Inside Congress’s Mind

Inside Congress’s Mind

In recent years, most would associate “intent skepticism” with the rise of modern textualism. In fact, however, many diverse approaches—legal realism, modern pragmatism, Dworkinian constructivism, and even Legal Process purposivism—all build on the common theme that a complex, multimember body such as Congress lacks any subjective intention about the kind of difficult issues that typically find their way into court. From that starting point, competing approaches have tended to focus on which inter­pretive method will promote appropriate conceptions of legislative supremacy and the role of the courts in our constitutional system. The debates, in recent years, between textualists and modern defenders of Legal Process purposivism (such as Professor Peter Strauss) nicely illustrate that emphasis.

A new generation of empirical scholarship, however, has raised questions about the intent skepticism that has long framed the interpretation debate. Most prominently, Professors Abbe Gluck and Lisa Bressman conducted an extensive survey of the understandings and practices of 137 members of the congressional staff who are engaged in legislative drafting. According to the authors, the resultant findings show, inter alia, that some interpretive approaches cannot be squared with legislative intentions while others nicely reflect such intentions. Ultimately, however, this Essay concludes that the study’s findings, although illuminating, do not alter the baseline of intent skepticism against which the statutory debate has proceeded. Indeed, the very idea of legislative intent remains unintelligible without a normative framework that structures what should count as Congress’s decision.


    1. Intent Skepticism Across the Board
    2. Legal Process Versus Textualism
      1. Legislative History
      2. The Chevron Doctrine(s)
    1. Norms of Statutory Usage
    2. Legislative History and the Construction of Legislative Intent
    3. The Chevron Doctrine



These days, one typically associates “intent skepticism” with the new textualism that began to press for judicial acceptance near the end of the twentieth century. 1 See, e.g., Bernard W. Bell, Legislative History Without Legislative Intent: The Public Justification Approach to Statutory Interpretation, 60 Ohio St. L.J. 1, 59 n.218 (1999) (linking textualism and intent skepticism); Linda D. Jellum, But That Is Absurd! Why Specific Absurdity Undermines Textualism, 76 Brook. L. Rev. 917, 919–21 (2011) (same); John David Ohlendorf, Textualism and the Problem of Scrivener’s Error, 64 Me. L. Rev. 119, 123–24 (2011) (same). For the article that identifies the phenomenon of “the new textualism,” see William N. Eskridge, Jr., The New Textualism, 37 UCLA L. Rev. 621, 623 (1990), which names and defines the Court’s movement toward a more text-based approach to interpretation. Because of the modern legislature’s complexity and the path dependence of its work, any such multimember lawmaking body will lack collective intent on any question worth worrying about. Hence, say the textualists, trying to cull legislative “intent” or “purpose” from snippets of legislative history invites the interpreter on a “wild-goose chase.” 2 Antonin Scalia, Judicial Deference to Administrative Interpretations of Law, 1989 Duke L.J. 511, 517 [hereinafter Scalia, Judicial Deference] (questioning collective legislative intent); see also, e.g., Frank H. Easterbrook, Text, History, and Structure in Statutory Interpretation, 17 Harv. J.L. & Pub. Pol’y 61, 68 (1994) [hereinafter Easterbrook, Text, History, and Structure] (“Intent is elusive for a natural person, fictive for a collective body.”). It is by now familiar that, for the textualist, this premise suggests that the best the interpreter can do is to ask how a reasonable person would read the text. 3 See, e.g., Frank H. Easterbrook, The Role of Original Intent in Statutory Construction, 11 Harv. J.L. & Pub. Pol’y 59, 65 (1988) [hereinafter Easterbrook, Original Intent] (articulating the “reasonable” reader framework).

The truth is, however, that intent skepticism also underlies most of textualism’s competitors. Theories as diverse as legal realism, 4 See Max Radin, Statutory Interpretation, 43 Harv. L. Rev. 863, 870–71 (1930) [hereinafter Radin, Statutory Interpretation] (laying out the realist position); see also infra text accompanying notes 44–45 (same). prag­matism, 5 See William N. Eskridge, Jr. & Philip P. Frickey, Statutory Interpretation as Practical Reasoning, 42 Stan. L. Rev. 321, 347–48 (1990) [hereinafter Eskridge & Frickey, Practical Reasoning] (defending modern interpretive pragmatism); see also infra text accompanying notes 46–48 (describing the pragmatist position on legislative intent). and Dworkinism 6 See Ronald Dworkin, Law’s Empire 318–50 (1986) (developing a coherentist approach to interpretation); see also infra text accompanying notes 49–53 (expanding on Dworkin’s view of legislative intent). all build out from the idea that interpreters cannot reasonably expect to identify what “Congress” actually decided about the litigated issue in hard cases. Likewise, the dominant philosophy of the post–New Deal state—the Legal Process purposivism that the new textualism has sought to replace—adopts intent skepticism. 7 See generally Henry M. Hart, Jr. & Albert M. Sacks, The Legal Process: Basic Problems in the Making and Application of Law (William N. Eskridge, Jr. & Philip P. Frickey eds., Foundation Press 1994) (1958) (developing a purposive approach to interpretation); see also infra text accompanying notes 54–61 (elaborating on the Legal Process approach). Only the prescription differs: Since a legislature lacks specific intentions on the questions that trouble interpreters, judges should indulge the normatively attractive presumption that “the legislature was made up of reasonable persons pursuing reasonable purposes reasonably.” 8 Hart & Sacks, supra note 7, at 1378.

The widespread acceptance of intent skepticism presents a difficult question for statutory interpretation theory: If leading theories such as textualism or purposivism cannot be justified as superior ways to identify some actual legislative intent or decision about the question at issue, how does each approach reflect the appropriate roles and functions of the diverse branches of government in a constitutional democracy such as ours? 9 For discussion of this development, see John F. Manning, The Supreme Court, 2013 Term—Foreword: The Means of Constitutional Power, 128 Harv. L. Rev. 1, 16–17 (2014) [hereinafter Manning, Foreword] (sketching statutory interpretation theorists’ responses to intent skepticism). Against a backdrop of intent skepticism, the concept of “legislative intent” is a metaphor that invites interpreters to think about how to attribute a decision to a complex, multiparty body that does not have a mental state. 10 See Victoria F. Nourse, A Decision Theory of Statutory Interpretation: Legislative History by the Rules, 122 Yale L.J. 70, 81–82 (2012) (making the point that intent is a common and sensible “metaphor” for decisions of corporate body like Congress). This task requires making normative judgments about the nature of legislative power and the role of the courts in our system of government. 11 See, e.g., Jerry Mashaw, As If Republican Interpretation, 97 Yale L.J. 1685, 1686 (1988) (“Any theory of statutory interpretation is at base a theory about constitutional law. It must at the very least assume a set of legitimate institutional roles and legitimate institutional procedures that inform interpretation.”); Jane S. Schacter, Metademocracy: The Changing Structure of Legitimacy in Statutory Interpretation, 108 Harv. L. Rev. 593, 593–94 (1995) (“To carry out its [interpretive] task, the court must adopt—at least implicitly—a theory about its own role by defining the goal and methodology of the inter­pretive enterprise and by taking an institutional stance in relation to the legislature.”). Textualists, for example, believe that focusing on semantic cues—the way a reasonable person would read the text—enables legis­lators to use their words to draw effective lines of legislative compromise that specify both the means and the ends of legislation. 12 See, e.g., Frank H. Easterbrook, Statutes’ Domains, 50 U. Chi. L. Rev. 533, 547–48 (1983) [hereinafter Easterbrook, Statutes’ Domains] (arguing that strict adherence to text enables the legislature to select the means as well as ends of legislation); see also John F. Manning, What Divides Textualists from Purposivists?, 106 Colum. L. Rev. 70, 103–09 (2006) (culling such theme from writings of leading textualists and from Supreme Court decisions). That theory reflects a conception of legislative power that places a premium on facilitating legislative compro­mise. 13 See, e.g., Rodriguez v. United States, 480 U.S. 522, 525–26 (1987) (per curiam) (concluding that since “no legislation pursues its purposes at all costs,” courts disserve legislative supremacy by “assum[ing] that whatever furthers [a] statute’s primary objective must be the law”); Bd. of Governors of the Fed. Reserve Sys. v. Dimension Fin. Corp., 474 U.S. 361, 374 (1986) (“Invocation of the ‘plain purpose’ of legislation at the expense of the terms of the statute itself takes no account of the processes of compromise . . . .”). Legal Process purposivists, in contrast, contend that because Congress enacts laws to make policy, legislative supremacy is better served by emphasizing policy cues from a statute’s structure, context, and history. 14 See, e.g., Archibald Cox, Judge Learned Hand and the Interpretation of Statutes, 60 Harv. L. Rev. 370, 370 (1947) (noting that some “purpose lies behind all intelligible legislation”); Felix Frankfurter, Some Reflections on the Reading of Statutes, 47 Colum. L. Rev. 527, 538–39 (1947) (“Legislation has an aim; it seeks to obviate some mischief, to supply an inadequacy, to effect a change of policy, to formulate a plan of government.”). They doubt that focusing on every word or semicolon in a statute will somehow advance Congress’s constitutional function. 15 See Max Radin, A Short Way with Statutes, 56 Harv. L. Rev. 388, 406 (1942) [hereinafter Radin, A Short Way] (“The legislature has no constitutional warrant to demand reverence for the words in which it frames its directives.”). Neither approach claims to find what Congress subjectively decided in any given case. Each tries to construct legislative outcomes in a way that advances its own theory of legislative supremacy. 16 It is, by now, widely accepted that there are multiple ways to look at the idea of legislative supremacy. See William N. Eskridge, Jr., Spinning Legislative Supremacy, 78 Geo. L.J. 319, 321–22 (1989) [hereinafter Eskridge, Spinning Legislative Supremacy] (noting the availability of competing conceptions of legislative supremacy); Daniel A. Farber, Statutory Interpretation and Legislative Supremacy, 78 Geo. L.J. 281, 282 (1989) (same).

Whatever one might think of the particular positions taken in these debates, the emergence of intent skepticism shifted the questions about interpretation away from debates about how best to find an unfindable legislative “intent” and toward a frank examination of our legal system’s structural and institutional commitments. This shift in emphasis has permeated questions large and small. In important work defending the Legal Process tradition against the new textualism, our honoree Peter Strauss offers two particularly crisp examples of this institutional approach. First, Professor Strauss urges interpreters to consult legislative history, but not because it can reveal what Congress actually intended on the difficult issues that present themselves to courts or agencies. 17 For elaboration of the discussion of legislative history, see infra section I.B.1. Instead, he reasons that if statutory indeterminacy leaves interpretive discretion, a court shows greater respect for the legislature by considering how legislative actors analyzed the problem at hand, especially when the alternative is the exercise of unguided judicial discretion about how best to fill in the blanks. 18 See Peter L. Strauss, The Courts and the Congress: Should Judges Disdain Political History?, 98 Colum. L. Rev. 242, 262 (1998) [hereinafter Strauss, The Courts and the Congress] (describing the exercise of judicial discretion in cases of indeterminacy). Second, in the never-ending debate about judicial deference to agency interpretations of law, Professor Strauss again eschews any ref­erence to actual legislative intent. 19 For elaboration of this account of Professor Strauss’s views on Chevron, see infra section I.B.2. Since most regulatory statutes do not specify when or how much courts should defer to agencies, Professor Strauss believes that judges should exercise case-by-case common law discretion to determine when it makes sense to give an agency what he calls “Chevron space.” 20 See Peter L. Strauss, “Deference” Is Too Confusing—Let’s Call Them “Chevron Space” and “Skidmore Weight,” 112 Colum. L. Rev. 1143, 1145 (2012) (coining and defining the term “Chevron space”). Though I differ with Professor Strauss on important particulars in both instances, 21 See John F. Manning, Chevron and the Reasonable Legislator, 128 Harv. L. Rev. 457, 467 (2014) [hereinafter Manning, Chevron] (expressing a preference for the “clean lines” of the categorical approach to deference); John F. Manning, Textualism as a Nondelegation Doctrine, 97 Colum. L. Rev. 673, 720–22 (1997) [hereinafter Manning, Textualism as Nondelegation] (arguing that interpretive norms crediting legislative history enable legislators to shift policy creation outside the process of bicameralism and presentment). his approach shows how a starting point of intent skepticism casts focus on the constitutional and institutional stakes of competing interpretive frameworks.

Until recently, I thought it obvious that fighting it out on those terms was more desirable than taking on the seemingly fruitless task of asking whether one interpretive method or another better captures Congress’s true “intent.” Recent empirical scholarship, however, has raised questions about that assumption by surveying the subjective understandings of the congressional staff who participate in real, live legislative drafting. 22 E.g., Lisa Schultz Bressman & Abbe R. Gluck, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part II, 66 Stan. L. Rev. 725 (2014) [hereinafter Bressman & Gluck, Part II]; Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part I, 65 Stan. L. Rev. 901 (2013) [hereinafter Gluck & Bressman, Part I]; Victoria F. Nourse & Jane S. Schacter, The Politics of Legislative Drafting: A Congressional Case Study, 77 N.Y.U. L. Rev. 575 (2002). The most extensive of the studies—by Professors Abbe Gluck and Lisa Bressman—samples 137 legislative drafters about their attitudes toward such diverse issues as the use of dictionaries, the semantic canons, the legislative history, and the cluster of practices governing the availability and implementation of the Chevron doctrine. 23 See Gluck & Bressman, Part I, supra note 22, at 919–24 (describing the methodology of their study). Gluck and Bressman’s findings suggest, for example, that staff regard committee reports as at least as important as the statutory text in framing legislative understanding of a bill. 24 See infra section II.B (discussing Gluck and Bressman’s survey results relative to legislative history). Of further interest, staff deem multiple factors relevant to when they want Chevron to apply—a conclusion, the authors say, that supports the subsequent refinements of the “Mead doctrine” and its more nuanced approach to the question of when Congress wants reviewing courts to defer to agency interpretations of law. 25 See infra section II.C (analyzing Gluck and Bressman’s findings concerning the Chevron doctrine); see also United States v. Mead Corp., 533 U.S. 218, 227, 230 (2001) (holding that courts should defer when agencies announce their decisions through “relatively formal administrative procedure[s] tending to foster . . . fairness and deliberation” or where legislative scheme gives “some other indication of . . . congressional intent” to delegate).

