Legal Theory Blog

Kumar, Penney, Schneier, & Albert on Adversarial Machine Learning Research

Ram Shankar Siva Kumar (Microsoft Corporation), Jon Penney (Harvard Law School; Harvard University - Berkman Klein Center for Internet & Society; Citizen Lab, University of Toronto), Bruce Schneier (Harvard University - Berkman Klein Center for Internet & Society; Harvard University - Harvard Kennedy School (HKS)), & Kendra Albert (Harvard Law School) have posted Legal Risks of Adversarial Machine Learning Research (International Conference on Machine Learning (ICML) 2020 Workshop on Law & Machine Learning) on SSRN.  Here is the abstract:

Adversarial machine learning is the systematic study of how motivated adversaries can compromise the confidentiality, integrity, and availability of machine learning (ML) systems through targeted or blanket attacks. The problem of attacking ML systems is so prevalent that CERT, the federally funded research and development center tasked with studying attacks, issued a broad vulnerability note on how most ML classifiers are vulnerable to adversarial manipulation. Google, IBM, Facebook, and Microsoft have committed to investing in securing machine learning systems. The US and EU are likewise putting security and safety of AI systems as a top priority.

Now, research on adversarial machine learning is booming but it is not without risks. Studying or testing the security of any operational system may violate the Computer Fraud and Abuse Act (CFAA), the primary United States federal statute that creates liability for hacking. The CFAA’s broad scope, rigid requirements, and heavy penalties, critics argue, has a chilling effect on security research. Adversarial ML security research is likely no different. However, prior work on adversarial ML research and the CFAA is sparse and narrowly focused. In this article, we help address this gap in the literature. For legal practitioners, we describe the complex and confusing legal landscape of applying the CFAA to adversarial ML. For adversarial ML researchers, we describe the potential risks of conducting adversarial ML research. We also conclude with an analysis predicting how the US Supreme Court may resolve some present inconsistencies in the CFAA’s application in Van Buren v. United States, an appeal expected to be decided in 2021. We argue that the court is likely to adopt a narrow construction of the CFAA, and that this will actually lead to better adversarial ML security outcomes in the long term.

Recommended.

Blank & Glogower on Progressive Tax Procedure

Joshua D. Blank (University of California, Irvine School of Law) & Ari D. Glogower (Ohio State University (OSU) - Michael E. Moritz College of Law) have posted Progressive Tax Procedure (New York University Law Review, forthcoming, 2021) on SSRN.  Here is the abstract:

Abusive tax avoidance and tax evasion by high-income taxpayers pose unique threats to the tax system. These strategies undermine the tax system’s progressive features and distort its distributional burdens. Responses to this challenge generally fall within two categories: calls to increase IRS enforcement and “activity-based rules” targeting the specific strategies that enable tax avoidance and evasion by these taxpayers. Both of these responses, however, offer incomplete solutions to the problems of high-end noncompliance.

This Article presents the case for “progressive tax procedure”— means-based adjustments to the tax procedure rules for high-income taxpayers. In contrast to the activity-based rules in current law, progressive tax procedure would tailor rules to the economic circumstances of the actors rather than their activities. For example, under this system, a high-income taxpayer would face higher tax penalty rates or longer periods where the IRS could assess tax deficiencies. Progressive tax procedure could also allow an exception for low-value tax underpayments, to avoid excessive IRS scrutiny or unduly burdensome rules for less serious offenses.

Progressive tax procedure could address the unique challenges posed by high-end tax noncompliance and equalize the effect of the tax procedure rules for taxpayers in varying economic circumstances. It could also complement the alternative approaches of increasing tax enforcement and activity-based rules while avoiding the limitations of relying exclusively on these responses.

After developing the normative case for progressive tax procedure, the Article illustrates how it could be applied in three specific areas: accuracy-related tax penalties, the reasonable cause defense, and the statute of limitations. These applications illuminate the basic design choices in implementing progressive tax procedure, including the types of rules that should be adjusted and the methods for designing these adjustments.

Interesting and recommended.

