Rebuilding the Federal Circuit Courts

Articles

Merritt E. McAlister

ABSTRACT—The conversation about Supreme Court reform—as important as it is—has obscured another, equally important conversation: the need for lower federal court reform. The U.S. Courts of Appeals have not seen their ranks grow in over three decades. Even then, those additions were stopgap measures built on an appellate triage system that had outsourced much of its work to nonjudicial decision-makers (central judicial staff and law clerks). Those changes born of necessity have now become core features of the federal appellate system, which distributes judicial resources—including oral argument and judicial scrutiny—to a select few. This Article begins to reimagine the courts in a time of surplus, not scarcity, and it offers a comprehensive framework for identifying why, where, and how to add judges to the federal appellate courts.

The existing distribution of judicial resources has created a problem: the courts have relied too much on procedural shortcuts that permit them to be highly selective in deciding which cases receive careful judicial attention. Ultimately, the distribution of appellate resources has been uneven across the country in ways that have a disparate impact on communities of color and poor communities.

To redress these systemic deficits, this Article urges Congress to engage in lower court reform by adding judges to the most underresourced federal appellate courts. In so doing, it offers a framework for identifying underresourced courts that would ensure periodic and consistent congressional review of judicial needs. These measures consider indicia of overdelegation, population growth, and caseload demands—all while incentivizing courts not to overrely on procedural shortcuts in the first place.

AUTHOR—Associate Professor of Law, University of Florida Levin College of Law. This work benefited from the thoughtful engagement of participants at faculty workshops at the University of Florida Levin College of Law, Washington University in St. Louis School of Law, and Northwestern the 2020 Duke Law Judicial Pritzker School of Law; and at Administration/Judicial Process Roundtable, the 2021 Southeastern Association of Law Schools Annual Conference, the 2021 Civil Procedure

Virtual Roundtable Workshop, and the 2021 Richmond Junior Faculty Forum. I am grateful to Seth Endo, Dan Epps, Andrew Hammond, Jason Nance, Rigel Oliveri, Judith Resnik, and Xiao Wang who read and improved earlier drafts. I thank Zach Clopton, Myriam Gilles, Marin Levy, Maya Sen, and Steve Vladeck for conversations that deepened and refined my approach to court reform. I appreciate the diligent and careful work of the editors of the Northwestern University Law Review for improving this work enormously. For excellent research assistance, I thank Jordan Grana. All mistakes, errors, and omissions that remain are my own.

INTRODUCTION .......................................................................................................... 1138 I. TWO TIERS OF FEDERAL APPELLATE JUSTICE ..................................................... 1147 A. How the U.S. Courts of Appeals Distribute Article III Resources ............ 1148 B. Criticisms and Suggestions for Reform ..................................................... 1161 II. WHY THE U.S. COURTS OF APPEALS NEED REFORM ........................................... 1163 A. Overuse of Shortcuts ................................................................................. 1165 B. Disparate Impact of the Triage System Across Geographic Circuits........ 1175

APPEALS ............................................................................................................ 1195 A. Judicial Objections to Adding Judges—and Why Congress Shouldn’t

Listen ........................................................................................................ 1196

B. How Courts and Congress Currently Think About Judicial Need—and

What’s Wrong with It ................................................................................ 1201 C. How to Add Judges to the Courts of Appeals ............................................ 1209 D. Why Adding Authorized Judgeships Matters ............................................ 1216 CONCLUSION ............................................................................................................. 1219 APPENDIX A: CIRCUIT-COMPOSITION DATA ............................................................... 1220 APPENDIX B: AUTHORIZED JUDGESHIPS OVER TIME ................................................... 1224

INTRODUCTION

Debates about Supreme Court reform1—as needed as they are—have overshadowed what is happening (and has been happening for decades) in

the lower federal courts.2 The Supreme Court receives a disproportionate amount of attention in modern court-reform debates.3 It is the rare federal case that makes its way to the Supreme Court; most cases draw their last breath in the U.S. Courts of Appeals. And most of these appeals are both utterly ordinary—in the sense that they do not involve the kinds of novel, contentious questions that make Supreme Court review likely—and also profoundly important to the individuals who litigate them.4 These are direct criminal appeals, prisoner and nonprisoner civil rights claims, habeas corpus petitions, Social Security disability appeals, and immigration and asylum appeals.5 Many of these appeals involve pro se appellants, as almost half of all appeals do.6 In most circuits, these cases—whether lawyered or not—are the great bulk of the work before each court every year. The courts of appeals, moreover, defer to lower court decisions in these ordinary appeals

in which oral argument is rare, decisions are poorly reasoned and recycled, and judicial attention is scarce.7

Within each of the U.S. Courts of Appeals, there are (at least8) two tiers of appellate procedure.9 The first tier is the traditional one, with an Article III judge reading briefs, hearing oral argument, and working on an opinion with her law clerks. The second tier, however, involves mostly nonjudicial decision-makers. Every circuit employs central staff attorneys who process, handle, and (essentially) decide large numbers of appeals with varying degrees of judicial oversight.10 These cases rarely receive oral argument, and they almost always end in nonprecedential or unpublished decisions.11 Certain classes of appeals—especially those involving pro se litigants—go through this second-tier process by default.12 In many circuits, the central staff themselves are responsible for “screening” cases to determine whether an appeal should receive first-tier or second-tier attention.13

The federal appellate courts’ distribution of resources through the tiering of judicial time and attention is a “triage” system supposedly born of necessity. Beginning in the 1960s, the federal appellate courts unilaterally created shortcut procedures to handle what they perceived as a flood of new

appeals outpacing the addition of new judges.14 Much of that flood involved work of a particular kind: these were filings from prisoners, criminal defendants, and civil rights claimants seeking to vindicate new statutory rights and newly articulated constitutional rights.15 This created what some viewed as a “crisis of volume,”16 but it was a crisis involving low-value or “peculiarly undeserving”17 and less meritorious18 appeals. As such, Article III judges decided this work needed less of their time and attention. The courts therefore created a second-tier appellate process—staffed with nonjudges—to help them handle the crisis.

