Congress, Statutory Interpretation, and the Failure of Formalism: The CBO Canon and Other Ways That Courts Can Improve on What They Are Already Trying to Do

I.  Why Interpretive Formalism Has Failed

There are at least two reasons for the enduring resistance to interpretive formalism in statutory interpretation. First, it turns out that most judges place a high value on retaining some flexibility for statutory cases.13 Interestingly, judges do not take this position with respect to other written instruments, such as contracts or wills.14 This distinction is some evidence that judges view their role in statutory interpretation as different from their role with respect to other written instruments15 —and/or that judges understand on some level that statutory drafting is more likely to be imperfect—even though textualists (including Justice Scalia16 ) argue that all of these written instruments are the same animal.

Second, the stakes of a formalist approach in terms of lost judicial power are unacceptably high for many judges. A formalist approach, by definition, requires that statutory interpretation rules be treated as “doctrines” like all other doctrines. Such an approach, properly understood, would give power to the Supreme Court to dictate rules of interpretation to lower courts. Many judges strongly resist that possibility.17 Further, so properly understood—that is, as a species of “real” law (that is, common law)—a formalist understanding of the rules of interpretation likewise would give Congress the power to legislate (that is, to command courts to apply) at least some interpretive rules.18 Case law, empirical work, and judicial writings all confirm that most judges have a visceral, highly negative reaction to such a proposition about congressional power (and even Supreme Court power) over statutory interpretation.19

In some cases, this reaction stems from a lack of confidence in Congress’s and/or the Court’s ability to provide worthy guidance.20 But in other cases, it seems to stem from a constitutional law–level intuition that choice of interpretive method is inherent in each individual judge’s power to judge. Federal judges do not even apply state interpretive rules when interpreting state statutes, Erie notwithstanding.21 This view of statutory interpretation doctrine as inherently “personal” is not compatible with formalism. It necessarily prevents the development of the kind of controlling doctrine that makes a transparent, predictable, rule-of-law regime possible. It also seems unique to this field.

Is there any other power derived from Article III that is so inherent in the individual judge that it does not answer to either Congress or the Court? (The answer is not “originalism” or a different constitutional theory. The relevant comparison is with not constitutional theory, but constitutional doctrine, like the tiers of scrutiny, the First Amendment balancing tests, and the many other implementing rules of the Constitution,22 which others have shown are in fact treated as precedent, even a species of common law.23 ) Query also why judges do not feel this way when courts and legislatures dictate other interpretive rules, such as Chevron deference,24 the parol evidence rule, the Uniform Commercial Code, or the thousands of rules of construction scattered throughout the US Code.25

But the fact is that judges do not want to do this type of interpretive formalism. And that means that it will not work on Congress either. An important ancillary consequence of the judicial unwillingness to follow through with formalism is that Congress views the courts’ interpretive rules as arbitrary and results oriented, and so not worth learning or coordinating with.26

This responds directly to Professor Manning’s point in the opening quotation, in which he argues that even if statutory interpretation doctrine is not connected to congressional reality, it “may still promote legislative supremacy by giving Congress the tools to draw effective lines of inclusion and exclusion.”27 The fact is that congressional drafters do not see the federal courts’ interpretive practice as sufficiently predictable or objective to coordinate with. Nor can the canons “promote legislative supremacy”28 if Congress cannot be assured in advance of which canons will be applied or, alternatively, that the canons that are applied reflect congressional assumptions.

Still, many modern formalists cling to the idea that their interpretive canons are linked to congressional intent.29 They claim a happy (unsubstantiated) coincidence between the system-
coordinating rules and congressional practice, because that is a comfortable place for jurists who are insecure about legitimacy and legislative supremacy. Scalia himself defended his approach on many occasions by arguing (inaccurately) that his favored textual canons were also justified on the ground that they best reflect how Congress actually drafts.30 Indeed, in a recent high-profile challenge to the Clean Air Act,31 Scalia was willing to abandon one of his own favorite textual rules—the presumption of consistent usage—in light of the very real-world (and nonformalist) recognition that the statute was “far from a chef d’oeuvre of legislative draftsmanship.”32

So what, then, would actually justify the system that we have? Why retain a litany of too many interpretive rules that predate the New Deal era and are at best generic assumptions about textual interpretation in general, not even statutes—much less American, federal statutes? Why preserve these rules if they do not actually serve the second-best coordinating, rule-of-law function they were supposed to serve, or even the third-best function of approximating congressional practice? Formalism does not justify it. A democratically focused link to Congress does not either, unless the rules are improved to do better.

Thus, improving the rules is the focus of the rest of this Essay, and if in the process we can eliminate some of the rules that do not add real value, we will be doing an additional service, one that should appeal to formalists as well.33

One final point: As alluded to in the Introduction, it seems possible that, at least some of the time, something entirely different is driving judges when they apply textualism’s tools—something that differs from what any common theory of interpretation has articulated as its justifying principle. Namely, many judges seem to approach their role, much of the time, as something akin to guardians of the US Code.34 Textualism’s linguistic doctrinal presumptions—which assume (and so impose) statutory consistence and omniscience on federal legislation—have the effect of cohering and perfecting the US Code when Congress has been inconsistent, sloppy, or unclear. Those canons impose those judicial or legal-system values of precision on Congress’s statutes, which in their natural state are almost never consistent, perfect, or omniscient. There are many reasons judges would and should do this—notice to the public being a salient one. But this “smoothing” function is one important cause of the disconnect between the rules applied and Congress’s practice, and also between what judges say they are doing and what they actually do.35

Judges who prefer this code-cohering role some of the time are neither formalist—because no one does this all the time or in predictable fashion—nor are they congressional-practice focused. Nor, importantly, are they passive interpreters. Making inconsistent law more consistent shapes and changes the US Code as much as the judges-as-legislative-partner model of Justice Stephen Breyer. Textualists have been resistant to acknowledging this active impact of their preferred doctrines.

I mention this code-cohering approach here to acknowledge that a Congress-tethered approach is not the only one that is attractive to judges, or actually being used by judges, and that could be justifiable in a system in which we are comfortable with judges tacking among different roles in different cases. (Formalism used inconsistently, on the other hand, is not justifiable in this way because the inconsistency defeats the entire point.) For the remainder of this Essay, however, I focus, perhaps oversimplistically, on just one consistent conception of the judicial objective in interpretation. And if we are choosing just one, a Congress-focused approach is what judges actually choose most of the time,36 at least in part because it has the most democratic legitimacy.

II.  The CBO Canon: A New Example of How Understanding How Congress Works Can Improve Doctrine

Now to get practical. Without a fully effectuated formalism to justify current practice, what are we left with? The broadest legitimating principle for our current interpretive regime is that the rules do a decent job of reflecting congressional practice, or, viewed slightly differently, that they are a set of shared norms against which we all draft and interpret legislation. Indeed, this latter principle is precisely what Professor Manning embraces. It is here that understanding Congress comes in and that we have room to improve, and the ability to do so.

I introduce here as one of many possible ways that understanding Congress can improve interpretive practice the CBO Canon. This canon—the presumption that an ambiguous statute should be construed in accordance with the assumptions made about the statute in calculating its “budget score” (its impact on the federal budget)—is a creature of my own devising (but with precursors in case law37 ). It derives from my empirical study of Congress with Professor Lisa Bressman.38 The CBO Canon is based on evidence, substantiated both in our work and in political science research, about the central influence of budgetary estimates on the final text and substance of legislation.39

This influence has democratic bona fides—it is a creature of Congress’s own creation. Congress has statutorily required the nonpartisan CBO to issue estimates of how much most proposed legislation will cost the federal government.40 That assumption is discussed at length with the high-level congressional policy staff drafting the bill, who then typically alter the statute continuously to achieve a particular budget-impact number desired by their bosses or the current administration. CBO’s assumptions are made official and public, and often are widely reported in the national media and commented on publicly by the president.41 Everyone knows what policy X is estimated to cost, or supposed to cost, and the proposed bill is drafted, and redrafted, to bring it and keep it within those numbers.

So how does knowledge about this functioning of Congress help statutory interpretation? Assume, for example, that a statute whose drafting process was heavily influenced by the CBO score later comes before a court in a dispute over the statute’s ambiguity. There may be various statements in the legislative record that support one side or the other, and there will always be various interpretive canons that could decide the case either way. But it is also evident that, read one way, the statute comports with the public budgetary assumptions relied on by all actors during the enactment process, while read the other way, it would knock those assumptions entirely out of whack. Why should a grammar canon of interpretation or a statement made on the Senate floor be a more appropriate tool, when reading the statute in light of the CBO score renders its meaning crystal clear?