Slocum & Gries on Corpus Linguistics & Judging

Brian G. Slocum (University of the Pacific - McGeorge School of Law) & Stefan Gries (UC Santa Barbara - Department of Linguistics) have posted Judging Corpus Linguistics (Southern California Law Review, 2020) on SSRN.  Here is the abstract:

The practice of legal interpretation has long sought legitimization through devices that seek to distance interpretations from the personal predilections of judges. Most notably, with the rise of textualism, courts have habitually relied on dictionary definitions to provide word meanings that are external to a judge’s own intuitions. Similarly, some scholars and judges have recently argued that corpus linguistics can provide especially powerful and objective information to judges about the ordinary meanings of statutory and constitutional texts. For instance, in their influential article, Judging Ordinary Meaning, Thomas R. Lee and Stephen Mouritsen argue that courts should “import” into the law of interpretation computer-aided means (primarily, corpus analysis) of determining “the sense of a word or phrase that is most likely implicated in a given linguistic context.” In the view of Lee and Mouritsen, statutory interpretation is an “empirical question” (the authors assert this more than forty times), which makes it natural that courts should rely on scientifically-based interpretive sources such as corpus linguistics.

The potential judicial adoption of interdisciplinary knowledge and techniques from fields such as linguistics is intriguing, and the resulting discussions from such proposals will enhance both the theory and practice of legal interpretation. Nevertheless, anyone advocating for the judicial adoption of a significant and novel interpretive source bears the burden of offering a compelling explication of the interpretive source and its role within the structure of interpretation. This demonstration should establish that the new interpretive source offers some comparative advantage to existing interpretive sources and is feasible in the sense that judges can competently use it. The advocate must therefore offer a compelling theory of how the interpretive source fits into existing processes of interpretation and explain whether the new interpretive source requires a new way of viewing those processes. With corpus linguistics, some of the issues that should be addressed therefore include:

(1) how corpus linguistic analysis is relevant to some objective of interpretation currently identified by judges, such as the determination of ordinary meaning;

(2) whether corpus linguistics should displace long-standing interpretive sources, such as dictionaries and textual canons;

(3) the extent to which corpus linguistics can take account of the relevant context of a statutory provision;

(4) to what extent determining statutory meaning is an empirical endeavor (with or without corpus linguistics); and

(5) whether judges have both the technical ability to conduct competent corpus analyses and sufficient linguistic expertise to evaluate the raw data and make judgments of the kind made by trained linguists.

In this short essay, in the spirit of offering general concerns about corpus analysis and legal interpretation, we largely focus on Lee and Mouritsen’s efforts in addressing the above issues. We argue that Lee and Mouritsen’s conceptualization of the potential role for corpus linguistics within legal interpretation is inadequate and underestimates the difficulty of judicial adoption of corpus analysis methods. Corpus analysis can provide useful information about the functioning of language, but it is crucial to neither understate the role of context in determining statutory meaning nor overstate the potential contribution of corpus analysis to legal interpretation.

Highly recommended.  Download it while it's hot!

Leib & Kent on the Law of Public Office

Ethan J. Leib (Fordham University School of Law) & Andrew Kent (Fordham University School of Law) have posted Fiduciary Law and the Law of Public Office: Suggestions for a Research Agenda (William & Mary Law Review, Forthcoming) on SSRN.  Here is the abstract:

A law of public office crystallized in Anglo-American law in the seventeenth and eighteenth centuries. This body of law — defined and enforced through a mix of oaths, statutes, criminal and civil case law, impeachments, and legislative investigations — imposed core duties on public officeholders: Officials needed to serve the public good, not their own private interests; were barred from acting ultra vires; could often be required to account to the public for their conduct in office; and needed to act with impartiality, honesty, and diligence. Office-holding came to be viewed as conditional, with officers removable for misdeeds. This law of public office reflected something that looks similar to modern fiduciary duties of loyalty and care.

In this Essay, we extend the historical record describing this law of public office, and make several new claims — historical and theoretical. First, there are strong reasons to suspect that the law of public office and private fiduciary duties developed together and influenced each other. During the critical centuries we explore, the duties of officeholders such as trustees, executors, and corporate directors were developing alongside the duties of officials such as tax collectors and government commissioners. Parliament and other actors repeatedly used the language of trust, trusteeship, guardianship, and account to define the law of public offices. And public law concerns about abuse of power and the need for honesty, fidelity, and altruism in service of others may have seeped from public law into private fiduciary law. Influential political theory about the monarchy and lesser magistrates was also using trust and related legal language to set forth a fiduciary conception of public office-holding; the theoretical developments in political theory not only drew from legal concepts but may have helped shape them, as well.