This Essay argues that the new empiricism does not undermine the intent skepticism that has framed so much of the discussion about how to read statutes. Despite what we have learned from the new studies, large—and, in my view, decisive—indeterminacy remains as the baseline for an essentially normative statutory interpretation debate. Put to one side the question whether one should equate the staff’s understandings with those of Congress itself. 26 See infra note 151 (raising questions about the relevance of “drafter’s” intent). Even if one takes the studies on their own terms, they clarify but do not alter the inescapably normative character of the interpretation debate. To say, for example, that legislators are more apt to consult a committee report than a bill to learn what they are voting for cannot tell us what legal significance to attach to the additional fact that those legislators choose to vote instead for the dry, technical, opaque statutory text that, the survey suggests, may not reliably capture the original deal. 27 See infra text accompanying notes 213–217 (considering the constitutional implications of relying on legislative history rather than statutory text). No one can identify “congressional intent” or the “legis­lative deal” as a matter of fact, unfiltered by normative, institutional considerations that tell interpreters what should count as such. The survey also leaves intact the largely institutional debate about appropriate judicial deference in administrative law. The staff survey suggests that staffers find lots of factors—ambiguity, agency lawmaking procedures, the importance of the interpretive question, and the like—relevant to their intention to embrace judicial deference to agency interpretations of statutes. 28 See infra notes 233–235 and accompanying text (discussing survey results on judicial deference). Given the multiplicity of unweighted factors that staffers deemed relevant, it seems appropriate to retain a healthy skepticism about whether the Court can reconstruct, in any case, a genuine legislative “intent” about when to delegate binding interpretive authority to an agency, and how much. 29 See infra section II.C (analyzing those results).

This Essay will defend the proposition that, despite the impressiveness of the new empirical learning, theories of statutory interpretation should continue to build on the intent skepticism that has long defined so many of the leading approaches. Part I catalogues the pervasiveness of intent skepticism in diverse theories of statutory interpretation, from the new textualism to the Legal Process school. Drawing on Professor Strauss’s work, it then uses the examples of legislative history and judicial deference to agency interpretations of law to illustrate the institutional approach that flows from such skepticism. Turning to the new empiricism, Part II argues that the results of the survey leave intact the basic questions of legislative supremacy and judicial power on which the earlier debates pivoted.

I. The Intent Skepticism Baseline

This Part has two objectives: The first is to identify the pervasiveness of intent skepticism as the starting point for reasoning about statutory inter­pretation across a range of different approaches—textualism, realism, pragmatism, Dworkinian constructivism, and even Legal Process purposivism. Quoting extensively from the most influential of the intent skeptics, section I.A outlines the standard descriptive and conceptual objections to aggregating any form of genuine legislative intent. Written from a diverse array of perspectives, the quoted passages reflect recurring questions about whether a complex, path-dependent, multimember legis­lative process produces a discernible intent on the hard interpretive questions that make their way to litigation. More fundamentally, these passages point out the deep conceptual difficulties in even settling on what should count as Congresss intent.

Section I.B pursues a related goal—namely, to use debates between textualism and Legal Process purposivism to illustrate the kind of insti­tutional reasoning that has emerged against the backdrop of intent skepticism. When Legal Process proponents (such as Professor Strauss) 30 See infra notes 80–98 and accompanying text (discussing Professor Strauss’s Legal Process views). debate textualists (such as Judge Frank Easterbrook or Justice Antonin Scalia) 31 See infra notes 32–43 and accompanying text (elaborating upon Judge Easterbrook’s and Justice Scalia’s theories of textualism). about the utility of legislative history or the appropriateness of the Chevron doctrine, the disagreement does not typically ask which approach better identifies the subjective intent of Congress. Rather, the debate focuses squarely upon which of the contending approaches can better promote a constitutionally grounded conception of legislative supremacy, reflect the proper role of the federal courts, or provide a sensible solution to a problem that neither the statute at issue nor the Constitution decisively resolves.

A. Intent Skepticism Across the Board

Today, intent skepticism mostly conjures up images of the new textualism that emerged in the last two decades of the last century. Textualism developed in reaction to a strongly intentionalist judicial status quo. 32 See John F. Manning, Second-Generation Textualism, 98 Calif. L. Rev. 1287, 1291–92 (2010) (considering interpretive practices that provoked the development of modern textualism). For almost its entire history, the Supreme Court has said that the touchstone of statutory interpretation is legislative intent. 33 See, e.g., Philbrook v. Glodgett, 421 U.S. 707, 713 (1975) (“Our objective . . . is to ascertain the congressional intent and give effect to the legislative will.”); Schooner Paulina’s Cargo v. United States, 11 U.S. (7 Cranch) 52, 60 (1812) (“[I]t has been truly stated to be the duty of the court to effect the intention of the legislature.”). In the post–New Deal state, the Court began to treat legislative history—especially committee reports and sponsor statements—as authoritative evidence of such intent. 34 See, e.g., Thornburg v. Gingles, 478 U.S. 30, 43 n.7 (1986) (treating such sources as authoritative evidence of legislative intent); N. Haven Bd. of Educ. v. Bell, 456 U.S. 512, 526–27 (1982) (same). The idea was that such clues from the legislative record would help judges get inside Congress’s head and “imaginative[ly] reconstruct[]” the way legislators would have resolved the issue before the court. 35 Richard A. Posner, Statutory Interpretation—in the Classroom and in the Courtroom, 50 U. Chi. L. Rev. 800, 817 (1983) (suggesting that the “task for the judge called upon to interpret a statute is . . . one of imaginative reconstruction,” which involves thinking his or her way “into the minds of the enacting legislators and imagin[ing] how they would have wanted the statute applied to the case at bar”).

In their campaign to discredit what they regarded as the illegitimate reliance on unenacted legislative history, the leading judicial textualists—those who defined the new textualism—argued that efforts to reconstruct “genuine” legislative intent were just “a wild-goose chase.” 36 Scalia, Judicial Deference, supra note 2, at 517; see also, e.g., Easterbrook, Text, History, and Structure, supra note 2, at 68 (“Intent is elusive for a natural person, fictive for a collective body.”). This claim is closely associated with Judge Easterbrook’s arbitrage of Arrovian social choice theory, which emphasizes the possibility that a multi­member legislature might have intransitive preferences that would cycle endlessly if not cut off in some way (i.e., Congress might prefer A to B to C to A). 37 See, e.g., Easterbrook, Statutes’ Domains, supra note 12, at 547–48 (citing Kenneth J. Arrow, Social Choice and Individual Values (2d ed. 1963)) (discussing how agenda-setting and logrolling negate the possibility of reconstructing genuine legislative intent). On that account, the majority’s “intentions” concerning its preferred policy might vary based on arbitrary factors such as the order in which policy alternatives were taken up or the point at which debate was cut off and a vote taken. 38 See id. (discussing the impact of agenda control).

The new textualists also cite more intuitive reasons for their intent skepticism. 39 Textualist legal philosopher Jeremy Waldron has thus written that the concept of a legislative will or intention “founders on the fact that a legislature comprises many people not just one person, and people with quite radically varying states of mind.” Jeremy Waldron, Law and Disagreement 42–43 (1999). They claim that the legislative process is too complex and path-dependent to permit judges, after the fact, to reconstruct what the legislature would have done about a policy question that it did not resolve explicitly in the statute. Bills take their final shape from a com­plex dance that may include multiple committees, behind-the-scenes logrolling, the threat of a Senate filibuster or presidential veto, the need to fight for scarce floor time, the need for unanimous consent to expe­dite votes in the Senate, and countless other factors that may or may not appear on the face of the record left in the bill’s aftermath. 40 See, e.g., Easterbrook, Original Intent, supra note 3, at 64–65 (discussing such complexities in the legislative process); Kenneth A. Shepsle & Barry R. Weingast, The Institutional Foundations of Committee Power, 81 Am. Pol. Sci. Rev. 85, 89 (1987) (same). The numer­ous veto gates erected by the rules of the two Houses build in a bias against enactment, so each bill has a thousand ways to die. 41 See Shepsle & Weingast, supra note 40, at 89 (noting that “veto groups are pervasive”). While the resulting complexities may not be “total bars to judicial understanding,” textualists like Judge Easterbrook see them as “so integral to the legis­lative process that judicial predictions of how the legislature would have decided issues it did not in fact decide are bound to be little more than wild guesses.” 42 Easterbrook, Statutes’ Domains, supra note 12, at 548. Textualists thus start from the basic intuition that “with respect to 99.99 percent of the issues of construction reaching the courts, there is no legislative intent.” 43 Antonin Scalia, A Matter of Interpretation 32 (Amy Gutmann ed., 1997) [hereinafter Scalia, Interpretation] (emphasis omitted).

Textualists, however, are hardly alone in their cynicism about the interpreter’s ability to read Congress. Whatever one may think about the various forms of constructive intent that have competed for the Court’s allegiance over the years, doubts about the existence of genuine legislative intent—an actual subjective congressional decision about the litigated issue—are very widely shared. Legal realism, for example, relied on such skepticism to justify the conclusion that judges are governmental officials who necessarily make policy in the course of statutory decisionmaking. As Professor Max Radin thus wrote:

That the intention of the legislature is undiscoverable in any real sense is almost an immediate inference from a statement of the proposition. The chances that of several hundred [legislators] each will have exactly the same determinate situations in mind as possible reductions of a given determinable, are infinitesimally small. The chance is still smaller that a given determinate, the litigated issue, will not only be within the minds of all . . . but will be certain to be selected by all of them as the present limit to which the determinable should be narrowed. In an extreme case, it might be that we could learn all that was in the mind of the draftsman, or of a committee of a half dozen . . . who completely approved of every word. But when this draft is submitted to the legislature and at once accepted without a dissentient voice and without debate, what have we then learned of the intentions of the four or five hundred approvers? Even if the contents of the minds of the legislature were uniform, we have no means of knowing that content except by the external utterances or behavior of these hundreds of [legislators], and in almost every case the only external act is the extremely ambiguous one of acquiescence, which may be motivated in literally hundreds of ways, and which by itself indicates little or nothing of the pictures which the statutory descriptions imply. It is not impossible that this knowledge could be obtained. But how probable it is, even venturesome mathematicians will scarcely undertake to compute. 44 Radin, Statutory Interpretation, supra note 4, at 870–71.

In other words, Congress does not legislate with the litigated issue in mind. And even if it did, the judge would have no way of knowing what legislators intended to do about it. 45 See id.

This theme also finds expression in the work of the leading modern pragmatists, Professors William Eskridge and Philip Frickey. They argue that judges in statutory cases should engage in “practical reasoning”—a form of pragmatic, dynamic, multifactor analysis that does not depend upon unearthing some decision actually made by the legislature. 46 Eskridge & Frickey, Practical Reasoning, supra note 5, at 347–48 (arguing that “creation of statutory meaning is not a mechanical operation,” that “interpretation will often depend upon political and other assumptions held by judges,” and that answers given by statutory interpreters are “driven by multiple values”). Professors Eskridge and Frickey identify the three foundational methodologies as intentionalism, purposivism, and textualism. See id. at 324–25. To clear the underbrush for their preferred approach, they express views of legislative intent that are nearly indistinguishable from those of Judge Easterbrook or Justice Scalia 47 See supra notes 42–43 and accompanying text (discussing textualists’ intent skepticism). :

It is hard enough to work out a theory for ascertaining the “intent” of individuals in tort and criminal law. To talk about the “intent” of the legislature, as that term is normally used, multiplies these difficulties, because we must ascribe an intention not only to individuals, but to a sizeable group of individuals—indeed, to two different groups of people (the House and the Senate) whose views we only know from the historical record. The historical record almost never reveals why each legislator voted for (or against) a proposed law, and political science scholarship teaches that legislators vote for bills out of many unknowable motives, including logrolling, loyalty or deference to party and committee, desire not to alienate blocks of voters, and pure matters of conscience. 48 Eskridge & Frickey, Practical Reasoning, supra note 5, at 326 (citation omitted).

Professor Ronald Dworkin, in turn, offers a particularly important conceptual critique of legislative intent—one that shows the impossibility of identifying a collective body’s “intent” as a fact of the matter, free of normative judgments about what should count as the body’s decision. In developing his constructivist “best reading” approach to statutory interpretation, 49 See Dworkin, supra note 6, at 337–54 (outlining a coherence-based approach to statutory interpretation). Professor Dworkin argues that even if a court could make a table of all of the legislators’ intentions, judges would lack a neutral metric for determining how to transform those intentions into a coherent whole. 50 Id. at 315–16. Thus, he asks:

Whose mental states count in fixing the intention behind [a statute]? Every member of the Congress that enacted it, including those who voted against it? Are the thoughts of some—for example, those who spoke, or spoke most often, in the debates—more important than the thoughts of others? What about the executive officials and assistants who prepared the initial drafts? What about the president who signed the bill and made it law? Should his intentions not count more than any single senator’s? What about private citizens who wrote letters to their congressmen or promised or threatened to vote for or against them, or to make or withhold campaign contributions, depending upon how they voted? What about the various lobbies and action groups who played their now-normal role? 51 Id. at 318.

The need for aggregation complicates things further:

Should [an interpreter] use a “majority intention” approach, so that the institutional intention is that of whichever group, if any, would have been large enough to pass the statute if that group alone has voted for it? Or a “plurality” intention scheme, so that the opinion of the largest of the three groups would count as the opinion of the legislature even if the other two groups, taken together, were much larger? Or some “representative intention” approach, which supposes a mythical average or representative legislator whose opinion comes closest to those of most legislators, though identical to none of them? If the last, how is the mythical average legislator to be constructed? There are many other possible ways of combining individual intentions into a group or institutional intention. 52 Id. at 320–21 (citation omitted).