These trends happened across all federal courts—and not just at the appellate level. As Professor Judith Resnik observed in 1984, federal courts innovated (lesser) procedures to handle low-value district court cases such as habeas corpus petitions, civil rights claims, and employment law matters.19 Congress added magistrate judges to handle increased volume instead of growing the ranks of Article III judges to carry that work.20 Professor Resnik argued that these procedural choices reflected value judgments that “embod[y] deeply held, albeit often unarticulated, views of human relationships, of the importance and difficulty of passing judgments on individuals’ conduct, and of the place of government in citizens’ lives.”21 Her observations hold true in other spaces within the federal judiciary; as this Article seeks to demonstrate, the “value-expressive function[]”22 of the shortcut procedures I detail here have now become firmly entrenched at the federal appellate courts.

In whatever way it’s sliced or diced, the triage system is fundamentally unequal as a matter of distributive justice.23 Perhaps because the courts have been somewhat shameless in defending their system,24 I am certainly not the first to call attention to its injustice. The late scholar Penelope Pether observed more than a decade ago: “[T]he appeals court processes themselves seem to place almost no priority on ‘justice.’”25 Not long thereafter, Professors William Richman and William Reynolds—the most prolific scholars on what I describe here as the second tier of the federal appellate the system—wrote a capstone entitled Injustice on Appeal about transformation they had observed over their careers.26

Nevertheless, the distribution of appellate resources has become firmly entrenched—so much so, in fact, that the tiers of appellate review appear impervious to the steady reduction in appellate courts’ caseloads since the early 2000s.27 To the extent that triage procedures are those deployed to tackle emergencies, one might have expected courts to rely less on shortcuts when the emergency they faced was no more. That has not happened. Courts, it appears, have not recalibrated in response to lessening demands on their time as their caseload volume has receded. On average, they issue more unpublished decisions today and hear fewer oral arguments than they did fifteen years ago when they saw, nationally, nearly 20,000 more appeals.28 The triage system has taken on a life of its own.

Let me make clear from the outset: I don’t argue that courts should abandon all triage. I don’t believe that every appeal demands equal judicial attention. Frivolous appeals, for example, certainly do not. But I do think courts have developed a dependence on the shortcuts they developed for a variety of human and institutional reasons—and that these entrenched habits will likely be difficult to change. What may have once been a much-needed efficiency has become an overused convenience that threatens to undermine core distributive justice and day-in-court values.29 There’s also reason to worry that judicial inattention may cause particularly vulnerable litigants to lose where more judicial attention (especially through appointed counsel and oral argument) might detect error (or identify needed lawmaking).30

Determining the optimal level of triage is a very difficult task. It is a question for which I do not have an answer. What I do argue, however, is that under the existing scheme, we can’t ask very busy courts to do more (e.g., write more precedential opinions and hold more oral argument) without more help. The data discussed here indicate that courts will do more with more judicial help. Not every court relies on shortcut procedures to the same extent—as will be abundantly clear by the end of Part II—and one feature of those courts that rely less on shortcuts is that they generally have fewer appeals per judge. That is reason to believe that service-oriented courts might more generously distribute their procedural goods (e.g., more oral argument and more reasoned decision-making) when they have more judicial help to do that work.

Central to my critique is the inexplicably uneven distribution of federal judicial attention across the circuits. Courts utilize oral argument and publication (or, really, unpublication) in different ways, despite sharing similar formal rules.31 These are disparities that our hierarchical judicial system cannot fix—these are not the kinds of circuit splits that receive

Supreme Court review.32 But these differences across the circuits have both a substantive effect—because there is a correlation between oral argument and favorable results on appeal—and also an unintended consequence: the existing two-tier system appears to disparately impact communities of color and poor communities. The latter—the risk of disparate impact—should not be terribly surprising given how and where these procedures began. Federal courts in the Deep South (the old Fifth Circuit and the Fourth Circuit) innovated procedural shortcuts, and they did so in response to concerns over a growing docket of habeas corpus and prisoner civil rights claims.33 And even when the Second Circuit began to loosen its strong tradition of holding oral argument, it innovated a devoted nonargument track to handle a bevy of immigration appeals in the wake of post-September 11 immigration reforms.34 Second-tier procedures and their impact on vulnerable litigants have always been linked. And they still are.

To the extent the triage process developed because demand (cases) outpaced supply (judges), one obvious—albeit very difficult—solution to the court’s workload problem would have been to add more judges. Dating back to the 1990s, however, judges have been divided over whether the courts should grow substantially or not.35 Increasing the number of authorized judgeships raises a number of challenges—especially for the manageability of courts, including the ease of en banc review. This Article charts a middle path—resisting both the urge to maintain the status quo and to double the size of the federal judiciary. It proposes a framework for periodic, predictable congressional review of judicial need based on a combination of both endogenous and exogenous factors that would empower courts to minimize disparities and reduce overreliance on nonjudicial decisionmakers.

To be sure, any judicial expansion (no matter how modest) raises serious political challenges that may be so fraught as to make this effort more pipedream than reality.36 Given its political difficulty, some scholars have turned to creative and more indirect alternatives. For example, Professor Marin Levy’s recent work seeks to make the possibility of senior status for federal appellate judges “more attractive, with the ultimate goal of creating vacancies to provide more judge power to the federal judiciary.”37 That proposal has the substantial advantage of bypassing the sticky wicket of Congress. But it offers only a partial solution to larger systemic problems that individual or even collective judicial retirements alone cannot fix. Professor Levy and I agree that the courts need to be “bigger.”38 I hope to find a way to bring Congress into that conversation while reducing the partisan concerns over judicial expansion. Although I do not wish to minimize the real political challenges of any expansion proposal, those challenges should not stop us from assessing judicial need. It should also not stop us for making such recommendations, which flows through the judiciary itself before reaching Congress. When the courts do not ask Congress for more judges, it may be easier to maintain the status quo and avoid the political thicket of court expansion, however needed and however modest.39

the current pathway

from critiquing

And the courts, indeed, have largely defended the status quo. The U.S. Courts of Appeals have not asked Congress for any meaningful number of additional judges over the last decade. Indeed, recently the Ninth Circuit asked Congress to create only an exceedingly modest number of additional

judgeships (two, down from a request for five in 2019).40 Busier and equally busy courts—the Fifth Circuit and the Eleventh Circuit, in particular—have made no requests for additional judges; indeed, the Eleventh Circuit has never done so.41 Today, the Eleventh Circuit sees about 500 more appeals per year than the entire old Fifth Circuit saw in the year before Congress split the Fifth Circuit into the Fifth and the Eleventh Circuits.42 At that time, the “jumbo” old Fifth Circuit had twenty-six judges.43 Congress gave the Eleventh Circuit twelve of those judges upon its creation, and at twelve judges the Eleventh Circuit has remained, even though it now handles more work than twenty-six judges together handled forty years ago.44

