the border, where the state treats foreigners as enemies outside the protections of law, has the ineluctable tendency to leak into the interior, as enforcement strategies and personnel expand the scope of their operations. And this is precisely what happened. Because the BORTAC troops had been trained and socialized into the lawless culture created by the Schmittian black hole of the American border, they were readily available for a lawless president who wished to use them to make an example of some protesters.

Third, foundational federalism and separation of powers problems of the sort frequently highlighted by the New Coke likely gave the federal agents good reason to believe that they would be able to kidnap and beat American citizens with impunity. For all of the difficulty that abused citizens have experienced in holding ordinary police officers to account for their misconduct (particularly their racist misconduct), at the very least, ordinary police are subject to some degree of local accountability: state and local prosecutors can punish them for their illegal force, they can be readily identified and subject to civil rights lawsuits, and pressure can be brought to bear on democratically accountable local authorities to impose restrictions on their behavior and institutional avenues for accountability of individual officers such as civilian complaint review boards. By contrast, the federal agents who did the kidnapping and beating in Portland are embedded in a massive, secretive, democratically unaccountable federal bureaucracy established [*30] for “national security” purposes. If Mr. David wants to see to it that the criminal agents who broke his hand are punished, to whom can he turn? If the people of Portland want to exercise democratic control over that punishment, can they? Will the thugs who brutalized Mr. David ever see any consequences? Will the rest of us ever know? The Supreme Court hasn’t been terribly receptive to BIVENS suits in a long time. Just this year the Court refused to find a cause of action for the parents of a child whom the Border Patrol shot and killed across the Mexican border (HERNANDEZ V. MESA, 589 U.S. ____ (2020)). At most, if we’re lucky, perhaps we’ll see an Inspector General report—or at least the unclassified parts thereof, surrounded by a sea of redactions—a few years into the Biden administration that tut-tuts the actions of those agents and those who sent them.

In short, the administrative state enabled those abuses, which were manifestly tyrannical under any reasonable definition of the word. But the problem runs much deeper than that single incident. It’s no coincidence that at least those of the federal agents whom the press identified were from one of the nation’s immigration agencies. For the immigration regime in particular is blatantly tyrannical, and can stand as perhaps the opposite pole of the administrative spectrum from the milquetoast parts of the bureaucracy like the EPA.

Here’s an example to serve as grist for the mill of Hamburger’s next Star Chamber comparison. The Attorney General has the power to refer immigration cases to himself. In doing so, that official can, and does, engage in retroactive lawmaking which casually shunts aside the fundamental interests of individuals and the reliance interests they had in preexisting law for the sake of politically motivated policy changes. For example, in MATTER OF A-B- (27 I&N Dec. 316 (A.G. 2018)), Jeff Sessions overruled a 2014 decision of the Board of Immigration Appeals which had held that a victim of domestic violence could be eligible for asylum on the basis of a theory of group-based persecution. But A-B is a real human person, who fled El Salvador through pains and risks that readers (and the writer) of this review are unlikely to be able to imagine, and who only became in danger of losing her access to asylum in the United States because of the Attorney General’s policy disagreement with the previous administration, and hence his decision to take the case for himself, vacate a decision of the Board of Immigration appeals in her favor on the basis of the 2014 ruling, and order her case reconsidered without the benefit of the legal rule that provided the basis for her asylum claim.

In other words, the Attorney General directly intervened in a quasi-judicial proceeding to issue a binding legal ruling—manifestly motivated by the anti-immigrant politics of the President who appointed him—which retroactively changed the law to subject a human being to the high likelihood of deportation, and possible extreme violence or death. Matter of A-B- squarely fits within a reasonable definition of tyranny, and, again, a tyranny distinctively connected to the vices of the administrative state and its avoidance of real judges in favor of bureaucratic adjudication in policymaking entities. For Hamburger’s sake, it also looks rather like the royal conduct the Old Coke condemned in the Case of Prohibitions—and Sunstein and Vermeule would have us believe that telephone justice isn’t allowed?

The claim that there’s some kind of bar on telephone justice might be the least plausible proposition in all of LAW AND LEVIATHAN, in view of the longstanding practice of agency head adjudication. To be precise, however, Sunstein and Vermeule don’t deny that there’s such a thing as agency head [*31] adjudication—their discussion of telephone justice carefully refers only to the president as the official who is prohibited (or at least vaguely discouraged) from interfering in ongoing adjudications. But the immigration legacy of the Trump administration illustrates the way that such a limited restriction is thoroughly toothless: Trump’s Attorneys General did nothing more than directly implement Trump’s own extremist immigration preferences through their exercise of the self-referral power—what possible difference did it make if Trump himself couldn’t do it under his own signature? Given the popularity of the theory of the “unitary executive” among presidents, executive branch officials, and many legal scholars, it is cold comfort indeed to hear that the President might personally be discouraged from tampering with adjudications, but that it is perfectly ordinary for high political officials who report directly to the President to get to outright take them over.

Even when the Attorney General doesn’t directly interfere in the cases, immigration adjudication is notoriously unfair. Some evidence of this can be gleaned from the fact that when Article III judges peel back the curtain they often shriek in horror at what they observe. For example, Judge Posner, in 2016 (CHAVARRIA-REYES V. LYNCH, 845 F.3d 275 (7th Cir. 2016)), explained that the case before him was “a typical botch by an immigration judge,” which he described as “no surprise” because the immigration adjudication is run by “the least competent federal agency” (845 F.3d 275 at 280). That outburst was occasioned by a case in which an undocumented immigrant was not told that he could apply for voluntary departure rather than be forcibly removed—a legally important mistake because, after he was forcibly removed, he became ineligible to petition to return to the United States on the grounds of hardship to his American spouse. Moreover, the immigration court proceeding was drastically procedurally deficient: to hear Judge Posner tell it, the judge “never invited him to present evidence on his own behalf,” didn’t even believe that he had an American family even though the government apparently conceded that he did, and, as a whole, offered him a proceeding that was “cursory, indeed farcical”—as Posner said, “railroaded” him (845 F.3d 275 at 281).

