A Theory of Constitutional Norms

The political convulsions of the past decade have fueled acute interest in constitutional norms or “conventions.” Despite intense scholarly attention, existing accounts are incomplete and do not answer at least one or more of three major questions: (1) What must all constitutional norms do? (2) What makes them conventional? (3) And why are they constitutional?

This Article advances an original theory of constitutional norms that answers these questions. First, it defines them and explains their general character: they are normative, contingent, and arbitrary practices that implement constitutional text and principle. Most scholars have foregone examining how norms are conventional or have relegated them to coordinating behavior, like rules requiring drivers to stick to one side of the road. By contrast, this Article argues that constitutional norms are constitutive conventions, which concretize values into practices; they are akin to conventions of etiquette that concretize concepts like “politeness.” Constitutional norms implement abstract principles, like the separation of powers, or indeterminate text, such as “advice and consent,” into specific behavior and action.

By understanding constitutional norms as constitutive conventions, this Article explains norms’ salient features, basic functions, and relationship to the Constitution. Norms are normative because they command respect and allegiance; they are contingent because they depend on political, social, and intellectual conditions to emerge and endure; they are arbitrary because they represent one of many possible ways of realizing constitutional text and principle; and they are constitutional because the values they implement arise from the Constitution itself. This Article animates its theory through case studies of three constitutional norms: blue slips, the norm against court-packing, and executive noninterference in law enforcement. It concludes by questioning the use of historical practice in constitutional interpretation. It suggests that when scholars and judges draw on norms that are intrinsically contingent and arbitrary, they embed unstated normative assumptions about the past and how it should constrain the future.


The political convulsions of the past decade have fueled acute interest in constitutional norms. Also known as “conventions” or “customs,” constitutional norms are common and recognizable. For example, from 1937 to the present, neither of the political branches has attempted to enlarge the size of the Supreme Court beyond nine justices.1  See infra Section IV.B; Joanna R. Lampe, Cong. Rsch. Serv., LSB10562, “Court Packing”: Legislative Control over the Size of the Supreme Court (2020). Similarly, presidents have, since the Founding, appointed principal officers during both intra- and inter-session Senate recesses and have not, until recently, interfered with agency adjudications or directed criminal investigations from the White House. Each of these practices constrains government officials, yet enjoys, at best, shallow textual foundations. Their prominence suggests that the Constitution in action is composed of, and sustained by, more than just the law on the books. Norms, in other words, form a central part of the “constitutional order”: the broader “set of rules, doctrines, and practices that structure political decisionmaking.”2Sanford Levinson & Jack M. Balkin, Democracy and Dysfunction 21 (2019).

Norm erosion, already a scholarly concern, accelerated during the Trump presidency, and public alarm about the future of democracy became routine.3There are countless examples, and these span the political spectrum: see, for example, Emily Bazelon, How Do We Contend with Trump’s Defiance of ‘Norms’?, N.Y. Times (July 11, 2017), https://www.nytimes.com/2017/07/11/magazine/how-do-we-contend-with-trumps-defiance-of-norms.html [perma.cc/P7AR-2LXS] (“Trump’s flouting of norms . . . has become a defining feature of his presidency. Along the way, he has exposed flaws in the structure of American governance that haven’t surfaced in modern times, mainly because no other president has probed them.”); Tom McCarthy, Donald Trump and the Erosion of Democratic Norms in America, Guardian (June 2, 2018, 8:24 AM), https://www.theguardian.com/us-news/2018/jun/02/trump-department-of-justice-robert-mueller-crisis [perma.cc/W3Z9-Z39E] (asking from the left whether Trump’s disregard for the Department of Justice’s independence represents “a constitutional crisis of some kind or even an erosion of the rule of law”); Michael Sean Winters, Opinion, Trump Threatens Norms That Make the Constitution Work, Nat’l Cath. Rep. (Nov. 6, 2017), https://www.ncronline.org/news/opinion/distinctly-catholic/trump-threatens-norms-make-constitution-work [perma.cc/VP42-QH7E] (“Perhaps the most damage done to the Constitution this year has less to do with any specific flouting of the rule of law than the president’s general disregard for that rule, and for the constitutional and democratic norms that make the rule of law possible.”). Although often associated with political and legal liberals, these anxieties have a conservative cast, as they reflect a commitment to the status quo and conviction about what interbranch comity or the rule of law should look like.4Dissenting voices were few but important. See Jedediah Britton-Purdy, Normcore, Dissent, Summer 2018, https://www.dissentmagazine.org/article/normcore-trump-resistance-books-crisis-of-democracy [perma.cc/DA63-FH47] (critiquing recent scholarship for its fixation on norms and comparative neglect of ideology); Corey Robin, Democracy Is Norm Erosion, Jacobin (Jan. 29, 2018), https://www.jacobinmag.com/2018/01/democracy-trump-authoritarianism-levitsky-zillblatt-norms [perma.cc/QKK2-YSYZ] (critiquing scholars of norms for prioritizing stability over democracy). The same preservationist instinct drives interpretive theories that transform historical practice into constitutional law.5  See infra Conclusion. And in four different cases during its October Term 2019, on issues ranging from agency structure6In striking down the Consumer Financial Protection Bureau’s single director structure with for-cause removal protection as unconstitutional, the Court split 5–4 over whether the agency was a “historical anomaly.” Seila L. LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2202 (2020). and the electoral college,7A unanimous Court emphasized that independent electors are historical “anomalies only” and held that states can penalize electors who break their pledge and vote for a presidential candidate other than the one who wins the state’s popular vote. Chiafalo v. Washington, 140 S. Ct. 2316, 2328 (2020). to the constitutionality of state8The majority, citing “200 years of precedent establishing that Presidents, and their official communications, are subject to judicial process,” rejected the president’s claim that Article II and the Supremacy Clause barred or required a heightened standard for state subpoenas. Trump v. Vance, 140 S. Ct. 2412, 2418 (2020). and congressional subpoenas,9“[A] significant departure from historical practice” pushed the majority to craft a new four-part standard for judicial review of congressional subpoenas. Trump v. Mazars USA, LLP, 140 S. Ct. 2019, 2031 (2020). the Supreme Court relied on past practice in precisely this way.

Such intense attention makes it all the more peculiar that we lack a general account of constitutional norms. This is not to say we lack definitions; we have many. Consider several leading ones:

  • “Conventions” are “maxims, beliefs, and principles that guide officials in how they exercise political discretion.”

These definitions are incomplete and inadequate because they do not answer at least one or more of three major questions. First, if norms or conventions play so many different roles in a constitutional order, is there something they all must do? Second, what, aside from informality, makes them conventional? Or are they just catch-all terms for non-legal rules? Third, what makes these practices constitutional?

This Article advances an original theory of constitutional norms that answers these questions. It defines them and explains their general character: they are normative, contingent, and arbitrary practices that implement constitutional text and principle. This definition is simple but powerful. It identifies a constitutional norm’s salient features, its basic function, and its relationship to the Constitution. And it expands on and distills intuitions either absent from or present but submerged in existing accounts. Drawing on this definition, this Article not only clarifies a fundamental aspect of the constitutional order, it also reveals deep tensions in the use of historical practice in constitutional interpretation.

The Article proceeds as follows. Part I critiques the recent literature on constitutional norms. In many precincts of the legal academy, the importance of norms has long been recognized. Private law scholars, for instance, have emphasized the ways that norms (or conventions as they are more often known in private law scholarship) help coordinate behavior in contracts, family law, and criminal law.10  See, e.g., Robert C. Ellickson, Order Without Law: How Neighbors Settle Disputes (1991); Eric A. Posner, Law and Social Norms 72, 89 (2002) (explaining the logic of family and criminal law through informal game theory); Richard H. McAdams, Beyond the Prisoners’ Dilemma: Coordination, Game Theory, and Law, 82 S. Cal. L. Rev. 209 (2009) (encouraging scholars to adopt coordination games in legal analysis); Ian Ayres, Playing Games with the Law, 42 Stan. L. Rev. 1291 (1990) (reviewing Eric Rasmusen, Games and Information: An Introduction to Game Theory (1989)). And for legal positivists—the dominant school of contemporary jurisprudence—law itself is grounded in convention. H.L.A. Hart famously placed conventions at the heart of modern legal systems,11H.L.A. Hart, The Concept of Law (3d ed. 2012). and many of his successors have remained committed to the same conventional approach.12  E.g., Joseph Raz, The Concept of a Legal System (2d ed. 1980); Joseph Raz, Practical Reason and Norms (Oxford Univ. Press 3d ed. 1999) (1975); Gerald J. Postema, Coordination and Convention at the Foundations of Law, 11 J. Legal Stud. 165 (1982). The most important critique of legal positivism’s conventionalist picture of law is chapter four in Ronald Dworkin, Law’s Empire (1986). While Dworkin’s critique has been generative, it has not displaced legal positivism as the dominant theory of law. American constitutional law scholarship, by contrast, has given conventions sporadic attention. While the British jurist A. V. Dicey observed nearly a hundred fifty years ago that conventions accompanied written and unwritten constitutions alike,13A.V. Dicey, Introduction to the Study of the Law of the Constitution (Liberty Fund 1982) (1885). the existence of norms or conventions is, as Adrian Vermeule wryly noted, a “revelation” that “bursts upon American constitutional scholars every other generation or so, and is lost in the succeeding generation.”14Adrian Vermeule, Conventions in Court, 38 Dublin U. L.J. 283, 283–84 (2015).

The current renaissance in norm scholarship, Part I argues, has taken three forms: strategic approaches, democratic alarmism, and thick institutional description. Each comes with characteristic limits. Strategic approaches embrace a game-theoretic view of constitutional norms that narrowly focuses on their role in coordinating the behavior of governmental agents. Democratic alarmists defend the necessity of norms but often ahistorically and without marking the boundaries between political and constitutional norms. And thick descriptivists offer rich accounts of norms within particular branches yet stop short of a broader definition. As a result, the existing literature lacks a general theory of norms that explains their functions, features, and relationship to the Constitution.

This Article addresses these issues by adopting a different strategy from current approaches. While recent work is uniformly inductive—it begins by describing examples of norms and then tries to distill their essence—this Article is primarily deductive. It treats the question “what is a constitutional norm?” as two separate inquiries. First, what makes a constitutional norm or convention (indeed, any social practice) conventional?15“Norm” and “convention” are used interchangeably in this Article. I stick primarily to “norm” to avoid confusion since “convention,” at least in American constitutional law, is often associated with the events of 1787 or any formal process for drafting a constitution. Because I want to surface and map the relationship between the philosophy of social conventions and constitutional theory, I use the terms “conventions” and “conventional” only in their philosophical sense. Nor is there is any systematic distinction between “norm” and “convention” in the literature. Scholars either use them as perfect substitutes or choose one for semantic reasons. See, e.g., Josh Chafetz & David E. Pozen, How Constitutional Norms Break Down, 65 UCLA L. Rev. 1430, 1434 n.14 (2018) (“[N]othing important hangs on the distinction (to the extent it exists) between constitutional norms and constitutional conventions.”); Renan, supra note 11, at 2196 n.34 (using “norm” instead of “gloss” or “convention” because “scholars have sometimes used [the latter two terms] to distinguish legally enforceable norms from extralegal norms”). As this Article shows, rejecting the term “convention” also means losing the philosophical insight the concept provides. Moreover, “gloss” and interpretive theories that rely on “historical practice” must be carefully distinguished from the practices themselves. The former endows the latter with significance that they do not independently possess. Second, what makes it constitutional? This strategy has important advantages over a more inductive approach. Focusing on particular examples can obscure what they share with conventions more generally; an exclusively inductive approach thus risks exceptionalism about constitutional norms, exaggerating their distinctiveness and inviting alarm when they begin to change or fade. Inductive approaches also lack specific criteria for identifying when a norm is constitutional. This problem flows downstream from confusion about conventionality. Explaining why a constitutional norm or convention is conventional provides an antecedent conceptual framework in which to place constitutionality. In this view, a constitutional convention is a convention first, constitutional second.

Part II explains what makes a norm or a convention conventional. Drawing on the philosophy of social conventions, this Part lays out what conventions do and what their common features are. Conventions generally fall into two categories: they either coordinate action or concretize values into practices. Coordinating conventions are famous; a familiar example is a rule requiring drivers to stick to one side of the road. Until now, scholars have either assumed that constitutional norms are coordinating conventions or have foregone asking what makes them conventional at all.16  See infra Part I. Both approaches are mistaken. Constitutional norms are always constitutive conventions. Just as the conventions of etiquette concretize the concept of “politeness,” constitutional norms implement otherwise abstract principles, like the separation of powers, or indeterminate text, such as “advice and consent,” into specific practices. Norms translate constitutional word into deed. The complete absence of constitutive conventions in recent work is therefore notable since constitution, much more so than coordination, helps make sense of what norms do in constitutional politics and why change provokes turmoil.17This sense of “constitution,” as a verb rather than a noun, also reflects an historical meaning that has been lost. Constitutional thought until the American Revolution often referred to the substance of a social order, not just a set of formal legal rules. See Graham Maddox, “Constitution,” in Political Innovation and Conceptual Change 50, 50 (Terence Ball, James Farr & Russell L. Hanson eds., 1989); see also Hanna Fenichel Pitkin, The Idea of a Constitution, 37 J. Legal Educ. 167, 168 (1987) (“[O]ur constitution is neither something we have nor something we are so much as something we do—or at any rate can do.”); cf. David Singh Grewal & Jedediah Purdy, The Original Theory of Constitutionalism, 127 Yale L.J. 664, 669–81 (2018) (reviewing Richard Tuck, The Sleeping Sovereign: The Invention of Modern Democracy (2016)) (outlining the modern theory of constitutionalism). Earlier thinkers imagined legal orders as established through practices. Larry D. Kramer, The People Themselves: Popular Constitutionalism and Judicial Review 9–13 (2004) (discussing the persistence of this view in English legal culture on the eve of the American Revolution). By shifting our focus from text to practice, the Article recovers this more expansive and richer view of a “constitution.” What this Article achieves through conceptual analysis, others have done through careful historical work. See, e.g., Jonathan Gienapp, The Second Creation: Fixing the American Constitution in the Founding Era (2018).

Understanding what norms do—concretizing values into practices—illuminates their key features. First, these practices, as their names suggest, are normative. By operationalizing constitutional text and principle, norms command respect and allegiance. And because constitutional norms are constitutive conventions, they are normative in a thicker sense than coordinating conventions; norms are not, or are not just, rules that meet functional needs, but rather practices that create the terrain of constitutional morality.18A constitutional norm can also coordinate action, but it must concretize values into practices. A given era’s constitutional norms reflect how people think constitutional text or principles should work.

Second, norms are contingent. If and when a norm emerges and how long it survives depends on a variety of historical forces. As others have observed, “constitutional norms are perpetually in flux.”19Chafetz & Pozen, supra note 19, at 1430. Because they are both weaker than law and depend on various intellectual, political, and social conditions for their survival, constitutional norms are inherently provisional.

Third, and most importantly, constitutional norms are arbitrary. A given norm represents only one of many possible ways of concretizing a principle into practice. Arbitrariness results directly from the indeterminacy of constitutional text and principles. It is no coincidence, for instance, that the separation-of-powers scholarship has been dominated by norm-talk. Conceptually amorphous,20  See, e.g., Peter L. Strauss, The Place of Agencies in Government: Separation of Powers and the Fourth Branch, 84 Colum. L. Rev. 573, 577–78 (1984) (discussing three different models of understanding the separation of powers). with weak textual foundations,21  See John F. Manning, Separation of Powers as Ordinary Interpretation, 124 Harv. L. Rev. 1939 (2011) (underlining the lack of clear textual guidance in many areas of separation-of-powers issues). separation-of-powers principles can be plausibly, if contestably, implemented by different norms. This constellation of features—normative, yet contingent and arbitrary—also suggests why constitutional norms have engendered both fascination and alarm. How could such a pervasive part of the constitutional order be intrinsically arbitrary? I postpone that question and its implications for constitutional interpretation until the conclusion.

Part III explains how constitutional norms are constitutional. It uses the conceptual structure built in the previous Part to define constitutional norms. They are normative, contingent, and arbitrary practices that implement constitutional text and principle. The latter half of the definition shows how these norms are constitutional: by virtue of the particular values they instantiate. For instance, a norm that required presidents to show respect for political opponents would not count as a constitutional norm under this Article’s theory. Even though the norm guides a constitutional actor’s—the president’s—behavior, it has little connection to the Constitution. And theories that connect a norm’s constitutionality to the identity of the actors they constrain risk being overinclusive since they do not offer a way of distinguishing between political and constitutional norms.22  See infra Part I. By contrast, this Article’s theory is practice-centric. Constitutionality stems from the nature of the practice itself, not those it directs. So while all constitutional norms channel a constitutional actor’s behavior, not every norm these actors follow is constitutional. Only when the practice itself is understood to implement constitutional text and principle is it a constitutional norm.

Part III then briefly considers a potential challenge to the distinction between constitutional law and norm. Conventionalist theories of law hold that law itself is a type of convention; why, then, keep the distinction between constitutional law and norm at all? This Part offers a modest defense of the distinction based on the difference between law as a “deep” convention and constitutional norms as “surface” conventions, as well as the ways each are enforced.

Part IV shows this Article’s theory in action through three case studies: blue slips, the norm against court-packing (“anti-court-packing”), and executive noninterference with law enforcement. Each of these norms implements constitutional text and principle, from senatorial “advice and consent”23U.S. Const. art. II, § 2. (blue slips), to judicial independence (anti-court-packing), to the president’s duty to “take [c]are that the [l]aws be faithfully executed”24  Id. art. II, § 3. (executive noninterference). As the case studies show, each is normative, contingent, and arbitrary. And they have all come under heavy pressure. Part IV also traces these norms’ trajectories—from their origins to their current crises—to expose their conventional characters. This historical approach is helpful for two reasons. First, these examples animate the Article’s theory; they show concretely how norms are contingent and arbitrary by tracking change over time. Second, the case studies illustrate the relationship between a norm’s contingency and arbitrariness. Because a norm’s underlying text and principle are indeterminate, the practices that emerge depend on context. Political events, ideology, and sheer chance shape the development of constitutional norms.25As Richard Primus has suggested, the practical stakes of constitutional cases can drive interpretation. Richard Primus, The Cost of the Text, 102 Cornell L. Rev. 1649, 1658 (2017). As conditions change, so do norms. When a norm transforms or collapses, it reveals the arbitrariness of prior arrangements: alternatives that we had previously rejected or not considered now become imaginable or even compelling.