Finally, Professor Dworkin notes that the answer to any counterfactual question about what Congress would have intended necessarily depends on the level of generality at which the question is framed, adding to the sense of arbitrariness in any attempt by judges to construct Congress’s actual intent. 53 Id. at 324–27.

Perhaps most surprising is the intent skepticism of the Legal Process scholars who defined the post–New Deal purposivism against which textualism has pushed. 54 See Hart & Sacks, supra note 7, at 1124 (articulating and defending such purposivism); see also, e.g., T. Alexander Aleinikoff, Updating Statutory Interpretation, 87 Mich. L. Rev. 20, 26–28 (1988) (discussing the influence of the Legal Process approach); William N. Eskridge, Jr. & Philip P. Frickey, Legislation Scholarship and Pedagogy in the Post-Legal Process Era, 48 U. Pitt. L. Rev. 691, 698–99 (1987) (same); supra text accompanying notes 12–16 (drawing the contrast between textualism and purposivism). For convenience, this Essay will refer to all of the post–New Deal writings that paved the way for the Legal Process materials under the rubric of the “Legal Process” approach. Although the Legal Process materials urged inter­preters to presume that a law’s enactors were “reasonable persons pur­suing reasonable purposes reasonably,” 55 See Hart & Sacks, supra note 7, at 1378. proponents of that approach never saw it as a road map to identify what Congress actually decided about the question in issue. Rather, those who laid the foun­dation for the Legal Process school took as a given that “[f]ew . . . legis­lators think in terms of the specific controversies which courts must settle by giving a statute one or another meaning.” 56 Cox, supra note 14, at 371–72. It was neither the practice nor the contemplated function of Congress to legislate at that level of granularity. 57 See, e.g., Jerome Frank, Words and Music: Some Remarks on Statutory Interpretation, 47 Colum. L. Rev. 1259, 1270 (1947) (arguing that our system of government is premised on idea that Congress “cannot itself enforce the statutes” it enacts but “must delegate that task to other governmental agencies”); Harry Wilmer Jones, Extrinsic Aids in the Federal Courts, 25 Iowa L. Rev. 737, 742 (1940) (“It will be agreed, of course, that the particular fact-situations presented for decision in actual cases are not foreseen by any of the enacting legislators, except in the rare instances in which legislation is aimed at particular individuals.”). Hence, as Professors Hart and Sacks themselves explained, “the overwhelming probability” in any hard case is “that the legislature gave no particular thought to the matter [before the court] and had no intent concerning it.” 58 Hart & Sacks, supra note 7, at 1182. And even if that was wrong, Professors Hart and Sacks found the prospect of reconstructing such intentions daunting:

[O]n what basis does a court decide what [the enacting] legislature . . . would have done had it foreseen the problem? Does the court consider the political structure of the . . . legislature? Does the court weigh the strength of various pressure groups operating at the time? How else can the court form a judgment as to what the legislature would have done? 59 Id. at 1183.

Hence, it became an article of faith among Legal Process purposivists that the object of interpretation was decidedly not to unearth a genuine (subjective) congressional decision about the case at hand. 60 See, e.g., Frederick J. de Sloovère, Extrinsic Aids in the Interpretation of Statutes, 88 U. Pa. L. Rev. 527, 538 (1940) (“If by ‘legislative intent’ is meant the minutiae of meaning in application to specific cases, then rarely does such intention exist.”); Frankfurter, supra note 14, at 539 (“[T]he judge . . . ought not to be led off the trail by tests that have overtones of subjective design . . . . We do not delve into the mind of legislators or their draftsmen, or committee members.”); Jones, supra note 57, at 742 (“‘Legislative intention’ is, indeed, a fiction in the great majority of cases, if the concept be taken to signify the will of the members of the legislative body as to the decision of each particular case which may seem to involve the application of a statutory direction.”). When the Court referred to legislative intent, it could not mean “what was actually in the minds of those who framed and passed the statute.” 61 Radin, A Short Way, supra note 15, at 419.

I have quoted at length from textualists, realists, pragmatists, Dworkinians, and Legal Process purposivists to show how pervasively intent skepticism frames the debate across leading theories of interpretation. 62 Indeed, the most full-throated theoretical defense of intentionalism in many years—that of Oxford Professor Richard Ekins—defends the concept of legislative intent as something that one could attribute to an ideal legislature, but not necessarily to a complex real-world legislature like Congress. See Richard Ekins, The Nature of Legislative Intent 218–44 (2012). The book’s philosophical argument is too intricate to do it full justice here, but a brief summary can convey the essentials. Professor Ekins acknowledges that it is “not sound” to try to aggregate the intentions of individual legislators into that of the body as a whole. Id. at 46. At the same time, the legislature can express a “joint intention” by adopting a procedure to select some “plan of action that coordinates and structures the joint action of the members of the group.” Id. at 47, 58. In a “well-formed legislature,” Professor Ekins writes, that procedure will be structured to yield a “reasoned choice” in the resultant legislation. Id. at 77. And the legislature’s “particular intention” will be evident in “the plan that the bill set out for the community, which there is good reason to expect to be coherent and reasoned, as if chosen by a sole legislator.” Id. at 224; see also id. at 247 (arguing that the interpreter’s job is “to understand the reasoned choice that finds expression in th[e] intended meaning” that can “be inferred from publicly available evidence”).
Notice that Professor Ekins’s treatment of legislative intent is constructed around an ideal legislature. See Donald L. Drakeman, Charting a New Course in Statutory Interpretation: A Commentary on Richard Ekins’ The Nature of Legislative Intent, 24 Cornell J.L. & Pub. Pol’y 107, 111 (2014) (“[Professor Ekins] describes the ‘central case’ (or what some social scientists would call the ‘ideal-type’) of the ‘well-formed legislature’ as ‘an institution capable of reasoned choice.’”). Professor Ekins himself suggests that the presumption of reasoned and coherent decisionmaking may extend to Parliament but not Congress because the latter “has many veto-players.” Ekins, supra, at 176. This reality means that Congress “has difficulty legislating well because veto-players may frustrate the coherence of the legislative act, making it less likely that proposals will be reasoned and workable.” Id. at 239. Although Ekins urges us still to assume that U.S. legislation reflects “a complex, reasoned, coherent scheme” from which a shared intent can be inferred, id. at 240, he never explains why one should indulge that assumption for a more chaotic legislature, like ours, that is not his ideal type.
That shared assumption is attractive not only because it rings true to those who regularly deal with the intricacies of cases hard enough to make it into a federal reporter but also because the assumption itself forces those who argue about interpretation to defend their preferred theories on openly normative grounds. Even if loose language by those who write on the subject may at times suggest otherwise, 63 See infra note 204 and accompanying text (listing some examples). the contest here is not about which theory of interpretation will most accurately uncover the actual decision that Congress intended to be made about the issue in this or that case.

To be sure, in any system of government that is predicated (as ours is) on some form of legislative supremacy, the outcomes produced by applicable rules of interpretation must be attributable to Congress in some way, at some level. 64 Joseph Raz thus has argued that if interpretive outcomes are not attributable to lawmakers’ decisions at least at some level, it would not “matter who the members of the legislature are, whether they are democratically elected or not, whether they represent different regions in the country, or classes in the population, whether they are adults or children, sane or insane.” See Joseph Raz, Intention in Interpretation, in The Autonomy of Law: Essays on Legal Positivism 249, 258 (Robert P. George ed., 1996). As Raz further contends, however, that minimum condition for legislative supremacy can be satisfied even if one rejects the notion that interpreters can identify the genuine or subjective intentions of the legislature in any given case. See infra note 78 (explaining Professor Raz’s position on minimum intention needed for legislative supremacy). But that does not—and, as Professor Dworkin showed—cannot mean discovering Congress’s actual, subjective “intent” as a fact of the matter. 65 See supra text accompanying notes 49–53 (laying out Professor Dworkin’s critique of legislative intent). Rather, as Professor Victoria Nourse has written, interpretive theorists properly use the concept of legislative “intent” only as a metaphor 66 See Nourse, supra note 10, at 81 (“The notion of congressional intent is built upon a metaphor . . . .”). —a fiction that underscores the idea that we are trying to construct or approximate a statutory decision in a way that makes sense of Congress’s role in our constitutional democracy. Interpreters, in other words, must impute constructive intentions to Congress through tech­niques that are meant to advance some conception of what legislative supremacy and judicial power properly entail under the U.S. Constitution.

Because this description of the enterprise sounds terribly abstract, it is helpful to think about it in terms of two competing claims about important matters of interpretation. Hence, the section that follows offers examples from prominent debates about legislative history and deference to agency interpretations of law. What the materials that follow show is this: When spared the pretense of trying to figure out—to make a factual judgment about—what Congress actually decided in any given case, interpreters have room to reorient the debate toward normative questions about the proper functioning of our system of government.

B. Legal Process Versus Textualism

The Hart and Sacks materials introduced the useful idea of “institutional settlement”—the notion that society avoids chaos by agreeing to “regularized and peaceable methods of decision” that the legal system accepts as binding. 67 Hart & Sacks, supra note 7, at 4–5. In almost any constitutional system (and certainly in one as old and complex as ours), the particulars of those methods and their application will be disputed. Because the rules of interpretation allocate policymaking authority among lawmakers and law appliers, debates about interpretive method necessarily and properly reflect assumptions about the appropriate institutional roles of the players involved. 68 See, e.g., Mashaw, supra note 11, at 1686 (discussing the structural constitutional foundations of interpretation theory); Schacter, supra note 11, at 593–94 (arguing that rules of interpretation require developing an “institutional stance” toward the legislature). If Congress has no actual intent on any hard question, then debates that focus on how best to identify such intent may submerge the background institutional stakes. A shift toward intent skepticism, in turn, may help bring those stakes to the surface.

The analytical divisions and, equally important, the common ground between prominent Legal Process adherents (such as Professor Strauss) and leading textualists (such as Judge Easterbrook and Justice Scalia) exemplify how much institutional settlement frames the debate in a post-intentionalist environment. The two sides have more in common than many may realize. Neither sees the debate as one about whether or how interpreters can identify “legislative intent.” Rather, the differences between textualists and Legal Process purposivists turn almost entirely on what a judge (or an agency) ought to do when statutory meaning runs out, as it so often does. The discussion below looks at two prominent examples: (1) the role of legislative history, if any, in statutory interpretation and (2) the proper level of deference, if any, owed by reviewing courts to agency interpretations of administrative statutes.

1. Legislative History. — As noted, for several related reasons, the new textualists argue that judges should not treat legislative history as authoritative evidence of legislative intent. In their early writings, text­ualists stressed that Congress has no collective legislative intent to unearth 69 See supra text accompanying notes 36–43 (discussing the textualist position). and that, even if it did, judges have no way of knowing which (if any) legislators actually relied upon any particular piece of legislative history, even premium items such as committee reports. 70 See, e.g., Wis. Pub. Intervenor v. Mortier, 501 U.S. 597, 620 (1991) (Scalia, J., concurring in the judgment) (questioning whether committee reports can properly claim to represent the common understanding of the enacting majority); Blanchard v. Bergeron, 489 U.S. 87, 97–100 (1989) (Scalia, J., concurring in part and concurring in the judgment) (same).

If legislative history is not a window into actual legislative intent, textualists believe that there are sound institutional, even constitutional, reasons not to use it. Given the volume and diversity of available legislative history, textualists fear that its use gives judges too much discretion to push their own preferred outcomes. 71 See, e.g., Antonin Scalia, Speech on Use of Legislative History 13 (Fall 1985–Spring 1986) (on file with the Columbia Law Review) (arguing that the use of legislative history “substantially increases, rather than reduces, the scope of judicial discretion”); see also Scalia, Interpretation, supra note 43, at 36 (invoking Judge Harold Leventhal’s quip that legislative history permits judges “to look over the heads of the crowd and pick out [their] friends”). In addition, if judges treat legislative history as authoritative evidence of statutory meaning, then legislators can make an end run around the constitutionally pre­scribed processes of bicameralism and presentment. 72 See, e.g., Bank One Chi., N.A. v. Midwest Bank & Tr. Co., 516 U.S. 264, 280 (1996) (Scalia, J., concurring in part and concurring in the judgment) (arguing that judicial reliance on legislative history permits legislators to delegate lawmaking authority to committees and other subunits of Congress); Thompson v. Thompson, 484 U.S. 174, 191–92 (1988) (Scalia, J., concurring in the judgment) (raising bicameralism and presentment concerns). Instead of bargain­ing to include particular measures in the statutory text, legislators have the ability to salt the legislative record with their preferred outcomes in the expectation that judges and administrators will treat those signals as reliable indicia of Congress’s decision. 73 See, e.g., Int’l Bhd. of Elec. Workers, Local Union No. 474 v. NLRB, 814 F.2d 697, 717 (D.C. Cir. 1987) (Buckley, J., concurring) (developing the point about end runs around the legislative process); Note, Why Learned Hand Would Never Consult Legislative History Today, 105 Harv. L. Rev. 1005, 1015–19 (1992) (making similar claim).