Alas for Mr. Chavarria-Reyes: Judge Posner was writing in dissent. The majority concluded that he could have no relief, for he failed to exhaust his administrative remedies before showing up in the Circuit Court. To rub battery acid into the rule of law wound, the way he failed to exhaust his administrative remedies was not by failing to appeal to the BIA, but by failing to first make before the BIA the specific argument he relied upon in the Circuit Court—in other words, the court applied a wavier rule similar to the kind of rule that it would have applied had the appeal been taken from a District Court rather than from the executive branch. This case is a stunning example of how administrative adjudication isn’t preliminary, and doesn’t at all resemble an executive function in any normal sense of those terms: the Court of Appeals is treating it exactly like it would treat a ruling from an Article III judge, except that if a real judge had behaved as badly as the immigration judge did, they might have actually remanded the case.

It’s particularly notable that the dissenting judge who raised the red flag about immigration court railroading was none other than Richard Posner, the arch-realist himself, who nonetheless apparently could not shed his Thompson-esque formalist socialization far enough to permit the immigration adjudication apparatus to ignore the strictures of procedural justice. Actually, Judge Posner was particularly famous [*32] for his judicial impatience with executive immigration railroading. Another highlight is BENSLIMANE V. GONZALES (430 F.3d 828 (7th Cir. 2005)), in which Posner begins his opinion for the court with a litany of cases in which the Seventh Circuit had responded angrily to immigration court abuses, and an astonishing statistic: the court had reversed a full 40% of the cases that had come before it from the BIA in the previous year. In that case, Judge Posner identified a “tension between judicial and administrative adjudicators,” and attributed it to the fact that immigration adjudication “has fallen below the minimum standards of legal justice” (430 F.3d 828 at 829-30).

Perhaps the apotheosis of the lawlessness of the immigration regime is the process of “expedited removal.” In that process, which applies to an increasing number of immigrants thanks to recent expansions in its usage by the Trump Administration, an immigrant entering the United States—or, as of the new policies, at any time within two years of their entrance —can, with a few narrow exceptions, be removed from the United States on the basis of an alleged irregular entry or fraudulent documents without any kind of judicial or even quasi-judicial administrative process whatsoever, and, if they reenter the United States later on, they become guilty of a felony (Koh 2018, 8 U.S.C. § 1225(b)(1); 8 U.S.C. § 1326).

The courts have been unwilling, or, more to the point, unable, Fullerian instincts notwithstanding, to put a stop to this abuse. To be sure, our well-socialized judges seem to wish they could do something about it. But they can’t. For example, the Seventh Circuit (KHAN V. HOLDER, 608 F.3d 325, 329-330 (7th Cir. 2010)) has upheld the “expedited removal” process while decrying its inherent arbitrariness:

The troubling reality of the expedited removal procedure is that a CBP officer can create the § 1182(a)(7) charge by deciding to convert the person’s status from a non-immigrant with valid papers to an intending immigrant without the proper papers, and then that same officer, free from the risk of judicial oversight, can confirm his or her suspicions of the person’s intentions and find the person guilty of that charge. The entire process-from the initial decision to convert the person’s status to removal-can happen without any check on whether the person understood the proceedings, had an interpreter, or enjoyed any other safeguards. To say that this procedure is fraught with risk of arbitrary, mistaken, or discriminatory behavior (suppose a particular CBP officer decides that enough visitors from Africa have already entered the United States) is not, however, to say that courts are free to disregard jurisdictional limitations. They are not, and we thus must align ourselves with the courts that have considered the issue and hold that we lack jurisdiction to inquire whether the expedited removal procedure to which the Khans were subjected was properly invoked.

The Trump administration’s plans to radically expand expedited removal are permissible, according to the District of Columbia Circuit in MAKE THE ROAD NEW YORK V. WOLF, 962 F.3d 612 (D.C. Cir. 2020) because Congress has decided that the Secretary of Homeland Security has “sole and unreviewable discretion” (8 U.S.C. § 1225(b)(1)(A)(iii)(I)) to decide which immigrants are subject to the expedited removal process. [*33]

The usefulness of the administrative state to strip away individual legal protections has been known by immigration enforcers since well before the New Deal. As Salyer (1995) describes—in a book which, incidentally, has been heavily cited in legal history and in law and society, but, as far as I can discern, passed largely unremarked in administrative law, except by Chin (2002)—Congress and the executive pioneered the strategy of transferring individual adjudications from courts to bureaucrats in order to permit the use of arbitrary power to enforce the Chinese Exclusion Act regime at the dawn of American immigration restrictionism.

On Salyer’s account, Chinese immigrants were fairly successful in convincing judges to apply at least a shadow of a simulacrum of legality to the process of ruling on the claims of Chinese migrants to be entitled to U.S. presence on the basis of the handful of exceptions to exclusion. They did so by actually enforcing judicial standards of evidence and minimal standards of procedural justice. In response, according to Salyer, Congress stripped the courts of jurisdiction and shifted the determination of admissibility and lawful presence to an executive immigration bureaucracy.

Unfortunately, the jurisdiction-stripping strategy worked: the bureaucrats, qua devoted implementers of exclusion policy rather than neutral adjudicators, were blatantly biased against migrants and successfully deprived them of any real opportunity to prove the lawfulness of their presence. For example, one of the standards by which someone otherwise entitled to be present could be excluded was if they were “likely to become a public charge.” But, once the courts were removed from the picture, the bureaucrats in whom discretion was vested to apply the public charge provision quite openly interpreted it in a deliberately broad fashion, and bragged of the number of successful exclusions they could achieve with this technique (Salyer 1995, 153–6). Much like today, the judges, unwilling to themselves dirty their hands with operating procedurally deficient immigration hearings, were perfectly happy to permit the bureaucracy to do the government’s dirty work.

In some sense, Salyer’s story is consistent with Sunstein and Vermeule’s: when Congress allowed them in the room, the judges did impose a kind of minimal legal morality on the process. And sometimes, even today, the real courts manage to wrest some minimal legal justice out of the immigration system. For example, in INS V. ST. CYR, 533 U.S. 289 (2001), the Supreme Court used aggressive strategies of statutory interpretation to narrowly read Congress’s intent to strip the courts of habeas jurisdiction in some immigration cases, declaring that “Congress must articulate specific and unambiguous statutory directives” to strip away habeas jurisdiction and drawing on the canon of constitutional avoidance to conclude that it had not done so. The Court reached this ruling despite the fact that the section of the statute it was interpreting was entitled—no kidding—“ELIMINATION OF CUSTODY REVIEW BY HABEAS CORPUS.”