One Essay cannot decisively establish whether the similarities in language, concept, and timing were mere coincidence or rather evidence of some conscious co-development in the law of public offices, political theory, and fiduciary law. Proving (or disproving) actual causal relationships will need to be the work of the future. We conclude with some potential implications for our research agenda, should further work continue to confirm our findings here. Fiduciary political theorists should be less anxious about drawing from private law models — and private law fiduciary theorists might need to be less insistent on the purity of the private sphere. As we show, during the critical periods when fiduciary law and the law of public office come into their own, the public-private distinction wasn’t yet creating the divide that exists today. Our research agenda invites more mutual learning — both historically and for law and institutions today.

Highly recommended.  Download it while it's hot!

Loza de Siles on Roger Taney

Emile Loza de Siles (Duquesne University School of Law; Technology & Cybersecurity Law Group; University of Maryland Global Campus) has posted Taney's Zombie: Chief Justice Roger B. Taney's Life in Maryland's Black Belt: Revelations About Dred Scott and the Still Undead Commitment to White Supremacy and Racial Hatred on SSRN.  Here is the abstract:

Roger Taney was born and raised in Maryland’s Black Belt, a region so committed to the institution of slavery that it sought to secede from Maryland and unite with slavery-entrenched Virginia. His first teachers included a well-credentialed, but unhinged man who so fervently believed that he, like Christ, could walk on water that he drowned in the attempt. Taney’s progenitors went from indentured servant to High Sheriff of Calvert County, whose duties included the “disposal” of the colonial Anglican church’s property of women, including free white women, and their mixed race children, and slaveholding landowner married into the illustrious colonial family of Francis Scott Key.

This grounding in white supremacy, slavery, and racial hatred followed Taney into his privileged rise through the Maryland legislature, into private practice, and then as Maryland attorney general and subsequently, as reward for his zealous support of Andrew Jackson in his rise to the Presidency, to the offices of U.S. attorney general, Secretary of Treasury, and finally, for almost thirty years, Chief Justice of the U.S. Supreme Court. During his long tenure as Chief Justice, the nation ran toward Civil War, and the Taney Court decided such monumental cases as the challenge of President Abraham Lincoln’s suspension of habeas corpus after a Maryland contemporary and acquaintance of Taney was arrested and held without charges for sabotage.

No decision of the Taney Court is more reviled than that of Dred Scott v. Sandford, an opinion written by Chief Justice Taney and in which his familial, social, and legal history of white supremacy and racial hatred is unstintingly revealed. The final irony in the life of this infamous Chief Justice is that on his deathbed in Washington, D.C., he urgently condemned the voter suppression taking place on that day in a constitutional vote in Maryland.

Today in the Black Belt and elsewhere in the nation, Taney’s zombie of white supremacy, racial hatred, and segregation walks undead, its evil spirit inflamed by hateful politically-motivated rhetoric, the imprimatur of such speakers, and cultural hatred and ignorance, and by failures of many of the avowed religious to condemn and cast out those who abominate the strength of our national values and identity. From voter suppression in North Carolina to horseback sheriff deputies leading a roped black man down to the unexamined use of predictive algorithmic systems disparately impacting the poor and people of color in bail, detention, parole, and sentencing decisions, the spirit that imbued Taney’s Dred Scott opinion and animated his life’s values remains alive within the nation’s culture and legal system.

Informed with this history and reflection, perhaps we as a nation can illuminate the evil at large and finally kill off that demon and conclude, at last, the Civil War.

Legal Theory Lexicon: Scalars and Binaries

Introduction

The law frequently requires that we answer a question "yes" or "no".  Was the plaintiff guilty?  Was the defendant negligent?  Was the trial court's finding of fact clearly erroneous?  These questions seem to demand a "yes" or "no" answer.

When an issue must be resolved in one (and only one) of two possible ways, we can call that issue "binary".  That is, the law frequently demands that we see the world in blacks and whites.  The real world (the world outside of the legal system) is frequently all shades of gray.  Our beliefs about whether the defendant was guilty, whether the defendant was negligent, or whether the trial court erred may be a matter of probabilities or degrees.  When an issue or finding is a matter of degree, we can call it a "scalar".