Why? Because that’s what a majority of the Eleventh Circuit’s judges have seemingly wanted. Current frameworks for evaluating judicial need begin with the courts themselves. They depend on a majority of each court’s active judges agreeing to ask Congress for more judicial help (and then on Congress having the political will to act on that recommendation).45 There has never been a majority of active judges on the Eleventh Circuit with the desire to make that request. For example, the Eleventh Circuit’s then-chief judge told Congress in 2002 during a hearing on judicial vacancies: “Although the Court leads the nation in number of cases per judge in most categories, twelve judges are enough.”46 That year, the Eleventh Circuit had

7,431 filings—three thousand more filings than the old Fifth Circuit with its twenty-six judges had in 1980.47 But “twelve judges are enough.” End of discussion.

To be sure, we owe some deference to the courts in determining how best to distribute Article III resources. No doubt, the Eleventh Circuit’s view of its own judicial need may have been shaped by having once been a jumbo court with twenty-six judges—a situation that former Fifth Circuit judges found unworkable.48 But deference to judicial self-perceptions of need cannot be both the beginning and the end of the discussion. At the very least, where Congress defers to judicial recommendations for adding authorized judgeships, the process for developing those recommendations should be open and transparent, and it should take into account dissenting voices and the possibility of judicial entrenchment.49

This Article will proceed in three Parts. Part I details the federal appellate triage system, including its constituent parts, existing criticisms, and suggestions for reform. Part II describes the toll that the tiers of federal appellate justice have taken on the federal system, examining the ways in which (at least some) courts have taken procedural shortcuts too far. In particular, it argues that maintenance of these systems may have a disparate impact based on race and, to a lesser extent, class. Part III offers a framework for Congress to distribute Article III resources by reimagining how we evaluate judicial need and resource allocation across the federal courts of appeals. It explains why and how some of the existing models fall short, and it offers alternatives that both rely less on majority rule in making judgeship recommendations and incentivize courts to avoid overreliance on nonjudicial decision-makers.

I. TWO TIERS OF FEDERAL APPELLATE JUSTICE

The modern federal appellate triage system generally divides the pathway for federal appeals in two: an oral-argument track to which Article III judges devote the most attention and a nonargument track to which nonjudges devote the most attention. These tracks, which I call “tiers”

because it more accurately reflects how courts prioritize (and not just separate) appeals, largely began in the late 1960s and early 1970s, when docket demands forced changes in internal court procedures.50 This Part explains how the tiers came to be, describes their constituent parts, and considers existing suggestions for reform.

A. How the U.S. Courts of Appeals Distribute Article III Resources

How the U.S. Courts of Appeals handle appeals today is not how it has always been done.51 Once upon a time, federal appellate judges did not have law clerks or central judicial staff. Instead, they read the briefs themselves, heard cases argued orally, and issued decisions they wrote themselves.52 With some nostalgia, scholars have described these as the “Learned Hand” days, when, in a land temporally far, far away, judges did the judging themselves with little to no help.53

The Learned Hand days are long gone. Between the 1960s and the 1980s, the federal appellate courts instituted a number of reforms—all court initiated54—to revise their internal decisional processes. Ostensibly, these reforms responded to what has been termed a “crisis of volume,”55 where supply (judges) did not keep up with demand (appeals). Indeed, during this period, appeals to the federal circuit courts swelled from 3,765 in 1960 to 37,524 in 1988,56 an increase of nearly 900%. At the same time, the number of federal appellate judges in the geographic circuits grew too—from 68 to

167 by 1990.57 But the pace of growth in cases dwarfed the growth in judges. Put differently, in 1960 litigants filed approximately 55 appeals per authorized judgeship; three decades later, they were filing 225.58

The volume of appellate activity at the federal level has recently receded,59 but the overall volume remains quite high—and higher than the 40,898 appeals filed in 1990, when Congress last added authorized judgeships to the U.S. Courts of Appeals.60 In the 2020 reporting year, litigants commenced 48,190 appeals across the geographic circuits.61 That reflects an approximately 18% increase over three decades—all during a time when the number of authorized judgeships has remained static.

At least some decisional and structural reforms were necessary to tackle demand. Courts are not efficient markets. As Professor Lawrence Friedman observed at the outset of the crisis: where “too many problems of a familiar kind arise”—that is, if too many cases are filed—“[t]he American judicial system . . . cannot . . . react to increased demand simply by giving additional service.”62 Structural limitations therefore informed how and in what ways the courts could adapt to the demand that the rise in appeals placed on them. Ultimately, the courts’ adaptation is the modern federal appellate triage system, which has three important constituent parts: using unpublished decisions, reducing oral argument, and relying on central staff. Each is crucial to the whole, and each is the subject of its own body of scholarly work and critique (as is the triage system itself). Below, I will touch briefly on unpublished decisions and the reduction in oral argument before devoting more time and attention to the use of central staff in courts.