But at a deeper level, Salyer’s history and expedited removal reveal the fundamental weakness in defenses of the administrative state without a robust separation of powers doctrine to ensure that the judicial role is protected. Much like the New Coke scholars have feared, in the absence of constitutional separation of powers protections or the kindness of Congress to supply some minimal procedural protections the thing that comes out of the other end of executive power is our tyrannical immigration system. Put differently, Sunstein and Vermeule claim [*34] that the Fullerian administrative rule of law is not traceable either to the Due Process Clause or to the APA, but to the judicial internalization of principles of legality. But if that’s true, then immigration law ought to be the perfect test case, for many of the core protections of APA do not apply in immigration court, and, thanks to the infamous plenary power doctrine, the Due Process Clause barely applies. If Sunstein and Vermeule are right, we ought to nonetheless see judges enforcing antityrannical legal morality in immigration law. But we don’t. The judicial unwillingness to defend the separation of powers has meant that when Congress and the President insist on creating a lawless immigration apparatus in the executive branch, with due process and the APA unavailable, the courts have found themselves with little left with which to defend individual rights.

The Attorney General’s Three Bodies

Given what I have said thus far, the reader will hopefully agree that it is more than merely coincidental that the circuit-level opinion to which opponents of administrative law pointed as their great hope for Justice Gorsuch to dismantle the entire apparatus when he took his Supreme Court seat came in an immigration case. In that case (GUITIERREZ-BRIZUELA V. LYNCH, 834 F.3d 1142 (10th Cir. 2016)), Gorsuch took the somewhat bizarre step of writing a concurring opinion to his own majority opinion, expressing his skepticism about the Chevron doctrine (and the even more dramatic BRAND X doctrine, according to which courts have to overrule their own precedential rulings in response to agency changes in statutory interpretation).

GUITIERREZ-BRIZUELA V. LYNCH is a fairly complex case. It involves the reconciliation of two immigration statutes which the court delicately described as existing in “tension”: the first which provides that people who enter the country illegally more than once are barred from being considered lawful residents until they’ve spent ten years outside the country; the second which provides the Attorney General with discretion to “adjust the status” of persons who are unlawfully present. The Tenth Circuit had read the two statutes together to conclude that the Attorney General retained the discretion to adjust status even of those who were subject to the ten-year bar. Two years later, however, the Board of Immigration Appeals issued a ruling concluding, “as a matter of policy discretion” (in the court’s words, 834 F.3d 1142 at 1144) that the two statutes should be read together to divest the Attorney General of discretion to adjust status for such immigrants.

Had the story stopped there, Justice Gorsuch would have been forced to affirm the BIA, for BRAND X establishes that an agency does, in fact, get to overrule court decisions when it disagrees with their interpretation of statutes. But the BIA wanted to go further: they wanted to apply this new ruling retroactively, to applicants whose cases arose before the agency made its decision reinterpreting the regulation. The Tenth Circuit, applying what Sunstein and Vermeule would doubtless (and rightly!) call the internal morality of administrative law, said no. Then—and this is how we get to GUITIERREZ-BRIZUELA—the Executive tried a new gambit: it wouldn’t apply the change in doctrine retroactively to the period before the BIA’s decision, but they would apply the Tenth Circuit’s reluctant concession of the change in doctrine retroactively to the period after the BIA’s decision but before the judicial concession. Are you still with me? GUITIERREZ-BRIZUELA was the case in which the Tenth Circuit shot that one down too. [*35]

In many ways, GUITIERREZ-BRIZUELA actually stands as a redemption of the Sunstein/Vermeule idea that judges can, and do, impose a kind of legal morality on the administrative process. Justice Gorsuch nicely articulates core rule of law concerns in his sublimely weird self-concurrence:

Transferring the job of saying what the law is from the judiciary to the executive unsurprisingly invites the very sort of due process (fair notice) and equal protection concerns the framers knew would arise if the political branches intruded on judicial functions. Under CHEVRON the people aren’t just charged with awareness of and the duty to conform their conduct to the fairest reading of the law that a detached magistrate can muster. Instead, they are charged with an awareness of CHEVRON; required to guess whether the statute will be declared “ambiguous” (courts often disagree on what qualifies); and required to guess (again) whether an agency’s interpretation will be deemed “reasonable.” Who can even attempt all that, at least without an army of perfumed lawyers and lobbyists? And, of course, that’s not the end of it. Even if the people somehow manage to make it through this far unscathed, they must always remain alert to the possibility that the agency will reverse its current view 180 degrees anytime based merely on the shift of political winds and still prevail. Neither, too, will agencies always deign to announce their views in advance; often enough they seek to impose their “reasonable” new interpretations only retroactively in administrative adjudications (834 F.3d 1142 at 1152).

In addition to highlighting a number of important rule of law issues, that passage also quite nicely captures the contrast between the sympathetic core of the administrative state—the part where it merely regulates the industrial activity of entities like Chevron that do indeed have an access to “an army of perfumed lawyers and lobbyists”—and its tyrannical periphery where people might have their lives uprooted or be sent away to potential death based on their inability to discern that an agency might retroactively reinterpret the law. (Why the perfume, though?)

Gorsuch, Fullerian jurist that he is, managed to moderate the executive’s capacity for retroactive tyranny as to that one lucky immigrant. In that sense, the case is in accord with how Sunstein and Vermeule would expect our judiciary to act. But any comfort we might take at knowing that Neil Gorsuch is standing at the breach to protect us from the executive must be limited; even as to the retroactive interpretation at issue in GUITIERREZ-BRIZUELA, other circuit courts came to the opposite conclusion (e.g. GARFIAS-RODRIGUEZ V. HOLDER, 702 F.3d 504 (9th Cir. 2012)). And the underlying legal issue demonstrates the troubling ways in which administration can warp ordinary legality.