This entry in the Legal Theory Lexicon is about the scalar/binary distinction in legal theory.  As always, it is primarily addressed to law students, especially first-year students, with an interest in legal theory.  If you have further questions about the distinction and the role it plays in legal theory, you will want to consult "Scalar Properites, Binary Judgments," by the great legal theorist, Larry Alexander.

What is the Scalar-Binary Distinction?

One way to explain the scalar-binary distinction is via an analogy with related distinction between scalar and discrete variables in mathematics.  A variable is scalar if it can assume any value within a range.  A variable is discrete if it can only assume some values.  For example, weight is a scalar variable.  If we way a group of humans, their weights can assume any value.  An infant could be 8.5 pounds, an adult 180 pounds, and any value in between is possible, including 80.8715 pounds or 13.025 pounds.  Technically, we also need to distinguish scalars (which have only one dimension) from vectors (which have mutliple dimensions).

Another metaphor for a continuous variable is a "spectrum."  For example, the frequency of a radio broadcase is a value on a spectrum.  Similarly, we can also use the term "continuum" to refer to the same idea that is captured by the notion of a scalar.  We also use the phrase "matter of degree" to get at the same basic notion.  More technically, sometimes the phrase "range property" is used to refer to scalars.

A discrete variable can only assume particular values.  For example, if we toss a coin three times and count the number of heads, the only possible outcomes are 3, 2, 1, and 0.  The number of heads cannot be 1.5 or 2.00786.  Binary variables are special cases of discrete variables; a binary variable can only assume one of two values.  If we toss a coin once, the only possible values are 1 and 0.  When someone says, "The answer to this question is either a 'yes" or a 'no," they are saying the question involves a binary judgment.  Multiple-choice questions are discrete, but not binary if there are more than two possible answers.

The law is full of binary judgments.  The defendant must be found "guilty" or "not guilty."  The plaintiff's injury was either "caused" or "not caused" by defendant's conduct.  The trial court's finding of fact was either "clearly erroneous" or "not clearly erroneous."  Sometimes, however, the law requires that a judgment be made on a scale.  For example, in a comparative negligence jurisdiction, the jury might be required to decide whether the plaintiff's injury was 100% caused by plaintiff's negligence or 0% caused by plaintiff or any value in between.

Scalar Properties, Binary Judgments

Sometimes, the law requires that we translate our beliefs about a scalar property into a binary judgment.  For example, we might think that the likelihood that a trial judge made an error in a finding of fact is a matter of degrees, ranging from a 0% chance of error to a 100% chance of error.  But when an appellate court reviews a trial court' judge's factual determinations it must make a binary judgment, either "clearly erroneous" or "not clearly erroneous."  This phenomenon, where the law demands a binary judgment about a scalar is pervasive in the law.

Vagueness and the Line-Drawing Problem in Binary Judgments

The demand for binary judgments results in a particular problem when there is no precise threshold for the binary categorization of the scalar property of the real world.  One example is the balancing test, which allows us to translate a complex set of scalar properties into a simple yes-no judgment; either the balance of interests tips one way, or the other.

Sometimes, there is a bright line, a precise threshold that allows us to translate our scalar determinations into binary judgmetns.  For example, in civil cases the burden of persuasion is usually "perponderance of the evidence."  One understanding of the meaning of that standard is that it requires that the evidence establish that the likelihood is greater than 0.5.  In other words, the plaintiff needs to show that the likelihod that defendant caused plaintiff's injury is greater than 50%.  This rule allows the translation of a scalar (probability) into a binary ("perponderance of the evidence" or "not perponderance of the evidence").  Let's call this kind of rule a "bright-line translation rule."

But in other cases, the translation rule does not provide a bright line, because the rule is vague.  For example, the clearly-erroneous rule is not associated with some precise bright-line probability of error.  Instead, the clearly-erroneous rule is vague.  In some cases, we are certain that the judge was in error and hence the finding of fact is definitely "clear error."  In other cases, we are very uncertain about the error; we might think the judge was wrong, but just barely so.  This would be a case where we would definitely say the error was not "clear error."  But there is no bright line, so there will be some cases on the borderline or in a zone of underdetermination.  In those cases, appellate judges may disagree about whether the error was clear, and even the same judge might make the call differently at different times.