1. Unpublished Decisions Chronologically speaking, the first of these structural shifts was the invention of the so-called “unpublished” decision.63 An unpublished decision is an opinion not designated by the court for inclusion in West’s Federal Reporter, and as a result it is not precedential.64 That is not to suggest that before the label “unpublished” arrived all federal appellate decisions were always published in the sense of being printed in a book, but it is to say that before then, courts didn’t make a precedential65 distinction between decisions that were printed in books and those that weren’t.66 Categorically nonprecedential decisions are a mid-twentieth-century invention.67

The reason for the innovation was straightforward68: if the courts could resolve some appeals without worrying about making law (that is, setting precedent), they could dispose of those cases more quickly (because it presumably takes more time to write lawmaking decisions).69 At least initially, it was thought that all unpublished decisions would still be reasoned decisions—even if not as thorough as their precedential counterparts—but that ideal has given way with time.70 After all, it’s more efficient to say nothing at all—especially when judges worry that the “sloppy”71 work in unpublished decisions might be cited back as persuasive authority.72 For this

reason, until a 2006 rule change,73 unpublished decisions were not citable in most circuits (even back to the issuing circuit).74

Today, more than 85% of federal appellate decisions on the merits—a category that includes decisions resolving appeals as of right and original applications to the appellate courts for relief75—are unpublished.76 The federal appellate courts do little formal, precedential lawmaking—at least compared to the volume of appeals they resolve—but they create a whole lot of poor, recycled lawmaking in “copy-paste,” nonprecedential unpublished decisions.77

Some circuits rely on the procedural shortcuts of unpublication more than others.78 Some of these variations are inexplicable, as we’d not expect any appellate court—outside of the D.C. Circuit, perhaps79—to have any greater need for formal lawmaking than any other similarly situated court. I’ll return to this important point in Section II.B, as it is foundational to understanding the disparate impact of the federal appellate triage regime.

Of the institutional reforms the courts undertook in response to their perceived caseload crisis, the use—and ever-expanding use—of unpublished decisions has been perhaps the most controversial reform.80 It has received

the most scholarly treatment,81 judicial debate (and defense),82 and rulemaking attention.83 But little of that critique has moved the needle. Today, unpublished decisions are as prevalent as they ever have been84— even though the courts see nearly 20,000 fewer cases than they did at the caseload zenith in 2005, when the unpublication rate was lower than it is today.85

2. Reduction in Oral Argument Around the same time that unpublished decisions emerged, the courts also started decreasing their reliance on oral argument.86 A 1979 rule change authorized resolution of appeals without oral argument when the appeal was frivolous, when the issues had already been “authoritatively decided,” or when the court’s decision-making process “would not be significantly aided by oral argument.”87 Federal rules permit the parties to make a statement

respecting a request (or not) for oral argument,88 but courts are opaque in their decisions to grant or deny such requests.89

The result is that oral argument has been on a precipitous decline ever since—and it has also gotten shorter. In the 1970s, oral argument of thirty minutes per side was routine.90 Two-thirds of all appeals were argued orally in the early 1980s; by 2011, only one-quarter of all federal appeals were argued orally.91 By reporting year 2020, oral argument had dropped to less than 20% across all circuits, with one circuit only hearing argument in 11% of appeals (the Eleventh Circuit) and only two circuits hearing argument in close to or more than 40% of appeals (the D.C. Circuit and the Seventh Circuit).92 And in most circuits, the time allotted for oral argument is fifteen minutes or less (with five minutes on the low end—for pro se litigants in the Second Circuit93—and twenty minutes on the high end).94

The reduction in oral argument has received its own criticism,95 though that criticism has paled in comparison to the academic hostility directed at the constellation of issues related to unpublication. Concerns over the reduction in oral argument mirror those over other components of the triage scheme, focusing on how a reduction in oral argument also reduces judicial engagement.96 Ultimately, in the words of one scholar, the courts now have

a “more efficient but more paper-driven bureaucratic process of appellate decision-making.”97

3. Central Staff Attorneys The final institutional reform was the development of central nonjudicial staff to handle the cases that would receive neither lawmaking treatment nor oral argument. The reduction in publication and oral argument were public-facing changes, but behind the scenes, the courts shifted their organizational structure to account for these new decisional processes. Scholars have observed that “[t]he most startling development in appellate courts in the past quarter century has been the very marked increase in the role played by nonjudicial personnel.”98 But the development shouldn’t have been startling to those watching closely: having created a class of inferior process, the courts needed a workforce to handle the cases relegated to the second tier.99 Enter the central staff attorneys.

To understand the role of the staff attorney, we must first consider how the position differs from the law clerk.100 Law clerks are recent law-school graduates who work at a judge’s side in chambers, and thus sometimes are known as “elbow clerks.”101 In 1930, Congress first authorized each federal circuit judge to hire a single law clerk—conditioned, at the time, on the approval of the Attorney General.102 By 1970, each circuit judge had two law clerks; by 1980, some had three.103 Today, most active circuit judges have four law clerks.104

Scholars trace the modern law clerk to Supreme Court Justice Horace Gray.105 When he was appointed to the Supreme Court in 1882, Justice Gray hired the first federal law clerk, continuing a practice he began while serving

as the chief justice of the Massachusetts Supreme Judicial Court.106 Even then, it was a position reserved for the elite: Justice Gray hired a “highly ranked new graduate of the Harvard Law School,” at the recommendation of his half-brother, Professor John Chipman Gray, who taught law at Harvard.107 From there, the practice of hiring a law clerk spread to other members of the Court (especially once Congress authorized funding for such positions, given that Justice Gray had paid for his clerk, initially, out of his own pocket).108 Those positions then, as they largely still are, were reserved for high-achieving white men at elite institutions.109

Originally, law clerks were “sources of inspiration and criticism”; they edited but did not draft the decisions the Justices wrote.110 Today, appellatelevel law clerks generally perform some combination or all of three tasks: selecting (or screening) cases; drafting bench memoranda to prepare the judge or justice for oral argument or decision; and drafting decisions and opinions in (both argued and unargued) cases.111 In reality, few, if any, law clerks serve merely as sounding boards for their judges, as judges have delegated substantial responsibilities to their law clerks. That’s certainly not all bad, of course, and perhaps is even, at times, quite desirable; judges are able to do more by delegating research assignments and drafting tasks to law clerks. They can give more attention to more cases that way. And even where the amount of delegation is great, law clerks remain under the direct supervision of a judge; a law clerk may be to a judge what an associate is to a law-firm partner.