The CBO Canon does not yet appear in the treatise books, but it is being invoked in statutory cases with increasing frequency, as a direct result of academic empirical research about how Congress works. A handful of federal court cases had mentioned the score over the past several decades,42 and it was first introduced as a canon in a 2012 blog post,43 applying empirical evidence about CBO’s influence on statutory drafting to the then-pending federal court challenge to the Affordable Care Act44 (ACA), King v Burwell.45 Surprisingly quickly, the canon then found its way into law and policy blogs,46 the politics of the ACA case, official congressional correspondence,47 the litigating briefs,48 and four of the lower court opinions.49 It continues to be invoked, even now that King has been decided, in new cases about the interpretation of different parts of the ACA.50

As noted, the uptake of a new canon of this nature is no special outlier. Both Justice Kagan and Judge Kavanaugh recently have expressed doubt on the well-worn rule against superfluities—the presumption that statutes are drafted without redundancies—in light of empirical evidence that congressional drafters are often redundant, sometimes carelessly, but frequently on purpose.51 Judge Henderson recently cited empirical work to argue that courts should not ignore legislative history in determining the extent of an agency’s discretion because Congress often actually uses legislative history, rather than text, to direct agencies’ work.52 Chief Justice Roberts himself wrote a groundbreaking opinion in King, which for the first time in recent memory took into account the particular unorthodox circumstances of a statute’s enactment in how it should be construed. That opinion adopted a canon-free, big-picture, functional, and structural approach to the statute. It asked: What are the various “interlocking” pieces of the law, and how do they work together? In so doing, the chief justice embraced an approach that much more closely mirrors how elected members themselves approach statutory text than do any of the familiar interpretive rules.53

III.  Why the Sudden Interest in Congress?

What might explain this apparent moment of mounting judicial interest in what Congress actually does? It is something of a puzzle that it comes now, just when Congress is at the peak of its dysfunctionality. It may be the case that the current state of affairs has highlighted what always has been the reality: that statutes are never perfectly drafted. But I think that it may signal a potentially deeper shift.

We appear to be on the cusp of a “third generation” of intellectual development in the field. That is, we are (finally) entering the post–“textualism vs purposivism” era. There is now virtually unanimous agreement among federal judges that text always comes first.54 Statutory interpretation as a “field of law” has become well established. The canons themselves have played an important role in “lawifying” the field and making statutory cases briefable as a discipline in court and teachable across virtually every law school in the country—because there are things that resemble doctrines to cite.55 Those earlier debates and the entrenchment of the canons were essential in setting the stakes, creating the relevant domain, and also making clear the importance of the task.

But those debates have taken us as far as they can go. Plenty of judges have grown up with this field by now and can see where the holes are. Explicit engagement with the gaps in the field remains rare, but it is becoming apparent that at least some judges are interested in thinking about how the doctrines might be improved. As noted textualist Judge Kavanaugh recently wrote, “[I]t is a mistake to think that the current mess in statutory interpretation is somehow the natural and unalterable order of things. . . . [W]e can do better.”56

A.    Even Rigorous Canon Application Can Be Profoundly Dissatisfying

Consider this example of a recent Supreme Court decision—heralded by commentators as a “classic case of statutory interpretation” and a “[b]attle of [the] canons”57 —to understand how profoundly dissatisfying the current approach is and so why there might be openness to new approaches.

Lockhart v United States58 concerned whether penalty enhancements in the Child Pornography Prevention Act of 199659 for offenders with a prior state conviction “relating to aggravated sexual abuse, sexual abuse, or abusive sexual conduct involving a minor or ward” applied to all prior abuse convictions or whether the limitation “involving a minor or ward” applied only to the last item on the list.60 The primary argument before the Court—in a case that involved a person’s liberty—was the applicability of two arcane textual-grammar canons that it is virtually certain no elected member of Congress or high-level policy staff considered when voting or drafting. The majority, per Justice Sonia Sotomayor, applied what she called a “timeworn textual canon,” the so-called “rule of the last antecedent,” which presumes that a modifier at the end of a series applies only to the last antecedent.61 The dissent, per Justice Kagan, would have had the case turn on the “series-qualifier canon,” a rarely applied presumption that “a modifier at the end of [a] list normally applies to the entire series.”62 Each justice cited Justice Scalia’s treatise of “approved” canons to justify her respective choice.63

At oral argument, Scalia lamented that some way was needed to choose between the equally applicable “dueling canons,” and so suggested yet a third canon, the rule of lenity,64 which was always his favorite tiebreaker policy canon. Justice Breyer wondered why the Court would not instead look to legislative history at least to break a canon tie, to which Scalia replied: “[Y]ou don’t think Congress can leave it to its staff to decide what a statute means,
do you?”65

This is perfectly fine statutory interpretation under the current regime, but why in the world, as a matter of legitimacy, would those two grammar canons be the methodologies that frame the debate in the case? Neither canon is normatively superior to the other, nor—because we do not have a formalist regime—is there any way to predict which one would have applied. Using these rules therefore serves neither democracy reasons nor predictability reasons. Judge Easterbrook likewise identifies Lockhart as an example of his “absence of method” argument.66

And with respect to Scalia’s concern about “staff,” many times comma placement—which some view as critical to the triggering of the last antecedent rule as a matter of formal grammar (a comma implies the modifier applies to the whole list)—is altered or added by Congress’s in-house grammarians and codifiers after a statute is passed.67 That is, changed by staff. Beyond grammar, staff also makes the nitty-gritty word choices in most situations. Staff often decides if statutory provisions are consistent or redundant.

The common textualist response to the central role of staff is that the staff’s role in statutory-text drafting is essentially “erased” by the formal vote.68 But even that argument does not explain the codification and cleanup process that takes place after statutes are passed and that can be central to statutory interpretation disputes.69

B.    Would a Canon like the CBO Canon Have Been More Satisfying?

Legislative history was only somewhat revealing in Lockhart. Kagan’s dissent noted that the Senate report referred only to “state child abuse law” in describing the section at issue.70 A canon like the CBO Canon might have assisted further. The provision at issue in Lockhart was scored as having no “significant impact on discretionary spending” because CBO expected “very few federal cases to be affected.”71 The CBO report likewise referred only to “crimes relating to child pornography.”72 If adding adult-related crimes would have significantly changed the score, such information could have been very useful in determining Congress’s understanding of the statutory text.

In the Gluck-Bressman study, the following comment about CBO’s influence on statutory text was typical: “In tax and spending programs you live and die by the score. We have a number in advance and we work back and retrofit the policy to the score. We send them draft after draft.”73 Newspapers are replete with quotes about CBO’s forceful impact on enacted text in statutes in areas ranging from health reform to Medicare, homeland security, and the FCC.74 Similarly, stakeholders often criticize proposed agency rules on the ground that they would not be consistent with the assumptions in the score.75

The CBO Canon will not be useful in every case, and perhaps it would not have been useful in Lockhart.76 My point is simply to illustrate how it could hypothetically apply when it is relevant. For instance, in the ACA case, King, the challengers’ reading of the statute to deny federal subsidies on certain insurance exchanges would have had a profound impact on the score, for a statute whose bottom-line amount was reported in the news media and scrutinized by the president and his opposition on a daily basis.77 There, the canon proved extremely powerful in revealing what Congress understood the statute to say.

There are compelling antecedents of this new canon. The CBO score had been referenced in a handful of lower federal court cases prior to 2012. Most frequently, those courts had found it helpful in determining the retroactive application of statutes. Because retroactive application significantly increases the costs of a statute in many instances, the score is extremely illuminating for such questions.78

Think of other statutory interpretation chestnuts that could have been aided by this tool. For instance, in one case that has been the subject of much academic and judicial treatment,79 the Court in Arlington Central School District Board of Education v Murphy80 divided over whether parents in an Individuals with Disabilities Education Act81 (IDEA) suit could recover expert consultants’ fees from the state as part of the statutory award of “reasonable attorneys’ fees as part of the costs.”82 At the time, CBO did not score for state impact, but CBO did seem to read the statute as the dissent did,83 and today, with state impact calculated, Arlington would be an excellent candidate for reference to the score.