The Article concludes by briefly considering its implications for the use of historical practice in constitutional interpretation. Practice-based theories presuppose that a practice should enjoy a legal status by virtue of its conventional one. From originalism in the “construction zone,”26  See, e.g., Keith E. Whittington, Constitutional Construction (1999); Randy E. Barnett, Interpretation and Construction, 34 Harv. J.L. & Pub. Pol’y 65, 65–72 (2011); Lawrence B. Solum, Originalism and Constitutional Construction, 82 Fordham L. Rev. 453 (2013). to “liquidation”27William Baude, Constitutional Liquidation, 71 Stan. L. Rev. 1 (2019). and “historical gloss,”28  E.g., Curtis A. Bradley & Trevor W. Morrison, Historical Gloss and the Separation of Powers, 126 Harv. L. Rev. 411 (2012); Curtis A. Bradley & Neil S. Siegel, Historical Gloss, Madisonian Liquidation, and the Originalism Debate, 106 Va. L. Rev. 1 (2020). to “unwritten constitutionalism,”29  See, e.g., Akhil Reed Amar, America’s Unwritten Constitution (2012); Charles L. Black, Jr., Structure and Relationship in Constitutional Law (Ox Bow Press 1985) (1969); David A. Strauss, The Living Constitution (2010); Laurence H. Tribe, The Invisible Constitution (2008). these theories all share this basic structure; they move from institutional “is” to constitutional “ought.” And just as in ethics,30  See, e.g., Rachel Cohon, Hume’s Moral Philosophy, in Stanford Encyclopedia of Philosophy (Edward N. Zalta ed., 2018), https://plato.stanford.edu/entries/hume-moral/#io [perma.cc/W6YL-PCTL]. the shift from “is” to “ought” in constitutional law is dubious and raises two critical questions: first, the descriptive question: What makes a constitutional norm or convention conventional? And second, the normative question: Why should a practice’s conventional status make it constitutional?

This Article answers the descriptive question, laying the foundation for addressing the normative one. The answer flows from an uncontroversial assumption: the nature of a potential source of law shapes our usage of it. That premise underlies, for instance, the turn to the philosophy of language to understand the character of statutory and constitutional text.31  See, e.g., Christopher R. Green, Originalism and the Sense-Reference Distinction, 50 St. Louis U. L.J. 555 (2006) (using Gottlob Frege’s distinction between the “sense” and “reference” of words); John Mikhail, The Constitution and the Philosophy of Language: Entailment, Implicature, and Implied Powers, 101 Va. L. Rev. 1063 (2015) (applying Gricean philosophy of language to constitutional theory); Lawrence B. Solum, Semantic Originalism, (Ill. Pub. L. Rsch. Paper, Paper No. 07-24, 2008), https://doi.org/10.2139/ssrn.1120244 (advancing a set of theses about the semantic content of the Constitution). In fact, the relevance of linguistic philosophy is so well accepted that a recent Court decision featured a debate about the “conversational conventions” of Title VII. Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1745 (2020). Because the current work has foregone asking what makes a norm conventional, practice-based theories have proliferated without a clear view of their legal materials. This Article’s philosophical investigation both fills that gap and extends beyond academic debate. As the Court’s October Term 2019 revealed, the political battles of the Trump era often led to litigation that required judicial resolution.32  See supra notes 6–9 and accompanying text. Even if conflict abates under a new president, in a world where courts routinely consult past practice, contestation over norms risks turning into legal battles.

This Article’s theory suggests that when scholars and judges draw on historical practice, they build their theories and decisions on vexed foundations. Once we recognize that constitutional norms are arbitrary and contingent, we realize that interpretive theories that privilege practice do so for reasons beyond the practices themselves. Whether it is judges applying “historical gloss” or scholars searching for “liquidated” meanings, all are implicitly relying on independent theories of legal normativity to ground their claims. In constitutional law and theory, something other than a practice itself is needed to justify its elevation into law. After all, how can we reliably turn to practices that themselves are historically contingent and conceptually arbitrary? Having raised this concern, this Article leaves a fuller reckoning for future work.

I.        Understanding Norms: Current Approaches

Recent legal scholarship on norms is marked by diversity in mood and method. There are three primary modalities: game theory, democratic alarmism, and thick institutional description.33There are two bodies of scholarship I do not include here for different reasons: (1) American political development (APD) literature on norms and (2) legal scholarship on historical gloss and constitutional interpretation. The first group is vast and better understood as a subset of political science. For this Article, the most relevant scholars of the genre share much in common with thick descriptivists. APD includes rich descriptive and theoretically informed work on particular branches of government. For exemplary work, see Josh Chafetz, Congress’s Constitution (2017) (exploring Congress’s various “hard” and “soft” powers and the way the successful exercise of these powers bolsters it institutional power against the other branches); Josh Chafetz, Democracy’s Privileged Few (2007) (comparing the special powers and privileges enjoyed by British and American legislators); Keith E. Whittington, Partisanship, Norms, and Federal Judicial Appointments, 16 Geo. J.L. & Pub. Pol’y 521 (2018) (tracking the escalating battles over judicial appointments and offering possible solutions given continued polarization). See also Julia R. Azari & Jennifer K. Smith, Unwritten Rules: Informal Institutions in Established Democracies, 10 Persps. on Pol. 37 (2012) (explaining that “informal institutions” complete, parallel, or coordinate). Despite their importance, I exclude this work largely because of a difference in emphasis. The project here is almost entirely theoretical and its results bear directly on modes of constitutional interpretation. Even Smith and Azari’s piece, which importantly highlights that informal practices can “fill gaps,” pays no attention to the normativity and arbitrariness of conventions and the implications for constitutional theory. This is the result of an important difference in audience. They speak to political scientists and their analysis is steeped in the language of rational choice theory. I am interested in a related but different question: what sort of practice is a constitutional convention and how do the features and functions of conventions bear on constitutional theory. The second excluded group—historical gloss and unwritten constitutionalism—includes both classic and recent work. For classics of unwritten constitutionalism, see, for example, sources cited supra note 33. For recent work on historical gloss, see, for example, Baude, supra note 31; Bradley & Morrison, supra note 32; Bradley & Siegel, supra note 32; Stephen M. Griffin, Against Historical Practice: Facing Up to the Challenge of Informal Constitutional Change, 35 Const. Comment. 79 (2020); Alison L. LaCroix, Historical Gloss: A Primer, 126 Harv. L. Rev. F. 75 (2013); Samuel Issacharoff & Trevor Morrison, Constitution by Convention, 108 Calif. L. Rev. 1913 (2020); Katherine Shaw, Conventions in the Trenches, 108 Calif. L. Rev. 1955 (2020); Daphna Renan, The President’s Two Bodies, 120 Colum. L. Rev. 1119 (2020). For an example of practice-based interpretation that is transparent about its normative analysis, see Jonathan S. Gould, Codifying Constitutional Norms, 109 Geo. L.J. 703 (2021). Except for the brief discussion in the conclusion, I leave the relationship between historical practice and constitutional interpretation for other work. Much of this scholarship is focused on courts and how they should handle historical, non-legal precedent. This can often bypass—Baude excluded, perhaps—what a constitutional norm is. This Part first maps this scholarly terrain, placing the present burst of interest in norms in context and describing the different approaches and priorities of these three modes. It then explains the two principal shortcomings with current work: (a) incomplete theory and (b) historically blinkered anxiety about conventional change. While these problems arise unevenly, they are connected: the lack of a general theory makes it hard to explain both conventional change and the reactions that attend it.

A.     Norms Scholarship: Past and Present

In the broader Anglophone world, scholarly interest in norms is long-standing. Dicey provided the first major account in his classic, Introduction to the Study of the Law of the Constitution. Dicey distinguishes between “the law of the constitution” and “conventions of the constitution.”34Dicey, supra note 17, at cxli. The former, he explains, “are in the strictest sense ‘laws’ since they are rules which (whether written or unwritten, whether enacted by statute or derived from the mass of custom, tradition, or judge-made axioms known as the Common Law) are enforced by the Courts.”35  Id. at cxl–cxli. By contrast, constitutional conventions comprise “understandings, habits, or practices which, though they may regulate the conduct of the several members of the sovereign power, of the Ministry, or of other officials, are not in reality laws at all since they are not enforced by the Courts.”36  Id. at cxli. One famous British example is “The King must assent to, or (as it is inaccurately expressed) cannot ‘veto’ any bill passed by the two Houses of Parliament.”37  Id. at cxlii (footnote omitted). Dicey includes other practices in his list of examples including “the House of Lords does not originate any money bill” and “[m]inisters resign office when they have ceased to command the confidence of the House of Commons.” Id. This rule has no explicit textual basis.

For Dicey, “[i]t [was] to be regretted that these maxims must be called ‘conventional,’ for the word suggests a notion of insignificance or unreality.”38  Id. at cxliii. While some of these conventions could be “trivial,” many are “as important as any laws”39  Id. and make up what he calls “constitutional morality.”40  Id. at cxli. Conventions, Dicey continues, are just as potent in a written constitution like America’s, where “stringent conventional rules, which, though they would not be noticed by any Court, have in practice nearly the force of law.”41  Id. at cxliv. Examples of such norms included the traditions of presidents not running for a third term prior to the passage of the Twenty-Second Amendment and a state’s electors casting their votes for the winner of a state’s popular vote. The latter convention, of course, was at issue in Chiafalo v. Washington, 140 S. Ct. 2316, 2320 (2020). As Part II will show shortly, Dicey’s phrase—“constitutional morality”—is telling. It states, ipse dixit, the normativity of conventions.

Dicey’s work did not go unnoticed. Contemporaries on both sides of the Atlantic—Woodrow Wilson42  See Woodrow Wilson, Constitutional Government in the United States 22 (1908). and James Bryce43  See James Bryce, Flexible and Rigid Constitutions, in 1 Studies in History and Jurisprudence 145, 233–34 (1901); James Kirby, A.V. Dicey and English Constitutionalism, 45 Hist. Eur. Ideas 33, 42, 44 (2019).—shared Dicey’s conviction that norms were central in American government. And no less an authority than James Thayer cited Dicey to critique judicial review.44  See James B. Thayer, The Origin and Scope of the American Doctrine of Constitutional Law, 7 Harv. L. Rev. 129, 130 (1893). But this early awareness of conventions and their importance did not endure. In the subsequent century, norms largely faded from legal consciousness.45Possible exceptions include several articles from the mid-1970s and 1980s exploring the idea of an “unwritten constitution.” See, e.g., Thomas C. Grey, Do We Have an Unwritten Constitution?, 27 Stan. L. Rev. 703 (1975); Lawrence Gene Sager, Fair Measure: The Legal Status of Underenforced Constitutional Norms, 91 Harv. L. Rev. 1212 (1978); Thomas C. Grey, The Constitution as Scripture, 37 Stan. L. Rev. 1 (1984); Thomas C. Grey, The Uses of an Unwritten Constitution, 64 Chi.-Kent L. Rev. 211 (1988); Suzanna Sherry, The Founders’ Unwritten Constitution, 54 U. Chi. L. Rev. 1127 (1987); Michael S. Moore, Do We Have an Unwritten Constitution?, 63 S. Cal. L. Rev. 107 (1989). These articles were part of a vibrant debate about constitutional interpretation in the wake of the Due Process Revolution. They emerged alongside and later responded directly to originalism. They were squarely focused on first-order interpretive questions, such as the existence of implicit normative principles in the Constitution or the sources judges might permissibly consult in deciding a constitutional case. These are related but different questions from the one I focus on here and Dicey and others considered a century ago: how to understand long-standing practices of constitutional government. The relationship between historical practice and constitutional interpretation, in my view, requires first sorting out the former.

Scholars have now returned to norms with enthusiasm. Mark Tushnet prefigured recent interest in his 2004 article, Constitutional Hardball. “[C]onstitutional hardball,” Tushnet explains, are practices “that are without much question within the bounds of existing constitutional doctrine and practice but that are nonetheless in some tension with existing pre-constitutional understandings.”46Mark Tushnet, Constitutional Hardball, 37 J. Marshall L. Rev. 523, 523 (2004). Tushnet’s intervention was prompted by political battles in the early 2000s between Democrats and Republicans, in both state and federal government, that departed from time-honored practices.47Two of Tushnet’s examples include Senate Democrats’ filibuster of President George W. Bush’s judicial nominations in 2002–03 and attempts by Colorado and Texas to redistrict between censuses. Id. at 524–27. These episodes, he argued, could be traced all the way back to Marbury v. Madison.48  Id. at 538–543. In exploring this tradition of conflict, Tushnet advanced a preliminary hypothesis: constitutional hardball was “a symptom of the possibility of a shift in the governing assumptions of a constitutional order.”49  Id. at 544. Norm erosion, in other words, accompanied political convulsion. This idea anticipated key themes of subsequent work: the language of games, the historicity of conventions, and the force of political incentives.

The literature has developed fitfully since Tushnet’s piece, with the bulk of it published this decade.50  See, e.g., Joseph Fishkin & David E. Pozen, Essay, Asymmetric Constitutional Hardball, 118 Colum. L. Rev. 915 (2018); Neil S. Siegel, Political Norms, Constitutional Conventions, and President Donald Trump, 93 Ind. L.J. 177 (2018). Its timing, from the tail end of the Obama Administration to the Trump Administration, is telling: scholarly interest followed the rise of political conflict and associated conventional breakdown. The academic reaction has been thoughtful but unprogrammatic. Scholars have thrown the kitchen sink at a phenomenon that sits squarely at the intersection of law and politics, defying either category. The resulting work can be parsed into three major strands: game theory-inspired work, thick institutional description, and democratic alarmism.51These labels do not track differences along one particular axis. Game theory-inspired articles deploy a common language, while thick institutional descriptions share a method of studying similar objects, and alarmists are united by sensibility. Nevertheless, the categories track common themes and help organize an otherwise disorganized field.

1.       Strategic Approaches

Strategic treatments of constitutional norms share a common premise about the behavior of agents. While none of the scholars who write in this vein explicitly use models, their work is informed by game theory. First, they all borrow the language of game theory. Second, by emphasizing strategy, they focus solely on coordinating conventions, which I explore further in Part II.

On the strategic view, constitutional actors are best understood as utility-maximizers. Their interactions with each other resemble a game. Norms exist because they redound to the benefit of everyone involved. They break down due to changed incentives. These scholars tend to focus on pitched constitutional battles, where the relevant actors, conventions, and political motives are relatively clear. Most work in this style imposes clarity on norms by flattening them. There is no need to prove that a given situation is a coordination game, let alone consider whether the convention is doing something other than solving a game. The result is often entirely functionalist: norms exist and survive only because agents think they are useful. When the incentives change, so do the norms.

There is diversity within this camp. Some scholars owe much to game theory, while others note its explanatory limits. Three examples illustrate the shared assumptions of the framework and its family differences. First, Adrian Vermeule’s work: in a 2005 piece, Vermeule revisits the court-packing battle of 1937 in an effort to explain why Supreme Court reform is so difficult.52Adrian Vermeule, Essay, Political Constraints on Supreme Court Reform, 90 Minn. L. Rev. 1154 (2006). His analysis is steeped in the language of game theory, as he attributes FDR’s failure to pack the Court to a “widespread perception that the court-packing plan was a disingenuous proposal,” and draws the broader lesson that “any political actor who seeks to change the rules in the middle of the game is untrustworthy, presumptively motivated by partisan advantage or a desire for unchecked power.”53  Id. at 1163. Vermeule’s subsequent scholarship sounds a similar message, from his co-authored theory of “constitutional showdowns”54Eric A. Posner & Adrian Vermeule, Constitutional Showdowns, 156 U. Pa. L. Rev. 991, 1033 (2008) (“In general, the most plausible case for the emergence of efficient custom involves conditions of symmetry and reciprocity, in which agents know that they will be on both sides of similar transactions over time and thus have an incentive to follow the rule that maximizes aggregate welfare for all concerned.”). to his comparative study of judicial enforcement of conventions, at home and abroad.55  See Vermeule, supra note 12.

Second, David Fontana, in his work on judicial politics during the Obama presidency, follows Vermeule, albeit less expansively. He explains the relative moderation of Obama judicial appointees and Obama’s missed opportunity to reshape the courts as failures of strategic action: “excessive cooperation with political forces that do not manifest the same behavioral patterns of cooperation.”56David Fontana, Cooperative Judicial Nominations During the Obama Administration, 2017 Wis. L. Rev. 305, 307. The Obama Administration and Senate Republicans were simply playing two different games, a mismatch the latter exploited. Fontana stops well short of Vermeule, however, in limiting his analysis to a specific context; he neither has nor purports to have a broader theory of norms.

Third, Josh Chafetz, Joey Fishkin, and David Pozen offer a careful use of the strategic style. Their contributions consider norms with greater generality than Fontana but with less abandon than Vermeule. For instance, Chafetz and Pozen, in a recent article, observe that constitutional norms break down, “[i]f changes in the institutional environment, the wider world, or the views of the relevant segments of the public raise the expected cost of adherence to a norm, such circumstances may arise with greater frequency and thereby weaken the norm’s regulative force.”57Chafetz & Pozen, supra note 19, at 1442. While using a strategic register, they also note the normative uncertainty of conventional breakdown: “After all, the mere fact that members of a community conform to certain behavioral regularities . . . does not make those behavioral regularities good.”58  Id. at 1445. Indeed, this entire section in Chafetz and Pozen’s piece—“Some Normative Implications of Norm Instability”—is one of the few instances in the literature where a general view of the normative consequences of conventional breakdown is considered. My project here is importantly different; I am interested in how a convention becomes normative in the first place. In other words, how does a norm make up our constitutional morality? See infra Section II.A.3. As I show in Part II, this sort of claim is only possible when we move beyond purely coordinating conventions.

Fishkin and Pozen show similar care in their work. On the one hand, they extend Tushnet’s idea of constitutional hardball to the present day. They argue that as political polarization has deepened, hardball has had a partisan tilt, with Republicans more likely to “play hardball” than Democrats.59Fishkin & Pozen, supra note 54. And when explaining why Democrats should avoid reciprocating Republican tactics, they turn to “two basic game theoretic models,” one of which predicts that “[r]amping up constitutional hardball . . . is a dangerous game to play over any extended period of time.”60  Id. at 979–80. On the other hand, they wisely note that “[g]ame theory itself cannot answer which model”—one counseling escalation and the other against—“is more plausible.”61  Id. at 980 n.261. These scholars thus couch their claims in game theory, while underlining its analytic and descriptive limits. Their concessions highlight the limitations of the strategic view.

2.       Democratic Alarmism

If the strategists buy clarity at the cost of completeness, democratic alarmists provide an account of constitutional norms without bounds. Instead of a common language, democratic alarmists are united by a shared sensibility: urgency fueled by the rise of domestic and global populism. To authors in this camp, the breakdown of constitutional norms represents not only a change in political culture but a threat to democracy itself. For example, when Majority Leader Mitch McConnell and Senate Republicans refused to hold any hearings on the Supreme Court nomination of Merrick Garland, these scholars saw an attack on the separation of powers, the divide between law and politics, and the uniformity of law. Democratic alarmists thus come closest to the journalistic register, as they try to make sense of the current presidency in both historical and theoretical terms. In the process, however, they tend to elide the difference between constitutional norms and broader political norms, which makes it hard to see what makes any given norm constitutional.