Instead of relying on legislative history, textualists want judges to listen for “the ring the [statutory] words would have had to a skilled user of words at the time, thinking about the same problem.” 74 Easterbrook, Original Intent, supra note 3, at 61. With some notable (and, I think, questionable) add-ons such as clear statement rules, 75 See generally John F. Manning, Clear Statement Rules and the Constitution, 110 Colum. L. Rev. 399 (2010) (discussing and criticizing the Court’s reliance upon clear statement rules). the new textualism thus gives close attention to the shared social and linguistic conventions that enable the relevant linguistic community to convey meaning. Textualists tend to move briskly to dictionary definitions, rules of grammar and syntax, and (since statutes are, after all, legal instruments) canons of interpretation or terms of art peculiar to the legal community. 76 See John F. Manning, Textualism and Legislative Intent, 91 Va. L. Rev. 419, 436–37 (2005) (discussing textualists’ use of specialized conventions); Thomas W. Merrill, Textualism and the Future of the Chevron Doctrine, 72 Wash. U. L.Q. 351, 372–73 (1994) (analyzing textualists’ focus on dictionaries, grammar, syntax, and semantic canons). Putting aside some careless statements in textualist writings, 77 See, e.g., Arizona v. Inter Tribal Council of Ariz., Inc., 133 S. Ct. 2247, 2257 (2013) (Scalia, J.) (articulating the “reasonable assumption . . . that the statutory text accurately communicates the scope of Congress’s pre-emptive intent”). textualists do not believe that legislators subjectively have in mind the content of dictionaries, the esoteric rules of syntax, or the specialized content of terms of art. 78 As Professor Raz has said, even if one denies the existence of subjective legislative intent, the demands of legislative supremacy are met as long as interpreters ascribe to legislators a constructive intention “to say what one would be normally understood as saying, given the circumstances in which one said it.” See Raz, supra note 64, at 268. Rather, textualists believe that if interpreters emphasize semantic cues, then legislators can reliably make use of those cues to record their lines of inclusion and exclusion, whether legislators actually paid close attention to textual meaning or not in any particular case. 79 See Manning, Foreword, supra note 9, at 25–26 (developing this defense of textualism).

The Legal Process position as articulated by Professor Strauss and others starts from similar structural premises but draws quite different conclusions. For Professor Strauss, interpreters cannot realistically expect to resolve pervasive indeterminacies by seeking genuine legislative intent on the litigated issue. “Congress,” he writes, “is a bureaucracy of tens of thousands, and too frequently acts on legislative behemoths no member can have read; the claim that it ‘knows’ anything is absurd.” 80 Peter L. Strauss, On Resegregating the Worlds of Statute and Common Law, 1994 Sup. Ct. Rev. 429, 438. What is more, “the difficulties of collective authorship . . . make[] problematic the idea that legislation can have an ‘intent’ if it emerges from the highly variable participation of 535 legislators divided among two Houses of Congress and their many committees and subcommittees, and assisted by innumer­able staff and lobbyists.” 81 Peter L. Strauss, The Common Law and Statutes, 70 U. Colo. L. Rev. 225, 236 (1999) [hereinafter Strauss, The Common Law and Statutes]. And even “[i]f we imagine a legislature with the best of will adopting statutory instructions, we know at once that as a human institution it will be imperfectly foresightful, unaware of all possible meanings the words it chooses could be given, and solipsistic.” 82 Strauss, The Courts and the Congress, supra note 18, at 244–45. Finally, he writes, if courts were to regard “themselves as bound in any legal sense by words spoken or written during Congress’s consideration of legislation,” they would run afoul of the constitutional principle that Congress “cannot limit the possible [interpretive] outcomes by any means other than the words it enacts.” 83 Id. at 250.

At the same time, to say that judges are not bound to enforce the legislative history is not to say that they are bound to ignore it. 84 See id. (“The problem arises when one moves from the proposition that courts are not bound by legislative history . . . to the conclusion that is usually inappropriate for them even to look to those materials for help in understanding and resolving an interpretive problem that may be before them.”). On the contrary, if statutes typically set a range of permissible meanings among which the interpreter must choose, then statutory interpretation will obviously entail policymaking discretion. To Professor Strauss, this reality poses the inevitable question of who will exercise that discretion and how. A judge who excludes all consideration of legislative history exercises that discre­tion on his or her own account. 85 Id. at 252 (emphasizing that “it is the judge’s discretion that will be exercised”). Even if one accepts that a committee report or sponsor’s statement cannot speak for Congress as a whole, “[w]hy would we prefer a judge operating within . . . a range [of policy­making discretion] to be indifferent or oblivious to information about the political history of th[e] legislation?” 86 Id.

Even if not binding, legislative history provides context that can help the judge to “see a complex statute as an integrated whole” or to under­stand technical nuances that may be unobvious to the layperson but may be “quite firmly settled in the public and private communities that deal with the statute on a daily and intimate basis.” 87 Id. at 253. At the very least, “[h]earings, debates, and reports will reveal what issues were brought to congressional attention, and what were not.” 88 Id. at 257. Being mindful of those clues, Professor Strauss argues, promotes the classic common law judicial function of reading statutes to suppress the mischief and promote the remedy that the legislation contemplated. 89 See id. at 256–58 (relying on Anglo American judicial tradition reflected in Heydon’s Case, 76 Eng. Rep. 637, 638 (K.B. 1584)); see also id. at 258 (“Considering the course of legislative development, to discover what kinds of problems were mentioned and what kinds were not, is the most natural means of accomplishing [an] understanding [of the mischief].”). In contrast, to exclude all consideration of political history, he worries, pits judges against Congress in a way that our constitutional system does not contemplate. 90 Though Professor Strauss does not fully spell this point out, judges’ heeding the signals sent by pivotal actors such as gatekeeping committees presumably minimizes the risk that Congress will find it necessary to overturn the interpretation thus rendered. See generally William N. Eskridge, Jr. & John Ferejohn, The Article I, Section 7 Game, 80 Geo. L.J. 523 (1992) (modeling dynamics of interpretation and triggers for legislative override). By increasing the possible need for a legislative override, a court that ignores the political history of a bill “impose[s] make-work on Congress” and “invite[s] repetitive struggles that, even if not intended, are wasteful of the limited resources of time and effort Congress has available to it for its legislative agenda.” 91 Strauss, The Courts and the Congress, supra note 18, at 255. This approach, he says, shows “disrespect for a coordinate branch of government that is . . . hard to justify under . . . well-established structural constitutional principles.” 92 Id. at 258.

All of this is classic institutional reasoning firmly within the Legal Process tradition. 93 Professor Strauss’s Columbia colleague Professor Harry Wilmer Jones, one of the greats in the Legal Process tradition, made a similar argument: “[T]he choice before the judges is that they must either derive the meaning of a statute solely from its language and from conjecture as to its purposes, or must accept as the ‘legislative intention’ the understanding of the committee experts and other interested legislators really responsible for its formulation.” Jones, supra note 57, at 743. That tradition, as noted, urges judges to read the legislative record to help them construct “reasonable purposes” to ascribe to legislators, whom the judges presume to be “reasonable.” 94 Strauss, The Courts and the Congress, supra note 18, at 265 (quoting Hart & Sacks, supra note 7, at 1125 ) (describing the “reasonable” legislator presumption); see also supra text accompanying notes 8, 55 (same). But as the Court’s most committed Legal Process Justice—Stephen G. Breyer—has made clear, the reasonable, purposive legislator is a “legal fiction that applies . . . even when Congress did not in fact consider a particular problem.” 95 Stephen G. Breyer, Active Liberty 88 (2005) [hereinafter Breyer, Active Liberty]. For the Legal Process purposivist, that fiction finds its justification not because it better captures some actual congressional decision 96 See, e.g., Strauss, The Common Law and Statutes, supra note 81, at 241–42 (stressing that “no one” could properly view the reasonable legislator assumption “as a description of American legislating in any but the most extraordinary setting”); Strauss, The Courts and the Congress, supra note 18, at 265 (“[O]ne could hardly suppose [Hart and Sacks] thought they were describing an actual state of legislative affairs.”); see also id. at 243 n.3 (forswearing the idea “that it is useful to employ bits and pieces of legislative reports or debates to resolve particular issues of meaning”). but rather because it provides “a workable method of implementing the Constitution’s democratic objective” of translating the “public will . . . into sound policy.” 97 Breyer, Active Liberty, supra note 95, at 101. Hence, Legal Process purposivism reflects “a normative statement prescribing proper attitudes for judges in their dealing with the work of legislatures, rather than a positive one describing what legislatures are.” 98 Strauss, The Common Law and Statutes, supra note 81, at 242.

2. The Chevron Doctrine(s). — A presumption of statutory indeterminacy also frames the ongoing debate between Legal Process purposivists and textualists over the Chevron doctrine. Chevron, of course holds that, at least in certain contexts, a reviewing court must accept an agency’s “reasonable” interpretation of a vague or ambiguous statute the agency is charged with administering. 99 Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843–44 (1984). Questions about both the justi­fication and scope of the Chevron doctrine have provoked considerable scholarly disagreement. 100 See John F. Manning & Matthew C. Stephenson, Legislation and Regulation 768–72 (2d ed. 2013) (outlining competing rationales for the Chevron approach). One part of the equation is easy: While the Administrative Procedure Act (APA) states that “the reviewing court shall decide all relevant questions of law, [and] interpret . . . statutory provi­sions,” 101 5 U.S.C. § 706 (2012).  most think that “Chevron deference” is nonetheless justified where the organic act delegates to the agency the discretion to resolve indeterminacies in the organic act’s operative terms. 102 See, e.g., Ronald M. Levin, Identifying Questions of Law in Administrative Law, 74 Geo. L.J. 1, 21 (1985) (“Sometimes . . . the legislature intended to make no decision on a particular substantive issue and to leave that issue to administrative creativity. In such a situation, a court’s refusal to use independent judgment actually fulfills Congress’ intent.”); see also Henry P. Monaghan, Marbury and the Administrative State, 83 Colum. L. Rev. 1, 6, 27 (1983) (contending, the year before Chevron was decided, that “[a] statement that judicial deference is mandated to an administrative ‘interpretation’ of a statute is more appropriately understood as a judicial conclusion that some substantive law-making authority has been conferred upon the agency” and that the court’s role is to “specify the boundaries of agency authority”). Where an organic act delegates interpretive discretion to an agency, the court “interprets” the organic act, first, by determining the existence of a delegation and, second, by deciding whether the agency has stayed within the bounds set by that delegation (i.e., has “reasonably” interpreted statutory terms about whose meaning reasonable people can differ). 103 See Einer Elhauge, Statutory Default Rules: How to Interpret Unclear Legislation 86 (2008) (explaining how to reconcile deference with the judge’s duty to interpret statutes). Professor Strauss offers a neat formulation of the point, observing: “[T]o the extent Congress empowers an agency to act using language of uncertain meaning, it may also empower the agency reasonably to determine that meaning within the resulting ambit of uncertainty, subject not to judicial redetermination but to judicial oversight of its judgment for reasonableness.” Peter L. Strauss, In Search of Skidmore, 83 Fordham L. Rev. 789, 791 (2014).

Even though the connection between deference and delegation is well accepted, confusion and disagreement have long surrounded the question of when to find that an organic act delegates the requisite interpretive lawmaking power to the agency that administers it. 104 Professors Thomas Merrill and Kathryn Tongue Watts have suggested that there was once an accepted interpretive convention for identifying a legislative intention to delegate lawmaking power to an agency. See Thomas W. Merrill & Kathryn Tongue Watts, Agency Rules with the Force of Law: The Original Convention, 116 Harv. L. Rev. 467, 472 (2002). If they are right, however, that convention seems to have faded from view long ago. The controversy perhaps reflects, in part, the fact that the Court has swung wildly among different formulae for identifying when deference is due. For two generations after the APA’s adoption—during the heyday of the Legal Process era—the Court applied the so-called HearstPackard approach. 105 Manning & Stephenson, supra note 100, at 753–55. The approach is named for the two leading cases of the period, Packard Motor Car Co. v. NLRB, 330 U.S. 485 (1947), and NLRB v. Hearst Publications, Inc., 322 U.S. 111 (1944). Under that pre–Chevron framework, the Court determined whether to “defer” to an agency interpretation of law based on multiple unranked factors: Did the question at issue involve a pure question of law or the application of law to fact? 106 Compare, e.g., Packard, 330 U.S. at 493 (holding that pure questions of law merit independent judicial judgment), with Hearst, 322 U.S. at 130–31 (holding that the application of broad standards to particular facts counsels in favor of deference). Did the interpretive question require agency expertise? 107 See, e.g., Aluminum Co. of Am. v. Cent. Lincoln Peoples’ Util. Dist., 467 U.S. 380, 390 (1984) (recognizing that “principles of deference have particular force” when the “subject under regulation is technical and complex” and the agency has “longstanding expertise in the area”); E.I. du Pont de Nemours & Co. v. Train, 430 U.S. 112, 134–35 & n.25 (1977) (concluding that an agency’s interpretation is entitled to “some deference” given the complexity of the statute or subject matter at issue). Did the agency help draft the legislation? 108 See, e.g., Zuber v. Allen, 396 U.S. 168, 192 (1969) (explaining that a court will “accord great weight to a departmental construction of [the agency’s] own enabling legislation,” particularly when “administrators participated in drafting”). Was the agency interpretation one of long standing? 109 See, e.g., SEC v. Sloan, 436 U.S. 103, 126 (1978) (Brennan, J., concurring) (crediting “administrative practice,” in part, because “assumptions which everyone shares . . . often go unspoken because their very obviousness negates the need to set them out”); Norwegian Nitrogen Prods. Co. v. United States, 288 U.S. 294, 315 (1933) (stating that “administrative practice, consistent and generally unchallenged, will not be overturned except for very cogent reasons if . . . the command is indefinite and doubtful”). Professor Colin Diver identified no fewer than ten factors that the Court found relevant in determining whether it had to accept an agency interpretation of law. 110 Colin S. Diver, Statutory Interpretation in the Administrative State, 133 U. Pa. L. Rev. 549, 562 n.95 (1985).