C.    Overblown Concerns about Judicial Competence

A common cop-out is the objection that judges are not up to this task, that judges cannot possibly understand the “sausage factory” of federal lawmaking. But judges reason from cues about Congress’s assumptions all the time. Consider the well-known example of Sutton v United Air Lines, Inc.84 In that case, the Court reasoned from the numerical assumptions in the preamble to the Americans with Disabilities Act of 199085 to determine which categories of disability Congress wanted the Act to cover;86 specifically, it reasoned from the fact that the number of Americans the preamble declared affected was too small if those requiring corrective eyewear were to be included. The CBO Canon requires essentially the same type of analysis.

There is also a broader judicial competence point here. Put simply, federal judges today are not interested in abdicating power over statutory questions. It is true that textualism is grounded in a legal-realism-meets-economics-inspired vision of Congress’s actions as irrational and so incomprehensible to judges.87 (The irony of course is that the textualist response has been to construct a second-best set of rules that assume the exact opposite state of affairs: a Congress whose texts are perfect.) But even if that view of the inscrutability of Congress might have been attractive thirty years ago, the federal courts today seem to want more, not less, say over statutory cases—which are, after all, the most frequent and often most important cases on the docket. As proof, one need only look to the recent destabilization of the Chevron rule for agency deference.88 One justification for Chevron—in which the Court voluntarily ceded much of its power over interpreting statutes to federal agencies—was this same idea of judicial incompetence, or at least a much lesser expertise or political role than that of agencies.89 But we now have a contingent of federal judges, including several on the Court, strongly signaling their desire to reclaim some of that power.90

A court that wants to retain control over statutory interpretation cases cannot justifiably do so while also proclaiming its own incompetence or its inability to understand Congress. Even those textualists embracing legislative supremacy cannot rationally do this. One cannot be a faithful agent to a master who one believes speaks nonsense. Moreover, to assume that Congress is incomprehensible may be to deny the Court’s role as secondary when interpreting Congress’s work—in other words, it makes the Court much more activist than it claims it wants to or should be.

This notion helps to explain Chief Justice Roberts’s opinion in King, the ACA case. That opinion went out of its way to evince the Court’s deep understanding of the complexities of the “interlocking” and “intertwined” pieces of the ACA.91 It made clear that the Court did not need or want the agency’s help. It explicitly referenced how the ACA’s unusual legislative process produced a statute with less “care” than courts would desire and referenced a cartoon about statutes being “too complicated to understand.”92 But in the end, the Court assumed that Congress acted rationally and that statutes are meant to be coherent; concluded that judges “must do [their] best”; and, despite it all, found the case to be an easy one in light of how the different sections of the statute were constructed to work together.93 King showed a Court flexing its muscle, a Court proving that it understands Congress. Not coincidentally, I think, King was also an opinion that deployed no textualist canons of interpretation.

I do not wish to imply that Scalia ever truly believed that judges are incompetent. His theory was really about keeping the courts out of the mess of politics, coming up with objective tools, and using strict interpretation to incentivize Congress to improve its own work product. But the numerosity and malleability of the canons have led to a never-ending chorus of criticism that the Court uses them politically and unpredictably. And any effort by the Court to use textualism as a tool of legislative improvement has been a near-total failure. The Gluck-Bressman study revealed a Congress that knows many of the canons and that also knows about the Court’s use of dictionaries and distaste for legislative history, but that cannot perfect its work, will never abandon legislative history, and does not use dictionaries itself.94

Congress is focused on what works for Congress. If the courts want to be in a dialogue with Congress, the courts need to take the first step and bend. And this is where formalist values come back, because it is possible that Congress ultimately may go along for the ride once the courts show they are serious and consistent about the effort.95

IV.  Beyond CBO: Ten New Rules for Congress-Focused Statutory Interpretation

Putting aside the CBO Canon, the Gluck-Bressman empirical study uncovered many other objective institutional features and cues derived from how Congress organizes itself. I stress the objective and institutional characteristics of these rules so that they are not put up against the straw man of the incomprehensibility of subjective legislative intent or the sausage factory. At least some of these cues are more helpful to understanding statutory ambiguity than are canons, dictionaries, and some legislative history. This Part highlights just a few—many more are detailed in the previous work—as a way of showing what is accessible and possible.

One critical point worth emphasizing at the outset is that the argument is not that these structural cues overcome clear statutory text. The CBO Canon, for instance, might meet objection because CBO scores are not voted on as statutory text is.96 That is not the question. The question is where to turn when text is ambiguous or runs out. Most of the canons are more disconnected from any congressional vote than CBO and the other institutional features the Gluck-Bressman study identified.

The study surveyed 137 congressional counsels, across different committees, offices, and political parties, over five months from 2011 to 2012.97 Every counsel was asked 171 questions, which aimed to examine the drafters’ knowledge and use of canons and legislative history as well as to elicit information about the legislative process and views of the role of courts in interpretation.98 We did not inquire about Congress’s self-organization—including CBO, the committee system, unorthodox lawmaking (the increasingly common deviations from the traditionally textbook legislative process),99 or other specialty offices inside of Congress—but our survey provided unlimited opportunities for qualitative explanations and additions. The structural and organizational influences on legislative drafting that I highlight here were volunteered time and again.

For instance, 15 percent of respondents volunteered that we had failed to inquire about the CBO score and emphasized its centrality to the drafting process.100 With respect to other important structural factors, 45 percent emphasized the committee system as the fundamental organizing principle of congressional drafting.101 The committee system, as elaborated below, turns out to be critical to understanding matters ranging from obstacles to linguistic consistency to determining agency delegation. As another example, more than 83 percent of staffers volunteered comments that understanding the role of the Offices of the Legislative Counsel—the nonpartisan drafting offices in the House and the Senate that turn policy deals into statutory language—is essential to understanding how statutes are written and what is often the distance between the big picture statutory policy decisions made by members and high-level staff and the individual words that wind up on the page.102

From a democracy perspective, it seems worth emphasizing that these institutional features are Congress’s own creation. CBO’s centrality has increased since the passage of the Statutory Pay-as-You-Go Act of 2010,103 which requires a budgetary estimate of a bill’s effects to accompany all covered legislation.104 Congress organized itself into committees. It has created a variety of statutory vehicles—omnibus, appropriation, and single-subject bills—that differ significantly from one another in structure, linguistic conventions, and legislative processes. It is Congress that created the various internal offices of nonpartisan staff with important roles in the drafting of statutory text.

The Constitution explicitly gives Congress control over its own procedures.105 As such, the federal courts have long protested that they have no role to play in the internal workings of Congress—but that is mere cover. The current approach to interpretation, because it ignores all of these institutional factors, actually seeks to smooth over the effects of the very institutional structures that Congress itself has erected, by imposing perfection and uniformity on statutes that the legislative process makes anything but perfect and uniform.

Perhaps courts are subtly trying to discourage these delegations and institutional constructs, but that may be giving the judges too much credit. It is a lot easier to overlook Congress’s internal structures than to think about their effect on legislation, but that does not mean that doing so is beyond the capacity of courts—as critiques like Professor Manning’s would imply.

A.    Committees, Type of Statute, Legislative History, and Underbelly Institutions

Here I detail a few more examples of structural, objective cues, beyond CBO, that are certainly as much within the competence of courts to absorb and employ as are the canons. Not every reality about the drafting process meets this description. For instance, a common refrain in the Gluck-Bressman survey was that the personal reputation of members, high-level staff, and administrators is important internally to judging the credibility of legislative history or the strength of an agency delegation.106 Courts could not possibly utilize such information. But courts—even formalism-preferring courts—can, for instance, determine whether the same committee drafted two sections of a bill that advocates are arguing should be treated consistently.

It should be evident that the doctrinal implications of these congressional cues are mostly for the linguistic, textual canons. Those canons make assumptions about consistency and legislative omniscience that an understanding of Congress’s institutional structures calls into question. The policy canons—like lenity and federalism—seem different. Although theorists, including textualists, argue that those policy presumptions are “shared” with Congress, they are not.107 But these canons still can find justification without having anything to do with Congress. There are reasons—normative, perhaps even in certain cases quasi-constitutional—why judges would apply, for example, a federalism canon as a tiebreaker even if Congress did not know or agree with it (in fact, many of the staffers in the Gluck-Bressman study were actually against a presumption in favor of preemption108 ). What the empirical work shows is simply that those policy decisions need to be justified for what they are—judicial lawmaking, judicial norm-and-constitutional-value imposition—and not something derived from Congress. That kind of judicial work of course can function alongside doctrines that aim to reflect Congress. If formalism is not the goal, having different kinds of interpretive tools in the bucket of potential options does not pose a legitimacy problem as long as the tools themselves are legitimate.