Neil Siegel is the primary exponent of this view among constitutional law scholars. In a pair of recent articles, Siegel has argued that President Trump’s consistent flouting of norms has exposed their importance and precarity.62Neil S. Siegel, supra note 54; Neil S. Siegel, After the Trump Era: A Constitutional Role Morality for Presidents and Members of Congress, 107 Geo. L.J. 109 (2018) [hereinafter Siegel, Constitutional Role Morality]. Indeed, for Siegel, the “[i]ncreasing disregard of political norms and constitutional conventions by candidates and elected officials” during the Trump era “is one indication that we have lost our way, and figuring out how to encourage greater respect for them may help us find our way back.”63Siegel, supra note 54, at 179. Siegel not only diagnoses our present ills but prescribes remedies. One of his articles develops a “constitutional role morality,” a set of ethical principles for guiding and constraining the behavior of elected officials.64Siegel, Constitutional Role Morality, supra note 66. These principles are drawn from various sources; Thomas Hobbes, Jean-Jacques Rousseau, Edmund Burke, James Madison, and Robert Post are all enlisted in the cause of constitutional democracy.65  Id. at 127–137. For Siegel, both the work of these diverse authors and the Constitution itself embody certain principles: “democracy as collective self-governance” and the pursuit of “well-functioning federal government.”66  Id. at 127–144. What either of those principles means remains fairly underdeveloped in theory or in practice, and Siegel acknowledges concerns about implementation. Nevertheless, he stresses the urgency of his project since “in a deeply polarized country in which politicians who hold high office too often act as if there are no non-legal role restraints, the immediate task is to develop the vision itself.” Id. at 170.

Other alarmists share Siegel’s diagnosis but take a more global view. Comparative political scientists Steven Levitsky and Daniel Ziblatt offer the most provocative account in How Democracies Die. Levitsky and Ziblatt draw on case studies from twentieth-century Latin America and Western Europe (their respective areas of expertise) to conclude that American democracy is under threat because of “[t]he erosion of our democratic norms.”67Steven Levitsky & Daniel Ziblatt, How Democracies Die 9 (2018). They claim that two norms in particular, “mutual toleration” and “forbearance,” have sustained democracy in America since its inception and that these are the very norms most threatened in the current moment.68  Id. at 102. To their credit, they do not present these norms as distinctly constitutional. Instead, they argue that these norms are basic principles of political morality, and the election of Trump has hastened their decline.69  Id. at 8. Just as their decline is due to political causes, their restoration must also be political, requiring both elite cooperation and popular mobilization. Aziz Huq and Tom Ginsburg sound a similar alarm. While they describe the erosion of norms as specifically “constitutional retrogression,”70Aziz Huq & Tom Ginsburg, How to Lose a Constitutional Democracy, 65 UCLA L. Rev. 78, 94–96 (2018). they too locate the revival of these norms in the “intersubjective understandings of elites and citizens” and American “[i]nstitutional pluralism.”71  Id. at 166–67.

The alarmists’ diagnoses and prescriptions thus vary. Levitsky and Ziblatt, perhaps because of their disciplinary bent, identify a political problem with political solutions. Siegel and Huq and Ginsburg instead cast the current moment as a constitutional crisis that requires political theory and action. These differences are important. They show that we lack a consistent way of distinguishing between constitutional norms and desirable political practices more generally. That lack of clarity makes it hard to tell whether a given crisis reveals cracks in the constitutional structure or merely political upheaval and transition since the evidence relied on—conventional breakdown—is itself unspecified.

The alarmists, at the same time, suggest something urgent about conventions—their normativity. For them, norms shape how people should behave,72  E.g., Siegel, supra note 54, at 179–180. something discounted or bracketed by many strategic approaches.73  See supra Section I.A.1. Yet, because the alarmists are so concerned with fashioning solutions to present crises, they barely examine why and how these practices command our respect and what they have to do with the Constitution. Normativity and constitutionality are assumed but unexplained. These concerns are different from asking whether the collapse of particular norms is troubling. The latter question is more specific and presentist than the former and only underlines the need for a more robust general theory.

3.       Thick Institutional Description

The third and final group stands apart. Thick descriptivists do not purport to diagnose a contemporary crisis of democracy or offer a general theory of norms. Instead, they use a common method—the close study of norms of particular episodes or branches—that reveals the ubiquity of norms. These scholars proceed inductively, as nearly every piece in this genre focuses on the behavior of certain actors in order to draw larger, tentative conclusions about the separation of powers or the Constitution more broadly. And because they work inductively, members of this group are the most historically minded of the three. In paying attention to change over time, they give a sense, if not an outright statement, of the contingency and arbitrariness of norms.

Thick descriptivists have studied all three of the coordinate branches, examining both the actors within them and the interactions between them. The result is a rich and broad-ranging body of work, with deep dives into previously obscure terrain and fresh reexaminations of well-trodden ground. Several scholars have opened a window into, for instance, the previously opaque Office of Legal Counsel (OLC), showing how executive branch lawyers understand their institutional role and interpret constitutional and statutory issues for the president.74  See, e.g., Curtis A. Bradley & Trevor W. Morrison, Presidential Power, Historical Practice, and Legal Constraint, 113 Colum. L. Rev. 1097 (2013) (defending a view of the executive as constrained by the publicity of legal discourse and recognition of historical practice); Trevor W. Morrison, Stare Decisis in the Office of Legal Counsel, 110 Colum. L. Rev. 1448 (2010) (cataloguing the use of stare decisis in the OLC and making a case for an effective but bounded role for it in executive legal interpretation); Daphna Renan, The Law Presidents Make, 103 Va. L. Rev. 805 (2017) (arguing that presidential control shapes the legal culture and practices of the OLC). Others have studied norms internal and external to the judiciary, from political restraint in jurisdiction stripping,75  E.g., Tara Leigh Grove, Article III in the Political Branches, 90 Notre Dame L. Rev. 1835 (2015). to practices that undergird judicial independence.76  E.g., Tara Leigh Grove, The Origins (and Fragility) of Judicial Independence, 71 Vand. L. Rev. 465 (2018). And most recently, Daphna Renan has written pathbreaking work on the norms governing the modern presidency.77Renan, supra note 11; see also Renan, supra note 37.

Professor Grove’s most recent work on the norms of judicial independence evinces the characteristic strengths and limits of this approach. In revisiting important episodes of interbranch conflict, she shows that “even when the constitutional text does not explicitly protect the judiciary from a court-curbing measure, a political norm has filled the gap.”78Grove, supra note 80, at 517. The similarity in language to Azari and Smith’s—“fill gaps”—is notable. Azari & Smith, supra note 37, at 41. Nevertheless, she warns that “it is crucial to recognize the historically contingent nature of these conventions.”79Grove, supra note 80, at 517. Her project demonstrates the idea that conventions are subject to change. If her empirical claim—that norms are contingent—is right, then we still have to answer further questions: Why are they contingent? And if they are, how does a constitutional order persist given this contingency?

The rest of this Article complements thick descriptive work and responds to theoretical issues raised and left unresolved by game theorists and alarmists alike. It provides a conceptual account that explains the structure and character of norms thick descriptivists have indexed. The theory and practice unearthed go hand in hand. The theory illuminates the forces driving conventional change, and the empirical work highlights the ubiquity and importance of norms in our constitutional order.

B.     The Need for a General Theory

The current approaches have two significant limitations. First, they offer an incomplete theory of norms. Second, they consider norms ahistorically and thus view conventional change with deep concern. These problems are joined at the hip. If the theories we have on offer cannot give a full picture of the nature and function of norms, then it is unsurprising that we are alarmed when conventions do change, since we lack a standpoint to evaluate the erosion of any particular norm. This Section examines these problems in greater detail; the next Part addresses them.

The first problem is incompleteness. Consider the mismatch between the strategists and the alarmists. When the latter worry that norm erosion is a sign of crisis, the former respond by pointing out changed incentives. The problem is that describing how context shapes the life of a norm does not make anxiety about its death intelligible. In fact, if we take a cynical view of the matter, explaining norms purely in terms of costs and benefits makes anxiety about their demise seem either naive or disingenuous.80My criticism here parallels Bernard Williams’s criticism of evolutionary psychologists and invisible hand explanations of ethics. Put simply, functional explanations that explain morality as a survival mechanism for the species are limited in two major ways: they can neither explain the persistence of many different moral rules that seem to have no connection to survival, and they have no account of how people themselves understand the norms they follow. Bernard Williams, Truth and Truthfulness: An Essay in Genealogy 27–35 (2002).

Another way of getting at the problem of incompleteness is to press current theories to explain what problems norms are actually solving. This question is especially hard for game theory-inspired work, which trades on the language of instrumental reason. Adrian Vermeule makes the strongest version of this argument when he claims that “[c]onventions always have a coordination component . . . . Requiring that there be ‘a reason for the rule’ . . . assumes away the problem of disagreement over good reasons that creates the need for rule-based coordination in the first place.”81Vermeule, supra note 18, at 289. In essence, any focus on normative argument is at best ancillary, since the point of a norm is to coordinate behavior.

But what exactly are norms coordinating? While some norms do solve so-called coordination problems,82Later on in the Article, I discuss a convention—blue slips—that likely solved coordination problems, at least at its inception. See infra Section IV.A. it is hard to identify the relevant problem or game for many of the norms we are interested in. Consider, for instance, the norm against court-packing.83One important wrinkle to this example is the convention against court-packing does not seem to have been in place at the time the event occurred. Of course, participants at the time invoked previous historical practice as evidence of the convention, but if the lack of a certain practice X necessarily means that there is a convention against doing X, then this threatens to stretch the idea of a convention too thin. Instead, conventions of forbearance are often actively forged, not born. See infra Sections IV.B–C. When President Franklin Delano Roosevelt proposed drastically enlarging the Court, the resulting debate had nothing to do with a game. Instead, critics charged him with attacking the Constitution itself, despite the fact that he was on firm legal ground.84  See, e.g., William E. Leuchtenburg, The Supreme Court Reborn: The Constitutional Revolution in the Age of Roosevelt 137–39, 146 (1995). As I will show, these commentators were primarily concerned with preventing a practice they thought violated a constitutional value—judicial independence—rather than preserving a coordination equilibrium. Moreover, our historical memory of the showdown and its lessons is distinctly normative. The episode is taught as an example of presidential “overreach”85William E. Leuchtenburg, When Franklin Roosevelt Clashed with the Supreme Court—and Lost, Smithsonian Mag. (May 2005), https://www.smithsonianmag.com/history/‌when-franklin-roosevelt-clashed-with-the-supreme-court-and-lost-78497994 [perma.cc/AKH7-2ZDG]. and a moment that threatened but ultimately bolstered judicial independence.86  See John Yoo & Robert J. Delahunty, The Foolish Court-Packing Craze, Nat’l Rev. (July 19, 2018, 6:30 AM), https://www.nationalreview.com/2018/07/court-packing-ideas-threaten-judicial-independence [perma.cc/8R97-KP39]. Or take the battle over Merrick Garland’s nomination to the Court: Mitch McConnell’s actions triggered a national debate over the role and nature of the confirmation process.87Lana Ulrich, Tracking the Controversy over Judge Garland’s Nomination, Nat’l Const. Ctr. (May 27, 2016), https://constitutioncenter.org/blog/tracking-the-controversy-over-judge-garlands-nomination [perma.cc/TCU7-ZLQ8].

These examples matter because they illustrate the limits of any approach that places incentives and strategy at the heart of norms. When norms are being built or dismantled, actors invoke constitutional values in their cause. These values may vary in their proximity to constitutional text, from disagreement about the meaning of the Appointments Clause,88U.S. Const. art. II, § 2, cl. 2. to structural concepts like judicial independence, but they remain recognizably constitutional. Accounts that do not explain the role of constitutional values risk reducing debates over norms to Kabuki theater. Agents might couch their arguments in constitutional language, but they are primarily motivated by political gain for a party89Daryl J. Levinson & Richard H. Pildes, Separation of Parties, Not Powers, 119 Harv. L. Rev. 2311 (2006). or branch. By contrast, this Article takes agents’ normative positions seriously. We can, at the same time, grant that actions taken in “bad faith”90David E. Pozen, Constitutional Bad Faith, 129 Harv. L. Rev. 885 (2016). are less likely to succeed as a political matter and that normative argument over a practice is irreducibly normative. We need a theory that connects the structure of norms to the claims people make in their defense.

The second issue in current work is its blinkered approach to change. History, or its absence, poses theoretical and empirical problems for current approaches. Theoretically, we can ask especially hard questions of the democratic alarmists. Is it the fact that norms change that should worry us? Or should we be worried about the breakdown of a particular norm? If the former, then we need an explanation of why conventional change is a bad thing, or at least deviant in some way. If the latter, we need both a specific defense of the particular norm at stake and reasons why the convention’s aftermath is worse than the status quo. At the very least, a blanket statement that norm erosion is either alien or inimical to democracy will not do.

The problem is also practical. As we have seen, scholars like Siegel, Levitsky, and Ziblatt are sharply presentist.91  See supra Section I.A.2. To the extent that constitutional alarmists turn to history, they do so in service of what is in their view a bleak present.92  See Josh Chafetz, Essay, Unprecedented? Judicial Confirmation Battles and the Search for a Usable Past, 131 Harv. L. Rev. 96 (2017) (explaining the challenges in crafting meaningful historical analogies). While this is a common way of using history, it misses an important lesson of historical inquiry—contingency. The norms they defend float free of their origins and are reified into permanent features of our constitutional order. And when these practices erode, we are told, it is symptomatic of broader decay.93Jack Balkin, Constitutional Rot and Constitutional Crisis, Balkinization (May 15, 2017, 2:02 PM), https://balkin.blogspot.com/2017/05/constitutional-rot-and-constitutional.html [perma.cc/DKG9-LBU3].

Yet the problem of ahistoricism is not limited to constitutional alarmists. Among those who use a strategic approach, only Chafetz and Pozen view conventional change as plausible.94Chafetz & Pozen, supra note 19. In an important recent article, Tamir extends Chafetz and Pozen’s work on the different ways norms change and offers his own account of how to counter it. See also Oren Tamir, Constitutional Norm Entrepreneuring, 80 Md. L. Rev. 881 (2021). I part ways with Tamir on the analogy between convention and law on a Hartian model. See infra Section III.B. Their examination thoughtfully lays out different forms of norm erosion (“destruction” and “decomposition”) and describes the conditions of change.95Chafetz & Pozen, supra note 19, at 1435–45. And they rightly encourage future normative and historical work precisely because of the “inherent instability of such norms.”96  Id. at 1458–59. Nevertheless, even their article explains change in largely functional terms. Whether and what sort of relation values might bear to conventional change remain open questions.

The gaps in our understanding of norms are not limited to intramural academic debate. Given the importance of historical practice in constitutional interpretation, incomplete theory has high stakes. Indeed, Larry Tribe’s description of certain rules and principles as the Constitution’s “dark matter” applies equally to norms.97Tribe, supra note 33, at 38. Like dark matter, norms are pervasive. The next Part begins developing an account of them.

II.      What Makes a Norm Conventional?

This Part explains what makes a practice conventional.98Recall that norms are conventions. I stick to the term “norm” only to avoid the confusion that the term “constitutional convention” might cause. Recent literature bypasses this question and goes directly to theorizing constitutional norms. This Part, by contrast, takes seriously the idea that a constitutional norm is first a convention. Understanding conventionality lays the groundwork for explaining how norms are constitutional.

This strategy also helps show why norms are valued for reasons beyond their utility. It does so by introducing the concept of a constitutive convention: practices that concretize principles. Current work either assumes that constitutional norms coordinate action or ignores their conventionality altogether. The former reduces their normativity to solving coordination problems, and the latter obscures the ways constitutional norms remain contingent and arbitrary.

I use philosophy to approach these issues. While scholars have sometimes drawn from other disciplines, such as economics or history, to discuss a particular norm, philosophy has gone untapped. This is a significant omission. Philosophers have long been alive to conventions99As early as 1738, David Hume explained both property and justice in conventional terms. Each emerged, he observed, from “a general sense of common interest; which sense all the members of the society express to one another, and which induces them to regulate their conduct by certain rules. . . . And this may properly enough be called a convention or agreement betwixt us, though without the interposition of a promise.” David Hume, 1 A Treatise of Human Nature 314–15 (David Fate Norton & Mary J. Norton eds., Oxford Univ. Press 2007) (1739). and have clarified their conceptual structure. While this might not matter in other areas of legal scholarship—you do not have to revisit Hart on the nature of law every time you discuss the Fourteenth Amendment—the lack of conceptual clarity is an issue for norms. Treating norms philosophically addresses that problem. When we grasp the basic structure of a convention, we can more easily see the seemingly confusing or unexplained aspects of norms.

This Part proceeds in three sections. First, it discusses coordinating conventions—practices that help coordinate action—and their analytic limits. It then introduces the idea of a constitutive convention, practices that help concretize principles. Finally, it explores norms’ features. It illustrates how constitutive conventions are normative in a thicker way than merely being useful. It shows how conventions are contingent. And it concludes by explaining how conventions are arbitrary.

A.     Coordination

The first thing conventions do is coordinate behavior. This function so dominates our understanding of them that David Lewis’s foundational book, Convention, is entirely devoted to it.100  Convention is the urtext in philosophy. It began as an intervention in a highly technical debate on the philosophy of language. The central question was whether language itself was conventional. In response, the philosopher W. V. Quine famously offered a naturalistic account. Quine’s arguments set the stage for Lewis’s monograph, which rejected Quine’s approach and advanced a conventionalist view of language. Lewis’s argument first defines conventions and then explains how language functions as a type of convention. I focus only on the first half of Lewis’s argument, which contains his definition. W.V. Quine, Foreword to David Lewis, Convention, at xi (1969); see also Thomas C. Schelling, The Strategy of Conflict (2d ed. 1980) (providing the classic account of focal points from which Lewis drew); David Singh Grewal, Network Power: The Social Dynamics of Globalization 60–62 (2008) (drawing on Schelling and Lewis to explain the conventionality of different practices of globalization). In his view, conventions exclusively solve coordination games. The latter “are situations of interdependent decision by two or more agents in which coincidence of interest predominates and in which there are two or more proper coordination equilibria.”101Lewis, supra note 104, at 24. Put more simply, coordination problems arise when actors have a mutual interest in acting the same way. This definition is intentionally broad. It describes equally well everyday situations like a couple’s decision about how to spend an evening102This refers to the canonical “Battle of the Sexes” coordination game, in which two players (a couple) are trying to decide how to spend an evening. While each partner prefers a different activity, both prefer spending time together to doing their preferred activities apart. Anatol Rapaport, Two-Person Game Theory 95–96 (1966); see also Russell Cooper, Douglas V. DeJong, Robert Forsythe & Thomas W. Ross, Communication in the Battle of the Sexes Game: Some Experimental Results, 20 RAND J. Econ. 568 (1989). or figuring out how to row a two-person boat103Hume, supra note 103, at 315. and more complex social problems such as choosing a common currency or language.