But the amount of delegation of judicial work within the federal appellate courts extends well beyond the law clerk. There’s another layer

within the federal appellate bureaucracy: the central staff attorney. The idea of a centralized nonjudicial staff appears to have originated in the Michigan Court of Appeals.112 These staff attorneys were “central” because they did not work for any one judge, nor did they work in any judge’s chambers.113 Although judges embraced the role of the law clerk, there was greater reluctance among the judiciary to embrace the role of central legal staff—at least at first.114 As the late Professor Daniel Meador and author Jordana Bernstein described: “Although those judges had long been accustomed to law clerks in their chambers, they were troubled by the idea of a centralized staff not under their close supervision. That arrangement . . . seemed to suggest the possibility of unduly delegating judicial work to non-judges.”115 But successful pilot projects among busy state appellate courts proved to judges’ satisfaction that such staff could “improve the courts’ productivity without apparent impairment of the judicial process.”116 And the continuing rise in caseloads across all appellate courts provided the necessary motivation for judges to overcome their early reluctance to embrace central staff.117 For proof of this widespread acceptance, look to the American Bar Association, whose standards relating to appellate courts embraced the idea of a central staff in 1977 (within a decade of the first robust experiment) and by 1985 featured extensive commentary on how best to utilize such staff.118 In the federal appellate system, the Fourth and Fifth Circuits appear to have been the first to experiment with central staff, hiring “habeas clerks” in the 1960s to handle “the explosion of cases brought by state prison inmates

seeking habeas corpus relief or relief under the civil rights statutes.”119 In other circuits, central staff emerged from “pro se deputy clerks of the clerk’s office, and law clerks to senior judges,” the latter of which “were stationed as a pool at the circuit headquarters.”120 Both shared similar responsibilities, focusing on pro se (which would include habeas corpus) litigation.121 As Professor Pether’s groundbreaking work revealed, the development of the staff attorneys’ offices—like all of the triage reforms—was intimately connected to a perceived flood of appeals from the system’s “have-nots” (namely, prisoners and civil rights claimants).122 This innovation permitted judges (and their elbow clerks123) to focus on the “relatively few important cases,” while staff attorneys handled “the larger, but more prosaic group of cases.”124

Initially, beginning in 1973, individual courts began requesting funds to support central-staff programs.125 By 1975, Congress officially authorized the appointment by a circuit’s chief judge of a “senior law clerk,” who would serve a centralized role.126 In 1982, Congress codified the central-staff programs that had developed across the circuits under that prior authorization.127 That new law authorized the chief judge to “appoint a senior staff attorney,” who could in turn appoint “necessary staff attorneys and

secretarial and clerical employees in such numbers as the Director of the Administrative Office of the United States Courts may approve.”128 That law remains in effect today.129 Initially, the Judicial Conference—the policymaking body of the federal courts130—adopted guidelines limiting the ratio of staff attorneys to authorized judgeships to reduce the risk of overdelegation.131 But those limitations were revisited over time as caseload demand continued to increase and judges did not.132

Just like with law clerks, it’s possible that more staff attorneys enable judges to spend more time on more cases. But law clerks and staff attorneys are distinct in ways that matter within the federal appellate system. First, and perhaps foremost, staff attorneys are far removed—physically and actually—from the oversight that judges exert over their elbow clerks.133 As is typical with any central staff, judicial oversight of the staff attorneys is, at best, only indirect.134 Second, staff attorneys focus on productive quantity over quality; staff attorneys are sometimes required to hit quotas in processing cases for decision.135 Third, staff-attorney positions carry far less cachet and prestige, perhaps because they work predominately on the court’s “low status” cases,136 thus potentially affecting the quality of applicants137 and the incentives that drive staff attorneys themselves.138

Structurally, staff attorneys do the work that judges don’t want to do so that judicial time may be spent on other, more “important” matters (that judges work on with elbow clerks). Their job, fundamentally, is to take work

off a judge’s plate and not add to it—including, perhaps, by identifying an issue that needs greater judicial scrutiny. In some circuits, there appears to be a norm against staff attorneys recommending reversal, for example.139 If law clerks enable judges, individually, to do more (e.g., hear more cases and write more precedential decisions), then we might think of staff attorneys as enabling courts, collectively, to do more (e.g., terminate more appeals more quickly).

As with each of these reforms, the rise of central staff is not without controversy.140 But reliance on and a need for central staff was inevitable; the courts never asked Congress for sufficient judicial resources to keep pace with the demands on their time.141 Put simply, someone needed to do the work that the caseload increase put on the courts, or else decision times would lag as dockets swelled. Central staff thus are essential to the U.S. Courts of Appeals’ ability to process and decide cases—and they remain essential today even as caseload volumes have stabilized.

4. Bringing It Together: The Triage System and the Tiers With these constituent parts established—unpublication, nonargument, and central staff—we have a fuller picture of the two tiers of federal appellate decision-making. There’s a first tier: oral argument, significant Article III attention, and (most likely) a published, precedential adjudication. And there’s a second tier: no oral argument, primarily central staff attention, and (most likely) an unpublished adjudication. The pivotal question, then, is how do courts decide which appeals belong to which tier? As I detail here, the answer is that through a combination of default rules and (mostly) nonjudicial review, courts divide the world into the procedural haves and have-nots.

At the risk of oversimplification, most circuits task nonjudicial staff— members of a central staff-attorneys’ office or attorneys who work for the clerk’s office—with an initial review of every appeal.142 That initial review

or “screen” may involve review for jurisdictional defects, relative complexity, and the triage decision itself—that is, whether the appeal should proceed to the first tier (judicial) or the second tier (central staff).143 Even after staff designate an appeal for second-tier treatment, however, a judge can reassign a case to the “regular” or first-tier process.144

Even in circuits that do not rely on central staff to make the (or at least an initial) triage choice, default processing rules route certain appeals to nonargument tracks. In the Third Circuit, the judges generally screen appeals in the first instance for oral argument (thus, for first-tier review), but certain cases (pro se appeals not involving direct criminal appeals and immigration appeals) go to dedicated nonargument panels by default.145 In the Second Circuit, all cases (even pro se appeals) are set for oral argument by default, except that “most immigration appeals are sent to a dedicated ‘NonArgument Calendar,’”146 an innovation the Second Circuit developed in response to a “flood of asylum claims . . . from 2002 through 2004.”147

Only one Circuit—the Tenth Circuit—appears not to rely on central staff to screen appeals or handle certain classes of appeals by default.148 Tenth Circuit judges make all screening decisions and decide, on an individual basis, to send some cases to the second tier, while setting other cases for oral argument.149 As you’ll see, the Tenth Circuit is the least busy

of the geographic circuits150—a fact that suggests that where judges have more time to devote to the triage decision, they do that important work.