I myself think there is plenty of space for that kind of judicial work when it comes to statutes. Erie’s long shadow over federal common law–making extends too far when it prevents judges from coming up with decision-making rules in federal statutory cases, as there is nothing about Erie’s prohibition that is relevant to the work of federal statutory interpretation.109 But that shadow has clouded perceptions of judicial lawmaking’s legitimacy, and so judges cling to the unnecessary fiction that these rules are somehow all related to congressional practice rather than sometimes crafted and imposed by courts atop of and in tension with legislation. They should not, and they do not have to under Erie. But the latter is a topic for another day.

Onward now to some specific recommendations to improve the Congress-court connection in statutory interpretation.

B.    Committees: Eliminate the Whole Act/Whole Code Rule and Related Presumptions

The organization of Congress into committees emerges from the Gluck-Bressman study as the most salient structural influence, and limitation, on how statutes are drafted and interpreted inside Congress.110 Most importantly, it poses a perhaps insurmountable obstacle to the accuracy of assumptions of linguistic consistency and legislative omniscience that permeate statutory interpretation doctrine. The idea that similar phrases mean the same thing across an entire statute or that variation of terms is meaningful even across multiple statutes does not comport with the structural separation of committees and the lack of communication between them, even when they work on the same statute. Different committees have different conventions as well. Even the Offices of Legislative Counsel, the nonpartisan drafting experts in both chambers who draft most complex statutory language, are assigned to different committees, and so they do not act as a unifying entity for purposes of linguistic consistency as one might assume.111

New Rule #1:  Absent clear evidence to the contrary, consistency presumptions should not be applied for exceedingly lengthy statutes, for different statutory sections within a single statute drafted by multiple committees, or across different statutes.

That is a plain enough rule for any judge to understand.

Committee jurisdiction plays an important role in many other interpretive presumptions, too numerous to chronicle here.112 But to understand this breadth, consider one example of another new rule that easily derives from understanding the organizational force of the committee system and that, as the
Gluck-Bressman study revealed, reflects congressional drafters’ understanding:

New Rule #2:  When there is doubt as to which agency is the intended recipient of delegated authority in a statute involving multiple agencies, the presumption is that the agency under the jurisdiction of the drafting committee takes the lead.113

Once understood, the presumption is pure common sense. The drafting committee wants to retain control over its statute’s implementation and can do so only if its own agency is the implementer.

Critics will surely ask: “How can federal judges possibly learn this information?” Well, the rule of the last antecedent and the presumption that ambiguities in the Bankruptcy Code should be construed in favor of the debtor114 are not innate either. Lawyers and judges learn interpretive presumptions in school, from other cases, from legal briefs. The same learning curve applies here.

C.    Not All Statutory Vehicles Are the Same

The survey, together with my recent work with Professor Anne Joseph O’Connell and Rosa Po,115 also reveals dramatic differences in structure and drafting conventions across different types of statutes. Omnibus bills are often completely separate statutes strung together.116 Appropriations bills place all substantive directives in legislative history per congressional rule; the text of the statute itself simply lists numbers.117 Many statutes no longer go through the conventional “textbook” process and so bypass traditional opportunities for cleanup.118 Statutes affecting the work of multiple agencies use common linguistic drafting provisions that have been virtually unknown to courts.119

New Rule #3:  Presumptions of linguistic consistency make no sense for omnibus bills.

New Rule #4:  Special attention should be paid to legislative history in appropriations.

It is no coincidence and is a striking fact that the only legislative history drafted by the Office of Legislative Counsel is appropriations history.120

New Rule #5:  Statutes that do not go through conference committee may not be as neatly drafted or checked for perfection as statutes that go through that process.121

New Rule #6:  Emergency bills are almost always vaguer than other statutes.

(Recall, for example, the post-9/11 Authorization for Use of Military Force,122 a single page, passed three days after the attacks, which has been used for the past decade to justify military acts certainly unanticipated when it was drafted.)

New Rule #7:  There is a common linguistic drafting convention when it comes to multiple-agency delegations.

Courts have yet to come up with a clear deference doctrine for overlapping delegations. But they have not looked to Congress. The Gluck-Bressman drafters reported that X “in collaboration [or conjunction] with” Y signals that the agencies are to work together; X “in consultation with” Y means agency X takes the lead.123 Indeed, a search of the US Code reveals that convention in hundreds of statutes,124 and the convention is simple to apply once learned.

D.    Dictionaries, Process, History, and Underbelly Institutions

New Rule #8:  Legislative drafters do not use dictionaries. Courts should not use them for ordinary, nontechnical terms.

Legislative drafters do not use dictionaries. Period.125 And yet, Professors James Brudney and Lawrence Baum have illustrated that courts have ignored Congress’s own statutory definitions in favor of dictionary definitions.126 It also has long been understood that dictionaries lag behind changes in ordinary speech; this makes them poor evidence of words with evolving meanings.127

Consider, moreover, some of the kinds of words that have been recent targets of judicial dictionary use: “even though”128 and “any”129 are examples sufficient to make the point. What does a court gain by consulting a dictionary for the meanings of these words other than a citation to an objective-looking source for purposes of opinion writing? It is one thing for a court to look up arcane legal terms or nonlegal technical terms in a specialized dictionary. But we should press our judges to tell us why they are looking up words whose meanings are overwhelmingly common (“any”!), whose meanings it is certain that no drafter ever thought would come into question. If a statute is ambiguous because it seems unlikely that “any” really means “any,” the dictionary is not going to tell you how Congress thought the statute should sensibly be read. We don’t need a dictionary to tell us how the average person understands “any.”

New Rule #9:  Expand and narrow the tool kit of legislative history to include markups, bipartisan colloquies, consensus history, and section-by-section summaries. De-emphasize grammar, codification placement, and section numbering.

Legislative history is a far more complex animal than courts give it credit for. We all know that not all legislative history is equal. Most judges (properly130 ) understand committee reports as generally more reliable than floor statements.131 This, by the way, is in itself more evidence that judges do make everyday attempts to learn about how Congress works and apply what they learn. But further research on Congress’s practices can teach us a lot more in this area. This is not to say that legislative history is the law. I fully agree with Judge Easterbrook that it is not; but certain types of legislative history may be more useful to address ambiguity than canons with no link to legislative practice.

First, the committee markup, which is the transcribed, section-by-section review of a bill and its disputed language that occurs publicly in committee, can be very illuminating. Ambiguities are often highlighted during the markup and resolved on the record. Former DC Circuit Judge Patricia Wald advised judges to look at markups decades ago;132 that task is today much simpler because Westlaw, as well as House and Senate committees themselves, often publishes them.

Second, there are certain kinds of floor statements that are broadly trusted inside Congress that have been thrown out or laughed at by courts and scholars (including this author!) as shams. The best example of this is the “scripted colloquy”—a staged conversation between members on the floor (often not even delivered “live” but simply inserted into the record).133 In the Gluck-Bressman study, staffer after staffer volunteered their importance (without being asked about colloquies). We learned that, because such colloquies are often between members of opposite parties, they are especially important because they aim to evince a shared agreement about statutory meaning.134 Consensus legislative history in general, we learned, especially legislative history across party lines, is considered particularly trustworthy.135

Space constraints do not permit additional examples, but I do wish to point out that there is an entire underbelly of Congress that barely has attracted any attention. The Law Revision Counsel arranges public laws into codified sections and often creates section numbers, titles statutory sections, and addresses grammatical issues.136 We all know how often those matters come up in statutory cases, and yet I cannot recall a single example of a time the Court recognized that this process occurs long after a statute is enacted without any involvement, supervision, or vote by members—and yet the changes wind up in the final product.137 Those formalists who insist on the exclusive importance of the text as voted on simply have not grappled with how Congress actually works.