Coordinating conventions solve these games and make cooperation possible. They are “coordination equilibria”—non-exhaustive solutions to situations where parties benefit more from working together than against each other.104  See Postema, supra note 16, at 174–75 (describing features of coordination equilibria). The classic example is the convention of driving on the right side of the road. Driving on the left side, so long as everyone does it, would work just as well. Conventions thus solve problems that permit multiple solutions. These solutions do not have to be equally effective to be conventions; instead, an equilibrium needs to be “only good enough so that everyone is ready to do his part if the others do.”105Lewis, supra note 104, at 50.

A general definition of a coordinating convention has two parts, a description of what they do and the conditions under which they emerge. Accordingly, a convention is regular behavior by members of a community among whom it is common knowledge that in a particular type of recurring situation, people generally follow the convention.106  Id. at 78. Lewis’s formal definition is the following:
A regularity R in the behavior of members of a population P when they are agents in a recurrent situation S is a convention if and only if it is true that, and it is common knowledge in P that, in almost any instance of S among members of P,
(1)  almost everyone conforms to R;
(2)  almost everyone expects almost everyone else to conform to R;
(3)  almost everyone has approximately the same preferences regarding all possible combinations of actions;
(4)  almost everyone prefers that any one more conform to R, on condition that almost everyone conform to R;
(5)  almost everyone would prefer that any one more conform to R’, on condition that almost everyone conform to R’,
where R’ is some possible regularity in the behavior of members of P in S, such that almost no one in almost any instance of S among members of P could conform both to R’ and to R.
Moreover, a convention persists because people expect each other to follow it. If this expectation were absent or people started converging on a different practice, the convention would no longer exist.

Each of these four conditions—preferences, mutual expectations, regularities of behavior, and common knowledge—are necessary for the existence of a convention. First, agents must have preferences. This is the uncontroversial idea that agents facing a coordination problem think some outcomes are better than others. Second, conventions require agents to share mutual expectations about their behavior. Only if an actor is sufficiently confident that others will act a certain way will they also conform their behavior. While a stylized economic model can specify exactly the level of confidence agents need to form a convention, real-world coordination problems are often too complex to give concrete thresholds. Instead, the level of certainty required for a convention will vary depending on the type of situation agents face.

Regularity of behavior and common knowledge—the third and fourth conditions—are explained by the idea of precedent.107“Precedent” here means a prior, observed pattern of behavior. It differs from the legal meaning of the word. When agents reach a convention by way of precedent, they rely on their shared awareness that a previous form of coordinated behavior achieved the desired outcome. From that shared awareness, agents facing an analogous situation in the future behave similarly if the knowledge is sufficiently widespread.108As Lewis notes:
Coordination by precedent . . . [is the] achievement of coordination by means of shared acquaintance with a regularity governing the achievement of coordination in a class of past cases which bear some conspicuous analogy to one another and to our present coordination problem. Our acquaintance with this regularity comes from our experience with some of its instances, not necessarily the same ones for everybody. . . . We acquire a general belief, unrestricted as to time, that members of a certain population conform to a certain regularity in a certain kind of recurring coordination problem for the sake of coordination.
Lewis, supra note 104, at 41.
Together, the third and fourth conditions teach us the following: over sufficiently long stretches of time, agents observe regularities of behavior that license analogical reasoning in similar future situations, and it becomes a part of the “common knowledge” in a community that you do X in situations of type Y.109  Id. at 57 (“Precedents also are a basis for common knowledge that everyone will do his part of a coordination equilibrium; and, in particular, past conformity to a convention is a basis for common knowledge of a tendency to go on conforming.”).

Legal scholarship, like the game-theoretic approaches discussed in Part I, pays close attention to coordinating conventions, and rightly so. Coordination is vital to social ordering, including law. Social life is fraught with problems that require solutions that are “good enough,” if not perfect. In these situations, it is more important that everyone settles on a solution, than that a solution be optimal in all respects.110In fact, pernicious conventions can endure because of compliance dependence even if people think a different practice would be better. See, e.g., Leonardo Bursztyn, Alessandra L. González & David Yanagizawa-Drott, Misperceived Social Norms: Women Working Outside the Home in Saudi Arabia, 110 Am. Econ. Rev. 2997 (2020) (providing experimental and survey data showing that women’s participation in Saudi labor markets is limited because their husbands wrongly believe other men disapprove of women working outside the home).

Yet for all its explanatory value, coordination does not exhaust our understanding of conventions. We still lack an explanation of how values relate to conventions. Lewis acknowledges as much: “The definition I gave of convention did not contain normative terms . . . . ‘[C]onvention’ itself, on my analysis, is not a normative term. . . . [C]onventions may be a species of norms: regularities to which we believe one ought to conform.”111Lewis, supra note 104, at 97. On this view, a convention might endure because people think they should adhere to it, but that is the extent of its normativity.

Just as this view is deficient in the constitutional realm, it also falls short at a general level. Social life is full of practices we find meaningful beyond the convenience they provide. While we might not care which side of the road we drive on, we attach value to conventional practices in areas like art and etiquette. And for constitutional theory in particular, a general theory has to make sense of the characteristic alarm that norm erosion provokes. This is especially important since few constitutional norms resemble solutions to coordination games; executive noninterference in administrative adjudication is not the same type of rule as driving on the right side of the road. These practices are often distinct from picking on which side of the road to drive. Happily, Andrei Marmor’s notion of constitutive conventions helps fill in the other half of our account.

B.     Constitution

Conventions do not only coordinate behavior; they also concretize principles into practice. These functions are not mutually exclusive.112  See infra Section IV.A. A convention can conceivably do both. For our purposes, however, implementing a principle as a practice is especially important. Andrei Marmor develops this idea in response to Lewis.113Indeed, Marmor begins with a definition that is essentially Lewisian: a convention is a social rule agents follow in a particular set of circumstances for a particular set of reasons, and there is at least one alternate rule they could follow in the same situations for the same reasons. More formally:
A rule, R, is conventional, if and only if all the following conditions obtain:
1.   There is a group of people, a population, P, that normally follow R in circumstances C.
2.   There is a reason, or a combination of reasons, call it A, for members of P to follow R in circumstances C.
3.   There is at least one other potential rule, S, that if members of P had actually followed in circumstances C, then A would have been a sufficient reason for members of P to follow S instead of R in circumstances C, and at least partly because S is the rule generally followed instead of R. The rules R and S are such that it is impossible (or pointless) to comply with both of them concomitantly in circumstances C.
Andrei Marmor, Social Conventions: From Language to Law 2 (2009).
His theory better tracks what conventions do in daily life and how people understand them. And it trains our attention on the salient features of conventions for constitutional theory.

Constitution is intimately linked to normativity. Constitutive conventions are practices we recognize that we should follow. The operative question here is where that “should” comes from. What does a convention do, beyond making cooperation possible, that commands our respect?

Constitution answers that question by highlighting the link between practices and their underlying values. Constitutive conventions are a type of constitutive rule114The philosopher John Searle discussed the concept of a constitutive rule in his book Speech Acts. John R. Searle, Speech Acts (1969). See also John R. Searle, Constitutive Rules, 4 Argumenta 51, 51 (2018) (“Constitutive rules create new forms of reality, with new powers, they typically require language, and they are the basis of human civilization.”).—rules that make up a particular sort of activity or social practice. Constitutive rules are ubiquitous: the rules of chess, the structure of a Greek tragedy, or the clauses of a written constitution.115Stephen Holmes importantly observed that constitutions themselves are a set of constitutive rules for a democracy. Against prevailing theories of constitutionalism that view these documents as purely constraints on action, Holmes argues that constitutions are enabling devices. Stephen Holmes, Precommitment and the Paradox of Democracy, in Passions and Constraint 134, 163 (1995). The idea that law enables freedom, rather than merely constraining it, is an old and distinguished one. See, e.g., John Locke, The Second Treatise of Government § 57, in Two Treatises of Government 265, 305 (Peter Laslett ed., Cambridge Univ. Press 1988) (“[T]hat ill deserves the Name of Confinement which hedges us in only from Bogs and Precipices.”). Nevertheless, they all play the same role: they bring into being and make a certain type of practice intelligible. They do so by linking together an underlying principle and a corresponding practice. For instance, consider the notion of etiquette or politeness. Telling someone to be polite, on its own, is vague and unhelpful. Instead, we have sets of practices that embody the idea of respect for another person. They comprise the conventions of etiquette. These practices often vary significantly. In one culture, etiquette might require taking off your shoes before entering a home, and in another, removing footwear may seem rude and perplexing. In either context, however, the structure of the convention remains the same: the convention constitutes its underlying value.

It is important to clarify what “constituting” means. This is not the strong ontological claim that these rules “create” certain actions or behavior. We could be dancing the waltz without knowing it. Rather, constitutive rules create the “particular social meaning or significance of the action in question.”116Marmor, supra note 117, at 34. In other words, unless the social convention of a waltz is in place, our dancing cannot be understood as a waltz. Constitutive rules thus make forms of social behavior meaningful. Nor do these rules exist only in isolation. Most social activities consist of many different constitutive rules that together form the “structure of rule-governed activity.”117  Id. This, as we will see, is important in the constitutional order, where a variety of conventions together guide and constrain behavior. Actors following constitutional conventions understand and defend their actions in constitutional terms.

Constitutive conventions are different from institutional rules, a distinction that is very important for constitutional norms.118Constitutional norms are a type of constitutive convention. The former includes “structured conventional games (like chess, tennis, soccer), forms of art, some practices of etiquette, [and] social ceremonies and rituals,” while the latter are “legal institutions (like legislatures, courts, administrative agencies), quasi-legal institutions (like political parties, sports leagues), and religious institutions (like a church).”119Marmor, supra note 117, at 35 (cleaned up). Both constitute meaningful social activity, but they diverge in terms of their strength.

The line between the two is porous. Conventions, when recognized by the relevant institutions, are replaced by institutional practices.120  See id. (“[S]ometimes conventional practices are replaced by institutional codification, and thus they may become institutional practices.”). When this happens, a convention loses its characteristic informality (and flexibility) and is transformed into an authoritative rule. As an example, consider the Court’s decision in NLRB v. Noel Canning.121573 U.S. 513 (2014). There, the Court, after surveying the practice of past presidents, decided that the Recess Appointments Clause embraced both inter- and intra-session appointments.122Id. at 538. The decision exemplifies institutional codification. A legal institution, the Supreme Court, empowered to declare law, turned past practice into a decisive rule.123  See Marmor, supra note 117, at 50–51. The transformation of convention into law is also central to the theory of common law. Oliver Wendell Holmes made this very argument in recounting the history of the law of contracts in The Common Law. See Oliver Wendell Holmes, The Common Law 223–60 (Harvard Univ. Press 2009) (1881). I thank Lev Menand for this point. This rule was then enforced by the relevant authorities: law enforcement and courts. A convention was thus transformed into law.

In contrast to institutional rules, conventions are informal. They lack the pedigree of institutional rules and therefore make weaker, more tentative claims to authority. Even when conventions command broad obedience and respect, we cannot point to a particular locus of power or procedure that makes a convention binding.124This fact, of course, squares with a key feature of conventions: compliance dependence, which I take up in the later discussion of arbitrariness. Keith Whittington is the rare mainstream constitutional theorist who cites Marmor’s work, and only for the idea of compliance dependence. In constitutional politics, the difference between conventions and institutional rules roughly tracks one between conventions and law. Constitutive rules in our constitutional order sort into conventions and law, with the latter enunciated by the typical actors (courts, legislatures, agencies) in the typical ways (judicial decisions, statutes, regulations).

Given how common constitutive conventions are and the central role they play in social life, it is easy to lose track of “what makes such rules conventional at all.”125Marmor, supra note 117, at 41. The vast difference between what coordinating and constitutive conventions do make it hard to see what they share in common. Yet conventions, regardless of their function, are united by a shared quality: “compliance dependen[ce].”126  See id. at 11. A convention is compliance dependent because one of the reasons we follow it is that others follow it too.127Again, the formal definition of compliance dependence is:
A reason for following a rule R is compliance dependent if and only if, for a population P in circumstances C,
1.   there is a reason for having R, which is also a reason for having at leas[t] one other alternative rule, S, and,
2.   part of the reason to follow R instead of S (in circumstances C) consists in the fact that R is the rule actually followed by most members of P in circumstances C. In other words, there is a reason for following R if R is generally complied with, and the same reason is a reason for an alternative rule if that alternative is the rule generally complied with.
Id. at 11.
Consider again, the example of a coordinating convention: driving on the right side of the road. In that case, a different rule—driving on the left side—could equally solve the coordination problem. The reason we drive on the right side is that everyone else is doing it. The value of a pure coordinating convention is thus entirely connected to the problem it solves. The same only partly applies to constitutive conventions, perhaps less obviously. There might be various reasons why we follow a rule of etiquette, but one major reason is that others follow the rule.

Compliance dependence is also important for understanding the arbitrariness of conventions since it explains the nature of the relationship between a rule and our reasons for following it. I turn to that shortly. But for now, it is enough to observe that compliance dependence makes clear why one rule prevails over other satisfactory ones. The successful rule simply enjoys enough support to become self-sustaining.

Finally, the notion of a constitutive convention carries with it several important observations about how conventions emerge and their flexibility over time. First, people do not always exercise equal influence over the construction of a convention. Some actors, by virtue of their social position, are better situated to construct and shape conventions. Take the law: “The conventions that determine what counts as law in the relevant legal system, are, first and foremost, the conventions of judges, particularly in the higher courts. . . . [O]ther legal officials can also play various roles in determining the content of such conventions.”128  Id. at 46. These other actors include agencies, police officers, and the like, and together they suggest a “division of labor” in the formation of conventions.129  Id. at 46–47. In this way, constitutive conventions, like coordinating conventions, emerge from the interaction of various agents, sometimes similarly situated but often not.

This view of a “division of labor” applies equally well to the constitutional context. There is a vast set of actors who are responsible for the construction of constitutional conventions including not only the obvious ones—the Supreme Court, the president, and Congress—but also other agents in and out of the government. These range from the OLC and the press to powerful institutions like law schools and their faculties.130  See Liora Lazarus, Constitutional Scholars as Constitutional Actors, 48 Fed. L. Rev. 483, 487, 491 (2020). Indeed, much recent scholarship can be understood as an active attempt by legal academics to influence the force and meaning of norms. By highlighting and defending constitutional norms, some scholars have tried to sustain, with varied success, the authority of past practice and to moor in place a constitutional order in flux. The politics of norms is part and parcel of constitutional politics writ large.

C.     Features of Norms

1.       Normativity

Conventions are normative. They prescribe behavior that “must be regarded as binding by the relevant population.”131Marmor, supra note 117, at 3. This is equally true for coordinating and constitutive conventions. While we do not attach any special value to driving on the right side of the road, we do think it is a rule we should follow, even if it is just because everyone else does. Similarly, when agents respect a constitutive convention, they do so partly because they think the behavior is socially required.132  Id. at 41. This explains why, for instance, we wear a suit to a job interview: it is just what you wear to those things. Both of these examples highlight the previous idea of compliance dependence—the fact that one of the reasons we follow a convention is the expectation that others will too.

Yet compliance dependence only gets us a thin notion of normativity. If the only reason we followed a convention were the expected compliance of others, then there would be little separating coordination and constitution. Even if a convention did either of those things, people would respect it for the same reasons. This would leave us with the same view of conventions as the strategic approaches: value depends entirely on function. This view implies that any concern about the collapse of constitutional conventions is mere handwringing.

Happily, conventions enjoy a thick idea of normativity. This thick view helps make sense of the anxiety of constitutional alarmists. Recall a central problem with an exclusively coordinating account of conventions: finding the relevant game. While conventions can serve as solutions to coordination games, “[f]or many types of familiar conventions . . . , this story does not make sense” since “there is no coordination problem that we can identify.”133  Id. at 22. And even in those cases where we can trace the emergence of a convention to a historical coordination problem, once the convention is in place it can persist for reasons “that are quite independent of the story of why and how the game . . . emerged.”134  Id. at 24. This exact problem prompted Marmor to add constitution as a separate function.135  Id. at 31. Constitution helps explain the relationship between a convention and its underlying value: the former embodies the latter in practice.136  Id. at 36–37.

Once the central role of values is kept in view, the broad conclusion is that conventions are irreducibly normative beyond mere compliance dependence. Constitutive conventions are thickly normative. We follow them because we attach importance to them beyond the way they enable cooperation. Failing to follow a convention, then, is not (or not just) bad strategy or being “foolish or wrong.”137  Id. at 15. It is also understood as transgression. Since “[c]onventions are rules of conduct, and they are normatively significant as such,”138  Id. when we fail to follow them—say by wearing pajamas to a funeral—we offend.139As we will see shortly, constitutional norms are normative because they implement constitutional values; these underlying values imbue the practice with normativity. While American constitutional theorists have not recognized the limits of coordinating conventions for constitutional norms, comparative law scholars have noted the limits without fleshing out alternatives. See, e.g., Joseph Jaconelli, The Nature of Constitutional Convention, 19 Legal Stud. 24, 44 (1999) (“If we consider for a moment the examples of convention that are given by David Lewis, . . . [i]t could be argued . . . that conventions of this type—where the element of underlying reason is exiguous—are not typically to be found in matters constitutional.”). This more expansive view of conventions better makes sense of the “wide variety of social functions” they serve, including but not limited to coordination.140Marmor, supra note 117, at 25. Thus, pajamas to a funeral not only evinces irrationality (why self-sabotage?) but also warrants condemnation (your outfit was disrespectful).

The idea that conventions are thickly normative also begins to explain why constitutional alarmists have reacted so strongly to the breakdown of long-standing practices. If constitutional norms only coordinated action, then alarmists bemoan the loss of a functional regime and nothing more. If, however, constitutional norms are constitutive, then alarmists are worried about the breakdown of a practice and its underlying value. Widespread norm erosion thus reflects the breakdown of a particular form of constitutional morality.

2.       Contingency

Conventions are contingent because their existence and survival depend on the state of the world.141Contingency might suggest that the Constitution makes complete internal sense, with values that are eternal and fixed but subject to the vagaries of a fickle and unprincipled real world. I reject such constitutional Platonism and do not mean to imply it (nor do I think these other authors take a Platonist view). Instead, my account tethers principles to concrete practices. That connection denies that principles can be cleanly distinguished from the practices that embody them; to the extent that these principles are fixed, they are empty. In other words, “separation of powers” and federalism simply are the practices that define them at any given time. As the world changes, so do conventions. This explains why conventional change is both possible and normal. A pure coordinating convention is the clearest example of this idea. When agents are not normatively attached to a convention—few people find meaning in driving on the right side of the road—it is easier to change their behavior. Pure coordination conventions do not “stick” any longer than the time it takes for a community to learn that people are behaving differently.142  See, e.g., Lewis, supra note 104, at 49–51.