Ultimately, whomever or whatever makes the decision—a judge, law clerk, staff attorney, or default rule—the federal appellate courts triage every appeal. That screening decision determines whether an appeal will receive first- or second-tier treatment. The choice to streamline appellate procedure may be a needed efficiency, an abandonment of traditional appellate values, or some combination of both. The next Section addresses some of the criticisms the triage regime has received.

B. Criticisms and Suggestions for Reform

Scholars began sounding the alarm about the federal appellate system’s revised decisional procedures early. The late Professor Charles Haworth’s 1973 article on the old Fifth Circuit’s “screening and summary procedures” warned that the court’s “dramatic innovations . . . may be the standard procedure for all appeals in both the state and federal systems within the next five years.”151 He was basically right.

Professor Haworth’s work also revealed a troubling fact: the Fifth Circuit’s procedures—however necessary they may have been—correlated with a statistically significant reduction in the reversal rate in the Fifth Circuit (and as compared to another circuit, the Third Circuit, which had not yet instituted such procedures).152 Before it began experimenting with summary procedures and screening in the late 1960s, the court’s median reversal rate had been between 27.1% and 27.4%.153 The rate dipped to 14.9% within two years of the Fifth Circuit’s “extensive use” of summary procedures.154 In the 2020 reporting year, the Fifth Circuit’s reversal rate was 5.8%, and the Eleventh Circuit’s reversal rate was 8.3%155 (recall that the latter was part of the Fifth Circuit when Professor Haworth was writing).

At the time of Professor Haworth’s work, every court that used summary or screening procedures continued to rely on judges to make those

decisions.156 What was most troubling, then, in Professor Haworth’s view was the use of summary affirmances—one-word decisions that said no more than “affirmed” or “enforced”—that circuit courts issued without oral argument.157 Although such “without comment” decisions have been on the decline in recent years,158 some courts continue to issue barely reasoned or circular decisions while stating that they issue only reasoned appellate decisions on the merits.159

From this early work on the effects of the federal appellate courts’ internal reforms, scholars have largely fallen into two camps of critique. The “alarmists,” I’ll call them, are those like Professors Richman, Reynolds, and Pether, who see the courts’ reforms as degrading foundational values and who call for substantial reform—for example, growing significantly the federal appellate bench or publishing all decisions.160 Some judges, too, undoubtedly fall into this category of critique; for example, by 1992, onefifth of federal appellate judges responding to a Federal Judicial Center survey agreed that “the quality of appellate justice has been unacceptably diminished by measures adopted by the courts to cope with rising caseloads.”161

On the other hand, the “necessary-evil institutionalists,” I’ll call them, recognize that some reform in response to caseloads was necessary—and perhaps even beneficial in some ways—but also that shortcuts can be taken too far. Many judges who defend the system fall into this camp.162 Professor

Levy’s careful work urging somewhat modest (but certainly needed) reforms to reduce reliance on categorical screening rules and increase staff attorney expertise may also fall into the institutionalist bucket.163 More recently, and in a more technologically advanced critique, Professor Ryan Copus has described how courts might use technology to better allocate judicial attention based on a highly contextualized assessment of the potential need for lawmaking and risk of error using “statistical precedent” and predictive algorithms.164 Professor Copus would rely on machine learning to make better triage choices and help courts more accurately distribute their attention.165

The courts themselves appear mostly indifferent to the criticisms their procedures have received—except in so far as they agreed to reform the ban on citing unpublished decisions in the early 2000s,166 which is the only significant reform undertaken to address concern over the outputs of the triage process.167 In subsequent years, the din has died down, and we’ve become largely inured to the status quo. The next Part provides reasons to challenge the status quo anew.

II. WHY THE U.S. COURTS OF APPEALS NEED REFORM

For better or, in my view, for worse, the U.S. Courts of Appeals weathered the “crisis of volume” they faced during the last century. Scholars have questioned whether that crisis was “real or imagined.”168 Regardless, the procedures the courts developed to tackle it have become the accepted

status quo. We have, in the words of one scholar, “defined down our appellate values” to adjust to this new normal,169 in which judicial resources are scarce and doled out sparingly to those deemed most deserving.

This Part takes stock of the toll the shortcuts have taken. I argue first that the courts have overrelied on these procedural shortcuts and, next, that their overreliance has imposed uneven burdens on the basis of race, class, and geography. Some regions of the country—in particular, the whitest and the wealthiest—benefit from more robust federal appellate process and have greater access to Article III resources. Those disparities are neither explicable nor the result of any transparent experimentation among the courts. These are also disparities that defy an easy solution; these are not differences in formal rules or lawmaking that might garner Supreme Court attention. Yet they are possibly outcome determinative in at least some cases—to the extent that greater judicial attention is more likely to result in favorable outcomes in any individual case.

Let me emphasize from the outset that I make no claim that any of these effects are intentional by the courts.170 This is a story of disparate impact, not disparate treatment.171 I do not seek to prove that racism or classism motivated the courts to adopt the federal appellate triage system nor that these systemic ills cause courts to maintain the systems they conceived. I do neutral procedures may argue, however, disproportionately harm people of color and poor communities, who tend to live where these procedures are most used—and this alone is a reason to rethink the distribution of Article III resources in the federal appellate system.

facially

these

that

A. Overuse of Shortcuts

The proliferation of the two-tiered system of appellate review has created convenient and useful shortcuts for the courts. But it has gone too far. That is a claim that is hard to prove as an empirical matter, but there are signs pointing in that direction. This Section considers two such signs. First, there has been a high number of certiorari grants in unpublished federal appellate decisions in recent years, especially from courts that rely to a greater extent on second-tier process. Second, the shortcuts have elevated judicial preference and convenience over systemic interests in equal access and equal treatment under the law.