Similarly, as detailed in the Gluck-Bressman study and already alluded to here, the Offices of the House and Senate Legislative Counsel are absolutely central to the production of statutory text and yet their important influence had virtually never been highlighted in legal literature until the study.138 Understanding their role drives home what we already knew about members—that they do not draft text—but also sheds new light on our understandings of accountable congressional staff. Staff directly accountable to members do not generally draft the nitty-gritty statutory text either. They make the policy and sketch out statutory contours, often in the form of policy “bullet points.” The nonpartisan Legislative Counsels then take over and turn the “asks” into statute-ese. The federal courts put all of the attention on the final text, whereas in Congress much of the attention, of the most senior and accountable people as well as of the members, is on the bigger picture. For what it is worth, understanding this fact has had a profound impact on my own thinking about statutory interpretation, and has reconfigured, and softened, my own views about the relevance of high-quality legislative history (typically drafted by the accountable policy staff, not the Legislative Counsels).

To that end, it is telling that one of the most important pieces of legislative history to most congressional insiders and legislation experts is the “section-by-section” summary that accompanies most statutes. The section-by-section is drafted by committee staff (the policy staff, not the Legislative Counsels) and describes what each section is doing. Anecdotal evidence suggests that most seasoned statutory players start with the section-by-section to understand the point of each section before turning to what is often the dense and unintelligible (because of cross-references and such) minutiae of the statutory text.139 The focus inside Congress is macro; but the focus inside courts, at least until King, has been micro.

Finally, the Congressional Budget Office is, of course, another internal congressional institution central to the drafting process but long overlooked by courts, which brings us full circle:

New Rule #10:  The CBO Canon: Ambiguities in statutes presumptively should be construed in accordance with the assumptions relied upon in calculating the budget score.

* * *

To be sure, Congress could make this a lot easier. Congress might create a list of these conventions or presumptions.140 But federal courts would still need to pay attention. At the moment, federal courts would be highly unlikely to hold themselves bound by such a congressional list, or even to defer to it. For instance, Congress has passed the Dictionary Act,141 which contains a series of common definitional presumptions, such as the presumption that singular statutory terms include the plural.142 Congress has not made application of that Act mandatory,143 and the courts apply it only when convenient. Justice Scalia once suggested that it would be unconstitutional for Congress to dictate interpretive rules to courts.144 It also undoubtedly would be a challenge to “update” interpretive presumptions if Congress’s own practices change over time; this is another reason that Congress’s own participation in this endeavor would be quite helpful.

On the other hand, the Supreme Court has now cited the House and Senate Legislative Counsel drafting manuals three times in support of the drafting presumptions they apply.145 And the US Code contains thousands of “rules of construction”—for example, the well-known presumption in ERISA146 that it shall not “be construed to exempt or relieve any person from any law of any State which regulates insurance”147 —that courts do apply without difficulty or objection,148 seemingly without recognizing they are of a piece with the notion that Congress can dictate the rules of interpretation.

The point is that, even as courts claim that Congress is inscrutable or resist Congress’s interpretive rules, courts simultaneously do in fact refer to congressional practice to legitimize the interpretive rules they employ. Indeed, in the absence of a formalist, rule-of-law justification, what choice do they have?

Conclusion

It has always been easiest, and most democratic, for federal courts to embrace the fiction that their interpretive doctrines derive from Congress itself. That is not the case, and may never fully be the case. But the formalist alternative—even as a second-best and even as attractive as that is to this writer—has even less possibility of complete success.

A task as central as statutory interpretation is to the everyday work of courts deserves stable normative and jurisprudential foundations. To reemphasize, there need not be only one acceptable approach to what makes interpretive rules legitimate. Courts can attempt to “do [their] best”149 to incorporate understandings of Congress that are amenable to legal doctrine—that is, the kinds of objective, structural cues I have detailed here—and thereby improve on what in reality they are already trying to do. But courts can also layer atop that effort external norms that have nothing to do with Congress as long as they are properly recognized as such. These are likely to be policy norms like federalism and lenity, but they could just as justifiably be language norms that judges currently grasp onto—such as consistency, perfection, and omniscience. Those language norms seem like “only” linguistic rules but of course embrace normative values of their own—for example, notice, objectivity, and predictability—that are highly valued by legal systems.

What needs to be understood, however, is that interpreting statutes to be consistent when they are not drafted that way is as much the imposition of judicial policy and norms that are external to how Congress works as is imposing a federalism doctrine. Textualists do not see the textual presumptions that way,150 but studying Congress reveals them for what they are. It does not mean these presumptions are illegitimate, but it does mean that they require a different theory of the Constitution or federal lawmaking to justify them. And that theory, as this Essay demonstrates, is not formalism.

What also needs to be understood is that an interpretive regime like ours that combines these different kinds of approaches to statutory interpretation is not a formalist regime, but it is not completely lawless either. The canons serve as loose frameworks, boundaries, reminders. They are a kind of very weak common law that is especially generous in giving judges the flexibility they need.151 Indeed, this is what Professor Karl Llewellyn liked about the canons all along, even though only his critique of them has been remembered.152 The key—and a critical area into which statutory interpretation theory has not really ventured—is how judges should choose one type of tool over another.

I grapple with that question in a separate project, but regardless of the answer, it is unquestionable that this terrain of canons includes a large component that relates to how Congress drafts. It always will. It is too easy to say that we can ignore the realities of congressional lawmaking because the canons play a formalist role regardless—they do not. It is likewise too easy to say that judges are not up to this task. They already are showing that they are.