Coordination games in the real world are not static. Imagine a warm-weather society where the convention of a mid-afternoon nap develops.143The siesta has evolved since its Roman origins when high mid-day heat and a primarily agricultural economy made it a useful practice. It’s Time to Put the Tired Spanish Siesta Stereotype to Bed, BBC: Worklife (June 9, 2017), https://www.bbc.com/capital/story/20170609-its-time-to-put-the-tired-spanish-siesta-stereotype-to-bed [perma.cc/C234-5RPG]. If the climate begins to cool and some actors realize they can forego the afternoon nap and conduct more business, others might follow, and the napping convention will collapse. Whenever any convention dissipates, it might or might not be replaced by another one. In each case it depends, and in many circumstances, it might take time for the common knowledge and mutual expectations necessary for a new convention to develop.144Highly salient events can scramble people’s common knowledge and upset previous conventions. See, e.g., Leonardo Bursztyn, Georgy Egorov & Stefano Fiorin, From Extreme to Mainstream: The Erosion of Social Norms, 110 Am. Econ. Rev. 3522 (2020) (providing experimental support for the claim that the election of Donald Trump has relaxed adherence to previous conventions against expressing xenophobic views publicly). But the underlying point remains: conventions are contingent because the world is.

That applies equally well to constitutional politics where constitutive conventions abound. A particular practice, say, executive noninvolvement in the Department of Justice, can be normatively important and also depend on political incentives for its survival. When these incentives change—a president discovers that they can flout them with impunity—the practice can also erode. When the world changes, we begin questioning past practices and can more easily imagine new ones.

3.       Arbitrariness

Conventions are arbitrary for two reasons.145Arbitrariness as defined here does not mean unreasoned or unjustified. As I explain below, a norm is arbitrary for different reasons than say, agency action under the Administrative Procedure Act. An “arbitrary and capricious” decision by an agency is one that does not offer reasons that can withstand judicial scrutiny. Motor Vehicle Mfrs. Ass’n. v State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 42–43 (1983). There might be many reasons to sustain a norm, even if other ones could plausibly concretize the same underlying principle or text. Thanks to Todd Aagaard for pointing out the administrative law context. First, conventions, coordinating and constitutive, are arbitrary because they are compliance dependent. Compliance dependence refers to the fact that one of the reasons we follow a convention is because others follow it too.146  See supra Section II.B. It describes the relationship between the reasons we have for following a practice and the practice itself. Nothing inherent about driving on one side of the road or wearing a suit marks it as the governing practice.

Second, constitutive conventions are arbitrary because of the nature of the underlying values and how they are cashed out in practice. These “needs, functions, or values . . . radically underdetermine the content of the rules that constitute the relevant social practice.”147Marmor, supra note 117, at 41. While certain “[n]orms of rationality, and basic moral norms” do not qualify as conventions since “they do not admit of alternatives,” many other values and principles can be realized through a variety of ways.148  Id. at 9–10. Social conventions implement values in intelligible action. To return to an earlier example, the basic norm of “being polite” can be practiced through shaking hands or making eye contact. As we know, however, etiquette ranges wildly between cultures, and being polite elsewhere might involve a downward gaze. “Politeness” is thus too abstract to specify the type of conduct required.

Arbitrariness does not imply, however, that the relationship between principle and practice is unidirectional. Instead, people enact principles in certain practices and then understand the principle in light of those very same practices. The relationship is dialectical. As Marmor puts it, “constitutive conventions tend to be in a constant process of interpretation and reinterpretation that is partly affected by external values, but partly by those same values that are constituted by the conventional practice itself.”149Id. at 48. Ironically, this view of conventions strongly resembles Dworkin’s theory of interpretive concepts. See Dworkin, supra note 16, at 45–86. Moreover, conventions develop over long periods of time even if, as in constitutional politics, we can identify discrete moments in time when a previous convention was abandoned or a future convention was first adopted.

Because constitutive conventions take time to develop, they typically have “a history, and the history tends to be socially significant.”150Marmor, supra note 117, at 49. It is no surprise, then, that arguments from history feature prominently during moments of constitutional change. Actors who challenge long-followed conventions usually meet resistance from defenders of the status quo. The former often insist that their changes are entirely consistent with the current practice, while the latter will assert a necessary connection between the status quo and the relevant constitutional principle. This pattern is played out in the examples considered in Part IV.

Arbitrariness is especially important for constitutional practices, where the underlying values are multiply realizable.151  See, e.g., Aziz Z. Huq & Jon D. Michaels, The Cycles of Separation-of-Powers Jurisprudence, 126 Yale L.J. 346, 352 (2016) (noting that, in the separation-of-powers context, doctrinal approaches that are “sensitive to the multiplicity of normative values . . . might well take the seemingly incoherent form of oscillating rules and standards”). The idea of a separation of powers, for instance, can be implemented in many different ways. The fact that much of the recent scholarship on conventions has concentrated in this area and the perennial divide between hard and fast allocations of power and functionalism speaks to the indeterminacy of this principle.152See id., for an example of this recent scholarship tracking the “zigzagging” between rules and standards in separation-of-powers law, and David E. Pozen, Self-Help and the Separation of Powers, 124 Yale L.J. 2 (2014), for a discussion of the limits of constitutional text for understanding separation-of-powers dynamics. See also Manning, supra note 25 (underlining the lack of a baseline in many areas of separation-of-powers issues). Conventions are practices that address that indeterminacy without dissolving it. Taken together, arbitrariness and normativity suggest something important about conventions and constitutional norms in particular: there are many norms that actors follow because they think such norms are right, without recognizing that alternate practices could work just as well if not better.

That conventions are at the same time arbitrary and normative may seem like a problem. Arbitrariness often has a negative normative valence. This likely stems from a view of morality as analogous to mathematics: we start with accepted normative premises and step by valid inferential step reach a sound conclusion. Arbitrariness destroys that picture. It suggests that a current practice, value, or rule could be otherwise. Yet just because we can do things differently does not mean that the current rule is not valuable. Rather, arbitrariness means that our normative imagination cannot be bound by the status quo. Defense of a convention, to wit, cannot simply rest on it being the way we have done things before. When alarmists warn us about norm erosion, they miss that a given constitutional morality is always particular and nonexhaustive because it is both the product of history and a single expression of capacious values.

III.     Constitutional Norms

This Part answers a second question: when is a norm constitutional? The answer lies in the fact that constitutional norms are a type of constitutive convention. These conventions concretize values into practices. Constitutional norms are normative, contingent, and arbitrary practices that implement constitutional text and principle.153Since I have defined constitutional conventions as constitutive conventions, I use “implement” instead of “constitute.” While constitutional conventions do concretize constitutional text and principle in practice, that formulation, for obvious reasons, is ponderous. And they enjoy the respect of actors in and out of government because they recognize them as constitutional. So, while they share the same features as all other conventions, constitutional norms are distinct because the things they concretize are constitutional.

After explaining what makes a norm constitutional, this Part concludes by defending the distinction between constitutional norm and law. Despite their importance to everyday constitutional practice, norms enjoy respect that falls short of the obedience individuals pay to law. Unconventional behavior is more common than lawbreaking precisely because unconventional acts are not illegal. And when we do want to stem conventional change, we often turn to law. For both conceptual and practical reasons, the distinction remains important.

A.     When Is a Norm Constitutional?

Part II explained that practices are conventional when they are normative, contingent, and arbitrary and either coordinate or concretize. As I have noted throughout, coordination fits poorly with constitutional norms. We rarely, if ever, can identify a coordination problem the practice solves.154  See Marmor, supra note 117, at 34. Instead, constitutional norms are constitutive. They link principle and practice. So, what makes these norms constitutional?

There are two possible answers: actor-centric and practice-centric. An actor-centric view holds that a given norm is constitutional when it involves constitutionally identifiable actors discharging a constitutional role. Given its breadth, the actor-centric view potentially covers a large swath of government. For instance, the rules and practices of executive branch lawyers155  See sources cited supra note 78. are constitutional norms because they shape the presidency. Similarly, the conduct of a textually specified actor—the Senate—in fulfilling a particular textual duty—advice and consent for treaties—qualifies as a constitutional norm, especially in the absence of applicable law.156  See, e.g., Amar, supra note 33, at 310–18; Whittington, supra note 10, at 1859. This definition also allows scholars to declare sets of practices as the “norms” of a particular branch, like Daphna Renan has done with the presidency.157Renan, supra note 11. Thus, under Renan’s framework, there is a shift from the Framers’ norm158  Id. at 2231. against the president speaking directly to the public to today’s “rhetorical presidency.”159  See generally Jeffrey K. Tulis, The Rhetorical Presidency (Princeton Classics ed. 2017). The practice-centric view focuses more squarely on the norm itself. It holds that a norm is constitutional when it implements constitutional principle or text. Constitutional norms are constitutive. These conventions concretize values, the way a handshake embodies politeness. Constitutional norms also instantiate constitutional principles. For example, executive noninvolvement in administrative adjudication respects at least two different principles—due process and judicial independence.160Vermeule, supra note 12. See also William Baude, Adjudication Outside Article III, 133 Harv. L. Rev. 1511, 1521 (2020) (emphasizing executive branch adjudication as subject to Due Process constraints). The first has an explicit textual basis;161U.S. Const. amend. V. the latter belongs to that category of constitutional values Charles Black called structural principles.162  See Black, supra note 33.

As the previous example shows, constitutional norms and their underlying principles do not need one-to-one correspondence. A given practice can implement multiple constitutional values. Conventions, more generally, operate the same way. The convention of a performer bowing in front of an audience after applause can represent several different principles at the same time: etiquette in showing gratitude, an appreciation of the hierarchy of patron and artist, and respect for the tradition of performance. Because conventions are both common and vital in making normative action possible, their multivalence is expected. Our ordinary lives are full of practices that are normatively significant in a number of ways at the same time. For our purposes, this means many norms are complex practices that can bear varied constitutional meanings.

The two ways of defining constitutional norms—actor-centric and practice-centric—are not mutually exclusive, but I use the latter for three reasons. First, the actor-centric view risks sweeping too broadly and pitching the relevant practice at too high a level of abstraction. It is hard to limit both who counts as “constitutional actor” and what qualifies as the discharge of a duty. Take, for instance, the shift to the “rhetorical presidency.” The president is obviously a constitutional actor.163In this case, the president sits comfortably in the “core” of the rule. H.L.A. Hart, Positivism and the Separation of Law and Morals, 71 Harv. L. Rev 593, 607 (1958). But what is the relevant constitutional duty they are discharging? The Take Care Clause164“[H]e shall take [c]are that the [l]aws be faithfully executed.” U.S. Const. art. II, § 3, cl. 5. is a possible option, but also a highly capacious one. If the rhetorical presidency is a constitutional convention since it involves a constitutional actor, the president, fulfilling their constitutional duty under the Take Care Clause, then an entire style of governance is a norm. This is not a decisive problem for this definition since we might want a theory of norms to capture shifts in the way politics works. After all, the idea of a “constitutional order” is a broad one.165  See supra note 2 and accompanying text. Yet range comes at the cost of precision; when a practice is defined too expansively, it is hard to see what the practice actually means.166The choice between a broader and more specific definition of a constitutional convention is analogous to the decisions different scholars of administrative constitutionalism have made in defining that concept widely or narrowly. On one end of the spectrum is Sophia Lee, who defines administrative constitutionalism “to include only agencies’ interpretation and implementation of the United States Constitution.” Sophia Z. Lee, From the History to the Theory of Administrative Constitutionalism, in Administrative Law from the Inside Out: Essays on Themes in the Work of Jerry L. Mashaw 109, 109 (Nicholas R. Parrillo ed., 2017). Others have taken much more expansive views. See, e.g., Gillian E. Metzger, Administrative Constitutionalism, 91 Tex. L. Rev. 1897, 1899 (2013) (expanding the notion to include “the statutes and legal requirements that create and govern the modern administrative state”). Like Lee, I opt for the narrower definition. It is more useful for grasping the interpretive issues reasoning from historical practice raises.

Second, the practice-centric view trains our attention on the structure of the norm. As I argued in Part II, constitutive conventions—of which constitutional norms are an example—have two parts: a practice and the underlying value it constitutes. The practice-centric view conditions a norm’s constitutional status on the nature of the value at stake. When the underlying value is either a recognizable constitutional principle, like federalism or the separation of powers, or constitutional text, then the practice is a constitutional norm. A practice-centric view expresses the following intuition: a norm is constitutional if a reasonable viewer seeing the practice thinks it may be required by the Constitution.167I thank Arjun Ramamurti for this formulation. Constitutional norms enjoy this respect from reasonable viewers because they form part of the constitutional morality of a given era. Taking the practice-centric view, then, has the virtue of showing how constitutional norms track the structure of conventions generally.

The final reason in favor of the practice-centric view involves constitutional interpretation. Sometimes courts have to decide whether a norm should inform their decisions. The practice-centric view focuses our attention on what this involves: judges using norms as a source of law. This reason, again, is not decisive. If the goal is to give a rich account of how the constitutional order works, the actor-centric view is sociologically attractive. But if we are interested in the interpretive consequences of using norms as law, the practice-centric model makes more sense. It tethers the practice to a constitutional principle or text and asks us how that relationship should bear on legal reasoning.168The interpretive consequences of this approach are only considered in the final Section of this paper. See infra Conclusion.

The distinction between the two views is not hard and fast since application and limits overlap. Any norm under the practice-centric view will also satisfy the actor-centric one. And many actor-centric norms will count as practice-centric ones. This means some constitutional norms can be framed either way. Presidential noninterference with the Department of Justice is one example. That norm is constitutional because it constitutes important constitutional principles and text like due process,169U.S. Const. amend. V. equal protection,170  Id. amend. XIV. and free speech and association171  Id. amend. I. All of these amendments are plausible textual hooks for this convention. Protect Democracy, No “Absolute Right” to Control DOJ: Constitutional Limits on white House Interference with Law Enforcement Matters (2018), https://s3.documentcloud.org/documents/4498818/2018-Protect-Democracy-No-Absolute-Right-to.pdf [perma.cc/‌6HYV-HYUF]. Historically, however, those who have followed the convention have linked it to the Take Care Clause. See infra Section IV.C. and because it informs how the president enforces federal law.

The practice-centric view can still be vulnerable to the charge of excess abstraction, though less so than the actor-centric view. Constitutional norms, even when we identify them in a practice-centric way, can vary in their normative and practical complexity. A norm can range from something as prosaic as attaching a slip of paper to a nomination to something as grand as respecting the structure and powers of a coordinate branch. Moreover, these norms can be nested: a relatively broad convention can be composed of smaller, constituent conventions. We can thus specify norms at various levels of abstraction. That analysis turns on several factors such as the complexity of the actors involved (a particular office within a branch or the branch itself), the relevant function (coordination or concretization), and the history and development of the practice.172This Article does not give a full account of what makes a constitutional convention complex or simple. For our purposes, it is enough to observe the complexity of constitutional conventions, describe relevant examples, and connect their character to that of conventions generally. The practice-centric view, unlike its counterpart, requires us to identify the underlying constitutional text or principle for a given practice. When we cannot (or the connection between practice and principle is highly attenuated), this counts against labeling it constitutional. So, while the shift to a “rhetorical presidency” is very important, the practice is better understood as a political norm, not a constitutional one.

Both approaches, however, recognize that constitutional norms are fundamentally constitutive conventions. The issue for constitutional scholarship is that normativity and arbitrariness sit together uneasily. Constitutional law scholars, understandably enough, are often focused on questions with clear and usable answers. Examples include “can a sitting president be indicted?”173Laurence H. Tribe, Yes, the Constitution Allows Indictment of the President, Lawfare (Dec. 20, 2018, 11:55 AM), https://www.lawfareblog.com/yes-constitution-allows-indictment-president [perma.cc/N632-9JUQ]. or “is West Virginia unconstitutional?”174Vasan Kesavan & Michael Stokes Paulsen, Is West Virginia Unconstitutional?, 90 Calif. L. Rev. 291 (2002) (cleaned up). And law professors have a comparative advantage in answering these questions. The questions are distinctly legal and invite traditional forms of analysis drawing from familiar sources: constitutional text, doctrine, statutes, and regulations. The question of normativity is either built into the question—what is normative is what is constitutional—or bracketed and addressed separately—what is constitutional and what is desirable? By contrast, I argue arbitrariness is an inherent feature of practices that remain normative. Just because a norm is, in a basic sense, arbitrary does not mean we should stop honoring it. Instead, constitutional conventions challenge us to live with contingency and uncertainty as facts of our constitutional order. They cannot be wished away.

B.     Constitutional Norms vs. Law

This Section considers the distinction between constitutional norm and law, which matters for several reasons. First, it is one of the few consistent threads in the norm scholarship. Despite its diversity, all recent scholars either recognize or assume that norm is different from law. If that assumption is unsound, then we are not analyzing anything special. Second, that distinction must hold for any consideration of the relationship between historical practice and constitutional interpretation. If norms—a form of historical practice—are not meaningfully distinct from laws, then there is nothing unique about reasoning from historical practice. Third, the distinction is yoked to the difference between conventions and institutional rules.175  See supra Section II.B. Conventions are different from institutional rules because the latter are ratified by authoritative institutions. If institutional imprimatur does not matter, then it is hard to explain why norms can be violated without regular penalties but laws cannot.

Conventionalist theories of law, however, put pressure on the border between norm and law. It is easy to see the problem in relation to Hart’s rule of recognition. In any given legal system, a rule of recognition solves the problem of what counts as law by “specify[ing] some feature or features possession of which by a suggested rule is taken as a conclusive affirmative indication that it is a rule of the group to be supported by the social pressure it exerts.”176Hart, supra note 15, at 94. Put simply, a legal system’s rule of recognition tells us how to identify a law as law.177  Id. Scott Shapiro offers a simple example of a rule of recognition. Consider a village society with a legal system. In such a society, “[i]f there is a doubt about, say, how many mates are acceptable, the rule of recognition can direct the parties to the authoritative list of rules on the rock in the town square, the past pronouncements of the village elder, the practice of other villages and so on, to determine the answer.” Scott J. Shapiro, What Is the Rule of Recognition (and Does It Exist)? (Yale L. Sch. Pub. L. & Legal Theory Rsch. Paper Series, Paper No. 181, 2009), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1304645 [perma.cc/KP62-8QFG]. Picking out what the rule of recognition is in any given legal system is hard because these rules “may take any of a huge variety of forms, simple or complex,” but Hart maintains that every legal system will have a rule of recognition.178Scholars have long tried to identify the rule of recognition in the American legal system. Indeed, Hart himself claims in passing that the American legal “system of course contains an ultimate rule of recognition and, in the clauses of its constitution, a supreme criterion of validity.” Hart, supra note 15, at 94, 106. Others have followed Hart’s suggestion and offered complex descriptions of the American rule of recognition. See Kent Greenawalt, The Rule of Recognition and the Constitution, 85 Mich. L. Rev. 621 (1987).