The underlying BIA decision rests on an absurd chain of legal fictions of a sort that can only arise in the administrative state. Let’s recall that the entire immigration adjudication system is nothing more than an extension of the Attorney General in the first place—that’s why Jeff Sessions could assign cases to himself and adjudicate them as an act of royal prerogative, and why John Ashcroft could order the BIA to do a speedup and decide a case every ten minutes. Now just contemplate the sheer bizarreness of a “judicial body” that really just operates as a [*36] bureaucratic arm of the Attorney General making a decision, as an exercise of policy discretion, to interpret a pair of statutes to divest the Attorney General of bureaucratic discretion. It’s all really just the Attorney General all the way down, but buttressed by a cascade of legal fictions that make it expedient for the Board of Immigration Appeals, with a homunculus of the Attorney General inside, to make a ruling commanding immigration judges, encompassing yet another Attorney General homunculus, to no longer exercise discretion to adjust the status of people subject to the ten year bar, but to do so in a quasi-legal form in which it is made to look like the BIA is actually making rulings about the powers of the full-sized Attorney General standing in the background—the Board’s boss, remember—and in virtue of having dressed up a use of discretion as a legal-ish limitation on discretion, acquire the power to force the Tenth Circuit to undo its own legal rulings.

In such a context, one really can’t blame then-judge Gorsuch for perceiving yet another little Attorney General climbing into the shell of the judiciary as well. And I cannot help but feel a certain sympathy for Hamburger’s comparisons of American administrative law to the bizarre and arcane judicial procedures of unreconstructed Stuart England. Although, actually, all this business with the Attorney General swapping hats (or, I guess, wigs) at lightning speed to do legislation in her judicial capacity via her appellate board as an exercise of policy discretion to strip herself of discretion in her bureaucratico-judicial capacity via her street-level officials reeks perhaps less of the Star Chamber and more of the Privy Council. The King, poor creature that he is, only has two bodies (Kantorowicz and Leyser 2016), but it appears that the Attorney General has at least three. And then if we swallow the unitary executive theory (we shouldn’t), all the Attorney General’s bodies get added to the thousands that the President must have anyway.

Such a proliferation of topsy-turvy legal fictions by the government can often serve as a clue that allows us to detect actually-existing tyranny. Outside of administrative law, my favorite example, which goes at least as far back as the colonial vice-admiralty courts that John Adams and other revolutionary leaders so loathed, is the use of in rem jurisdiction to seize property without providing real procedural protections to its owners, a tradition that continues today in civil asset forfeiture cases captioned something like “U.S. vs. $50,000 in unmarked bills.” And, of course, I cannot mention tyrannical legal fictions without referring back to the incident noted above in which the government turned Border Patrol special forces troops into Federal Protective Service security guards, and then turned the streets of Portland into a federal courthouse, a clever bit of virtual badge-swapping and jurisdiction extending which directly led to people getting stuffed into unmarked vans. Here’s one more border-related legal fiction: the Border Patrol famously claims—via Congressionally unscrutinized regulation, naturally—the authority to operate within 100 miles of any border, and claims the authority to do things like operate suspicionless immigration checkpoints anywhere in that territory. “Border,” however, includes things like Lake Michigan—and so the ACLU estimates that some 2/3 of the U.S. population lives in the Border Patrol’s bizarre border zone (“ACLU Factsheet on Customs and Border Protection’s 100-Mile Zone”). [*37]

There’s a wisdom in that popular strain of thought that fears excessive legal sophistication (maybe that’s what Justice Gorsuch was going for with the perfume?), for laypeople are savvy enough to recognize that when the government starts pretending that a suit against a person is really a suit against property, or that the streets of Portland are really a courthouse, or that the delightfully named town of Normal, Illinois, is really part of the border in virtue of being just within 100 air miles of the tippy-toe of Lake Michigan, there is almost certainly some nasty business going on. Sunstein and Vermeule’s main interlocutor, Hamburger, also identifies the danger of such fictions; for example, he points out that Stuart-era prerogative tribunals got away with compelled testimony on the fiction that their “proceedings were not really criminal in nature” (Hamburger 2014, 160). Likewise, at least since the Chinese Exclusion era, the Supreme Court (FONG YUE TING V. UNITED STATES, 149 U.S. 698 (1893)) has maintained the fiction that deportation is not a criminal penalty, and consequently has permitted endless abuses that would not be permitted in the criminal process, such as the gross non-neutrality of the “judges” and a widespread lack of counsel, even as deportation (a) is frequently imposed as a consequence of crime, (b) triggers later criminal consequences like prison time for reentering, and (c) is accompanied by the burdens of the criminal system, most notably lengthy detention in harsh conditions (Stumpf 2014).

But in the case of GUITIERREZ-BRIZUELA V. LYNCH, the aversion to extreme legal fictions that Gorsuch, Hamburger, and I seem to share actually conceals an answer to the puzzle that Sunstein and Vermeule can help us solve if we examine the one place in LAW AND LEVIATHAN where immigration law makes an appearance. To wit, a discussion of the ACCARDI V. SHAUGHNESSY, 347 U.S. 260 (1954) case (p. 66), which held that an agency had to follow its own procedural rules in a deportation. ACCARDI is actually quite striking, and serves as an exemplar of the emptiness of the Fullerian morality of law as applied to immigration cases. The Attorney General had been delegated discretion by statute to suspend deportations, and had in turn delegated that discretion by regulation to the Board of Immigration Appeals. Mr. Accardi applied for such a suspension, but the Attorney General personally—rather than through the BIA—put Accardi’s name on a list of “unsavory characters” who were to be deported. The BIA obediently declined to suspend the deportation. The Supreme Court, however, held that when the Attorney General promulgates a regulation delegating some of his discretion to the Board of Immigration Appeals and requiring them to exercise their own judgment in using it he couldn’t just turn right around and exercise that discretion himself over their heads.

The holding in ACCARDI sounds nice and lawlike, considered on its own. And it might contribute to an explanation of why the Attorney General couldn’t just send a memo declaring that adjustment of status discretion would henceforth no longer be exercised in favor of people subject to the ten-year bar: it had to be dressed up in some kind of legal act to get past the ACCARDI principle.