In roughly the past three decades, the Court has changed course at least twice. In 1984, as noted, it adopted the Chevron approach, which seemed (to some, at least) to embrace a categorical rule requiring review­ing courts to defer to reasonable agency interpretations of ambiguity in an organic act. 111 See Gary Lawson & Stephen Kam, Making Law Out of Nothing at All: The Origins of the Chevron Doctrine, 65 Admin. L. Rev. 1, 59–60 (2013) (explaining the D.C. Circuit’s role in propagating that view of the Chevron case); see also infra note 223 (describing the Court’s formulation of the categorical position on Chevron). In 2001, United States v. Mead Corp. held that Chevron deference is available only when the agency resolves statutory ambiguity through “relatively formal administrative procedure[s] tending to foster the fairness and deliberation” that one would expect to accompany a proper legislative delegation of interpretive lawmaking power. 112 533 U.S. 218, 230 (2001). An organic act, the Court said, might signal such a delegation “in a variety of ways, as [through its grant of] power to engage in adjudication or notice-and-comment rulemaking, or by some other indication of a comparable congressional intent.” 113 Id. at 227. In the absence of such processes, the Court could still properly invoke Chevron deference in “any other circumstances reasonably suggesting that Congress . . . thought of [the particular kind of agency action] as deserving the deference.” 114 Id. at 231. Even if Chevron deference is not available, Mead indicates that a litigant may invoke Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944), which holds that a reviewing court should give an agency interpretation the weight that it deserves in light of “the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade.” Mead, 533 U.S. at 221. Whether or not Skidmore means more than that a reviewing court should allow itself to be persuaded by a persuasive agency, Skidmore deference is surely not as robust as Chevron deference is. Though the Court in Mead did not spell out those circumstances, a subsequent opinion sug­gests that “the interstitial nature of the legal question, the related expertise of the Agency, the importance of the question to adminis­tration of the statute, the complexity of that administration, and the careful consideration the Agency has given the question over a long period of time” may justify Chevron deference even if the agency has invoked procedures too informal to qualify for Mead’s safe harbor. 115 Barnhart v. Walton, 535 U.S. 212, 222 (2002).

The Court’s approach to what some have called “step zero” 116 See generally Cass R. Sunstein, Chevron Step Zero, 92 Va. L. Rev. 187 (2006) (introducing that concept). —the judicial determination of whether a given organic act confers interpretive lawmaking power upon the agency—might surely be understood as a question of when Congress intends to delegate such power. 117 For a particularly clear expression of that position, see City of Arlington v. FCC, 133 S. Ct. 1863, 1882 (2013) (Roberts, C.J., dissenting) (emphasizing that the availability of Chevron deference “is defined by congressional intent”). Yet whatever the background legislative understanding about deference and delegation may once have been, it would be facetious for judges today to treat the availability of deference as a question of genuine legislative intent. 118 See, e.g., Lisa Schultz Bressman, Reclaiming the Legal Fiction of Congressional Delegation, 97 Va. L. Rev. 2009, 2024–25 (2011) (characterizing intent to delegate as a legal fiction); Sunstein, supra note 116, at 248 (same). Organic acts typically say nothing explicit about the question. In no opinion has the Court premised its application of Chevron on the existence of legislative history suggesting that Congress preferred or disfavored a deferential approach under a given organic act. And even if one takes at face value the Court’s occasional suggestion that Congress legislates with awareness of the Court’s settled rules of interpretation, 119 See Cannon v. Univ. of Chi., 441 U.S. 677, 699 (1979) (deeming it “not only appropriate but also realistic to presume that Congress was thoroughly familiar with . . . unusually important precedents” establishing rules of construction). the Court’s rules here—hardly pellucid in the first place—have changed time and again, rendering settled interpretive practice a questionable basis for inferring congressional intent. 120 See Manning, Chevron, supra note 21, at 459–63 (describing various regime changes in the Chevron family of doctrines). If the organic acts do not speak clearly to whether the court or the agency has final authority to flesh out the open-ended terms of an organic act in specified circum­stances, then the judiciary necessarily exercises delegated discretion to determine whether the Chevron framework applies or not.

Viewed in that light, the major fights over deference have almost always turned on institutional arguments about when a court should impute a delegation of interstitial interpretive lawmaking authority to an agency administering an organic act. If Justice Scalia represents textualism on this point, he has made clear that deference does not depend on decoding actual legislative instructions about delegation. 121 Scalia, Judicial Deference, supra note 2, at 517 (disclaiming any intent-based justification for Chevron). Describ­ing “the quest for the ‘genuine’ legislative intent [as] . . . a wild-goose chase,” Justice Scalia opines that in “the vast majority of cases” Congress “didn’t think about [deference] at all.” 122 Id. Instead, he stresses, “any rule adopted in this field represents merely a fictional, presumed intent.” 123 Id. From that starting point, he pushes a version of textualism that prefers clear and predictable rules over more judgmental standards. 124 See generally Caleb Nelson, What Is Textualism?, 91 Va. L. Rev. 347 (2005) (developing this strand of textualism). Under this account of textualism, “[w]hat is of paramount importance is that Congress be able to legislate against a background of clear interpretive rules, so that it may know the effect of the language it adopts.” 125 Finley v. United States, 490 U.S. 545, 556 (1989). Justice Scalia thus prefers Chevron’s categorical approach over a multifactor approach not because it better captures actual legislative intent, but rather because it establishes a clear background presumption against which Congress can legislate. 126 See Scalia, Judicial Deference, supra note 2, at 517 (articulating that view of Chevron). 

Professor Strauss, who may be the law professoriate’s leading Chevronologist, writes in the same spirit of intent skepticism, 127 See Peter L. Strauss, One Hundred Fifty Cases per Year: Some Implications of the Supreme Court’s Limited Resources for Judicial Review of Agency Action, 87 Colum. L. Rev. 1093, 1120 (1987) (noting that the Chevron Court fashioned its presumption of agency delegation “unconnected to congressional wishes reflected in any given law”). disagreeing with textualists mainly about why and when the judiciary should posit a fictive legislative intent to delegate interpretive lawmaking power to an agency. In the early days of Chevron, he joined Justice Scalia in defending the broad, categorical reading of Chevron. Professor Strauss believed that, in the absence of a clear legislative signal about delegation, Chevron provided a sensible means of managing an overloaded Supreme Court docket and promoting uniformity in federal law. 128 Id. at 1121 (“Rather than see Chevron just as a rule about agency discretion, in other words, it can be seen as a device for managing the courts of appeals that can reduce (although not eliminate) the Supreme Court’s need to police their decisions for accuracy.”). If administrative statutes typically allow a range of possible meanings, then ascribing primary interpretive responsibility to reviewing courts “virtually assure[s]” diverse interpre­tations of federal law across circuits, given the Supreme Court’s “practical inability” to take enough cases to correct inconsistent readings by the lower federal courts. 129 Id. In contrast, requiring judges to defer to an agency’s reasonable interpretations of ambiguity establishes a focal point around which to promote uniformity that the Court itself might be unable to impose through its limited docket. 130 See id. (“By removing the responsibility for precision from the courts of appeals, the Chevron rule subdues this diversity, and thus enhances the probability of uniform national administration of the laws.”). On that view, Chevron can be defended not “just as a rule about agency discretion,” but “as a device for managing the courts of appeals that can reduce (although not eliminate) the Supreme Court’s need to police their decisions for accuracy.” 131 Id.

More recently, Professor Strauss moved away from Chevron’s cate­gorical approach and toward Mead’s more fact-specific approach 132 See Peter L. Strauss, Courts or Tribunals? Federal Courts and the Common Law, 53 Ala. L. Rev. 891, 893 (2002) [hereinafter Strauss, Courts or Tribunals?] (explaining the institutional basis for preferring a contextual approach). —one that frees the judiciary to “tailor deference to variety” in a complex admin­istrative state. 133 United States v. Mead Corp., 533 U.S. 218, 236 (2001). His evolution does not reflect a changed view about the availability (or not) of genuine congressional intent. Rather, Professor Strauss ties his endorsement of Mead to a common law vision of the federal courts. 134 See Strauss, Courts or Tribunals?, supra note 132, at 893 (explaining why Mead’s case-by-case approach is consistent with traditional common law). Simply put, if Congress does not specify what kind of deference it expects in the typical administrative statute, then Professor Strauss thinks it desirable for judges to use their discretion to foster “case-by-case development of an imperfect statutory framework to resolve a difficult issue of federal administrative law—that is, the classic common law approach.” 135 Id. Once again, this is pure Legal Process reasoning, rooted in an impulse to craft a sensible and institutionally sensitive response to statutory indeterminacy. 136 Professor Strauss’s approach again resonates with Justice Breyer’s similar endorsement of the “institutional virtue[]” of doctrinal flexibility “to allocate the law-interpreting function between court and agency in a way likely to work best within any particular statutory scheme.” Stephen Breyer, Judicial Review of Questions of Law and Policy, 38 Admin. L. Rev. 363, 371 (1986). From the Legal Process perspective, in other words, Chevron’s one-size-fits-all approach is problematic because it is “seriously overbroad, counterproductive and sometimes senseless.” Id. at 373. And because Congress typically expresses no clear intention about the appropriate form of judicial deference (if any), nothing precludes the judiciary from crafting a presumed intent that imputes to Congress an institutionally sensible approach to the administration of federal regulatory law and policy. Strauss, Courts or Tribunals?, supra note 132, at 893.

II. The Old Skepticism Meets the New Empiricism

In a debate framed largely around the premise of intent skepticism, scholars have periodically asked whether it is possible to zero in better on Congress’s actual decisions by gathering and better analyzing evidence of the way Congress works. In the early days of the fight over legislative history, Chancellor Nicholas Zeppos, for example, urged judges to look at the legislative record to determine, if possible, which legislators might have had access to a committee report and when. 137 See Nicholas S. Zeppos, Legislative History and the Interpretation of Statutes: Toward a Fact-Finding Model of Statutory Interpretation, 76 Va. L. Rev. 1295, 1359 (1990). Dean Daniel Rodriguez and Professor Barry Weingast have counseled interpreters to identify and rely on legislative history generated by “[p]ivotal legislators,” who “have the strongest incentives to communicate reliably the act’s meaning” rather than engage in “cheap talk.” 138 Daniel B. Rodriguez & Barry R. Weingast, The Positive Political Theory of Legislative History: New Perspectives on the 1964 Civil Rights Act and Its Interpretation, 151 U. Pa. L. Rev. 1417, 1448 (2003). Judge Robert Katzmann has called upon judges to credit the key pieces of legislative history that legislators them­selves regard “as essential in understanding [statutory] meaning.” 139 Robert A. Katzmann, Judging Statutes 52 (2014). And Professor Nourse has intriguingly written that judges can learn a lot by filtering legislative signals through explicit congressional rules of practice governing matters such as the import of conference reports. 140 See Nourse, supra note 10, at 95 (arguing interpreters should resolve discrepancies between pre- and post-conference committee-bill language in light of “congressional rule[] [that] the conferees ha[ve] no power to change the text in any significant way”).

Following in the footsteps of an important but smaller-scale study of congressional staff published roughly a decade earlier, 141 Nourse & Schacter, supra note 22. Professors Gluck and Bressman have undertaken a lengthy survey of 137 congres­sional drafting staff. The authors pose 171 questions that “covered topics ranging from the role of canons such as the presumption against preemption, expressio unius, and Chevron deference, to legislative history, the legislative process, and the way that staffers perceive the responsibilities of courts and agen­cies in statutory interpretation.” 142 Gluck & Bressman, Part I, supra note 22, at 906.

Professors Gluck and Bressman carefully frame the purpose of the study and the use to which they hope the statutory interpretation com­munity will put the results: To the extent possible, they want their findings to shed light on whether “legislative drafting practice” confirms or disproves hypotheses that inform “modern interpretive practice.” 143 Id. at 907. They pursue this labor-intensive objective because, they say, “drafting ‘reality’ has been a central component of virtually all of the theoretical and doctrinal debates.” 144 Id. at 909. Professors Gluck and Bressman do not identify themselves with classic intentionalism. 145 Indeed, at one point, they describe “the notion of a single ‘congressional intent’” as a “fiction” that “is most certainly false” as an empirical matter. Id. at 915. But they plainly invoke intentionalist reasoning—the subjective expectations of legislative drafters—to criticize important interpretative techniques and to articulate grounds for embracing others. For instance, Professors Gluck and Bressman suggest that their findings about legislative practice undermine the basis for textualists to claim support in the “faithful-agent model” that underlies “most interpretive approaches.” 146 Id. at 907. The authors acknowledge that their study does not necessarily negate the validity of any particular theory of interpretation, but rather casts upon its proponents the burden of articulating a justification other than the faithful-agent theory for any method that is not borne out by the legislative facts on the ground. See id.

Because the Gluck and Bressman study unearthed “a treasure trove of information about key influences on the actual drafting process,” 147 Id. at 908. one cannot help but ask whether this knowledge undermines the intent skepticism that has framed so much of the interpretation debate. The study is too extensive to consider all of its findings. But three important examples will give a flavor of the challenges that the study presents. First, Professors Gluck and Bressman argue that many of the tools textualists use to identify statutory meaning—including consistent-usage canons and dictionaries—lack support in the realities of legislative drafting practice. 148 See infra text accompanying notes 155–165 (discussing survey results on canons and dictionaries). Second, relying on the respondents’ descriptions of legislative practice and attitudes, Professors Gluck and Bressman suggest that the empirics of the legislative process undermine the textualist position that legislative history is a poor proxy for “legislative intent.” 149 See infra text accompanying notes 193–209 (examining findings on legislative history). Third, Professors Gluck and Bressman argue that much about the Court’s current approach to judicial “deference” to agency interpretations of law matches up with actual legislative preferences, at least as reflected in the assumptions of the drafters. 150 See infra text accompanying notes 230–235 (sketching Gluck and Bressman’s conclusions about deference doctrines). Professors Gluck and Bressman argue that staffers do not subscribe to Chevron’s one-dimensional assumption that Congress uses ambiguity in an administrative statute to signal a delegation of interpretive lawmaking power. They find, instead, that staffers’ views correspond more closely to the more complex set of presumptions embodied in Mead and the other, more context-sensitive cases that have come on line with it.