The problem is that many legal positivists think the rule of recognition itself is conventional. Their arguments take different forms, variously stressing the coordinating,179Marmor, supra note 117, at 164. epistemic,180  See id. at 162 n.17. and constitutive181  Id. at 165. functions of the rules of recognition. Whichever view we choose, the resulting challenge is the same: if the very foundation of a legal system is a convention, can we meaningfully distinguish between law and norm?

We can respond in four ways. First, law might be conventional, but it is in a special way. Hart’s discussion of pre-law and law-bound societies suggests there are alternative forms of social ordering to law.182  See Hart, supra note 15, at 90–93. But conventions are not monolithic either. Marmor himself distinguishes between “deep” and “surface” conventions. The former include bedrock parts of social life, including language, art, and games; the latter include more specific examples such as English, Bauhaus, and chess.183Marmor, supra note 117, at 59–62. Deep conventions make surface conventions possible, and “in following surface conventions one also follows, albeit indirectly, the deep conventions that underlie it.”184  Id. at 63.

If we accept this view, it explains the relationship between constitutional norms and law as one of entailment: when you practice a constitutional norm, you are, in an attenuated sense, obeying a constitution and its law. Nevertheless, constitutional norms and law, even if they are fundamentally conventional, have different pedigrees—laws are created by authoritative institutions, while norms often emerge organically. This difference explains their disparate social strengths. So when we deploy the law–norm distinction, we do not commit ourselves to any strong position on the nature of law. And if law is in fact conventional, then the distinction is merely a shorthand for the difference between deep and surface conventions.

Second, if the distinction is not sound, then that result does not square with our ordinary experience at all. We know there are distinct differences in how we identify things like statutes and court decisions, which have defined institutional contours, and how we pick out norms, whose outlines are far less clear. If the problem of a stable divide between laws and norms still survives, then it applies equally to all legal scholarship. Every scholar of constitutional norms depends on this distinction. Without it, judicial decisions and statutes become the same as the norm against court-packing.

Third, abandoning the law–norm distinction deprives us of the ability to develop a more finely grained picture of our constitutional order. The very point of introducing the distinction is to account for patterns and regularities in constitutional politics that are governed by rules that are not laws. We need the concept of a constitutional norm as distinct from law in order to explain these practices.

Finally, law is one of our few tools for stemming norm erosion, and collapsing the distinction between the two can obscure that. This Article does not give a way of sorting between “good,” “bad,” and “neutral” conventional breakdowns. The answer always depends. But the theory developed so far does take conventional change as a given. Sophisticated analysts have argued that one effective response to harmful conventional change is “anti-hardball,” which encases norm in law.185  See Fishkin & Pozen, supra note 54, at 981–82. For this response to work, law and norm must be meaningfully different. Even if it is a difference of degree and not kind, the fact that disobeying law comes with regular and effective penalties while flouting norm does not is meaningful for prescriptive purposes.186  See Tamir, supra note 98, at 887–88, 945 (observing the analogy between formal law and conventions and suggesting the latter clarifies the former). I part with Tamir on what bears emphasis in the comparison between law and convention. Where he stresses similarity, I press difference. For these reasons, philosophical and practical, I retain the distinction.

IV.     Norms at Work

Thus far, my argument has been theoretical. Now the theory is put to work. This Part animates constitutional norms through three case studies: blue slips, the convention against court-packing (anti-court-packing), and presidential noninterference with the Justice Department. Despite ostensible differences, all share the same underlying character: they are normative, contingent, and arbitrary practices that implement constitutional text and principle.

Here, I offer three further observations. First, when a convention is constitutional, it is also normative.187The converse, of course, is not true. This is a premise of constitutional theory: if we think the Constitution requires us to do something, then it is something we think we should do. As each of the following examples shows, actors following a constitutional norm defend the practice in constitutional terms. If a widely respected practice is justified in constitutional language, then it is good evidence that the practice may be a norm.188This is neither a sufficient condition nor always true. For instance, no one under age 35 has ever become president. This is not a convention. Instead, people are following a clear legal rule. U.S. Const. art. II, § 1, cl. 5 (“[N]either shall any person be eligible to that Office who shall not have attained to the Age of thirty five Years . . . .”).

Second, constitutional norms, like all conventions, have irregular “life cycles.” They can emerge organically during relatively calm political periods or at moments of political upheaval. They can endure undisturbed for a long time or undergo small changes while keeping the broader practice intact. And norms can end. They can die out or be transformed into institutional practices, exchanging their malleability for greater endurance and authority.

Third, a norm’s contingency and arbitrariness require a historical lens. For contingency, this is obvious. If a norm changes over time, then it is clearly contingent. Arbitrariness, however, is harder to see. In theory, we should be able to see how a practice is arbitrary since internal arbitrariness is a purely conceptual relation between principle and practice. Yet arbitrariness is not always obvious; in any given period, the convention often reflects the prevailing wisdom about how constitutional government should work.

Fortunately, a norm’s contingency reveals its arbitrariness. When norms come under pressure, the relationship between principle and practice unravels. This can happen in several different, but related, ways. First, an underlying value may be reinterpreted in a way that decouples it from past practice. Second, an alternative norm may be proposed that purports to better fit the underlying value. Third, the political world that made compliance not only appropriate but also attractive can change, creating incentives for actors to discard the practice. Whichever way a norm erodes, the resulting insight is the same: the norm bears no necessary relation to its underlying value and other ways of realizing constitutional text or practice are possible.

A.     Blue Slips for Judicial Nominations

The senatorial blue slip is a constitutional norm that has implemented the Advice and Consent Clause189“The President . . . shall nominate, and by and with the Advice and Consent of the Senate . . . appoint . . . Judges of the supreme Court, and all other Officers of the United States . . . .” U.S. Const. art. II, § 2, cl. 2. and separation of powers and federalism values since the early twentieth century. It is a practice of the Senate Judiciary Committee by which home-state senators exert influence in the selection of federal judges.190Senate blue slips are different from blue slips in the House. House blue slips are grounded in Article I, Section 7 of the Constitution—the Origination Clause—which provides that “All Bills for raising Revenue shall originate in the House of Representatives.” U.S. Const. art. I, § 7, cl. 1; James V. Saturno, Cong. Rsch. Serv., RL31399, The Origination Clause of the U.S. Constitution: Interpretation and Enforcement 1 (2011). Blue slips in this paper refer exclusively to the Senate’s, unless otherwise indicated. When the Senate Judiciary Committee is considering a judicial nominee for either a circuit or district court vacancy, the home-state senators are provided with blue slips of paper on which they can indicate approval or disapproval. The practice is an “informal custom”191Scholars have linked it to the broader tradition of “senatorial courtesy.” Brannon Denning, for example, has called the blue slip a “result of the Senate Judiciary Committee’s institutionalization of ‘senatorial courtesy.’ ” Brannon P. Denning, The “Blue Slip”: Enforcing the Norms of the Judicial Confirmation Process, 10 Wm. & Mary Bill Rts. J. 75, 76 (2001). Senatorial courtesy refers to a set of informal practices that shape interactions among senators and between the Senate and the president. Id. Whether blue slips are best seen as a form of senatorial courtesy—one where executive deference crosses party lines—or as a separate practice altogether, they are still conventions. of the Senate and is not codified in its rules.192Comm. on Rules & Admin., Standing Rules of the Senate, S. Doc. No. 113–18 (1st Sess. 2013). The basic procedure—the use of a blue slip by home-state senators—has remained intact throughout its recorded existence. Its effect on the committee’s decision on whether to advance a nominee to a full vote by the Senate, however, has varied with different committee chairs.193  See, e.g., Denning, supra note 195, at 78.

The origins of the blue slip are obscure. The few archival studies date the practice as far back as 1913, with the first confirmed blue slip appearing in the sixty-fifth Congress in 1917.194Sarah A. Binder, Where Do Institutions Come From? Exploring the Origins of the Senate Blue Slip, 21 Stud. Am. Pol. Dev. 1, 7–8 (2007) (observing that solicitations of home-state senators for judicial nominations “became routine in 1913” at the start of the sixty-third Congress); Mitchel A. Sollenberger, Cong. Rsch. Serv., RL32013, The History of the Blue Slip in the Senate Committee on the Judiciary, 1917–Present 5 (2003). I follow Binder’s 1913 dating since she is the only scholar who has tracked solicitation of home-state senators—and not just the blue slip as a proxy—in the executive dockets of the Senate Judiciary Committee. Binder, supra, at 7. The best account—Sarah Binder’s—suggests that it first emerged as a tool for reducing uncertainty in the Senate (a coordination device).195Binder, supra note 198, at 1. The 1910s were a period of robust institutionalization and formalization as congressional workloads “burgeoned.” The blue slip was nonpartisan—home-state senators could be from either party—and fit with the idea of a unified government trying to develop a “clear record of the home senators’ views on pending nominees” in an effort “to facilitate confirmation.”196  Id. at 10.

While the blue slip may have begun as an attempt to make the nomination process more efficient, it was transformed into a constitutive norm. Senator James Eastland gave the blue slip its modern shape. Before Eastland, a negative blue slip did not, on its own, sink a nomination.197Barry J. McMillion, Cong. Rsch. Serv., R44975, The Blue Slip Process for U.S. Circuit and District Court Nominations: Frequently Asked Questions 3 (2017). Negative blue slips during this period still carried weight but did not prevent a nominee from going to the Senate floor. For instance, in 1917, Senator Thomas W. Hardwick’s objected to U.V. Whipple’s nomination to the U.S. District Court for the Southern District of Georgia as “personally offensive and objectionable,” and although Whipple made it out of committee, he was voted down in the Senate. Alex Seitz-Wald, The Dubious Century-Old U.S. Senate ‘Blue Slip’ Custom May Finally End, NBC News (Oct. 14, 2017, 7:05 AM), https://www.nbcnews.com/politics/congress/dubious-century-old-u-s-senate-blue-slip-custom-may-n810571 [perma.cc/T2SM-BMLC]. Similarly, Senator Paul Douglas’s objections to two district court nominations by President Truman—a fellow Democrat—in 1951 led to the committee siding with Douglas but still sending the nominations to the Senate floor, where they were also rejected. This is a particularly notable pre-1956 example since Douglas was a consistent supporter of President Truman’s legislative agenda. Paul H. Douglas Award for Ethics in Government, Inst. Gov’t & Pub. Affs., https://igpa.uillinois.edu/ethics#section-1 [perma.cc/F793-AER3]; Sheldon Goldman, Picking Federal Judges: Lower Court Selection from Roosevelt Through Reagan 75 (1997). But when Eastland became chair of the Judiciary Committee in 1956, he turned the blue slip into a veto: the committee would not move forward on a nominee without positive blue slips from both home state senators.198McMillion, supra note 201, at 3. Eastland’s reasons for adopting this policy are unclear,199I suspect that Eastland sought to control judiciary appointments in the wake of Brown v. Board of Education, 347 U.S. 483 (1954). First, leading civil rights groups, including the National Association for the Advancement of Colored People (NAACP) and Americans for Democratic Action (ADA) both directly petitioned then-Senator Lyndon B. Johnson to abandon the seniority rule that gave Eastland control of the committee after the death its previous chair, Senator Kilgore. 3 Robert A. Caro, The Years of Lyndon Johnson: Master of the Senate 783 (2002). Second, Eastland fiercely opposed civil rights legislation and used procedural maneuvers in the Senate to combat it. Id. at 842–43, 874–875, 902–903 (documenting Eastland’s efforts in delaying and defanging the 1957 Civil Rights Act). Finally, Eastland used his position as chair to bargain with liberal presidential administrations over judicial appointments. Eastland most famously opposed President Kennedy’s nomination of then-lawyer Thurgood Marshall to the Second Circuit. According to one contemporary account, Eastland bargained over Marshall’s nomination by telling Attorney General Robert Kennedy, “You tell your brother if he gives me Cox, I will give him his [******].” Robert Shogan, Ex-Mississippi Sen. Eastland Dies at Age 81, L.A. Times (Feb. 20, 1986, 12:00 AM), https://www.latimes.com/archives/la-xpm-1986-02-20-mn-9797-story.html [perma.cc/C7LY-2FBT]. but his successors entrenched it. Depending on their priorities and the broader ideological climate, chairs have adjusted the practice while retaining its basic shape. Senator Kennedy briefly restored the blue slip to its pre-1956 strength200  See Mitchel A. Sollenberger, Judicial Appointments and Democratic Controls 100 (2011). to push through confirmations of more minorities, women, and liberals to the federal bench during the Carter presidency,201Ryan C. Black, Anthony J. Madonna & Ryan J. Owens, Qualifications or Philosophy? The Use of Blue Slips in a Polarized Era, 44 Presidential Stud. Q. 290, 294 n.8 (2014); see also Sollenberger, supra note 198, at 11 (documenting Sen. Kennedy moving forward with the nomination of James E. Sheffield, an African-American attorney, for a West Virginia district court seat, despite Senator Harry Byrd’s negative blue slip). but from the 100th Congress until our current 116th Congress, the blue slip has remained a fixture of senatorial judicial politics.

Historically, senators have invoked three different constitutional grounds to justify the norm: the Advice and Consent Clause,202“The President . . . shall nominate, and by and with the Advice and Consent of the Senate, shall appoint . . . Judges of the supreme Court, and all other Officers of the United States . . . .” U.S. Const. art. II, § 2, cl. 2. the separation of powers, and federalism. Senator Paul Douglas of Illinois gave one of the clearest statements of the perceived relationship between the blue slips and underlying constitutional text and principle during a nomination battle with President Truman. First, Douglas insisted that debates at the Founding made clear that “[t]he phrase ‘with the advice and consent of the Senate’ was not intended to be lightly construed.”20397 Cong. Rec. 12,838 (1951) (statement of Sen. Paul Douglas). The history, on Douglas’s construal, showed that the Advice and Consent Clause was a “relatively late . . . compromise” in which “the Senate was expected to play an active part in selecting Federal judges.”204  Id. The blue slip thus implemented advice and consent, helping the Senate play its “active part.” Later senators have rehearsed the same claims. During Obama’s second term, Senator Patrick Leahy explicitly invoked constitutional text. In language that seemingly nearly repeated the definition of a constitutive norm, Leahy claimed that blue slips “help[ed] make constitutional ‘advice and consent’ a reality.”205Patrick Leahy, ‘Blue Slips’ Help Make Constitutional ‘Advice and Consent’ A Reality, U.S. Sen. Patrick Leahy of Vt. (Apr. 7, 2014), https://www.leahy.senate.gov/press/blue-slips-help-make-constitutional-advice-and-consent-a-reality [perma.cc/32RW-RDYY] (cleaned up). And Senator Orrin Hatch, at virtually the same time, maintained that the blue slip helped “make meaningful ‘advice and consent’ a reality.”206Orrin Hatch, Protect the Senate’s Important ‘Advice and Consent’ Role, Hill (Apr. 11, 2014, 8:00 AM), https://thehill.com/opinion/op-ed/203226-protect-the-senates-important-advice-and-consent-role [perma.cc/3PCY-GYZB].

Second, Douglas defended the blue slip on separation-of-powers grounds. Given the role of the judiciary as “the arbiter of grave and basic disputes” between the political branches, the blue slip helpfully divided authority between them.20797 Cong. Rec. 12,838 (1951) (statement of Sen. Paul Douglas). Senator Richard M. Russell made nearly identical arguments a year before Douglas in rejecting the nomination of M. Neil Andrews to a Georgia district court. He maintained that disregard for senatorial courtesy and the blue slip as an “action taken to be in derogation of the rights of individual Senators and of the dignity of the Senate as a coordinate branch of the Government. It is contrary to custom, and in defiance of the constitutional powers of the Senate.” 96 Cong. Rec. 12,105 (1950) (statement of Sen. Richard Russell). Third, Douglas and later senators urged that the blue slip embodied important federalism values. “However excellent [the president’s] general knowledge,” Douglas asserted, “[he] does not have the detailed knowledge of the qualifications, background, and record of judges in a particular State” that the Senators from that state have.20897 Cong. Rec. 12,838 (1951) (statement of Sen. Paul Douglas).

Senators after Douglas have made similar arguments from federalism. For instance, even as he tinkered with the blue slip, Senator Kennedy acknowledged the unique role of home-state senators in the nominations process: “Appointments to [lower federal] courts . . . have been of special interest to individual Senators because Federal judicial districts are drawn within the boundaries of individual States.”209  The Selection and Confirmation of Federal Judges: Hearing Before the S. Comm. on the Judiciary, 96th Cong. 3 (1979) (statement of Sen. Edward M. Kennedy, Chair, S. Comm. on the Judiciary). During that same hearing, Senator Paul Laxalt, who opposed the change, framed the blue slip less as a senatorial privilege and more as a “responsibility” for home-state senators “to call these tough shots within our States.”210  Id. at 26 (statement of Sen. Paul D. Laxalt, Member, S. Comm. on the Judiciary).

As the blue slip has become obsolete, last-ditch defenses of the practice have again sounded in federalism. In opposing a Ninth Circuit nominee to a seat in California, both Senator Dianne Feinstein and former Senator and current Vice President Kamala Harris underlined that the nominee was “not a part of California’s legal community,” and added that “[h]e attended law school and clerked for two federal judges on the East Coast.”211Press Release, U.S. Sen. for Cal. Dianne Feinstein, Feinstein, Harris on Daniel Bress Nomination (June 19, 2019), https://www.feinstein.senate.gov/public/index.cfm/press-releases?id=495F52C1-2247-441F-89A7-A354E0FF9EC1 [perma.cc/H72Q-JWDP]. The nominee, Daniel Bress, was confirmed despite two negative blue slips. See Hailey Fuchs, Senate Confirms Trump Judicial Nominee to California-based 9th Circuit, Wash. Post (July 9, 2019), https://www.washingtonpost.com/politics/senate-confirms-trump-judicial-nominee-to-california-based-9th-circuit/2019/07/09/34671d6c-a27f-11e9-b8c8-75dae2607e60_story.html [perma.cc‌/RSP8-977P]. According to the senators, the nominee was “not familiar with the complicated, California-specific issues that regularly come before the Ninth Circuit.”212U.S. Sen. for Cal. Dianne Feinstein, supra note 215.

The historical development of the blue slip shows it was clearly contingent. From its initial shift from a coordinating to a constitutive convention to its later iterations under different chairs to its recent obsolescence,213John Crawley & Patrick L. Gregory, Senate Blue Slip Custom ‘Essentially Dead,’ Feinstein Says, Bloomberg L. (Mar. 28, 2019, 11:00 AM), https://news.bloomberglaw.com/us-law-week/senate-blue-slip-custom-essentially-dead-feinstein-says [perma.cc/4PRZ-BBA2]. the blue slip has evolved. Depending on the ideologies of various norm entrepreneurs214  See Tamir, supra note 98. and the political environment of a given era, the blue slip varied in strength until polarization rendered it untenable.