• 13See Abbe Gluck and Richard Posner, Statutory Interpretation on the Bench: A Survey of 35 Judges on the Federal Courts of Appeals *18 (forthcoming 2017) (on file with author).
• 14Gluck, 120 Yale L J at 1970–76 (cited in note 5).
• 15I set aside the Constitution for purposes of this discussion because it is a very different kind of written instrument than a statute, contract, or will.
• 16See Antonin Scalia and Bryan A. Garner, Reading Law: The Interpretation of Legal Texts 49–239 (Thomson/West 2012) (listing rules for “all types of legal instruments”).
• 17See Gluck and Posner, Statutory Interpretation on the Bench at *37–41 (cited in note 13).
• 18See Gluck, 54 Wm & Mary L Rev at 757, 802–03 (cited in note 5). But see id at 766 (noting that there may be some interpretive rules that have constitutional foundations over which Congress could not legislate); Nicholas Quinn Rosenkranz, Federal Rules of Statutory Interpretation, 115 Harv L Rev 2085, 2139–40, 2156 (2002) (“Some interpretive techniques are constitutionally required and may not be abrogated by Congress.”).
• 19See Gluck and Posner, Statutory Interpretation on the Bench at *37–41 (cited in note 13). Many states have attempted to legislate interpretive rules, only to meet resistance from state and federal courts alike. Gluck, 119 Yale L J at 1785–97 (cited in note 3); Gluck, 120 Yale L J at 1940–59 (cited in note 5) (collecting cases); Scalia and Garner, Reading Law at 244–45 (cited in note 16) (suggesting this would be unconstitutional).
• 20See Gluck and Posner, Statutory Interpretation on the Bench at *37–41 (cited in note 13).
• 21See generally Gluck, 120 Yale L J 1898 (cited in note 5).
• 22Mitchell N. Berman, Constitutional Decision Rules, 90 Va L Rev 1, 50–78 (2004). See also generally Richard H. Fallon Jr, The Supreme Court 1996 Term—Foreword: Implementing the Constitution, 111 Harv L Rev 54 (1997) (describing the decision rules of constitutional doctrine).
• 23See generally Henry P. Monaghan, The Supreme Court 1974 Term—Foreword: Constitutional Common Law, 89 Harv L Rev 1 (1975). The one possible comparator here might be stare decisis. Compare generally, for example, Richard H. Fallon Jr, Stare Decisis and the Constitution: An Essay on Constitutional Methodology, 76 NYU L Rev 570 (2001), with Michael Stokes Paulsen, Abrogating Stare Decisis by Statute: May Congress Remove the Precedential Effect of Roe and Casey?, 109 Yale L J 1535 (2000).
• 24See generally Chevron U.S.A. Inc v Natural Resources Defense Council, Inc, 467 US 837 (1984).
• 25See Gluck, 120 Yale L J at 1972, 1989–90 (cited in note 5).
• 26Bressman and Gluck, 66 Stan L Rev at 774–77 (cited in note 7).
• 27Manning, 115 Colum L Rev at 1942 (cited in note 1) (emphasis omitted).
• 28Id.
• 29See Lockhart v United States, 546 US 142, 148–49 (2005); Miranda McGowan, Do as I Do, Not as I Say: An Empirical Investigation of Justice Scalia’s Ordinary Meaning Method of Statutory Interpretation, 78 Miss L J 129, 170–88 (2008) (cataloging Scalia’s attention to legislative intent).
• 30See, for example, Antonin Scalia, Judicial Deference to Administrative Interpretations of Law, 1989 Duke L J 511, 516–17. For other textualists making similar arguments, see Manning, 2011 S Ct Rev at 154–56 (cited in note 6) (citing others’ arguments and arguing that “the legal community shares a rich set of established background conventions that apply in recognizable situations”).
• 31Pub L No 88-206, 77 Stat 392 (1963), codified as amended at 42 USC § 7401 et seq.
• 32Utility Air Regulatory Group v Environmental Protection Agency, 134 S Ct 2427, 2441 (2014).
• 33See Brett M. Kavanaugh, Book Review, Fixing Statutory Interpretation, 129 Harv L Rev 2118, 2121, 2144–56, 2159–62 (2016) (arguing for fewer canons).
• 34See, for example, Scalia and Garner, Reading Law at 252 (cited in note 16) (“[T]he body of the law should make sense, and . . . it is the responsibility of the courts, within the permissible meanings of the text, to make it so.”). Strict textualists are not the only ones to take this view. See, for example, Lozman v City of Riviera Beach, Florida, 133 S Ct 735, 744 (2013) (Breyer) (“Consistency of interpretation of related state and federal laws is a virtue in that it helps to create simplicity. . . . [T]hat consideration here supports our conclusion.”).
• 35See Gluck and Bressman, 65 Stan L Rev at 905, 909 (cited in note 7) (noting the discrepancy between the “model of the judge as a ‘faithful agent’ of the legislature” and most judges’ acknowledgement that “many of the canons on which they rely are ‘fictions’”).
• 36See Gluck and Posner, Statutory Interpretation on the Bench at *18 (cited in note 13).
• 37See notes 43 and 77.
• 38See Bressman and Gluck, 66 Stan L Rev at 763–65, 782 (cited in note 7).
• 39See id at 763–64; Philip G. Joyce, The Congressional Budget Office: Honest Numbers, Power, and Policymaking 224 (Georgetown 2011); Walter J. Oleszek, Congressional Procedures and the Policy Process 72 (CQ 9th ed 2014).
• 40See generally The Statutory PAYGO Process for Budget Enforcement: 1991-2002 (Congressional Research Service, Dec 30, 2009), archived at http://perma.cc/KC9W-44W6; The Statutory Pay-as-You-Go Act of 2010: A Description (Office of Management and Budget), archived at http://perma.cc/G287-685R. Nonpartisan does not mean noncontroversial. Some disagree with the philosophy underlying the economic models that CBO utilizes. Bruce Bartlett, The Trouble with Dynamic Scoring, 139 Tax Notes 309, 311 (2013) (advocating different economic assumptions for CBO models). But this does not undermine the centrality of CBO’s influence, and, in fact, the attention paid to its modeling approaches evinces its importance.
• 41See, for example, David M. Herszenhorn and Robert Pear, Democrats See Hope on Health Bill (NY Times, Dec 9, 2009), online at http://www.nytimes.com/2009/12/10/health/policy/10healthbill.html (visited Oct 21, 2016) (Perma archive unavailable) (stating that the Affordable Care Act’s fate “hinged on the results” of a CBO analysis); Dylan Matthews, No, the CBO Report Doesn’t Mean Immigration Brings Down Wages (Wash Post, June 19, 2013), archived at http://perma.cc/4JZM-AVTU (discussing the CBO reports and related academic papers as “being taken seriously by members of the Obama administration’s economic team as they consider the effects of immigration reform”).
• 42See, for example, ABKCO Music, Inc v LaVere, 217 F3d 684, 690 (9th Cir 2000) (considering a CBO report as well as the Congressional Record in determining the extent to which an amendment changed the Copyright Act); Hendricks v Bowen, 847 F2d 1255, 1258–59 (7th Cir 1988) (referring to a CBO report in determining that the plaintiff should not receive attorney’s fees from the Government under the relevant legislation); Berman v Schweiker, 713 F2d 1290, 1298–99 (7th Cir 1983) (referring to a CBO report in determining whether the relevant act applied retroactively with regard to awarding attorney’s fees). There appears to have been one mention by the Supreme Court. See Heckler v Turner, 470 US 184, 206–07 (1985) (presuming, in interpreting a statute, that Congress relied on a CBO description of the legislation’s impact).
• 43Abbe Gluck, The “CBO Canon” and the Debate over Tax Credits on Federally Operated Health Insurance Exchanges (Balkinization, July 10, 2012), archived at http://perma.cc/VM2X-ATU3.
• 44Patient Protection and Affordable Care Act, Pub L No 111-148, 124 Stat 119 (2010).
• 45135 S Ct 2480 (2015).
• 46Timothy Jost, Tax Credits in Federally Facilitated Exchanges Are Consistent with the Affordable Care Act’s Language and History (Health Affairs Blog, July 18, 2012), archived at http://perma.cc/3QZJ-7ZBK; Jonathan Cohn, The Legal Crusade to Undermine Obamacare—and Rewrite History (New Republic, Dec 4, 2012), online at http://newrepublic
  .com/article/110770/legal-challenge-obamacare-insurance-exchanges-full-holes (visited Oct 21, 2016) (Perma archive unavailable).
• 47See, for example, Douglas W. Elmendorf, Director of the Congressional Budget Office, Letter to Representative Darrell E. Issa, Chairman of the Committee on Oversight and Government Reform *1 (Dec 6, 2012), archived at http://perma.cc/AX56-LX4J.
• 48See, for example, Defendants’ Memorandum in Opposition to Motion for Preliminary Injunction, Halbig v Sebelius, Civil Action No 13-00623, *21 (DDC filed Sept 27, 2013).
• 49King v Sebelius, 997 F Supp 2d 415, 430–31 (ED Va 2014); Halbig v Sebelius, 27 F Supp 3d 1, 24–25 (DDC 2014); Halbig v Burwell, 758 F3d 390, 407–08 (DC Cir 2014); King v Burwell, 759 F3d 358, 374 (4th Cir 2014).