Assuming arguendo that the Gluck and Bressman survey captures the views of the staff and of the legislative body more generally, 151 In assessing the survey’s implications, one might ask at the threshold why the subjective understandings of the staff are even relevant to questions of statutory meaning. Professors Gluck and Bressman did not interview members of Congress—a choice they made, in part, because they had a “pragmatic” sense that few legislators would agree to be interviewed and, in part, because they thought it “theoretical[ly]” significant that the legislators themselves “do not draft statutory text.” Gluck & Bressman, Part I, supra note 22, at 923. But it is not clear why the staffers’ intentions or understandings about drafting practice should matter if they lack the power to enact legislation. Consider a hypothetical variation on Gluck and Bressman’s recurrent claim that legislative drafters subscribe to understandings and practices that frequently differ from the established conventions used by the legal community to decode legal texts. See, e.g., id. at 910–11 (noting that staffers do not know or act upon the premises of many of the established linguistic canons applied by the Supreme Court). In particular, imagine that one discovered that the committee charged with producing the final draft of the Constitution at the Philadelphia Convention had its own understandings of interpretive norms—many of which did not line up with the legal community’s established conventions for decoding legal texts. Would anyone seeking constitutional meaning attach significance to the framers’ idiosyncratic assumptions—at least without trying to ascertain whether the ratifiers shared the same understanding of relevant practice? Compare, e.g., Raoul Berger, Government by Judiciary: The Transformation of the Fourteenth Amendment 8 (2d ed. 1977) (expressing the traditional view that the “intention of the framers . . . is as good as written into the text”), with Vasan Kesavan & Michael Stokes Paulsen, The Interpretive Force of the Constitution’s Secret Drafting History, 91 Geo. L.J. 1113, 1135–39 (2003) (noting that the center of gravity has moved away from framers’ intent and toward either ratifiers’ understanding or original public meaning). To be sure, the legislative process on which Professors Gluck and Bressman have cast their focus may differ in important respects from the constitution-making process. For instance, if modern legislators do not really read the text anyway, see Gluck & Bressman, Part I, supra note 22, at 968, I suppose one could argue that staffers advise members of Congress about a bill’s meaning based on the staffers’ understandings of their own drafting practices. Still, given the complexity and diversity of the staff views and practices that Professors Gluck and Bressman have spread across two lengthy articles, see Bressman & Gluck, Part II, supra note 22, at 758 (noting the “number of categories [of staff] and diversity of practice across them”), one might also wonder whether the advice received by members of Congress likely reflects a thoroughgoing application of what staffers perceive to be the practices and understandings of their staff colleagues. That larger question, however, is for another day. its findings nonetheless confirm rather than undercut (a) the ineffability and contingency of legislative intent and (b) the corre­sponding impos­sibility of treating intent as a “fact of the matter” that can be made sense of without a normative frame of reference. First, the staff’s dismissal of dictionaries and of consistent-usage canons highlights what (one hopes) interpreters of any school would acknowledge—that members of Congress do not, and need not, know the precise content of what they are voting for in order to give it legal significance. Rather, even the obscurest norms of interpretation can promote a defensible version of legislative supremacy by providing off-the-rack rules that enable Congress to draw lines of inclusion or exclusion, whether or not its drafters have consciously done so in any particular case. Second, Gluck and Bressman’s findings about legislative history confirm Professor Dworkin’s insight that one cannot identify the fact of “legislative intent” without a prior theory of what intentions interpreters should attribute to Congress. Third, the complexity of staff attitudes about Chevron and Mead confirm the difficulty of moving even from in-depth knowledge of staff attitudes to a firm set of legislative intentions about any particular case. As the analysis that follows indicates, Professors Gluck and Bressman acknowledge many of these complexities, even if their findings occa­sionally suggest that their survey results tell us what a faithful agent would do or what democracy or legis­lative supremacy requires. The analysis that follows means only to show how theory-dependent those claims turn out to be and how greatly value judgments suffuse the choice of what to attribute to Congress, no matter how well we can know the minds of its drafters.

A. Norms of Statutory Usage

The Gluck and Bressman survey explores the extent to which legis­lative drafters subjectively know and subscribe to norms of statutory usage applied by the Court to determine statutory meaning. Interestingly, Professors Gluck and Bressman found that staffers were aware of and relied on certain semantic canons, including the dreaded expressio unius canon and the almost equally controversial ejusdem generis maxim. 152 While not able to identify these canons by their Latin names, most of the staffers surveyed knew of and embraced the concepts behind the negative implication canon (expressio unius) and the word association canons (noscitur a sociis and ejusdem generis). See Gluck & Bressman, Part I, supra note 22, at 932–33. These findings are somewhat surprising given the controversy that has long surrounded the Latin canons. See, e.g., David L. Shapiro, Continuity and Change in Statutory Interpretation, 67 N.Y.U. L. Rev. 921, 927 (1992) (noting widespread criticism of the expressio unius canon). The survey also revealed, however, that not all semantic tools of construc­tion captured the respondents’ subjective expectations or practices. 153 Gluck & Bressman, Part I, supra note 22, at 954. Nor, say Professors Gluck and Bressman, can many of those same canons be justified “as drafting-teaching tools”—norms that, over time, will induce staffers to write more carefully in accordance with the semantic framework specified by the Court. 154 Id.

Many of the canons rejected by the drafting staff have been staples of the Court’s shift toward textualism in the past quarter century. 155 See Manning, Foreword, supra note 9, at 69 (discussing the Court’s increased reliance on tools of construction that carefully parse statutory text). For example, Professors Gluck and Bressman found that the “rule against superfluities” (which, as one might think, presumes that Congress does not use words superfluously) 156 See, e.g., Milner v. Dep’t of the Navy, 131 S. Ct. 1259, 1268 (2011) (rejecting a proposed interpretation of statutory language because the construction would render other language in statute “superfluous and so depriv[e] that amendment of any effect”); Corley v. United States, 556 U.S. 303, 314 (2009) (rejecting the government reading because “it renders § 3501(c) nonsensical and superfluous”). does not reflect drafting reality. 157 Gluck & Bressman, Part I, supra note 22, at 934–35. Nor do consistent-usage canons (such as the “whole act rule”), 158 Id. at 936–37. which presume that Congress uses the same words or phrases consistently across different parts of the same statute or even the U.S. Code. 159 See, e.g., Taniguchi v. Kan Pacific Saipan, Ltd., 132 S. Ct. 1997, 2004–05 (2012) (articulating that presumption); Gustafson v. Alloyd Co., Inc., 513 U.S. 561, 570 (1995) (same). The less frequently invoked “whole code rule” presupposes consistency in phraseology across statutes. See infra text accompanying note 176 (giving example of the rule’s application). And despite the proliferation of dictionaries in Supreme Court opinions in the past quarter century, 160 See, e.g., Jeffrey L. Kirchmeier & Samuel A. Thumma, Scaling the Lexicon Fortress: The United States Supreme Court’s Use of Dictionaries in the Twenty-First Century, 94 Marq. L. Rev. 77, 86 (2010) (charting the explosive growth in the Court’s use of dictionaries). staffers apparently do not consult dictionaries to ascertain the meaning of the words they choose. 161 See Gluck & Bressman, Part I, supra note 22, at 938–40 (discussing dictionary use). In short, Gluck and Bressman’s findings make clear that a Court that purports to be text-sensitive uses many conventions that are disclaimed by the staffers who draft the text being decoded.

Professors Gluck and Bressman are measured in the implications they draw from their findings about what they call the “rejected canons.” 162 Id. at 954. They argue that their results compel proponents of “most versions” of the “faithful-agent” theory to find another justification for parsing the statu­tory text. 163 Id. And even if textualists claim to be looking for “objective intent” rather than subjective legislative understandings, the rejected canons “do not actually effectuate even general, ‘objective’ congressional expectations.” 164 Id. Professors Gluck and Bressman allow that judges might still invoke them based on “rule of law arguments that the canons help judges coordinate systemic behavior or cohere the U.S. Code.” 165 Id. at 951. But courts do not typically rely on such justifications because they “are difficult to reconcile with the faithful-agent paradigm that modern judges find so attractive” and because the judicial imposition of rule-of-law values might “appear ‘activist.’” 166 Id.

Neither legislative intent nor related concepts such as faithful agency, however, are facts in the world, waiting to be discovered. As noted, legislative intent, no less than faithful agency or legislative supremacy, is a construct that tries to make sense of the constitutional roles of various governmental actors in the face of indeterminacy. Two of Gluck and Bressman’s findings, in particular, help to illustrate the point.

First, consider the survey’s results concerning the family of consistent-usage canons, such as whole act or whole code rules. It turns out that a single committee may use words consistently. 167 See id. at 936 (contrasting bills generated by a single committee with those generated by multiple committees). But in today’s Congress, the “increasing tendency” is “to legislate through omnibus or otherwise ‘unorthodox’ legislative vehicles” that reflect the handiwork of “multiple committees.” 168 Id. Because “congressional committees are ‘islands’” that have distinct knowledge bases and drafting practices, most of the staffers surveyed by Professors Gluck and Bressman question the factual validity of presuming consistent usage in omnibus bills whose sections originate in different committees. 169 Id. Even less are courts justified in presuming consistent usage across different statutes passed at different moments in time. 170 Id.; see also Posner, supra note 35, at 812–14 (denouncing this family of canons for assuming legislative “omniscience”).

A second important finding relates to the use of dictionaries. Despite a dramatic spike in judicial consultation of dictionaries, legis­lative staffers do not look up words in the dictionary when they are trying to ascertain statutory meaning. 171 Gluck & Bressman, Part I, supra note 22, at 938. Professors Gluck and Bressman thus conclude that “continued judicial reliance on dictionaries simply cannot be justified on the ground that Congress knows the definitions that will be used.” 172 Id. at 955. Though Professors Gluck and Bressman do not focus on technical diction­aries in particular, it seems safe to stipulate further that their general findings concerning dictionaries apply with at least equal force to the often-arcane technical sources that the Court and its most ardent textualists routinely use to look up the detailed legal connotations of terms of art. 173 See, e.g., Beck v. Prupis, 529 U.S. 494, 504 (2000) (invoking Black’s Law Dictionary and Ballentine’s Law Dictionary to determine the common law elements of “civil conspiracy” for purposes of construing that term in the RICO statute); Kolstad v. Am. Dental Ass’n, 527 U.S. 526, 535 (1999) (using Black’s Law Dictionary and a treatise on tort law to define the meaning of “malice or reckless indifference” under 42 U.S.C. § 1981a(a)(2012)); Molzof v. United States, 502 U.S. 301, 307 (1992) (relying on Black’s Law Dictionary and The Cyclopedic Law Dictionary to define “punitive damages” for purposes of an exclusion contained in the Federal Tort Claims Act). The Court often uses even more obscure sources than law dictionaries to determine the meaning of a term of art. See, e.g., Rosemond v. United States, 134 S. Ct. 1240, 1245–48 (2014) (consulting three nineteenth-century criminal law treatises to determine the reach of accomplice liability for purposes of 18 U.S.C. § 2 (2012), which imposes criminal liability upon one who “aids, abets, counsels, commands, induces, or procures” commission of a federal crime); Air Wis. Airlines Corp. v. Hoeper, 134 S. Ct. 852, 860–61 (2014) (determining the meaning of a statutory “reckless disregard” standard by parsing the way the Court’s First Amendment libel cases apply the actual malice standard). It seems fair to surmise that few, if any, congressional staffers (even the lawyers) look up the legal arcana that the Court so often reads into technical statutory terms. Cf. Gluck & Bressman, Part I, supra note 22, at 952 (noting even their lawyer respondents did not know many technical canons invoked by the Court).

Accepting the Gluck and Bressman findings about staff practices or understandings, however, does not answer the question whether those practices can draw justification from complementary ideas of faithful agency and legislative supremacy. 174 The idea of legislative supremacy itself is contested. See generally Eskridge, Spinning Legislative Supremacy, supra note 16 (discussing competing conceptions of legislative supremacy); Farber, supra note 16 (same). Thus, one cannot decouple a theory of interpretation from faithful agency without first establishing what legislative supremacy entails in our system of government. In fact, both sets of results deepen rather than resolve the baseline statutory indeterminacy that requires judges to construct rather than find legislative intent. First, consider the way the Court applied the consistent-usage canon in its much-discussed decision in West Virginia University Hospitals, Inc. v. Casey. 175 499 U.S. 83 (1991). To simplify, the Court in Casey concluded that the power granted by 42 U.S.C. § 1988 to shift “attorney’s fees” did not extend to “expert fees” because numerous other statutes explicitly shifted both “attorney’s fees” and “expert fees.” 176 Id. at 88–90 (discussing that statutory pattern of usage). For the Court, this pattern of usage was clear and decisive. According to Professors Gluck and Bressman, however, the Court should have paused over the fact that thirty-seven of the forty-one comparison statutes upon which the Court relied—including the four upon which it relied most centrally—originated in different committees from the one that produced § 1988. 177 Bressman & Gluck, Part II, supra note 22, at 781. Because different committees do not use common drafting practices or attach the same meanings to words, Professors Gluck and Bressman suggest that the Court in Casey should have rejected the “consistent-usage presumption[]” on the facts before it. 178 Id.

Second, one can hardly pretend that the Court captures real-world staff knowledge when it relies on the arcana in legal dictionaries or, at times, in even more obscure sources than that. In CTS Corp. v. Waldburger, for example, the Court relied on law dictionaries to draw the fine distinctions between “statutes of repose” and “statutes of limitations” for purposes of an environmental statute that preempts state “statutes of limitations” governing certain pollution-related injuries. 179 134 S. Ct. 2175, 2182, 2186 (2014) (construing the Comprehensive Environmental Response, Compensation, and Liability Act of 1980, 94 Stat. 2767, as amended, 42 U.S.C. § 9601 et seq. (2012)). The relevant section preempts any state “statute of limitations” in a state law action “for personal injury, or property damages . . . caused or contributed to by exposure to any hazardous substance, or pollutant or contaminant, released into the environment from a facility, if the applicable limitations period for such action . . . provides a commencement date which is earlier than the federally required commencement date . . . .” 42 U.S.C. § 9658(a)(1). Or consider Justice Scalia’s famous dissent in Moskal v. United States, which relied on law dictionaries, Blackstone’s Commentaries, old criminal law treatises, and obscure state and federal cases to hold that the crime of “falsely mak[ing]” a security entails forging a document, and not creating a genuine document with false information. 180 498 U.S. 103, 121–26 (1990) (Scalia, J., dissenting). Gluck and Bressman’s findings about dictionaries make it hard to imagine congressional staffers repairing to the nearest law library to scour these obscure sources for the fine points of technical common law meaning.