The blue slip’s contingency also reveals that it is arbitrary. The various obituaries written about the blue slip are telling, as critics of both political persuasions have celebrated its death. David Lat, for instance, has highlighted the need for more federal judges and greater judicial efficiency given that “the vast majority of cases heard by federal courts are not political.”215David Lat, Opinion, Good Riddance to Blue Slips, N.Y. Times (May 9, 2018), https://www.nytimes.com/2018/05/09/opinion/senate-judicial-nominees-blue-slips.html [perma.‌cc/PW5K-UJ47]. And Kevin Drum cast recent developments as part of a “long, crooked road” to more robust majority rule in Congress.216Kevin Drum, Blue Slips Are Finally Dead, Mother Jones (Feb. 28, 2019), https://www.‌motherjones.com/kevin-drum/2019/02/blue-slips-are-finally-dead [perma.cc/C2MH-L2J2].

Notably, the emphasis on silver linings is not accompanied by any suggestion that something of constitutional importance has been lost. This absence makes sense. Blue slips represented only one way of implementing the “Advice and Consent” Clause and bore a tenuous relation to federalism values. After all, appellate judges often hear claims arising from different states because circuit courts encompass multiple states. The blue slip is thus a prototypical constitutional norm: a normative, contingent, and arbitrary practice that has implemented constitutional text and principle.

B.     Anti-Court-Packing

Anti-court-packing is roughly the norm against “manipulating the number of Supreme Court seats primarily in order to alter the ideological balance of the Supreme Court.”217Joshua Braver, Court-Packing: An American Tradition?, 61 B.C. L. Rev. 2747, 2748 (2020). My account here does not challenge Braver’s learned argument about the novelty of Roosevelt’s plan. See id. at 2802. Through a careful examination of previous historical episodes when the Court’s size was changed, Braver claims that there never was a tradition of “court-packing.” See id. at 2750–51. He is up against what he terms the “standard history of court-packing,” which highlights several instances of court-packing during the nineteenth century. See id. at 2753. If Braver is right, then 1937 was the first constitutional showdown over court-packing. See id. at 2802. This leaves the conceptual argument that the anti-court-packing convention was forged at that moment, untouched. Braver concedes as much since he observes Tara Grove’s argument “that there was no norm against court-packing until the 1950s . . . may still hold,” regardless of the credibility of the “standard history.” See id. at 2753 n.11. Born during the climax of the New Deal Revolution,218  See Alan Brinkley, The Debate over the Constitutional Revolution of 1937: Introduction, 110 Am. Hist. Rev. 1046 (2005). anti-court-packing has implemented a nearly century-long commitment to judicial independence.219  See Grove, supra note 80, at 532. As Grove has persuasively shown, anti-court-packing is one among several practices that constitute this principle.220  Id. at 467–68. Other conventions here include compliance with federal court orders and respect for judicial tenure. Id. When the political branches forbear from expanding the Court for partisan purposes, they express respect for judicial independence. Judicial independence is an uncontroversial constitutional principle, even if its exact content is contested. Whether we define it as noninterference from the political branches or a statement that judicial decisionmaking is itself apolitical, it is clearly a constitutional principle and connected to the separation of powers. And contemporaries warn that violating the norm would be “anti-constitutional”221  E.g., Neil Siegel, The Anti-Constitutionality of Court-Packing, Balkinization (Mar. 26, 2019, 10:29 AM), https://balkin.blogspot.com/2019/03/the-anti-constitutionality-of-court_36.html [perma.cc/SMC9-8FFL]. and would leave a “semi-permanently tainted Supreme Court.”222  See Braver, supra note 221, at 2798. For defenders of the norm, judicial independence seems to entail anti-court-packing.

The perceived entailment makes it hard to see how the norm is contingent and arbitrary. But as with so many norms, anti-court-packing’s past and present are instructive.223The court-packing crisis boasts its own impressive body of secondary work. This Section draws primarily from four major accounts of the genesis and timeline of the 1937 crisis. See William E. Leuchtenburg, The Origins of Franklin D. Roosevelt’s “Court-Packing” Plan, in The Supreme Court Reborn, supra note 88, at 82; William E. Leuchtenburg, FDR’s “Court-Packing” Plan, in The Supreme Court Reborn, supra note 88, at 132; Marian C. McKenna, Franklin Roosevelt and the Great Constitutional War: The Court-Packing Crisis of 1937 (2002); Jeff Shesol, Supreme Power: Franklin Roosevelt vs. the Supreme Court (2010). First, contingency: anti-court-packing was hard-won and its birth was by no means guaranteed. The conflict over the Court had both long-term and proximate causes. Seen in the longue durée, the events of 1937 are unsurprising. Criticism of the judiciary was common in the various quarters of the American left224I use the term “left” to capture a broader range of the American political spectrum than the term “progressive.” As more than a half-century of historical scholarship has shown, the Progressives comprised a diverse and often times loosely organized group of reformers drawn from various elements of American society, concerned with the social, political, and economic consequences of industrialization. The movement’s internal diversity and its lack of a discrete institutional form thus make it hard to describe it as “leftist” in a conventional sense (there were, after all, Republican Progressives) or use it as a catchall term for antijudicial sentiment. Arthur S. Link, What Happened to the Progressive Movement in the 1920’s? 64 Am. Hist. Rev. 833, 836 (1959). See generally Robert H. Wiebe, The Search for Order, 1877–1920 (1967) (situating the Progressives in a broader transformation of the United States from localism to an organized, industrial society); David M. Kennedy, Overview: The Progressive Era, 37 Historian 453 (1975) (reviewing scholarly attempts to conceptualize the Progressive movement); Daniel T. Rodgers, In Search of Progressivism, 10 Revs. Am. Hist. 113 (1982) (arguing debates over the elusive characteristics of progressivism provide less insight than inquiries into the context of surrounding progressivism). as early as the 1890s.225William G. Ross, A Muted Fury: Populists, Progressives, and Labor Unions Confront the Courts, 1890–1937, at 1 (2016). This early burst of outrage was fueled by judicial decisions invalidating labor reforms and redistributive legislation.226Stuart S. Nagel, Court-Curbing Periods in American History, 18 Vand. L. Rev. 925, 928–30 (1965). Court crises have been a regular feature of American politics since the early Republic. Id. at 925–26. What perhaps distinguishes 1937 is its extended prelude, during which discontent with the judiciary became a part of the country’s political vocabulary. Despite ebbs and flows,227  See Melvin I. Urofsky, Myth and Reality: The Supreme Court and Protective Legislation in the Progressive Era, Y.B. Sup. Ct. Hist. Soc’y, 1983, at 68–70 (observing that the White Court did reluctantly embrace a large role for government in the 1910s). voices ranging from the socialist lawyer and failed judicial candidate Louis B. Boudin228L.B. Boudin, Government by Judiciary, 26 Pol. Sci. Q. 238, 238, 264 (1911) (warning that judicial review had pushed the nation into “the condition of ‘judicial despotism’”). to establishment figures like Roscoe Pound229Roscoe Pound, The Causes of Popular Dissatisfaction with the Administration of Justice, in Report of the Twenty-Ninth Annual Meeting of the American Bar Association Held at St. Paul, Minnesota 395 (1906) (urging that judicial doctrine be more responsive to public opinion). and Felix Frankfurter230Felix Frankfurter, Diary Entry of Felix Frankfurter (Oct. 27, 1911), in From the Diaries of Felix Frankfurter 113 (Joseph P. Lash ed., 1975) (calling the judiciary the “meanest, most selfish force in resisting just reforms and perpetuating public abuse in [the] administration of [our] laws”). all expressed frustration with a conservative judiciary.

Alongside academic commentary were serious political proposals backed by a coalition of labor, populists, and Progressives all critical of courts.231Ross, supra note 229, at 28–29. These proposals included popular recall of state judges,232Stephen Stagner, The Recall of Judicial Decisions and the Due Process Debate, 24 Am. J. Legal Hist. 257, 265 (1980). the elimination of the labor injunction, and a constitutional amendment allowing congressional override of federal judges.233  See David P. Thelen, Robert M. La Follette and the Insurgent Spirit 172–73 (Oscar Handlin ed., 1976); Nancy C. Unger, Fighting Bob La Follette 289–90 (2000). The political, social, and intellectual currents that converged in 1937 thus suggest an alternate narrative that might have emerged had FDR “won” the battle, along with the war: in overturning important pieces of the New Deal, the Court pushed its luck too far and a president armed with sufficient political will realized a dream a half century in the making, the reassertion of democracy over juristocracy.

In any event, Roosevelt’s court-packing plan failed, albeit narrowly, due to a series of political blunders, shrewd maneuvers by his opponents, and sheer accident. The basic timeline is well-known.234  See, e.g., Brinkley, supra note 222. In response to decisions striking down liberal state and federal legislation,235No single case, but rather a large set of them, persuaded Roosevelt and his attorney general, Homer Cummings, to pursue court-packing. They included Panama Refin. Co. v. Ryan, 293 U.S. 388 (1935); Norman v. Balt. & Ohio R.R. Co., 294 U.S. 240 (1935); United States v. Bankers’ Tr. Co., 294 U.S. 240 (1935); Nortz v. United States, 294 U.S. 317 (1935); Perry v. United States, 294 U.S. 330 (1935); A.L.A. Schechter Poultry Corp. v. United States, 295 U.S. 495 (1935); Louisville Joint Stock Land Bank v. Radford, 295 U.S. 555 (1935); Humphrey’s Ex’r v. United States, 295 U.S. 602 (1935); United States v. Butler, 297 U.S. 1 (1936); Carter v. Carter Coal Co., 298 U.S. 238 (1936); Morehead v. New York ex. rel. Tipaldo, 298 U.S. 587 (1936). Roosevelt introduced the “Judicial Procedures Reform Bill” in February, and the battle lasted until July, when he eventually relented after Justice Roberts’s “switch” in West Coast Hotel Co. v. Parrish.236300 U.S. 379 (1937). The leading historical accounts emphasize that Roberts’s “switch” occurred in late 1936 during a judicial conference that preceded the court-packing plan. See, e.g., Barry Cushman, Rethinking the New Deal Court (1998); Richard D. Friedman, Switching Time and Other Thought Experiments: The Hughes Court and Constitutional Transformation, 142 U. Pa. L. Rev. 1891, 1950 (1994). For Felix Frankfurter, the timing of Roberts’s switch was evidence of the Court’s independence. See Felix Frankfurter, Mr. Justice Roberts, 104 U. Pa. L. Rev. 311 (1955). And despite subsequent attack, the authenticity of Roberts’s 1945 memo explaining his decisions in Tipaldo and Parrish remains intact. Richard D. Friedman, A Reaffirmation: The Authenticity of the Roberts Memorandum, or Felix the Non-Forger, 142 U. Pa. L. Rev. 1985 (1994). For the purposes of this Article’s argument, the fact that the “switch” happened several weeks before the court-packing plan was announced is less important than the fact that Parrish, along with other events, undermined Roosevelt’s case for Court reform. Several factors conspired together to sink the plan. First, Roosevelt erred in framing the plan as a response to phantom docket congestion.237  See Shesol, supra note 227, at 325. Even after the botched delivery and remedial honesty about the bill’s motivations, the public split evenly for and against the bill.238  Id. at 330–31. Second, Roosevelt’s opponents, in the judiciary in particular, countered his plan in several ways. Chief Justice Charles Evans Hughes, an accomplished politician in his own right, provided a letter to the Senate Judiciary Committee exposing docket congestion as a sham;239James F. Simon, FDR & Chief Justice Hughes: The President, the Supreme Court, and the Epic Battle over the New Deal 390 (2012). Justice Van Devanter—a conservative stalwart—retired;240Shesol, supra note 227, at 446–48. and finally, Justice Roberts joined the four liberals in Parrish. Even with these moves, however, a compromise bill241  See William E. Leuchtenburg, Comment, FDR’s Court-Packing Plan: A Second Life, A Second Death, 1985 Duke L.J. 673. that would have tipped the Court in Roosevelt’s favor was viable well into the summer. The nail in the proverbial coffin was the death in July of Senate Majority Leader Joe Robinson, who took any chance of passage with him.242Even in mid-June, Robinson likely had the necessary votes for the compromise bill. Shesol, supra note 227, at 474–76.

Roosevelt’s plan, then, lost in the court of public opinion and the halls of Congress, but not in the forum of principle. From the ill-advised decision to frame court-packing as a solution to a nonexistent problem, judicial retrenchment, and Chief Justice’s Hughes deft politicking, anti-court-packing owed its birth to a blend of skill and luck. And it took a literal act of God—the death of Robinson—to bring it into being. The norm we have today, while durable and venerated, has been contingent from its very conception. And as it is often the case, it is better for a convention to be lucky than to be good.

But is anti-court-packing arbitrary? This is often the hardest condition to satisfy, especially when we have associated a practice with a principle for as long as we have anti-court-packing with judicial independence. It is even more so when the line between the two is so direct so as to seem deductive. As renascent arguments for court-packing show, however, shifting political conditions can help us question received truths. From the refusal to hold hearings for Judge Garland’s nomination to the bitter battle over Justice Brett Kavanaugh’s confirmation,243  See Ron Elving, What Happened with Merrick Garland in 2016 and Why It Matters Now, NPR (June 29, 2018, 5:00 AM), https://www.npr.org/2018/06/29/624467256/what-happened-with-merrick-garland-in-2016-and-why-it-matters-now [perma.cc/S64V-LFFE]; Tom McCarthy, Q&A: Brett Kavanaugh’s Controversial Confirmation Battle Explained, Guardian (Oct. 5, 2018, 9:55 AM), https://www.theguardian.com/us-news/2018/oct/05/kavanaugh-confirmation-battle-explainer [perma.cc/9UT3-J72Q]. judicial reform is once again a serious concern for legal liberals. This burgeoning interest has taken various forms, including proposals for changing the docket of the Supreme Court,244  See, e.g., Jack Balkin, Constitutional Rot Reaches the Supreme Court, Balkinization (Oct. 6, 2018, 10:50 AM), https://balkin.blogspot.com/2018/10/constitutional-rot-reaches-supreme-court.html [perma.cc/G3JP-6J48]; Reforming the Court, New Am., https://www.newamerica.org/political-reform/events/reforming-court [perma.cc/U7XM-ACGZ]. stripping its jurisdiction,245  See, e.g., Samuel Moyn, Resisting the Juristocracy, Bos. Rev. (Oct. 5, 2018), https://bostonreview.net/law-justice/samuel-moyn-resisting-juristocracy [perma.cc/AS2E-EFZL]; Eric Segall, Yes, It’s Time to Reform the Supreme Court—but Not for the Wrong Reasons, Salon (Dec. 4, 2018, 7:00 AM), https://www.salon.com/2018/12/04/its-time-to-reform-the-supreme-court-but-not-for‌-the-wrong-reasons [perma.cc/R2YE-7F3E]. setting term limits246  See, e.g., Bruce Ackerman, Opinion, Trust in the Justices of the Supreme Court Is Waning. Here Are Three Ways to Fortify the Court, L.A. Times (Dec. 20, 2018, 3:15 AM), https://www.‌latimes.com/opinion/op-ed/la-oe-ackerman-supreme-court-reconstruction-20181220-story.html [perma.cc/RGR9-Y3QJ]; Ezra Klein, Ruth Bader Ginsburg’s Death Is a Tragedy. The Supreme Court’s Rules Made It a Political Crisis, Vox (Sept. 18, 2020, 9:25 PM), https://www.vox.com/policy-and-politics/2018/12/26/18155093/ruth-bader-ginsburg-supreme-court-term-limits [perma.‌cc/8S2Q7342]; David Leonhardt, Opinion, The Supreme Court Needs Term Limits, N.Y. Times (Sept. 18, 2018), https://www.nytimes.com/2018/09/18/opinion/columnists/brett-kavanaugh-supreme-court-term-limits.html [perma.cc/XW57-8Y8Y]. and voting rules,247  See, e.g., Moyn, supra note 249; Jed Shugerman, Balanced Checks, Slate (June 20, 2012, 6:19 PM), https://slate.com/news-and-politics/2012/06/supermajority-voting-on-the-supreme-court.html [perma.cc/ARW7-TCGD]. and even expanding the Court.248  See, e.g., Ian Ayres & John Fabian Witt, Opinion, Democrats Need a Plan B for the Supreme Court. Here’s One Option, Wash. Post (July 27, 2018), https://www.washingtonpost.com/opinions/democrats-need-a-plan-b-for-the-supreme-court-heres-one-option/2018/‌07/27/4c77fd4e-91a6-11e8-b769-e3fff17f0689_story.html [perma.cc/3DKS-3AK5]; Jamelle Bouie, Opinion, Mad About Kavanaugh and Gorsuch? The Best Way to Get Even Is to Pack the Court, N.Y. Times (Sept. 17, 2019), https://www.nytimes.com/2019/09/17/opinion/kavanaugh-trump-packing-court.html [perma.cc/84UT-ZFCQ]; Daniel Epps & Ganesh Sitaraman, How to Save the Supreme Court, Vox (Oct. 10, 2018, 11:25 AM), https://www.vox.com/the-big-idea/2018/9/6/17827786/kavanaugh-vote-supreme-court-packing [perma.cc/2V6R-CUX7]; Michael Klarman, Why Democrats Should Pack the Supreme Court, Take Care (Oct. 15, 2018), https:‌//takecareblog.com/blog/why-democrats-should-pack-the-supreme-court [perma.cc/7GWX-RUNT].

If anti-court-packing has endured because the interbranch bargain it reflects has been tolerable, then the recrudescence of arguments for court-packing suggests that compromise has grown less attractive. One prominent strand of this thinking is nakedly partisan. It assumes the Court is as political as any other branch and justifies court-packing as a corrective to conservative judicial power.249Bouie, supra note 252. By rejecting judicial independence as mere ideology, it attacks anti-court-packing at its roots. On this view, even the principle underlying the practice is dubious.

To see arbitrariness, however, a second view is more illuminating. Daniel Epps and Ganesh Sitaraman’s recent Court reform proposals exemplify this position. Both the “Supreme Court Lottery” and “the Balanced Bench” would enlarge the Court, either directly or by expanding the eligible roster of judges.250Daniel Epps & Ganesh Sitaraman, How to Save the Supreme Court, 129 Yale L.J. 148, 181–205 (2019). The argument that follows could equally work with a different argument for court-packing in service of judicial independence. Rivka Weill, Court Packing as an Antidote, 42 Cardozo L. Rev. 2705 (2021) (arguing for court-packing as a way to defend the legitimacy and independence of the Supreme Court); see also Thomas M. Keck, The Supreme Court Justices Control Whether Court-Packing Ever Happens, Wash. Post (Nov. 19, 2018), https://www.washingtonpost.com/outlook/2018/11/19/supreme-court-justices-control-whether-court-packing-ever-happens [perma.cc/LK9F-69VT] (rejecting the idea that “tinkering” with the size of the Court threatens judicial independence); Danielle Root & Sam Berger, Ctr. for Am. Progress, Structural Reforms to the Federal Judiciary 13 (2019), https://americanprogress.org/wp-content/uploads/2019/05/JudicialReform-report-1.pdf [perma.cc/2QZF-PFA8] (similarly floating the idea that an expanded Court would be less partisan); cf. Marin K. Levy, Packing and Unpacking State Courts, 61 Wm. & Mary L. Rev. 1121 (2020) (cataloguing important examples of court-packing in state courts). They pitch their plans as a “hardball” means toward “anti-hardball” ends that would “lower the temperature of political disputes.”251Epps & Sitaraman, supra note 254, at 172. The “balanced bench” approach, in particular, is cast as an attempt to “bring[] back the possibility of a Supreme Court that is not wholly partisan” and elevate judges with a “reputation for fairness, independence, and centrism.”252  Id. at 193.