• 50See, for example, Defendants’ Memorandum in Opposition to Plaintiff’s Motion for Summary Judgment, United States House of Representatives v Burwell, Civil Action No 14-01967, *3–4, 12, 18 (DDC filed Jan 15, 2016) (available on Westlaw at 2016 WL 452190); Brief of Appellants, Ohio v United States, Civil Action No 16-3093, *12, 51 (6th Cir filed Apr 4, 2016). But see Ohio v United States, 154 F Supp 3d 621, 642 (SD Ohio 2016) (claiming that “the CBO does not and cannot authoritatively interpret federal statutes”).
• 51See Yates v United States, 135 S Ct 1074, 1096 (2015) (Kagan dissenting); Loving v Internal Revenue Service, 742 F3d 1013, 1019 (DC Cir 2014) (Kavanaugh); Kavanaugh, Book Review, 129 Harv L Rev at 2161–62 (cited in note 33).
• 52See Council for Urological Interests v Burwell, 790 F3d 212, 233 (DC Cir 2015) (Henderson dissenting in part).
• 53King, 135 S Ct at 2485–87 (Roberts). See also Abbe R. Gluck, Imperfect Statutes, Imperfect Courts: Understanding Congress’s Plan in the Era of Unorthodox Lawmaking, 129 Harv L Rev 62, 87–88 (2015).
• 54See The Scalia Lecture: A Dialogue with Justice Elena Kagan on the Reading of Statutes 8:28 (Harvard Law School, Nov 18, 2015), online at http://today.law.harvard.edu/in-scalia-lecture-kagan-discusses-statutory-interpretation (visited Oct 21, 2016) (Perma archive unavailable) (recording a conversation between Justice Kagan and Professor Manning in which Kagan describes the predominance of textualism); Kavanaugh, Book Review, 129 Harv L Rev at 2118 (cited in note 33).
• 55See Abbe R. Gluck, The Ripple Effect of “Leg-Reg” on the Study of Legislation & Administrative Law in the Law School Curriculum, 65 J Legal Educ 121, 123–24, 143–44 (2015).
• 56Kavanaugh, Book Review, 129 Harv L Rev at 2121–22 (cited in note 33).
• 57Evan Lee, Opinion Analysis: Battle of Statutory Interpretation Canons Ends in Defeat for Convicted Sex Offender (SCOTUSblog, Mar 1, 2016), archived at http://perma.cc/2EJY-7NXN.
• 58136 S Ct 958 (2016).
• 59Pub L No 104-208, 110 Stat 3009-26, codified as amended in various sections of Title 18.
• 60Lockhart, 136 S Ct at 961, quoting 18 USC § 2252(b)(2).
• 61Lockhart, 136 S Ct at 962–63, citing Barnhart v Thomas, 540 US 20, 26 (2003), and Federal Trade Commission v Mandel Brothers, Inc, 359 US 385, 389 n 4 (1959) (applying the rule and citing earlier cases as examples of its application).
• 62Lockhart, 136 S Ct at 970 (Kagan dissenting) (quotation marks omitted).
• 63Id at 962–63 (Sotomayor), citing Scalia and Garner, Reading Law at 144 (cited in note 16); Lockhart, 136 S Ct at 970 (Kagan dissenting), citing Scalia and Garner, Reading Law at 147 (cited in note 16).
• 64Transcript of Oral Argument, Lockhart v United States, No 14-8358, *32–33 (US filed Nov 3, 2015) (available on Westlaw at 2015 WL 7188394).
• 65Id at *39–41 (emphasis added).
• 66Easterbrook, 84 U Chi L Rev at 85–90 (cited in note 9).
• 67See Positive Law Codification (Office of the Law Revision Counsel), archived at http://perma.cc/B45N-HHFM (explaining the role of the Law Revision Counsel in making “technical changes . . . in [ ] wording and organization to integrate” new provisions into the US Code after new laws have been enacted). In Lockhart, however, it appears the relevant serial comma was added before the vote. See Child Pornography Prevention Act of 1995, S Rep No 104-358, 104th Cong, 2d Sess 4, 41–42 (1996).
• 68See Bressman and Gluck, 66 Stan L Rev at 742 (cited in note 7).
• 69See Positive Law Codification (cited in note 67) (describing the cleanup process that occurs prior to codification).
• 70Lockhart, 136 S Ct at 974 (Kagan dissenting).
• 71S Rep No 104-358 at 24–25 (cited in note 67) (reproducing CBO’s cost estimate letter).
• 72Id at 24.
• 73Bressman and Gluck, 66 Stan L Rev at 764 (cited in note 7).
• 74See, for example, Herszenhorn and Pear, Democrats See Hope on Health Bill (cited in note 41) (describing the CBO score as crucial to debates around health-care reform); Kimberly J. Morgan and Andrea Louise Campbell, The Delegated Welfare State: Medicare, Markets, and the Governance of Social Policy 122–23 (Oxford 2011) (noting that the structure of the 2003 Medicare prescription drug benefit was designed in response to CBO scoring implications); Kate Tummarello, “It Just Drives You Crazy”: Why Inflexible Budget Rules Are Keeping Lawmakers from Selling Billions of Dollars Worth of Government Airwaves (Politico, Oct 7, 2015), archived at http://perma.cc/2NXP-K6R9 (describing CBO’s influence over FCC legislation and characterizing “a bad score from the CBO” as a “looming threat” in a “budget-conscious” Congress).
• 75See, for example, Department of Health and Human Services, Centers for Medicare & Medicaid Services, Medicare Program; Hospital Inpatient Prospective Payment Systems for Acute Care Hospitals and the Long-Term Care Hospital Prospective Payment System and FY 2012 Rates; Hospitals’ FTE Resident Caps for Graduate Medical Education Payment, 76 Fed Reg 51476, 51664 (2011), amending 42 CFR §§ 412, 413, 476 (summarizing public comments that expressed criticism of the proposed regulations on the grounds that they were inconsistent with CBO scoring of the original statute); Bureau of Land Management, Rental Fees, Mining Claim Recordation, and Assessment Work, 58 Fed Reg 38186, 38191 (1993), amending 43 CFR §§ 3730, 3820, 3860 (describing a public comment that referred to CBO scoring as evidence of congressional intent regarding an exemption for small miners).
• 76It would not have been useful, for instance, if adding adult crimes would not have significantly changed the score.
• 77See, for example, Jon Perr, CBO: Obamacare Subsidies Apply to Both State and Federal Exchanges (Daily Kos, July 25, 2014), archived at http://perma.cc/FLL2-KG2T (describing widespread deliberation about CBO scoring among members of Congress from both parties).
• 78See, for example, Gay v Sullivan, 966 F2d 1124, 1129 (7th Cir 1992) (calling the score “persuasive evidence of congressional intent in this case” because retroactive application of the statute would cost a lot more); Nunes-Correia v Haig, 543 F Supp 812, 815–16 (DDC 1982) (citing the CBO score, which accounted for retroactive application, in holding that the statute at issue applied retroactively).
• 79See, for example, Robert A. Katzmann, Judging Statutes 86–91 (Oxford 2014); Richard A. Posner, Divergent Paths: The Academy and the Judiciary 108 n 44 (Harvard 2016).
• 80548 US 291 (2006).
• 81Pub L No 101-476, 104 Stat 1141 (1990), codified as amended at 20 USC § 1400
  et seq.
• 82Arlington, 548 US at 293, quoting 20 USC § 1415(i)(3)(B).
• 83Handicapped Children’s Protection Act of 1985, HR Rep No 99-296, 99th Cong, 1st Sess 10–11 (1985); Handicapped Children’s Protection Act of 1985, S Rep No 99-112, 99th Cong, 1st Sess 3–4 (1985), reprinted in 1986 USCCAN 1798, 1800–01.
• 84527 US 471 (1999).
• 85Pub L No 101-336, 104 Stat 327.
• 86Sutton, 527 US at 484–87.
• 87See Gluck, 129 Harv L Rev at 107 (cited in note 53); John F. Manning, Textualism and Legislative Intent, 91 Va L Rev 419, 430 n 34, 444 n 84 (2005).
• 88Gluck, 129 Harv L Rev at 94 (cited in note 53).
• 89Chevron U.S.A. Inc v Natural Resources Defense Council, Inc, 467 US 837, 842–43, 865–66 (1984).
• 90See City of Arlington, Texas v Federal Communications Commission, 133 S Ct 1863, 1879 (2013) (Roberts dissenting); Perez v Mortgage Bankers Association, 135 S Ct 1199, 1208 n 4 (2015) (Sotomayor); Texas Department of Housing and Community Affairs v Inclusive Communities Project, Inc, 135 S Ct 2507, 2526–29 (2015) (Thomas dissenting).
• 91King, 135 S Ct at 2485–87 (providing three pages of detail).
• 92Id at 2492.
• 93Id.
• 94See Gluck and Bressman, 65 Stan L Rev at 933–34, 938–39, 954, 964–90 (cited in note 7).
• 95See Katzmann, Judging Statutes at 53–54 (cited in note 79) (suggesting that canons based on “shared conventions” could be crafted). See also Bressman and Gluck, 66 Stan L Rev at 789–90 (cited in note 7) (explaining that “loudly communicated principles of interpretation” may be necessary to provide “some common language . . . for the branches to interact”).
• 96Indeed, Judge Easterbrook made precisely this objection in Ameritech Corp v McCann, 403 F3d 908, 913 (7th Cir 2005) (“[The CBO’s] view—on which Congress did not vote, and the President did not sign—cannot alter the meaning of enacted statutes.”). The more relevant concern is that entrenching a CBO “canon” would increase what some believe is the already disproportionate influence that CBO has on the process. This is the same kind of argument as the notion that looking to legislative history inappropriately strengthens the role of congressional staff. The CBO Canon has some advantages over legislative history, including that CBO is nonpartisan and that the score does not share legislative history’s problem of courts having to guess whether anyone knew of it.