1. Supreme Court Review of Unpublished Decisions The federal appellate courts do not always do a very good job of predicting when published, precedent-making decisions are needed and when they are not.172 In her 2021 testimony before the presidential commission on Supreme Court reform, Professor Resnik revealed that annually approximately one-seventh of the Supreme Court’s grants of certiorari come from unpublished decisions.173 We might think that if the triage system was working as it should, the Supreme Court would rarely, if ever, have the need to grant certiorari review of any unpublished decision. Unpublished decisions should not need error correction, nor should they be used where lawmaking is needed.

Resnik’s back-of-the-envelope calculations went back three years; the trend holds if we go back five years more (to October Term 2013). Isolating opinions from grants of certiorari from the geographic federal courts of appeals alone, 70 of 435 certiorari grants (16.1%) involved unpublished adjudications in the circuit courts, 13 of 79 opinions reviewing state court decisions (16.5%) involved unpublished adjudications in the lower state court, and 4 of 33 certiorari grants from the U.S. Court of Appeals for the Federal Circuit (12.1%) came from unpublished decisions.174

The clear majority (more than 75%) of these cases were merits cases— not summary adjudications. Only 14 of the 70 (20%) ended in per curiam

summary decisions in which the Court reached the merits of a certiorari petition without the benefit of oral argument. The rate of review for unpublished decisions is highly variable across the geographic circuits. In the First Circuit, for example, none of the Supreme Court’s 13 certiorari grants over the last eight years involved an unpublished adjudication. That may not be terribly surprising, given that the First Circuit has among the highest publication rates in the country.175

Compare, then, the First Circuit’s track record to that of the Eleventh Circuit: over the last eight years, nearly half (48.8%) of the 43 certiorari grants from the Eleventh Circuit have involved unpublished decisions. No other circuit comes close; indeed, that rate of review for unpublished decisions from the Eleventh Circuit is more than two standard deviations (12.8%) from the mean rate (13.9%) of review of unpublished decisions across all geographic circuits. Table 1 identifies total certiorari grants since 2013 by geographic circuit and isolates those arising from unpublished adjudications.

Number of Cert. Grants

Number from Unpublished Decisions

Percentage from Unpublished Decisions

3.3%

8.3%

3.8%

27.7%

17.8%

48.8%

16.1%

12.8%

I don’t want to overstate the case, but these numbers—especially for the Fifth and Eleventh Circuits—seem surprising. It’s possible, of course, that some unpublished decisions are unpublished because they apply settled

circuit law that the Supreme Court later decides to review (where it did not review the precedent-making decision). At least some of the certiorari grants fall into this category; the Supreme Court’s decision to review Bostock v. Clayton County,176 an unpublished Eleventh Circuit decision, is one prominent example.177 But that’s hardly always the case.178 And, in any event, it’s hard to see why that particular practice—issuing an unpublished decision noting that an argument was foreclosed by precedent or granting review to overturn settled circuit law—would be any more common in one circuit than it would be in any other circuit with a high rate of unpublished decisions.

In further support of this intuition, I note that the outcomes in Supreme Court decisions reviewing unpublished decisions are not significantly different from the outcomes in decisions reviewing published lower court decisions. Table 2 identifies the rates of reversal, vacatur, and affirmance across published and unpublished decisions over the last eight years. The Court appears to be taking unpublished decisions to reverse settled law no more often than it is in published decisions and vice versa. Nor, for that matter, is the Court reviewing unpublished decisions simply to correct error.

Published Decisions

Unpublished Decisions

Reversals

Reversal Rate

Vacaturs

Vacatur Rate

Affirmances

Affirmance Rate

48.0%

30.5%

51.8%

24.7%

23.5%

What’s more, anecdotally speaking at least, the Eleventh Circuit’s recent en banc votes seem to reflect some internal concern over potential abuse of the unpublication shortcut. Although the vast majority of the Eleventh Circuit’s en banc votes since 2015 have involved published

decisions, there were seven votes in unpublished decisions,179 including a vote in a controversial police-involved shooting and qualified immunity decision in which, after argument, the panel disposed of the appeal in a single-paragraph unpublished decision.180 (I should note that the panel’s decision in Bostock was among the seven en banc votes, too.)181 The plural of anecdote is not data, of course, but it is telling that such a decision received circuit-wide recognition—and indeed even provoked a published, reasoned response from some of the original panel members.182

All of this underscores what Professor Resnik has suggested: at a minimum, there appears to be some “disjuncture between what lower courts understand to be cases of legal significance and what four members of the Supreme Court believe worthy of review.”183 That raises, in Professor Resnik’s view, “puzzles about how federal appellate judges decide (or bargain) over unpublication.”184 Is this evidence of inconsistent application of publication criteria across the circuits? Perhaps—especially because almost all circuits share essentially the same basic standard for determining publication worthiness.185 Is this evidence of “strategic publication”186—that is, use of unpublication to shield favorable and unfavorable legal rules from further review (by colleagues or a higher court)? Again, perhaps.

To make more than tentative conclusions, we’d need to know more about the inputs, including, for example, the relative rates of seeking certiorari review from unpublished versus published decisions. One might expect the rate for seeking certiorari to be higher for published decisions, given that published decisions make law and unpublished decisions do not. On its face, it would be hard for a nonprecedential decision to satisfy the Supreme Court’s standard for review—a standard that requires “compelling reasons” for the grant of certiorari, including importance, conflict with

another court, or a significant departure from judicial norms.187 Decisions that make no law, by definition, cannot create a real conflict in precedent because they do not bind. They also can’t be all that important—except to the parties themselves, of course. And if they represent such a significant “depart[ure] from the accepted and usual course of judicial proceedings”188 to call for the Supreme Court to use its supervisory powers, well, then, that’s reason enough to rethink the enterprise.