To me, all of this suggests that the Gluck and Bressman survey either leaves intact or even reinforces the baseline of intent skepticism from which textualism and so many other philosophies proceed. What have we learned? First, the legislative staff—and presumably the legislators for whom they work—did not consciously use the phrase “attorney’s fee” to mean something different from a phrase like “attorney’s fee and expert fee.” Second, it seems most unlikely that either the staff or their legis­lative principals rushed out to ascertain the subtle common law dis­tinctions between “a statute of limitations” and “a statute of repose” or sweated the technical details of the “falsely making” offense. All of this reinforces the assumption, shared by so many approaches, that Congress almost surely failed to consider or resolve any of those interpretive questions at anything like that level of specificity. 181 This supposition accords with Gluck and Bressman’s finding that members of Congress relate to statutes “at the more abstract level of policy rather than at the granular level of text.” Gluck & Bressman, Part I, supra note 22, at 940.

Again, this clears the underbrush to argue about how competing interpretative approaches might or might not serve conceptions of legislative supremacy or faithful agency. From that starting point, one could easily defend both the consistent-usage canon and judicial reliance even upon technical dictionaries. Imagine that the Court rejected the consistent-usage canon, eliminating the presumption that Congress means something different when it says “attorney’s fee” in one place and “attorney’s fee and expert fee” in another. This shift would make it costlier for Congress to use even obvious differences in wording, as one might expect, to express differences in policy. But if the Court presumes as a matter of law that such obvious differences are deliberate, then Congress has a cheap and predictable way to draw lines between statutes that merely award attorney’s fees and those that award both kinds of fees. Similarly, the Court might forgo any presumption that an apparent technical word or phrase imports the full measure of its technical meaning. But that would just make it harder for Congress to tap, for example, into the richness of the law’s many specialized terms. Conversely, if the Court presumes as a matter of law that Congress “knows and adopts the cluster of ideas” that accompany a legal term of art, then legislative drafters have a cheap way to incorporate and reference the common law complexities of a recognizably legal term. 182 Morissette v. United States, 342 U.S. 246, 263 (1952). The opposite presumption—that terms of art do not reflect their common law meaning—might correspond better to the staff’s subjective lack of knowledge of the content of those terms. But by requiring drafters to speak clearly when they want to adopt the technical meaning of a technical term, that presumption would raise the cost of enacting legislation that embraces the “recondite connotations” of the common law. Frankfurter, supra note 14, at 537. If neither presumption affirmatively captures legislative intent—if staffers neither affirmatively embrace nor consciously reject a technical term’s common law meaning—it is not clear why the costlier “clear statement rule” approach to terms of art is preferable to the one in effect today.

It is not my aim here to defend those canons or any particular understanding of faithful agency and legislative supremacy. Rather, the purpose of this discussion is to show that even if the “rejected canons” do not correspond to conscious or subjective staff expectations, those canons may still promote legislative supremacy by giving Congress the tools to draw effective lines of inclusion and exclusion. 183 In a moment, I will discuss the competing possibility of relying on legislative history. See infra text accompanying notes 191–219. To be sure, Professors Gluck and Bressman worry that organizational impediments—such as Congress’s siloed committee structure—prevent legislators from realizing the drafting benefits that a consistent-usage canon might otherwise provide. 184 See Bressman & Gluck, Part II, supra note 22, at 738–39 (“The committee system divides policymakers into ‘silos’ that do not communicate with one another, a fragmentation exacerbated by the separate and different roles that noncommittee leadership staff and personal staff play in the drafting process.”). At least some evidence, however, suggests that recent enhancements in the professionalization, and the contemplated role, of entities such as the Congressional Research Service may increase the likelihood of coherence, even across provisions in omnibus statutes and in different statutes enacted over time. 185 See Jarrod Shobe, Intertemporal Statutory Interpretation and the Evolution of Legislative Drafting, 114 Colum. L. Rev. 807, 856–59 (2014) (discussing improvements in staff resources and drafting practice). Whether or not such improved coordination has (or will) come to pass, the so-called “rejected canons” still make available “off-the-rack” rules for Congress to express itself more precisely, even if staffers do not necessarily take advantage of the tools provided. 186 Easterbrook, Statutes’ Domains, supra note 12, at 540.

Finally, Professors Gluck and Bressman assert that the Court applies its canons too inconsistently to establish an effective semantic baseline or toolkit. 187 Gluck & Bressman, Part I, supra note 22, at 951 (arguing judges “are notoriously inconsistent in their application of the canons, a fact that undermines the efficacy of any canons ostensibly targeted to provide coherence, notice, or consistency”). That concern is a fair one; 188 I have worried about it myself. See Manning, Foreword, supra note 9, at 71 (discussing inconsistencies in the Court’s application of usage canons). consistency is difficult for any multimember body to achieve. 189 See Frank H. Easterbrook, Ways of Criticizing the Court, 95 Harv. L. Rev. 802, 823–31 (1982) (using social-choice theory to examine aggregation problems on multimember courts). But since all communication requires a baseline of social and linguistic practice, 190 See Gerald C. MacCallum, Jr., Legislative Intent, 75 Yale L.J. 754, 758 (1966) (“The words [a legislator] uses are the instruments by means of which he expects or hopes to effect . . . changes [in society]. What gives him this expectation or this hope is his belief that he can anticipate how others (e.g., judges and administrators) will understand these words.”). Gluck and Bressman’s concern goes well beyond the efficacy of the rejected canons. Taken to its logical end, the premise that the Court is incapable of consistency casts doubt on the entire interpretive enterprise. Whether and to what extent that doubt is justified is a topic for a different (and much larger) paper.

B. Legislative History and the Construction of Legislative Intent

Professors Gluck and Bressman also question the foundations of the textualist critique of legislative history. Textualists, as noted, argue that legislative history does not constitute a legitimate basis for establishing meaning because Congress does not enact it through the processes prescribed by Article I, Section 7 of the Constitution. 191 See supra text accompanying note 72 (sketching the textualist position). They also question the empirical proposition that legislative history reflects legislative intent. 192 See supra text accompanying note 70 (same). For two reasons, Professors Gluck and Bressman argue that their survey forecloses any empirical basis for the claim that the text provides a superior source of statutory meaning.

The first reason derives from the staff dynamics of legislative policy-making and statutory drafting. To simplify, Professors Gluck and Bressman find that it is “not uncommon” for legislators to strike a policy deal “before pen is put to paper.” 193 Bressman & Gluck, Part II, supra note 22, at 740. Often at that point, the policy staffers who work for the members of Congress or committees formulate “bullet points” or “rough outlines” of the text, which they transmit to another set of staffers—those who work in one or the other House’s nonpartisan Office of Legislative Counsel. 194 Id. at 740–41. In contrast with the policy staff, the Legislative Counsel staffers who draft the legislation do not answer to (viz. are not hired or fired by) the legislators who frame the policy. 195 Id. at 741. In this environment, Professors Gluck and Bressman find it significant that “legislative history . . . , such as committee reports, . . . is drafted by those staff with more policy expertise and greater direct accountability to the members than the staff who may draft the text.” 196 Id. at 741.

The second reason relates to what members of Congress and the staffers who advise them are likely to read. In a famous opinion, Judge Posner once suggested that, especially for technical statutes, “the slogan that Congress votes on the bill and not on the report strikes us as pretty empty.” 197 Archer–Daniels–Midland Co. v. United States, 37 F.3d 321, 323–24 (7th Cir. 1994) (Posner, C.J.) (citation omitted). Indeed, he wrote, a member of Congress “would have great difficulty figuring out the purport of [a technical term] without the aid of the committee reports.” 198 Id. Gluck and Bressman’s findings appear to substantiate Judge Posner’s hypothesis. According to the survey results, members of Congress and their staff are much more likely to learn about the contents of a bill by reading the legislative history than by reading the text itself. 199 Gluck & Bressman, Part I, supra note 22, at 969. In advising members of Congress, moreover, staffers will have a nose for what kind of legislative history is likely to be the best reflection of what Congress decided. 200 Id. Indeed, Gluck and Bressman’s findings suggest that legislative history was “the most valued” tool of construction, with “92% [of their respondents reporting] that legislative history is a useful tool for courts to consider . . . to determine legislative intent.” 201 Id. at 975.

Professors Gluck and Bressman acknowledge that the complexity of the process they describe “may drive home” the difficulty “of discerning collective intent.” 202 Bressman & Gluck, Part II, supra note 22, at 742. Still, they suggest that their survey has revealed enough about legislative intent to rule out, for example, “democracy-based, faithful-agent” justifications for textualism. 203 Id. In incautious moments, textualists have said that they follow the text because it captures the legislative compromise. 204 See, e.g., Ragsdale v. Wolverine World Wide, Inc., 535 U.S. 81, 93–94 (2002) (“[A]ny key term in an important piece of legislation . . . [is] the result of compromise . . . .”); Hallstrom v. Tillamook County, 493 U.S. 20, 29 (1989) (“Giving full effect to the words of the statute preserves the compromise struck by Congress.”). I have so implied as well. See, e.g., John F. Manning, Competing Presumptions About Statutory Coherence, 74 Fordham L. Rev. 2009, 2039 (2006) (defending “textualists’ systemic presumption that the clear semantic import of the enacted text reflects compromise (rather than judicial correctable legislative inadvertence)”). I outline below what I hope is a better reason for following the statutory text rather than the legislative history. See infra text accompanying notes 213–217. Professors Gluck and Bressman note, however, that “the particularly granular level” at which today’s cases often read statutory text is simply “disconnected from the way in which members and congressional policy staff engage in the drafting process.” 205 Bressman & Gluck, Part II, supra note 22, at 743. The technicality of statutory texts makes it difficult for policy staffers to confirm that the Legislative Counsel “accurately translate[d] their deals” or “reflect[ed] the[] intentions” of the policymakers who made them. 206 Id. And the realities of the legislative process preclude reliance on the premise that “directly elected members, and even high-level staff, are involved in the details of statutory text.” 207 Id. Hence, if nothing else, the facts on the ground foreclose the conclusion that textualism better captures actual legislative outcomes. Instead, a preference for text over legislative history just “boils down to a very spare formalism.” 208 Gluck & Bressman, Part I, supra note 22, at 969. Professors Gluck and Bressman argue that, rather than accept that result, sophisticated interpreters should investigate the more relevant and complex question of what forms of legislative history do or do not reflect the shared understanding of drafters in a complex legislative environment. 209 For example, Gluck and Bressman’s findings suggest that legislators view “committee and conference reports . . . as particularly reliable [in] that they are viewed as evidence of a shared consensus.” Id. at 978. More than half of their respondents, moreover, asserted that legislative history in the appropriations context is particularly important because “the purpose of the committee report . . . is essentially to legislate—that is, to direct where the money appropriated is going.” Id. at 980. Given their nuanced findings, Professors Gluck and Bressman suggest that it would be “rather eas[y]” for courts to “implement many of our respondents’ insights” by giving some types of legislative history more weight than others in the interpretive process. Id. at 989.

The facts found by Professors Gluck and Bressman, however, cannot answer the question whether text or legislative history better reflects “legislative intent” because that question remains, at bottom, a normative one. Recall Professor Dworkin’s admonition that one cannot have a theory of legislative intent without first articulating a theory of adju­dication—that is, without a legal theory that tells a judge or other interpreter how to determine whose intentions count and in what way. 210 See supra text accompanying notes 49–53 (showing the absence of any neutral way to aggregate individual legislative intentions into a collective whole). Professors Gluck and Bressman point to several texts as potentially relevant for identifying the legislative “deal”: (a) the bullet points (or outlines) prepared by the policy staff for the drafting staff; (b) the com­mittee report crafted by the policy staff; and (c) the dry technical text drafted by the Legislative Counsel. Professors Gluck and Bressman contend that their survey “provides new reasons to question the current focus on text as the best route to legislative intent.” 211 Bressman & Gluck, Part II, supra note 22, at 742. Similarly, they argue that their findings preclude textualists from claiming “that a text-based approach best respects legislative bargains.” 212 Id.

Again, what counts as “legislative intent” or the relevant “legislative bargain[]” is simply not a “fact of the matter” that can be established empirically. Any such conclusion depends on normative premises—value judgments that necessarily precede and make sense of the facts. One could easily construct a narrative in which the dry, formal, technical work product of the unaccountable Legislative Counsel can better claim to represent the democratically chosen legislative outcome. Why? Article I, Section 5, of the Constitution gives Congress authority to structure (within constitutional limits) the way it does its business. 213 U.S. Const. art. I, § 5; see also Noel Canning v. NLRB, 134 S. Ct. 2550, 2574 (2014) (holding that Congress enjoys considerable discretion in establishing its rules of procedure). Pursuant to its rules, Congress determines the texts on which it chooses to vote. 214 Given constitutional requirements for enacting legislation, it is fair to impute to Congress some sort of signaling purpose when it chooses the texts on which to vote. See U.S. Const. art. I, § 7. Professors Gluck and Bressman say that the policy staff’s “‘bullet points’ may be the strongest evidence of congressional intent,” but the fact remains that Congress neither votes for them nor even makes them available to the “public.” 215 Bressman & Gluck, Part II, supra note 22, at 741. Indeed, the very strength of Gluck and Bressman’s findings highlights a puzzle about legislative behavior: If legislative reports are produced by the more accountable staff (the ones closer to the deal) and if those reports contain the policy narratives on which legislators base their votes, why does Congress not typically incorporate such materials, in effect, as glossaries to the statutory text itself? 216 Indeed, Congress sometimes does just that. See Manning, Textualism as Nondelegation, supra note 21, at 730 n.245 (collecting examples). Even if there is no procedure at present for doing so as a matter of course, presumably Congress could devise (though perhaps not muster the votes to adopt) a procedure that routinely either compiles committee reports or sponsors’ explanations into an appendix to each bill or provides for their incorporation by reference. Certainly, given the Court’s shift toward (though not all the way to) textualism over the past three decades—and its increased skep­ticism of legislative history in that period—one might at least ask why legislators do not incorporate key legislative history into their bills if it plays such a central role in the articulation and communication of the deal. 217 Almost a sixth of the Gluck and Bressman respondents surveyed “volunteered . . . that there is a level of important legislation-related detail that is simply inappropriate for statutory text.” Gluck & Bressman, Part I, supra note 22, at 974. These respondents stressed that the omission of statutory details was not driven by a political inability to “agree[] on those details,” but rather reflected “a perception of what modern statutory language ‘should look like’ and, relatedly, how much detail statutory text is supposed to have.” Id. That concern is somewhat hard to understand. The statute is a text; so is the legislative history. Presumably, it should make little difference if the committee reports appear on one shelf in the U.S. Statutes-at-Large as an enacted statutory appendix rather than on another in the U.S. Code Congressional & Administrative News.