But notice how empty this restraint is. The bare Fullerian proviso that an agency must obey its own rules does not prevent it from writing rules that carry with them near-limitless discretion on behalf of high-level officials. It merely has to write those rules down first. Thus, notwithstanding ACCARDI, under the existing rules the Attorney General still has the power to engage in ex cathedra deportations—the precise behavior that the ACCARDI court disapproved—just so long as the [*38] Attorney General utters some magic words to the effect of “I hereby refer this case to myself” first. And the Attorney General can still procure a systematic increase in deportations by imposing unrealistic case-completion standards on adjudicators through the mechanism of personnel management. It’s hard to see the rule applied in ACCARDI as anything like a limitation on the arbitrary discretion of executive officials when it may be so trivially evaded in the very context in which it arose.

Toward An Egalitarian Administrative Rule of Law

The structure of administrative law seems to me to be but one piece of a three-part formula that, when taken together, leads to the danger of tyranny:

1. The conjunction of legislative, executive, and judicial powers within agencies that may exercise them together in the pursuit of policy or even nakedly political goals (this is the administrative law part);

2. The vesting of those agencies with the power to affect fundamental interests of vulnerable individuals—particularly, but perhaps not exclusively, in agencies whose mission is to directly wield state violence against natural persons (Border Patrol, not PTAB); and

3. The ultimate control of those agencies by one person who may, with greater or lesser degrees of legal formality, singlehandedly formulate policy and command those agencies to execute it (or, in the words of legal scholars, “the unitary executive”).

When you combine those three features, it’s easy to see where Charles I and George III stand: a president may decide that he dislikes Muslims, and hence may decree that immigrants from a bunch of predominantly Muslim countries are not allowed; he may decide to deter the exercise of the lawful right of asylum by taking helpless children, separating them from their families, and locking them up in cages; or he might decide that he dislikes a number of cities who have offered him insult and send federal SWAT teams to beat people up in some of them, while having his Department of Justice declare others among them “anarchist jurisdictions” and seek to withhold law enforcement funds from them. And this three-part formula can help us explain why it is ludicrous to suppose that the Clean Air Act is the pathway to Charles I; but it is equally ludicrous to suppose that ICE is not.

It is as a result of observing many cases of actually-existing tyranny—starting with the abuses associated with the war on terror, such as the creation of the legal black hole of Guantanamo Bay, to which accused terrorists were condemned for indefinite periods on nothing more than executive say-so, and the extrajudicial assassination of U.S. citizens via hellfire missiles launched from drones, again, on nothing more than executive diktat—but continuing through the Trump administration’s immigration abuses—that some critics of executive power have focused on the unitary executive, and with it, a broad conception of the scope of presidential power—item 3 in the tyranny menu above—as the source of the danger (e.g. Nourse 2018).

Getting rid of the notion of the unitary executive would at least permit Congress to impose some constraints on the ability of political officials to insist that the [*39] administrative state cast aside individual rights to serve policy goals. A robust, professionalized civil service insulated from political control probably wouldn’t be able to be immediately warped by an incoming presidential administration for tyrannical purposes.

But there’s also a truth that the libertarian legal scholars whom Sunstein and Vermeule most closely associate with the New Coke have seen. Dismembering the administrative state—eliminating element 1 of the tyranny menu—would also prevent a healthy amount of actually-existing tyranny. Imagine, for example, that the Department of Justice and the Department of Homeland Security were not allowed to exercise legislative and judicial powers. When some President desired to change immigration policy to exclude Muslims, that President would have to convince Congress to agree to do it. Moreover, that policy would have to be debated in a public forum, it wouldn’t be enacted on the basis of a secretive Homeland Security review process which allowed Donald Trump to whitewash his blatant and openly-admitted anti-Muslim animus as some kind of antiterror policy. When that President decided to change the rules under which asylum claims were decided, again, Congress would have to agree, and the adjudications would have to be conducted in front of judges who are independent, who have life tenure, and who have been socialized into legal culture.

Some critics of the administrative state (e.g. Ginsburg and Menashi 2010) have specifically identified the relationship between items 1 and 3 of the tyranny menu, arguing that resurrecting the nondelegation doctrine (and thus killing off the administrative state) is a precondition for the nontyrannical unitary executive. As I am, personally, rather fond of getting to breathe cleanish air, I’d rather lose the unitary executive than lose the bureaucracy, but either way the point remains that the elements of the menu together, not separately, are the real pathway to tyranny.

However, there’s an option we haven’t considered yet: item 2 on the tyranny menu. It should not be controversial that the threats to the rule of law are more salient where there are real and serious individual interests at stake. But here’s an idea that might be much more controversial: we might consider giving the administrative state more leeway in the core of its application, namely, to the economic regulation that, from the New Deal onward, motivated the expansion of executive power in the first place. By contrast, on the administrative law periphery, where real human beings are subject to state violence, we should impose much sterner constraints.

Individuals and their serious interests of life and limb and liberty only show up in a significant way a handful of times in LAW AND LEVIATHAN. I’ve mentioned one of them (ACCARDI) already. Early in the book, Sunstein and Vermeule briefly hint that critics of the lawless incarceration of accused terrorists in Guantanamo Bay ought to be included in “the New Coke” (pp. 24–25), but then never take the problem up again. From one perspective, it’s surprising that they mention Guantanamo at all. After all, military prisons operated on foreign soil to hold foreigners accused of terrorism don’t look a whole lot like notice and comment rulemaking or ALJ adjudications. The fact that it came to mind to them at least sufficiently for them to class critics of Guantanamo in the same category as critics of the SEC and the EPA reveals that they at least dimly perceive that the problem of administrative law is part of a broader problem of arbitrary executive power to lawlessly do serious harm to private persons. But it is disappointing that they don’t follow up on it. [*40]

Actually, Guantanamo Bay is a lot closer to administrative law than one might expect. The War on Terror was not Gitmo’s first Schmittian rodeo: a decade before 9/11 it made an appearance in—you guessed it—immigration law. As Gerald Neuman (1996) recounts in his well-known article on “anomalous zones,” the U.S. government held Haitian refugees there during the first Bush administration, denied them access to counsel, and even went so far as to feed them disinformation about the legal standards under which asylum might be granted. The Eleventh Circuit held that the refugees lacked any avenue for judicial review of their treatment under either the Administrative Procedure Act or the Immigration and Naturalization Act under the color of the claim that any procedural rights they might have did not attach until they reached the United States: sending the Coast Guard to interdict them at sea and house them in a military base in Cuba didn’t count (HAITIAN REFUGEE CENTER. V. BAKER, 953 F.3d 1498 at 1505-1509 (11th Cir. 1992)). One cannot help but suspect that the executive branch applied lessons from its successful evasion of judicial review with the Haitian refugees to its later adventures in lawless incarceration during the war on terror.