Smooth and Bumpy Laws

The relationship between binaries and scalars can be conceptualized as one case of a more general phenomenon--which Adam Kolber has called "smooth and bumpy laws."  HIs idea is that legal input and outputs can be "smooth" (gradual) or "bumpy" (uneven).  Scalars are smooth, but binaries are bumpy.  This conceptualization allows the insight behind the scalar-binary distinction to be extended to situations in which the output is discontinuous or uneven in some other way, but not a true binary.

Conclusion

This entry in the Legal Theory Lexicon introduces the scalar-binary distinction and illustrates some of the issues and problems that the distinction illuminates.

Related Lexicon Entries

• Legal Theory Lexicon 020: Causation
• Legal Theory Lexicon 024: Balancing Tests
• Legal Theory Lexicon 026: Rules, Standards, and Principles
• Legal Theory Lexicon 054: Standards of Review

Bibliography

• Larry Alexander, Scalar Properties, Binary Judgments, 25 J. Applied Phil. 85 (2008), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=829326.
• Adam Kolber, Smooth and Bumpy Laws, 102 Cal. L. Rev. 655 (2014), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1992034.
• Adam Kolber,  The Bumpiness of Criminal Law, 67 Ala. L. Rev. 855 (2016), https://ssrn.com/abstract=2684203.
• Tim Mulgan, Critical Notice of Jeff McMahon, The Ethics of Killing: Problems at the Margin of Life,” 34 CANAD. J. PHIL. 443 (2004).

(Last revised on August 2, 2020.)

Legal Theory Bookworm: "Taking Back the Constitution" by Mark Tushnet

The Legal Theory Bookworm recommends Taking Back the Constitution: Activist Judges and the Next Age of American Law by Mark Tushnet.  Here is a description:

What Supreme Court justices do is far more than just “calling balls and strikes.” The Court has never simply evaluated laws and arguments in light of permanent and immutable constitutional meanings. Social, moral, and yes, political ideas have always played into the justices’ impressions of how they think a case should be decided. Mark Tushnet traces the ways constitutional thought has evolved, from the liberalism of the New Deal and the Great Society to the Reagan conservatism that has been dominant since the 1980s. Looking at the current crossroads in the constitutional order, Tushnet explores the possibilities of either a Trumpian entrenchment of the most extreme ideas of the Reagan philosophy, or a dramatic and destabilizing move to the left. Wary of either outcome, he offers a passionate and informed argument for replacing judicial supremacy with popular constitutionalism—a move that would restore to the other branches of government a role in deciding constitutional questions.

And from the reviews:

“Tushnet persuasively demonstrates how judges’ preexisting views influence how they decide cases. He rightly calls for a reinvigoration of the role of Congress and the public in constitutional decision-making.”—Caroline Fredrickson, author of The Democracy Fix and Under the Bus

 

"Readers will find Mark Tushnet's new book bracing, exhilarating, and (for some) maddening. It is controversial in the best way, providing a framework for specific arguments (like court-packing) that many readers will have encountered but perhaps not considered systematically."—Josh Chafetz, author of Congress's Constitution

Download of the Week: "The Distinction Between Private Law and Public Law" by Dagan & Zipursky

The Download of the Week is The Distinction Between Private Law and Public Law by Hanoch Dagan & Benjamin C. Zipursky.  Here is the abstract:

Twentieth Century legal theorists – especially the American legal realists – provided vigorous critiques of the putative distinction between private law and public law, persuading jurists that the distinction relied upon legal formalism. In the latter third of the Twentieth Century, the law and economics school constituted a focused and dominant version of the legal realist capture of private law theory. By the 1990s, however, new forms of private law research developed, many of which we characterize as instances of “neo-realism.” And philosophical theorists brought private law theory full circle, re-embracing the distinction between private law and public law. These views are depicted here as forms of “neo-conceptualism,” and they are in many ways the polar opposite of neo-Realism. In this Chapter, Dagan and Zipursky each defends a view of the private/public distinction that falls in between the neo-realists and the neo-conceptualists. Dagan’s relational justice theory is a form of neo-realism highly receptive to the deontological concerns of the neo-conceptualists. Zipursky’s pragmatic conceptualism is a form of neo-conceptualism sensitive to the pragmatic and modernist concerns of legal realism.