Professors Gluck and Bressman acknowledge that the statutory text alone comes up for a vote. 218 Bressman & Gluck, Part II, supra note 22, at 742. But they view any “vote-on-the-text-based approach” as mere “formalism” rather than as a way to capture the “legislative bargain[].” 219 Id. at 742–43. Again, I think that this conclusion does not fully acknowledge that what counts as the relevant legislative “intent” or “bargain” is only partly a factual question. Those constructs—and that is what they are—necessarily depend on normative conclusions about what we should treat as meaningful legislative outputs within our particular system of government. In this case, one might deem it significant that Congress’s own procedural choices result in the framing of a dull, technical, formal final text on which the body as a whole votes. Surely it is not mere formalism to equate the relevant legislative outcome with the text on which Congress chooses to act rather than the other texts that it gener­ates in the same process but chooses not to bring before the body as a whole. Whatever facts one may learn about the way the legislative process works, the question of what counts as Congress’s decision depends heavily—indeed, inevitably—on institutional value judgments about what should count as such.

C. The Chevron Doctrine

The survey’s findings about the scope and availability of deference also reaffirm the continued relevance of intent skepticism as a central organizing principle in interpretation doctrine. I have argued in prior writing that the debate over appropriate judicial deference to agency interpretations of statutes depends on the imputation of constructive legislative intent. 220 See generally Manning, Chevron, supra note 21 (outlining imputed intent theory). Congress rarely speaks directly to the question of when and to what degree it wishes to delegate interpretive lawmaking discretion to an agency. The Court has, therefore, cycled through different deference regimes based on functional assessments of when a “reasonable” legislator would want a reviewing court to defer. 221 See id. at 459–63 (examining approaches to imputed intent in different doctrinal periods). Hence, as discussed above, textualists tend to favor what I will call the broad view of Chevron 222 Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984); see also supra text accompanying notes 121–126 (discussing the textualist view of Chevron as a clear background rule against which Congress can legislate). one that categorically treats ambiguity in an organic act as a basis for judicial deference to the agency’s reasonable interpretations of the act. 223 The crispest articulation of the purist position on Chevron appears in Smiley v. Citibank (South Dakota), N.A., which stated:
“We accord deference to agencies under Chevron . . . because of a presumption that Congress, when it left ambiguity in a statute meant for implementation by an agency, understood that the ambiguity would be resolved, first and foremost, by the agency, and desired the agency (rather than the courts) to possess whatever degree of discretion the ambiguity allows.”
517 U.S. 735, 740–41 (1996); see also United States v. Mead Corp., 533 U.S. 218, 240–45 (2001) (Scalia, J., dissenting) (elaborating and defending the purist view of Chevron).
Professor Strauss, as noted, prefers a more fact-bound Mead doctrine, 224 See supra text accompanying notes 132–136 (discussing Professor Strauss’s preference for Mead). which presumes that Chevron applies only in relatively formal agency proceedings but also provides a safety valve in “other circumstances reasonably suggesting that Congress . . . thought of [the agency action] as deserving . . . deference.” 225 Mead, 533 U.S. at 231. And then there are twists such as the “major questions” doctrine, which denies Chevron deference on any matter that is so important that Congress could not have intended to delegate its resolution to an agency. 226 See King v. Burwell, 135 S. Ct. 2480, 2488–89 (2015) (applying the major questions doctrine); Manning & Stephenson, supra note 100, at 819–20 (discussing that doctrine).

The proper approach to deference regime is much discussed, 227 Mead, for example, has given rise to a broad and extensive body of scholarship. See, e.g., David L. Franklin, Legislative Rules, Nonlegislative Rules, and the Perils of the Short Cut, 120 Yale L.J. 276, 319–24 (2010); Thomas W. Merrill, The Mead Doctrine: Rules and Standards, Meta-Rules and Meta-Standards, 54 Admin. L. Rev. 807, 814–15 (2002); Russell L. Weaver, The Emperor Has No Clothes: Christensen, Mead, and Dual Deference Standards, 54 Admin. L. Rev. 173, 175 (2002). Scholarly writing on the major questions doctrine is also extensive. See, e.g., Elhauge, supra note 103, at 103–04; Jody Freeman & Adrian Vermeule, Massachusetts v. EPA: From Politics to Expertise, 2007 Sup. Ct. Rev. 51, 71–78; Abigail R. Moncrieff, Reincarnating the “Major Questions” Exception to Chevron Deference as a Doctrine of Noninterference (or Why Massachusetts v. EPA Got It Wrong), 60 Admin. L. Rev. 593, 597 (2008). but only rarely in terms that focus on what Congress actually intended. 228 But see Merrill & Watts, supra note 104, at 526 (arguing that, until recently, Congress had an established convention for signaling its intention to delegate interstitial lawmaking authority to agencies). Professors Gluck and Bressman, however, argue that their survey findings show that “Chevron now seems too text- and court-centric, in light of our findings, to actually capture congressional intent to delegate.” 229 Bressman & Gluck, Part II, supra note 22, at 729. Instead, they pull from their findings a story in which legislative staffers prefer Mead and the major questions doctrine to Chevron in its pure form.

At the outset, they conclude that the evidence of drafting assumptions does not support Chevron’s equating ambiguity with delegation. Apparently 82% of staffers surveyed knew Chevron by name; 58% said that it plays a role in their drafting; 31% connected ambiguity with judicial deference; and 29% said that Chevron made them think about how precisely they needed to frame their policies. 230 Gluck & Bressman, Part I, supra note 22, at 995–96. But other findings, say the authors, contradict Chevron’s core premise that Congress intends ambiguity as a signal of delegation. In particular, Professors Gluck and Bressman stress that while 91% of respondents stated that delegation constitutes one reason for statutory ambiguity, an even higher percentage cited other contributing factors, such as “lack of time (92%), the complexity of the issue (93%), and the need for consensus (99%).” 231 Id. at 997. From this, the authors conclude that while “Chevron now seems to be a relatively fixed point in many of [the staffers’] drafting practices,” the survey shows that “the doctrine’s assumptions are not entirely reflective of [legislative] intent.” 232 Id.

More importantly, Gluck and Bressman’s results further indicate that many or most staffers subscribe to subsequent refinements that have limited Chevron’s domain. In particular, the authors deemed Mead “a ‘big winner’” in their survey because 88% of the staffers told them that authorizing notice-and-comment rulemaking is always or often relevant to whether staffers intend to delegate interpretive discretion to agencies. 233 Id. at 999. In addition, the survey’s respondents cited numerous other factors as relevant to whether they intended to delegate gapfilling authority to agencies—including the “everyday” nature of the question (99%), the need for agency expertise (93%), the consistency and duration of the agency interpretation (66%), the subject matter of the statute (60%), the agency’s participation in drafting (50%), and common control of Congress and the Presidency at the time of enactment (40%). 234 Id. at 1000–04. Three-fifths, moreover, “intend[ed] for Congress, not agencies to resolve [major questions].” 235 Id. at 1003.

For three reasons, the Gluck and Bressman survey does not alter the baseline that Congress has failed to express a meaningful intention about delegation and thus deference. It therefore leaves intact the Court’s authority to base its rules of deference on functional considerations—to determine what kind of deference a hypothetical reasonable legislator might prefer. First, the fact that staffers ascribed ambiguity to multiple reasons other than delegation is consistent with Chevron’s core premise that one cannot know why Congress did not resolve the “precise question at issue”:

Perhaps [Congress] consciously desired the [agency] to strike the balance at this level, thinking that those with great expertise and charged with responsibility for administering the provision would be in a better position to do so; perhaps it simply did not consider the question at this level; and perhaps Congress was unable to forge a coalition on either side of the question, and those on each side decided to take their chances with the scheme devised by the agency. 236 Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 865 (1984).

The Court added that, for purposes of its decision to give an agency Chevron deference, “it matters not which of these things occurred.” 237 Id. In other words, in harmony with Gluck and Bressman’s findings, Chevron recognized that ambiguity can come from many sources. And since we cannot know, even now, why Congress leaves gaps in any particular statute, the Court felt it could fill the void with a presumption that makes sense of the relative institutional competencies of the agency and the reviewing court. In other words, in the absence of any firm indication of relevant legislative intent, the Court ascribed to Congress what the Court regarded as reasonable choices about how to allocate interstitial lawmaking authority. That focus on relative institutional com­petence has been evident in every one of the post–New Deal deference regimes embraced by the Court.

Second, other findings confirm how difficult it is to discern a subjective legislative intention about a topic such as the appropriate trigger for deference. The Gluck and Bressman survey revealed that staffers found many different factors “relevant” to their intention to delegate. But the respondents did not rank or weight the multiple factors that they found relevant. Many of the factors—including important ones such as the presence of a major question or the agency’s role in drafting—were deemed relevant by far fewer than all of the staffers surveyed. Hence, if 60% of the staffers do not intend to delegate power over a major question but 40% have a different view, how can an interpreter know which type of staffer did the drafting or how to aggregate the disparate views of different staffers who might participate together in writing a statute? (The question itself assumes that staffer intent is legally relevant—a proposition that is itself a normative rather than factual one.) 238 See supra note 151 (raising questions about the relevance of “staffer’s” intent). In short, given the complexity, multiplicity, and unranked character of the factors deemed relevant, there is little reason to believe that a court would have an easy time applying these factors to determine a decisive legislative intent about the appropriateness of deference. In such cases, as in others, the determination of whether to defer will reflect an exercise of judicial discretion to decide if deference makes sense in the circumstances.

Third, the questions posed by Professors Gluck and Bressman left a critical ambiguity in the respondents’ answers. The Methods Appendix published with the two articles shows that, although Professors Gluck and Bressman invited staffers to add freely to their responses, the survey framed many of the specific questions about deference in terms of whether legislative drafters meant to give the agency gap-filling authority. 239 See Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpretation from the Inside: Methods Appendix, 65 Stan. L. Rev. 901 app. at 36–37 (2013), [] (listing survey questions). One question, for example, asked: “To what extent is it relevant to the agency’s gap-filling authority that the agency participated in drafting the legislation?” 240 Id. at 37. Another gave multiple choices to the query, “[w]hat kinds of statutory ambiguities or gaps do drafters intend for the agency to fill?” 241 Id. The framing of such questions potentially leaves in the background what may be the true choice faced by staffers whose texts will inevitably leave residues of ambiguity: “To the extent that statutory ambiguity leaves interpretive discretion, would you rather have that discretion exercised by an agency or a reviewing court exercising independent judgment to fill in these inevitable statutory gaps?” Given Gluck and Bressman’s conclusion that many staffers were reluctant to delegate gap-filling authority to courts, 242 See Bressman & Gluck, Part II, supra note 22, at 776 (explaining that “our respondents” rejected the “notion . . . that Congress tends to delegate at least in some measure to courts”). it is difficult to know what the staff would intend if faced with the choice between conferring discretion upon an agency or a court—typically the choice when a statute does not speak clearly to a question that one or the other must implement. At a minimum, this ambiguity confirms Professor Dworkin’s sense that the identification of subjective legislative intent inevitably depends on the way an interpreter frames the question of what was intended. 243 See supra text accompanying note 53 (discussing Dworkin’s view that the answer to any question about “legislative intent” depends on the level of generality at which one poses the question).

Gluck and Bressman’s study reveals important things about staff attitudes toward judicial deference in administrative law. Most importantly, it confirms that staffers have diverse, complex, and (as yet) unweighted attitudes about the consequences of ambiguity and the desirability of delegation and deference. These findings seem to reinforce the Court’s starting from a baseline of intent skepticism when it has assessed the relative institutional merits of different deference regimes across the years.


For many years, intent skepticism has been an organizing feature in the debate over statutory interpretation. Legal Process theorists like Professor Strauss, no less than realists, textualists, pragmatists, and con­structivists have started from the assumption that the absence of subjective legislative intent impels interpreters to think about interpretive problems in normative terms. And much of the debate has proceeded along the lines of asking which theories of interpretation best capture the relative institutional competencies and responsibilities that our system of government allocates among our lawmakers and implementers.

In recent years, a new empiricism has sought to look into the legislative process to learn facts that cut through what sometimes feels like stalemate in these debates. The findings have been illuminating. If nothing else, they compel modern textualists—who sometimes speak loosely about the justifications for their approach—to clarify the norm­ative basis for their preferences. Most importantly, though, Gluck and Bressman’s findings force us to reckon with the fact that there is no way to derive legislative intent from the brute facts of the legislative process. Every decision about how to filter the facts requires a theory of adjudication, a normative theory of relative institutional roles. Peering inside Congress’s mind reveals that the more we know, the more we understand how hard it is to identify congressional intent.