The other significant moment (leaving aside some mentions of KISOR V. WILKIE, 588 U.S. __ (2019), a VA benefits case) in which individual liberty interests show up in LAW AND LEVIATHAN is toward the end, where Sunstein and Vermeule discuss the 2019 case of GUNDY V. UNITED STATES, which permitted Congress to permit the Attorney General to decide whether and how to retroactively apply sex offender registration requirements to offenders who had been convicted prior to the enactment of the act. Because sex offenders are our most despised criminals, they, like immigrants, are a particularly vulnerable class of people whose treatment seems like a natural place to go looking for tyranny. Yet Sunstein and Vermeule do not seriously engage with the implications of GUNDY for their Fullerian framework. The Attorney General was allowed to implement a retroactive criminal punishment (well, a quasi-criminal quasi-punishment, since the Supreme Court held that registration requirements are not criminal punishments in the 2003 case of SMITH V. DOE, 538 U.S. 84 (2003)—another unbelievable fiction) that imposed affirmative obligations on individuals. Such a thing strikes at the absolute core of the Fullerian internal morality of law—no King Rex who was seriously trying to make rules that convicted sex offenders were actually supposed to be able to follow would regulate them that way. Must a convicted sex offender who has already served their sentence hire an “army of perfumed lawyers” to parse the Federal Register on a daily basis to figure out if some new registration requirement suddenly applies to them? Yet Sunstein and Vermeule’s discussion of GUNDY (pp. 122-124) focused not on that serious rule of law issue but on the Court’s aggressive reading of the underlying statute to avoid the nondelegation doctrine.

The fact that administrative law’s defenders as well as its critics have largely missed the possibility of intervening on item 2 of the tyranny menu strikes me as a symptom of a broader problem, what I have elsewhere called a disjuncture between the private law and the public law conceptions of the rule of law (Gowder 2020, 144–5). The New Coke to whom Sunstein and Vermeule address themselves are really concerned about business regulation. Many of them secretly or not-so-secretly want to bring LOCHNER back (LOCHNER V. NEW YORK, 198 U.S. 45 (1905)). The administrative state is identified with the New Deal era of economic regulation, a legal period which two of the leading libertarian scholars have described as “depravity” (Lawson and Calabresi 2018). And from the [*41] normative standpoint of business freedom comes the private law conception of the rule of law, which is largely about things like investment-backed expectations and reliance interests, property rights and occupational licenses.

It is in that context that we should read Richard Epstein’s (2020) review of LAW AND LEVIATHAN, which declares that “the most indefensible omission in this entire book is the large number of breakdowns in the administrative law system that genuinely matter to the health and welfare of a firm.” With all due respect to Professor Epstein, I’d rather they attend to the breakdowns that matter to the health and welfare of living human beings.

The New Coke’s allergy to economic regulation is a serious weakness more broadly. If they achieved their goals and the administrative state fell to pieces, it would be vanishingly unlikely that Congress, torn apart by longstanding political dysfunction, could fill in the gaps—there just wouldn’t be environmental regulation, or securities regulation, or labor regulation. (Maybe some states would do it for about five minutes, until the same scholars convinced the Supreme Court to strike those laws down under the dormant commerce clause.) But, because the EPA and the SEC and the FDA and the like actually carry out valuable social functions (unlike, if I may be blunt, ICE), the New Coke fall prey to Sunstein and Vermeule’s (pp. 34–35) critique about their inattention to the costs of sacrificing the administrative state. A more persuasive version of the New Coke might combine a real concern with excessive executive power with a real concern for government effectiveness.

From the rule of law standpoint, the main weakness of the New Coke is that ordinary economic regulation—even lawless ordinary economic regulation—typically just isn’t a path to anything like tyranny. To illustrate, let’s return to Richard Epstein’s (2019, 785–6), complaints about the Patent Trial and Appeals Board. He observes that the head judge of the PTAB has the power to tinker with the composition of ongoing adjudicative panels and at least one head liked to show up to speak in front of an advocacy group with a particular position on the validity of technology patents. That’s a blatant violation of the norms of judicial propriety, to be sure. But ultimately, the stakes are low: they’re just distributive fights between companies. Nobody is going to be locked up in a cage. Nobody is going to die. Probably, in most cases, nobody who is already rich will even find themselves not rich if they lose. There may be some broader economic consequences to incentives for innovation, but it’s hard to predict what they might be, and, at any rate, those are the policy issues we hire our political officials to handle. And the political scientists who may read this review will surely observe that regulated industries have the built-in ability to defend themselves in the legislative and regulatory process (Justice Gorsuch’s perfumed lobbyists), and indeed stand as a quintessential example of Mancur Olson’s (1971) great insight about the collective advantages of small groups with narrow interests.

To be fair to Epstein, some industrial regulation might be turned to tyrannical purposes. Epstein’s (2019) other complaint, about SEC determinations that can destroy careers, is a good example. (But even there, stockbrokers are a lot less politically vulnerable than sex offenders and immigrants, and having one’s finance career ended is a lot less terrible than being deported back to a genocidal dictatorship or being imprisoned for not knowing about a retroactive registration requirement.) And certainly a motivated president might use industrial regulation as a tool of arbitrary power. Indeed, we may have seen a very recent attempt to do [*42] just that: Donald Trump has been complaining for years about alleged social media “censorship” of his allies, and FCC Chairman Ajit Pai has recently announced that he intends to engage in a rulemaking to “clarify” Section 230 of the Communications Decency Act, the law that insulates these companies from civil liability for their content moderation activities (Reid 2020). In such cases, we should be vigorous indeed in subjecting administrative action to legal scrutiny. But it is vanishingly unlikely that day-to-day industrial regulation—the ordinary activities of the EPA, FDA, EEOC, FTC, FCC, and the rest of the alphabet soup of agencies that govern the conduct of the national economy—will pose a realistic threat of tyranny.