• 97See Gluck and Bressman, 65 Stan L Rev at 919–21 (cited in note 7).
• 98Id at 919–20.
• 99See generally Barbara Sinclair, Unorthodox Lawmaking: New Legislative Processes in the U.S. Congress (CQ 4th ed 2012). See also generally Abbe R. Gluck, Anne Joseph O’Connell, and Rosa Po, Unorthodox Lawmaking, Unorthodox Rulemaking, 115 Colum L Rev 1789 (2015) (extending the concept to the theories and doctrines of legislation and administrative law).
• 100Bressman and Gluck, 66 Stan L Rev at 763 (cited in note 7).
• 101Id at 747–48.
• 102Id at 739.
• 103Pub L No 111-139, 124 Stat 8, codified at 2 USC § 931 et seq.
• 104Statutory Pay-as-You-Go-Act § 4, 124 Stat at 9–15, codified at 2 USC § 933. See also The Statutory Pay-As-You-Go Act of 2010 (cited in note 40).
• 105US Const Art I, § 5.
• 106Gluck and Bressman, 65 Stan L Rev at 985–86, 995 (cited in note 7).
• 107See generally id.
• 108See id at 944.
• 109See Gluck, 54 Wm & Mary L Rev at 760–70 (cited in note 5).
• 110Bressman and Gluck, 66 Stan L Rev at 747–49 (cited in note 7).
• 111Id at 746–47.
• 112See Gluck and Bressman, 65 Stan L Rev at 936–37 (cited in note 7).
• 113Id at 1007.
• 114Id at 909 n 9, 930.
• 115See Gluck, O’Connell, and Po, 115 Colum L Rev at 1794–95, 1800–03 (cited in note 99).
• 116Gluck and Bressman, 65 Stan L Rev at 936 n 104 (cited in note 7).
• 117See, for example, Rule XXI(2)(a)(2), Constitution, Jefferson’s Manual, and the Rules of the House of Representatives, HR Doc 113-181, 113th Cong, 2d Sess (GPO, May 21, 2015), archived at http://perma.cc/2239-T47M, as amended by H Res 5, 114th Cong, 1st Sess, in 161 Cong Rec H7 (daily ed Jan 6, 2015); Rule XVI(1), Senate Manual Containing the Standing Rules, Orders, Laws, and Resolutions Affecting the Business of the United States Senate, S Doc 113-1, 113th Cong, 1st Sess (GPO 2011), archived at http://perma.cc/XXT8-TPRQ, as amended by S Res 16, 113th Cong, 1st Sess, in 159 Cong Rec S270 (daily ed Jan 24, 2013).
• 118See Gluck, O’Connell, and Po, 115 Colum L Rev at 1800 (cited in note 99) (“[I]n the first year of the 112th Congress, fewer than 10% of enacted laws went through the ‘textbook’ legislative process.”).
• 119See Gluck and Bressman, 65 Stan L Rev at 1006–07 (cited in note 7).
• 120Id at 980.
• 121See Bressman and Gluck, 66 Stan L Rev at 762–63 (cited in note 7).
• 122Pub L No 107-40, 115 Stat 224 (2001).
• 123Gluck and Bressman, 65 Stan L Rev at 1007–08 (cited in note 7). See also Bressman and Gluck, 66 Stan L Rev at 781 (cited in note 7) (illustrating an application of the convention).
• 124For examples of these conventions in use, see 29 USC § 3141(b)(3)(A)(iv)(I)–(II), (b)(3)(A)(vi)–(vii), (d)(1), (d)(5), (e)(2) (describing the Secretary of Labor’s obligations “in conjunction with the Secretary of Education” as well as state entities in implementing performance requirements in a work training program); 15 USC § 4623(a) (permitting Department of Energy national laboratories to participate in research and development projects “in conjunction with the Department of Defense” or other research institutions); 42 USC § 294q(d)(5)(B) (requiring the National Health Care Workforce Commission to make recommendations for grant recipients “in collaboration with the Department of Labor and in coordination with the Department of Education and other relevant Federal agencies”); 42 USC § 285a-10(b)(1) (directing the relevant Department of Health and Human Services agency to establish an education program “in collaboration with the Director of the [National Institutes of Health]”); 42 USC § 300hh(b) (requiring the Secretary of Health and Human Services to establish an interagency agreement for managing a public health emergency “in collaboration with” a host of other federal agencies).
• 125See Gluck and Bressman, 65 Stan L Rev at 931–32, 938 (cited in note 7) (illustrating that fewer than 20 percent of respondents knew of or used the assumptions of the dictionary presumption and quoting a respondent claiming that “no one uses a freaking dictionary”).
• 126See James J. Brudney and Lawrence Baum, Oasis or Mirage: The Supreme Court’s Thirst for Dictionaries in the Rehnquist and Roberts Eras, 55 Wm & Mary L Rev 483, 557–58 (2013).
• 127See Zoe Kleinman, How the Internet Is Changing Language (BBC News, Aug 16, 2010), archived at http://perma.cc/5V8X-RH9H.
• 128See Bruesewitz v Wyeth LLC, 562 US 223, 235 n 43 (2011).
• 129See United States v Gonzales, 520 US 1, 5 (1997); Ali v Federal Bureau of Prisons, 552 US 214, 218–19 (2008).
• 130See Gluck and Bressman, 65 Stan L Rev at 977–78 (cited in note 7) (indicating that most of the survey respondents considered committee and conference reports more reliable than most other kinds of legislative history); Bressman and Gluck, 66 Stan L Rev at 757 (cited in note 7) (relating that many respondents dismissed floor statements by party leaders as “spin”).
• 131See, for example, In re Silicon Graphics Inc Securities Litigation, 183 F3d 970, 977 (9th Cir 1999) (“[W]e first look to the conference report because, apart from the statute itself, it is the most reliable evidence of congressional intent.”). See also Anita S. Krishnakumar, Dueling Canons, 65 Duke L J 909, 991–92 (2016) (“Committee reports . . . are widely considered to be the most authoritative form of legislative history.”).
• 132See Patricia M. Wald, Some Observations on the Use of Legislative History in the 1981 Supreme Court Term, 68 Iowa L Rev 195, 202 (1983).
• 133Gluck and Bressman, 65 Stan L Rev at 986–87 (cited in note 7).
• 134Id at 977–78.
• 135Id.
• 136See notes 67 and 69 and accompanying text. I am working on an article on this topic with Jesse Cross.
• 137For an example of a litigant making this argument, however, see Brief for the Respondents, Sebelius v Auburn Regional Medical Center, No 11-1231, *25 n 7 (US filed Oct 16, 2012) (available on Westlaw at 2012 WL 5078760) (arguing that a section title emphasized by an amicus “appears only in the margin notes of the Statutes at Large, not in the Public Laws,” and so was “made by codifiers not Congress”).
• 138For new work prompted by the study, see generally Jarrod Shobe, Intertemporal Statutory Interpretation and the Evolution of Legislative Drafting, 114 Colum L Rev 807 (2014). An earlier mention of the Legislative Counsels can be found in Victoria F. Nourse and Jane S. Schacter, The Politics of Legislative Drafting: A Congressional Case Study, 77 NYU L Rev 575, 588–89 (2002), but that article—based on a qualitative study of only the Senate Judiciary Committee—stated that the Office of Legislative Counsel had no substantial role, a claim seriously undermined by our study.
• 139See Gluck and Bressman, 65 Stan L Rev at 972–73 (cited in note 7).
• 140See Katzmann, Judging Statutes at 53 (cited in note 79) (suggesting the possibility of a set of canons based on shared conventions).
• 14161 Stat 633 (1947), codified as amended at 1 USC § 1 et seq.
• 1421 USC § 1.
• 143It applies “unless the context indicates otherwise.” 1 USC § 1.
• 144Scalia and Garner, Reading Law at 245 (cited in note 16).
• 145Bressman and Gluck, 66 Stan L Rev at 751 (cited in note 7), citing Carr v United States, 560 US 438, 463 (2010) (Alito dissenting), United States v O’Brien, 560 US 218, 233–34 (2010), and Koons Buick Pontiac GMC, Inc v Nigh, 543 US 50, 60–61 (2004).
• 146Employee Retirement Income Security Act of 1974, Pub L No 93-406, 88 Stat 829, codified as amended at 29 USC § 1009 et seq.
• 14729 USC § 1144(b)(2)(A).
• 148See Gluck, 54 Wm & Mary L Rev at 801–04 (cited in note 5).
• 149King, 135 S Ct at 2483.
• 150See John F. Manning, Federalism and the Generality Problem in Constitutional Interpretation, 122 Harv L Rev 2003, 2006–08, 2048–52 (2009) (distinguishing policy presumptions from text-based ones).
• 151See Jonathan R. Macey, Promoting Public-Regarding Legislation through Statutory Interpretation: An Interest Group Model, 86 Colum L Rev 223, 264–66 (1986) (arguing that the canons of statutory construction “enhance, rather than limit,” judicial discretion, even though they can be ranked and ordered on the basis of legitimacy).
• 152Karl N. Llewellyn, Remarks on the Theory of Appellate Decision and the Rules or Canons about How Statutes Are to Be Construed, 3 Vand L Rev 395, 398–99 (1950).