Loza de Siles on AI & Discriminatory Bias

Emile Loza de Siles (Duquesne University School of Law; Technology & Cybersecurity Law Group; University of Maryland Global Campus) has posted AI, on the Algorithmic State of the Nation: Artificial Intelligence Unleashed and Civil Rights on SSRN.  Here is the abstract:

In the classic movie "O Brother, Where Art Thou?" set in the turn-of-the-20th-century South, an ominously unstoppable lawman and his posse track escaped prisoners with bloodhounds and rifles, finally catching up with those flawed and comical heroes at an old home place. About to be summarily executed, our heroes protest that they have been recently pardoned by the governor and that their imminent lynching would be illegal. Their arguable leader Ulysses Everett McGill exclaims, “It ain’t the law!” Unfazed, Sheriff Cooley, cruel eyes hidden behind lifeless black shades, juts his implacable jaw skyward and scoffs, “The law? The law is a human institution.” With those immortal words, the lawman at once justifies and pardons the illegality of the looming nooses on the grounds that the law is intrinsically fallible, being a human creation, and that fallibility is an irrefutable inevitability.

Artificial intelligence is a human institution. It, machine learning, and all its algorithmic system variants (collectively, “artificial intelligence” or “AI”) are flawed and are subject to the discriminatory biases of those people who and organizations that design, code, test, and use them, including the judiciary, the administrative state, and other government institutions and private companies. Much of the multitudinous Big Data upon which algorithms feed and operate are likewise flawed and reflect discriminatory biases, including the historical biases baked into and persisting in the data from redlining and systematized racism in the Jim Crow era.

AI deploys rapidly, at scale, and often worldwide. Consequently, discriminatory biases in AI has a tremendous and rapid potential for harm to the people who are the subjects of those systems, their families and communities, and to the societies of which they are a part. Because algorithmic determinations may be used, in turn, as inputs for yet other algorithmic systems, the harms propagate. The harms may be irreparable. The scale and gravity of the impact of discriminatory bias in machine learning and other artificial intelligence and algorithmic systems are enormous and represent potentially existential and global threats to a more perfect union of humanity.

Artificial intelligence is a human institution, but that does not mean that we should accept its misdesign and misuse as inevitable failings that may be simply written off as the cost of “doing business.” Where people and organizations invent and seek to deploy such tools, they and public policy leaders must find ways of minimize the amplified risks of new and irreparable harms associated with those tools.

How then can discriminatory biases in machine learning and other artificial intelligence systems be detected and eradicated? How can we ensure that, despite their status as a “human institution,” algorithmic systems are purged of their human-derived flaws and rendered just? These questions and others resonate, leading to the ultimate question of algorithmic justice: What is the Algorithmic State of our Nation, and how may it be ascertained and continually improved?

Wadhia & Walker on Chevron Deference & Immigration

Shoba Sivaprasad Wadhia (The Pennsylvania State University (University Park) – Penn State Law) & Christopher J. Walker (Ohio State University (OSU) - Michael E. Moritz College of Law) have posted The Case Against Chevron Deference in Immigration Adjudication (Duke Law Journal, Vol. 70, 2021 Forthcoming) on SSRN.  Here is the abstract:

The Duke Law Journal’s fifty-first annual administrative law symposium examines the future of Chevron deference—the command that a reviewing court defer to an agency’s reasonable interpretation of an ambiguous statute the agency administers. In the lead article, Kristin Hickman and Aaron Nielson argue that the Supreme Court should narrow Chevron’s domain to exclude interpretations made via administrative adjudication. Building on their framing, this Essay presents an in-depth case study of immigration adjudication and argues that the case against Chevron has perhaps its greatest force in this immigration context. That is because much of Chevron’s theory for congressional delegation and judicial deference—including agency expertise, deliberative process, and even political accountability—collapses in the immigration adjudication context.

As for potential reform, Hickman and Nielson understandably focus on the Supreme Court. We too explore that judicial option, but we argue that it is a mistake to focus just on courts when it comes to immigration law and policy. The political branches can and should act to narrow Chevron’s domain. First, our proposal should be part of any comprehensive immigration reform legislation, which may well become a key legislative initiative after the presidential election. Second, the Executive Branch can and should embrace this reform internally—by not seeking Chevron deference for immigration adjudication and by turning to rulemaking instead of adjudication to make major immigration policy. Shifting the default from adjudication to rulemaking for immigration policymaking is more consistent with Chevron’s theoretical foundations—to leverage agency expertise, to engage in a deliberative process, and to increase political accountability.