In conclusion, our conception of executive power and the permissibility of the administrative state needs to shift with the private stakes involved. For many cases in core post-New Deal industrial regulation, a kind of weak Fullerianism like that described by Sunstein and Vermeule is more than sufficient to protect the relatively unimportant individual interests at stake, especially in light of the high stakes that the polity at large often has in getting the overall policy right (especially in environmental law), and in light of the fact that the regulated “individuals” are typically sophisticated actors with both substantial compliance and substantial lobbying resources.

Thus far, I think I am stating a view shared by most defenders of administrative law. It is on that basis, I believe, that administrative law scholars typically accept practices like agency head adjudication that seem utterly bizarre to the rest of us. Thus, even as Richard Epstein (2019) criticizes the PTAB for stacking the adjudicator panels, mainline administrative law scholars like Christopher Walker and Melissa Wasserman (2019) offer precisely the opposite criticism: the PTAB is defective for not having full-fledged agency-head adjudication, and hence depriving the head of the Patent and Trademark Office of the power to control the policy which its adjudications instantiate.

I can accept Walker and Wasserman’s implicit normative assumption that executive control of policy is more important than adjudicative fairness when it is proposed that high political officials dictate the outcomes of patent validity disputes. But I cannot accept it when such officials start deciding which refugees will be locked up in an ICE prison and then sent away to face torture or death at the hands of some dictator. When the direct application of state violence is involved, or where those who are subject to the depredations of the executive are least capable, because of their lack of perfumed lawyers and lobbyists, of defending themselves, the avoidance of tyranny requires far more robust protections.

That simple notion—provide more formal legal protections to those who are the most vulnerable to the state’s power—ought not to be controversial. It’s a basic moral intuition. It nicely coheres with the historical progress of American jurisprudence since the New Deal, in which we have recognized that the government ought to have a relatively freer hand in economic regulation than in other areas. More importantly, it’s also an abstraction of the idea underlying modern procedural due process doctrine since MATHEWS V. ELDRIDGE 424 U.S. 319 (1976) that the weight of the individual interest needs to be taken into account in determining how protective the government must be of procedural niceties like a fair hearing. [*43]

Even though (as Sunstein and Vermeule point out), the Due Process Clause doesn’t apply to every administrative decision, because not every way the administrative state might harm one counts as a deprivation of liberty or property, something like the MATHEWS doctrine can still serve as a moral critique of the government’s failure to provide procedural protections to individuals who face its might and as an expression of the fundamentally egalitarian ideas at the heart of the rule of law (Gowder 2016b, chs. 1- 4). If we could impose it on our administrative state, we could keep things like environmental and securities regulation but still avoid executive tyranny.

And yet that basic egalitarian core of our rule of law ideal is completely ignored both by the New Coke libertarians and by the Sunstein and Vermeule-style defenders of the administrative state. Both sides in the debate make this foundational error because the entire conversation among a generation of administrative law scholars has been laser-focused on business law. Because Sunstein and Vermeule’s participation in this artificially narrow literature leads them to ignore the ways in which the administrative state actually inflicts tyranny on the vulnerable, LAW AND LEVIATHAN is an intelligent, carefully argued, and in many ways insightful defense of administrative law’s compatibility with the rule of law that nonetheless manages to be totally disconnected from reality.


My thanks to Erin Delaney, Andy Koppelman, Jim Speta, and Matt Spitzer for conversations and comments on this review.


ACCARDI V. SHAUGHNESSY, 347 U.S. 260 (1954).


AUER V. ROBBINS, 519 U.S. 452 (1997).

BENSLIMANE V. GONZALES (430 F.3d 828 (7th Cir. 2005).


CHAVARRIA-REYES V. LYNCH, 845 F.3d 275 (7th Cir. 2016).


DHS V. THURAISSIGIAM, 591 U.S. __ (2020).

FONG YUE TING V. UNITED STATES, 149 U.S. 698 (1893).

GARFIAS-RODRIGUEZ V. HOLDER, 702 F.3d 504 (9th Cir. 2012).

GUNDY V. UNITED STATES, 588 U.S. ___ (2019). [*44]

GUITIERREZ-BRIZUELA V. LYNCH, 834 F.3d 1142 (10th Cir. 2016).

HAITIAN REFUGEE CENTER. V. BAKER, 953 F.3d 1498, 1505-1509 (11th Cir. 1992).

HERNANDEZ V. MESA, 589 U.S. ____ (2020).

INS V. ST. CYR, 533 U.S. 289 (2001).

KHAN V. HOLDER, 608 F.3d 325, 329-330 (7th Cir. 2010).

KISOR V. WILKIE, 588 U.S. __ (2019).

LOCHNER V. NEW YORK, 198 U.S. 45 (1905).

LUCIA V. SEC, 585 U.S. __ (2018).

MAKE THE ROAD NEW YORK V. WOLF, 962 F.3d 612 (D.C. Cir. 2020).

MATHEWS V. ELDRIDGE, 424 U.S. 319 (1976)


SMITH V. DOE, 538 U.S. 84 (2003).

UNITED STATES V. JU TOY (198 U.S. 253 (1905).



“ACLU Factsheet on Customs and Border Protection’s 100-Mile Zone.” n.d. American Civil Liberties Union. Accessed September 28, 2020.

Campbell, Stanley W. 1970. THE SLAVE CATCHERS: ENFORCEMENT OF THE FUGITIVE SLAVE LAW, 1850-1860. Chapel Hill: University of North Carolina Press.

Carroll, Rory. 2018. “US Border Patrol Routinely Sabotages Water Left for Migrants, Report Says.” THE GUARDIAN, January 17, 2018, sec. US news.

Chin, Gabriel J. 2002. “Regulating Race: Asian Exclusion and the Administrative State.” HARVARD CIVIL RIGHTS-CIVIL LIBERTIES LAW REVIEW 37:1: 1-64.


Department of Justice Executive Office for Immigration Review’s Statistical Yearbook. 2001. available online at

Dicey, Albert Venn. (1885) 1982. INTRODUCTION TO THE STUDY OF THE LAW OF THE CONSTITUTION. Indianapolis, IN: Liberty Fund.