Their proposals are interesting because they invoke judicial independence in the name of expanding the Court. In other words, the very value that anti-court-packing is meant to concretize is enlisted to pack the Court.253With respect to this inversion, arbitrariness is similar to the phenomenon of ideological drift, where “an argument or trope” shifts from one political valence to another over time. Both concepts reveal the indeterminacy of political and legal concepts in practice. See, e.g., J.M. Balkin, Ideological Drift and the Struggle over Meaning, 25 Conn. L. Rev. 869, 870 (1993); David E. Pozen, Transparency’s Ideological Drift, 128 Yale L.J. 100, 105–07 (2018). Of course Epps and Sitaraman’s proposal, if implemented, might fail to deliver. This could happen because they misjudge the tit-for-tat dynamics of Court expansion or because judges dig further into partisan positions instead of moderating their views. But their claims are serious and intelligible. And for arbitrariness, that is what counts. Anti-court-packing might ultimately prove more effective at constituting judicial independence than a finely tuned court-packing proposal, but a norm can be better than other plausible options and still be arbitrary. It is the existence of other plausible ways—indeed, even opposite ones—of construing the underlying value that is characteristic of norms.

C.     Executive Noninterference in the Department of Justice

Executive noninterference in the Department of Justice (DOJ) is just as conventional as the blue slip or anti-court-packing but more complex. Whereas the prior two conventions consisted of one discrete practice, executive noninterference comprises several. It is thus a good example of a nested convention: a relatively broad convention composed of constituent practices.254  See supra Section III.A.

Executive noninterference refers to long-standing presidential forbearance from involvement in specific cases and investigations. This norm implements the president’s duty to “take [c]are that the [l]aws be faithfully executed.”255U.S. Const. art. II, § 3. Noninterference consists of a set of White House and law enforcement agency policies channeling and restricting communications between them. For example, the White House can learn information about a criminal investigation only on a need-to-know basis when it is required for the “performance of the President’s duties.”256U.S. Dep’t of Just., Just. Manual § 1-8.600 (2021).Similarly, only specific members of the Office of the Counsel to the President, the vice president, and the president themself can begin contact with the Justice Department about ongoing criminal cases. In addition, direct legal advice to the Executive Office of the President257The Executive Office of the President includes the Council of Economic Advisers, the Council on Environmental Quality, the Domestic Policy Council, the Gender Policy Council, the National Economic Council, the National Security Council, the Office of Intergovernmental Affairs, the Office of Management and Budget, the Office of National Drug Control Policy, the Office of Public Engagement, the Office of Science and Technology Policy, and the Office of the United States Trade Representative. Executive Office of the President, White House, https://www.whitehouse.gov/administration/executive-office-of-the-president [perma.cc/7D89-SGWE]. must be channeled through the Office of Legal Counsel.258  See U.S. Dep’t of Just., Just. Manual § 1-8.600 (2021). And White House staff have been prohibited from “even [asking] for a status report” about some “pending matter[s]” in the Justice Department.259See Memorandum from Jack Quinn, Counsel to the President, Kathleen Wallman, Deputy Counsel to the President, and Stephen Neuwirth, Associate Counsel to the President, to White House Staff, Contacts with Agencies (Jan. 16, 1996), https://clinton.presidentiallibraries.us/items/show/27001 [perma.cc/Q3N4-B9NL].

In practice, these policies mean that informal communications, like a call from a White House official to the DOJ inquiring about specific matters, are forbidden. These restrictions are self-imposed and “prophylactic” and are meant to create procedural regularity and formality between the White House and the country’s chief law enforcement arms.260  See White House Communications with the DOJ and FBI, Protect Democracy (Mar. 8, 2017), https://protectdemocracy.org/agencycontacts [perma.cc/R9M8-MK85]. By contrast, communications between the president and the Justice Department over general policy matters such as enforcement priorities are open and informal, as they are with other executive branch agencies.261  Id.

From its inception, executive noninterference has been defended in constitutional terms. When he first articulated the norm, Attorney General Griffin B. Bell explicitly framed it as way to effectuate the president’s Take Care duties. Observing that the president is “charged by the Constitution with the duty to . . . ‘take care that the laws be faithfully executed,’” Bell linked it to an institutional division of labor, in which “the President has delegated certain responsibilities to the Attorney General.”262Griffin B. Bell, Att’y Gen. of the U.S., An Address Before the Dep’t of Just. Laws. (Sept. 6, 1978), transcript available at https://www.justice.gov/sites/default/files/ag/legacy/2011/08/23/09-06-1978b.pdf [perma.cc/4AFF-X7FK], at 4–5. “Although true institutional independence [was] . . . impossible,” Bell insisted the president “[was] best served if [government lawyers were] free to exercise their professional judgments.”263  Id. at 5. Subsequent attorney generals and White House counsels264Memorandum from Donald F. McGahn II, Counsel to the President, to All White House Staff, Communications Restrictions with Personnel at the Department of Justice (Jan. 27, 2017), https://www.politico.com/f/?id=0000015a-dde8-d23c-a7ff-dfef4d530000 [perma.cc/LXS7-W8TB] (White House Counsel for the Trump Administration reiterating commitment to the convention as consistent with “the President’s constitutional obligation to take care that the laws of the United States are faithfully executed.”). have made similar arguments. They continue to link these policies to the Take Care Clause, explaining that noninterference “recognizes the President’s ability to perform his constitutional obligation to ‘take care that the laws be faithfully executed’ while ensuring that there is public confidence that the laws . . . are . . . enforced in an impartial manner.”265Memorandum from the Att’y Gen. to Heads of Dep’t Components and United States Att’ys, Communications with the White House (Dec. 19, 2007), https://www.justice.gov/sites/default/files/ag/legacy/2008/04/15/ag-121907.pdf [perma.cc/65DE-7XE5]. And the policy has been taken very seriously in both the White House and the Justice Department.266U.S. Dep’t of Just., Just. Manual § 1-8.100 (2019) (noting that it is “a fundamental duty of every employee of the Department to ensure that these principles are upheld”); Quinn, Wallman & Neuwirth, supra note 263 (urging that these policies “must be strictly enforced” with clear guidelines on things White House staff “should” or “should not” do). In other words, these norms form a crucial part of the constitutional culture of the presidency.267  See Reva B. Siegel, Community in Conflict: Same-Sex Marriage and Backlash, 64 UCLA L. Rev. 1728, 1730 (2017) (observing that “[c]onstitutional culture, on this account, includes the understandings about role” that guides action).

Noninterference, as a product of its time, is contingent. Forged and articulated in the wake of Watergate, the norm has endured for more than forty years, despite moments of pressure. While the Reagan, Clinton, and Obama presidencies all restricted communications with senior officials in the agencies and the White House, the George W. Bush Administration significantly relaxed them, allowing the White House far greater access to the Justice Department. Political pressure returned the norm to its historical strength late in the Bush Administration, but the brief interregnum is revealing.268  See Isaac Arnsdorf, Sessions Faces Decision on Politicizing Justice Department, Politico (Jan. 9, 2017, 9:05 PM), https://www.politico.com/story/2017/01/jeff-sessions-attorney-general-justice-233382 [perma.cc/4J3Q-XS95]. Like the blue slip, then, noninterference has endured but with moments of real change.

Current and past pressure on the norm also show how it is arbitrary. When Attorney General Alberto Gonzales expanded White House access, he defended the move by invoking an expansive vision of executive power inspired by Justice Antonin Scalia.269  Id.As the norm has been violated repeatedly in the Trump Administration,270  Protecting Independent Law Enforcement, Protect Democracy, https://protectdemocracy.org/protecting-independent-law-enforcement/tracker [perma.cc/W4H5-UC8N] (listing examples of White House interference with Justice Department enforcement actions during the Trump era). critics of the norm have sought to delegitimize the practice altogether.271  See, e.g., Mario Loyola, Trump’s DOJ Interference Is Actually Not Crazy, Atlantic (Feb. 27, 2020), https://www.theatlantic.com/ideas/archive/2020/02/trumps-doj-unitary-executive/607141 [perma.cc/UY28-2HP9]; Carson Holloway, Opinion, No Easy Task for a President to ‘Abuse’ His Authority over the Justice Department, Hill (Mar. 2, 2018, 11:00 AM), https://thehill.com/opinion/white-house/376434-no-easy-task-for-a-president-to-abuse-his-authority-over-the-justice [perma.cc/SR6T-EPT9]. As they see it, the very idea that the president could unlawfully interfere with law enforcement, even in specific matters, is a solecism. Instead, they envision a “unitary executive,”272The literature on the unitary executive is vast and now spans both originalist statements of the position and historiographical surveys. See, e.g., Steven G. Calabresi & Christopher S. Yoo, The Unitary Executive (2008); Steven G. Calabresi & Kevin H. Rhodes, The Structural Constitution: Unitary Executive, Plural Judiciary, 105 Harv. L. Rev. 1153 (1992); see also Stephen Skowronek, John A. Dearborn & Desmond King, Phantoms of a Beleaguered Republic (2021) (outlining the history of the theory); cf. Gillian E. Metzger, The Constitutional Duty to Supervise, 124 Yale L.J. 1836 (2015) (emphasizing an executive duty to foster effective administration); Zachary S. Price, Enforcement Discretion and Executive Duty, 67 Vand. L. Rev. 671 (2014) (taking a limited view of the president’s enforcement discretion). For recent work critiquing the historical foundations of unitary executive theory, see Daniel D. Birk, Interrogating the Historical Basis for a Unitary Executive, 73 Stan. L. Rev. 175 (2021); Jane Manners & Lev Menand, The Three Permissions: Presidential Removal and the Statutory Limits of Agency Independence, 121 Colum. L. Rev. 1 (2021) (recovering the origins of removal protections as removal permissions); Peter M. Shane, The Originalist Myth of the Unitary Executive, 19 U. Pa. J. Const. L. 323 (2016). whose power as the Constitution’s highest law enforcer includes the ability to intervene in specific cases and investigations.

Their position, though by no means uncontested,273Many have pushed back strongly against the Trump Administration’s attack on noninterference. For academic arguments questioning the legality of such behavior, see, for example, Bruce A. Green & Rebecca Roiphe, Can the President Control the Department of Justice?, 70 Ala. L. Rev. 1 (2018) (collecting historical evidence of prosecutorial independence); Daniel J. Hemel & Eric A. Posner, Presidential Obstruction of Justice, 106 Calif. L. Rev. 1277 (2018) (defending the applicability of obstruction of justice statutes to the president); Andrew McCanse Wright, The Take Care Clause, Justice Department Independence, and White House Control, 121 W. Va. L. Rev. 353 (2018) (linking the convention of noninterference to prevention of unconstitutional conduct). The opposition has also included former Justice Department officials in Republican administrations. Statement in Response to Attorney General Barr’s Address at Federalist Society, Checks & Balances (Nov. 22, 2019), https://checks-and-balances.org/statement-from-co-founders-and-additional-members-of-checks-balances [perma.cc/E7RR-3ELA] (rejecting the historical credibility of an “autocratic vision of executive power”). is not out of the legal mainstream. Even opponents concede that noninterference is ultimately conventional274Vikram David Amar, Two Constitutional Lessons Worth Remembering: Norms Are Different from Legal Rules; and Improper Intent Matters but Is Hard to Establish, Verdict (Feb. 18, 2020), https://verdict.justia.com/2020/02/18/two-constitutional-lessons-worth-remembering [perma.cc/7EXB-VTBG] (explaining that non-interference is ultimately a nonlegal practice). and depends on a legal and political culture to sustain it.275Jack Goldsmith, Independence and Accountability at the Department of Justice, Lawfare (Jan. 30, 2018, 2:16 PM), https://www.lawfareblog.com/independence-and-accountability-department-justice [perma.cc/Y464-RB7Q] (explaining the various nonlegal and cultural bases of Justice Department independence). When critics of a constitutional norm attack it in terms that their opponents concede as intelligible, that is a clear sign the practice is arbitrary. The practice might still be worth fighting for and defending, but it cannot rest its case on its “constitutionality.” Other glosses on the underlying text and principles are now firmly “on the wall”276“On the wall” refers to constitutional arguments within the mainstream of legal and academic discourse at any point in time. J.M. Balkin, Agreements with Hell and Other Objects of Our Faith, 65 Fordham L. Rev. 1703, 1710, 1735 (1997) (using the terms “off the wall” and “on the wall” to describe constitutional arguments); Jack M. Balkin, From Off the Wall to On the Wall: How the Mandate Challenge Went Mainstream, Atlantic (June 4, 2012), https://www.theatlantic.com/national/archive/2012/06/from-off-the-wall-to-on-the-wall-how-the-mandate-challenge-went-mainstream/258040 [perma.cc/FJ6G-23FS] (discussing the terms “off the wall” and “on the wall” as they pertain to constitutional arguments). and the norm’s ultimate survival up for grabs.


A theory of constitutional norms does not tell us everything about their consequences. But this Article is an important first step. Just as constitutional theory has turned to the philosophy of language to understand the nature of constitutional text,277  See supra note 35 and accompanying text. this Article has enlisted the philosophy of conventions to grasp the character of constitutional practice. It explains that norms are normative, contingent, and arbitrary practices that implement constitutional text and principle. And it shows this theory in action, animating norms in various institutional settings. For judges and scholars who increasingly look to historical practice as a source of law, this Article raises hard questions.

First, reasoning from norms means relying on a practice that is not, on its own, the final word on constitutional text or principle. After all, norms are intrinsically arbitrary. They remind us that just because things have always been one way does not mean they have to be and vice versa.278  See Leah M. Litman, Debunking Antinovelty, 66 Duke L.J. 1407, 1407 (2017) (rejecting judicial skepticism of novelty as assuming either “the mistaken Madisonian premise that Congress reliably exercises the full scope of its constitutional powers” or that prior failures to enact statutes are clear signs of constitutional doubt). When norms are litigated, then, it is rarely a simple question of deciding whether the practice is permissible under the Constitution. Instead, institutional role matters. For instance, there is the initial question of when conventions can be litigated. This will turn on justiciability, touching factors like standing (“Can an institutional actor like Congress bring this claim?”)279Wilson C. Freeman & Kevin M. Lewis, Cong. Rsch. Serv., R45636, Congressional Participation in Litigation: Article III and Legislative Standing 6–13 (2019). and political question doctrine (“Is this a question courts are best suited to answer?”).280Compare the majority opinion and dissent in Zivotofsky v. Clinton, 566 U.S. 189 (2012).

If a norm does end up in court and a judge points to it as a source of law, then its institutional role is pivotal. History is rarely decisive on its own, until a court deems it so. Decisions that ratify a norm as the authoritative gloss on constitutional text are a quintessential example of courts creating law. This also means that in such cases, invoking history can obscure the exercise of judicial discretion. Given that a particular norm is an arbitrary way of realizing constitutional text and principle, this Article’s theory suggests that when judges transform a norm into law, they do so for policy decisions about institutional design and democratic theory. Clarifying the persistence of that discretion and the factors that guide it is an important next step.281  See David E. Pozen & Adam M. Samaha, Anti-Modalities, 119 Mich. L. Rev. 729, 731 (2021) (describing forms of constitutional argument “no reputable constitutional decisionmaker wishes to be associated with” because they run counter to constitutional norms.).

Second, the theory has implications for theories of constitutional interpretation. In particular, originalism and unwritten constitutionalism282  See sources cited supra note 33. consult historical practice in deciding thorny constitutional questions. Yet they do so in different ways and with different understandings of what consultation entails. For originalism, historical practice is common in the “construction zone.”283  See sources cited supra note 30. Originalists, however, disagree about what exactly should happen there.284Robert W. Bennett & Lawrence B. Solum, Constitutional Originalism 26 (2011). Unwritten constitutionalists similarly vary on fundamentals: is the unwritten constitution purely conceptual285  See Strauss, supra note 33; Tribe, supra note 33. or does it also comprise historical practice?286  See Amar, supra note 33. Insofar as both originalism and unwritten constitutionalism rely on norms, this Article’s theory challenges any strong claims to constitutional authority and determinacy by both camps.

Third, this Article underlines the interaction between ideology and politics in the constitutional order. Historical practices inform and are shaped by the constitutional culture they inhabit. Some of our most important and hallowed traditions depend on constitutional theories to sustain them. When their conditions of possibility change, they do too. Just as the norms against executive noninterference in law enforcement came of age during a period of skepticism of presidential authority, it has been endangered by the rise of unitary executive theory. So while these practices are normative, they can only be so when actors justify them in terms that others recognize. This means when ambient constitutional assumptions change, as they might in our era, the roster of possible responses also expands or narrows. If unitary executive theory triumphs, then saving the norm of noninterference by trying to further legalize it will be difficult. Under a strong unitary executive, after all, presidents cannot “interfere” with investigations, they only “intervene.” The interaction between ideology and institutions is thus a key dynamic of the constitutional order.

The Article’s theory means that the Constitution, in practice, is always a developing project. This might trouble any vision of constitutionalism that sees its only virtues as fixity and stability. But for those less wedded to formalism and more congenial to pragmatism, who understand the Constitution as “intended to endure for ages to come, and, consequently, to be adapted to the various crises of human affairs”287McCulloch v. Maryland, 17 U.S. (4 Wheat.) 316, 415 (1819). this Article’s theory throws new light on an old truth.

* Academic Fellow and Lecturer in Law, Columbia Law School. I am grateful for comments from Dhruv Aggarwal, Akhil Amar, Kate Andrias, Anya Bernstein, Josh Chafetz, Richard Friedman, Jonathan Gould, Maeve Glass, Jamal Greene, David Grewal, Michael Heller, Daniel Hornung, Jeremy Kessler, Issa Kohler-Hausmann, Jane Manners, Lev Menand, Samuel Moyn, Robert Post, David Pozen, Arjun Ramamurti, Aziz Rana, Daphna Renan, Daniel Richman, Cristina Rodríguez, Ganesh Sitaraman, Oren Tamir, Karen Tani, John Witt, Josh Zoffer, and participants in the Columbia Academic Fellows Workshop. Many thanks as well to the careful and conscientious editors of the Michigan Law Review, in particular Latazia Carter, Ruby Emberling, Dashaya Foreman, Michaela Olson, Ahmed Rizk, Julianna St. Onge, Emily Statham, Nia Vogel, and Lauren Yu.