Eaglesham, Jean. 2015. “Fairness of SEC Judges Is in Spotlight.” WALL STREET JOURNAL, November 23, 2015, sec. Markets.

Epstein, Richard A. 2019. “Structural Protections for Individual Rights: The Indispensable Role of Article III - or Even Article I - Courts in the Administrative State.” GEORGE MASON LAW REVIEW 26:3: 777–810.

———. 2020. “Leviathan’s Apologists” (review of LAW AND LEVIATHAN). LAW AND LIBERTY.

Fraenkel, Ernst. 1941. THE DUAL STATE: A CONTRIBUTION TO THE THEORY OF DICTATORSHIP. Translated by E. A. Shils. New York: Oxford University Press.

Fuller, Lon. 1964. THE MORALITY OF LAW. New Haven: Yale University Press.

———. 1978. “The Forms and Limits of Adjudication.” HARVARD LAW REVIEW 92:2: 353-409.

Getter, Lisa, and Jonathan Peterson. 2003. “Speedier Rate of Deportation Rulings Assailed.” LOS ANGELES TIMES, January 5, 2003.

Ginsburg, Douglas H., and Steven Menashi. 2010. “Nondelegation and the Unitary Executive.” UNIVERSITY OF PENNSYLVANIA JOURNAL OF CONSTITUTIONAL LAW 12:2: 251–76.

Gowder, Paul. 2016a. “Democratic Paternalism, Not Libertarian Paternalism.” GEORGETOWN JOURNAL OF LAW AND PUBLIC POLICY 14: 747–87.

———. 2016b. THE RULE OF LAW IN THE REAL WORLD. New York: Cambridge University Press.

———. 2020. “The Dangers to the American Rule of Law Will Outlast the Next Election.” CARDOZO LAW REVIEW DE NOVO 2020: 126–64.

Graff, Garrett M. November/December 2014. “The Green Monster: How the Border Patrol Became America’s Most Out-of-Control Law Enforcement Agency.” POLITICO MAGAZINE, November/December 2014.

Hamburger, Philip. 2014. IS ADMINISTRATIVE LAW UNLAWFUL? Chicago: University of Chicago Press. [*46]

Ismay, John. 2020. “A Navy Veteran Had a Question for the Feds in Portland. They Beat Him in Response.” THE NEW YORK TIMES, July 20, 2020, sec. U.S.

Jain, Amit. 2019. “Bureaucrats in Robes: Immigration Judges and the Trappings of Courts.” GEORGETOWN IMMIGRATION LAW JOURNAL 33:2: 261–326.

Jawetz, Tom, Philip E Wolgin, and Claudia Flores. 2020. “5 Immediate Steps to Rein in DHS in the Wake of Portland.” Center for American Progress.

Johnson, Allen. 1921. “The Constitutionality of the Fugitive Slave Acts.” YALE LAW JOURNAL 31:2: 161-182.

Kantorowicz, Ernst H., and Conrad Leyser. 2016. THE KING’S TWO BODIES: A STUDY IN MEDIAEVAL POLITICAL THEOLOGY. First Princeton classics edition. Princeton Classics. Princeton: Princeton University Press.

Koh, Jennifer Lee. 2018. “When Shadow Removals Collide: Searching for Solutions to the Legal Black Holes Created by Expedited Removal and Reinstatement.” WASHINGTON UNIVERSITY LAW REVIEW 96:2: 337–94.

Lawson, Gary, and Steven Calabresi. 2018. “The Depravity of the 1930s and the Modern Administrative State.” NOTRE DAME LAW REVIEW 94:2: 821–66.

Levy, Jacob. 2018. “Law and Border.” Niskanen Center Blog. July 25, 2018.

Murphy, Colleen. 2005. “Lon Fuller and the Moral Value of the Rule of Law.” LAW AND PHILOSOPHY 24:3: 239-262.

Neuman, Gerald L. 1996. “Anomalous Zones.” STANFORD LAW REVIEW 48:5: 1197-1234.

Nourse, Victoria. 2018. “The Special Counsel, Morrison V. Olson, and the Dangerous Implications of the Unitary Executive Theory.” ADVANCE: THE JOURNAL OF THE ACS ISSUE BRIEFS 12: 43–60.

Olmos, Sergio, Mike Baker, and Zolan Kanno-Youngs. 2020. “Federal Agents Unleash Militarized Crackdown on Portland.” THE NEW YORK TIMES, July 17, 2020, sec. U.S.

Olson, Mancur. 1971. THE LOGIC OF COLLECTIVE ACTION. 2nd ed. Cambridge, MA: Harvard University Press.

Pilkington, Ed. 2020. “‘These Are His People’: Inside the Elite Border Patrol Unit Trump Sent to Portland.” THE GUARDIAN, July 27, 2020, sec. US news.

Posner, Eric A., and Adrian Vermeule. 2010. THE EXECUTIVE UNBOUND: AFTER THE MADISONIAN REPUBLIC. New York: Oxford University Press. [*47]

Reid, John. 2020. “FCC Set to Tackle Social Media Liability Limits, Pai Says.” BLOOMBERG LAW, October 15, 2020.


Stumpf, Juliet P. 2014. “Civil Detention and Other Oxymorons.” QUEEN’S LAW JOURNAL 40:1: 55–98.

Sunstein, Cass R., and Richard H. Thaler. 2003. “Libertarian Paternalism Is Not an Oxymoron.” THE UNIVERSITY OF CHICAGO LAW REVIEW 70:4: 1159–1202.

Thompson, E. P. 1975. WHIGS AND HUNTERS: THE ORIGIN OF THE BLACK ACT. London: Allen Lane.

Vermeule, Adrian. 2009. “Our Schmittian Administrative Law.” HARVARD LAW REVIEW 122:4: 1095–1150.

———. 2017. “A Christian Strategy.” FIRST THINGS: A MONTHLY JOURNAL OF RELIGION AND PUBLIC LIFE, no. 277 (November): 41–46.

Waldron, Jeremy. 2012. DIGNITY, RANK, AND RIGHTS. Oxford University Press.

© Copyright 2021 by author, Paul Gowder.