That the Supreme Court—or at least four Justices on the Court—see fit to review any nonprecedential decision is at least some sign that the triage system isn’t working quite as it is supposed to work (at least not all of the time). In an ideal world, no unpublished decisions should ever meet the criteria for certiorari worthiness set out in Supreme Court Rule 10. Those cases relegated to the second tier of federal appellate review are supposed to be the easy ones about which there is little to no room for disagreement over the correctness of the outcome. If it is otherwise, then these cases deserve more attention. Whatever we are to make of the rate of Supreme Court review of unpublished decisions, it is at least evidence that the Supreme Court isn’t ignoring unpublished decisions—the product of the underbelly of the U.S. Courts of Appeals—and neither should we.

2. Convenience and Judicial Preference The triage system emboldens judges to be picky. They choose which appeals are worthy of their attention and which are not. That’s an invitation that can be misused and abused. And it is difficult to assess such misuse or abuse because of the volume of appellate activity and the difficulty of finding it all.189 To the extent that the Supreme Court reviews a nontrivial number of unpublished decisions each year, that is some quantitative evidence of misuse or abuse. This Section considers these risks qualitatively.

Judges tend to defend their selective distribution of appellate resources as a matter of efficiency. For example, the current chief judge of the Eleventh Circuit once observed that providing full process for certain categories of cases—“routine cases based on settled precedent,” such as “petitions to

review deportations and denials of Social Security benefits”—“would only delay decisions that should be speedy.”190 This judge suggests that shortcut procedures are in the interests of the litigants themselves—downplaying, perhaps, how such procedures are also easier for the judges.

Less charitably—and perhaps more revealingly—another

judge appeared to acknowledge the judicial motivation for continued reliance on shortcuts. The former chief judge of the Fifth Circuit said the federal judiciary was at risk of getting “dumbed-down” if it had to focus too much on the “overwhelming number of routine or trivial appeals”; that judge observed, “The situation is like that of a competitive tennis player forced to spend the bulk of his time rallying with novices.”191 The triage system thus ensures that judges can keep their skills sharp by working only on the hard cases that truly deserve their attention. The court’s staff, on the other hand, can spend their time “rallying with novices.”192

The system of appellate justice that has emerged—one of selective distribution of appellate resources—is no longer a needed response to a crisis. There is no longer any correlation between caseload volume and the use of unpublished decisions193 that are a hallmark of the second-tier federal appellate system. The federal appellate caseload has fallen since its highwater mark in 2005, yet the number of unpublished decisions continued to increase until leveling off at around 87% to 88% nationwide in the last five years.194 Because of that trend, the use of unpublished decisions no longer correlates with caseload volume—the stated cause of the need to rely on unpublished adjudications in the first place195—but it does positively correlate with a rise in pro se or unrepresented appeals.196 Across the country, the number of pro se appellants has risen in the last two decades from around 40% to around 50%.197 Correlation does not equal causation, but caseload volume alone can no longer account for the continued reliance on unpublished decisions. Consider this: in 1992, when the U.S. Courts of

Appeals heard around 48,000 appeals (roughly the volume they hear today198), the courts were publishing approximately 29% of their work nationwide.199 Recently, they have published only 13%.200

Let’s consider another possibility: the courts simply like the procedures they’ve adopted. These procedures—the ability, in essence, to operate as choosy courts of last resort and not workmanlike courts of mandatory jurisdiction201—have freed federal appellate judges from the need to deal closely with the humdrum of routine work. They can instead spend their time and energy on more “important” cases—however one defines important. From a human-labor perspective, that makes sense. Like any of us, judges desire to make their jobs more interesting, more fun, or—more cynically— easier and less work.202 Judges, after all, are workers.203 But they are workers in what is essentially a dead-end job, and we should consider their behavior in that context.204 (Of course, there’s no more glorious dead-end job than being a judge on a U.S. court of appeals—except, perhaps, being a Justice on the U.S. Supreme Court.)

Enter, then, “meritless appeals.”205 The specter of meritless appeals, and the sense that the federal courts are awash in useless, boring, and frivolous work, is a trope that supports the maintenance of the two-tiered system itself. Federal judges have said for decades that meritless appeals are a major problem.206 That the federal appellate courts were beset with meritless appeals first emerged as a narrative in the 1960s207—a boom time, perhaps not coincidentally, for federal civil and criminal constitutional and statutory rights as well.208 Indeed, the term “meritless” was used just twenty-five times in the federal appellate courts’ available decisions issued before 1960.209 It has been used in more than 27,000 decisions since the beginning of 1960.210 Just because an appellant may not win doesn’t mean a case is unworthy of judicial attention—or that it is worthy of being branded with the derisive label “meritless.” Almost all appeals in the federal system are losers. The average annual outright reversal rate across the geographic circuit courts is less than 9%.211 And fewer than 1% of appeals earn a remand—another favorable result.212 Few of us would defend a system that bestows meaningful judicial scrutiny on fewer than 10% of filed appeals.213 And yet, that is the process we have in some of the circuit courts.

We too often use “meritless” to denote “frivolous.” But those should be distinct concepts, as Professor Alexander Reinert has convincingly argued in the most substantial work to date on meritless litigation.214 Professor Reinert

argues that courts, Congress, and scholars alike have elided the difference between frivolous and meritless litigation—and they have done so to the judicial system’s detriment. By substantially undervaluing meritless litigation, courts and Congress have promoted reforms that stymie legal innovation and undermine efforts to use the legal system as a neutral forum for dispute resolution.215 Put succinctly, our system ultimately needs and depends on meritless appeals to define and develop the contours of legal rules. It can do without frivolous work that wastes judicial time with outlandish and unsupportable requests, but it needs meritless litigation for the system to function well.

treatment for

their gender dysphoria violates

Take, for example, a claim by a prisoner that the state’s failure to provide the Eighth Amendment’s prohibition on cruel and unusual punishment. Such a claim requires the prisoner to establish that the state has acted with “deliberate indifference to [a] serious medical need[].”216 Had the prisoner sought treatment in 1976 when the U.S. Supreme Court announced that standard, or shortly thereafter, the claim would have been seen as frivolous as a matter of existing law—because gender dysphoria (or any analogous condition) would not have been considered a serious medical need at that time.217 The diagnosis didn’t even exist yet.218 Today, there is a circuit split over whether the failure to treat gender dysphoria with gender affirmation surgery violates the Eighth Amendment.219 That example underscores Professor Reinert’s point: “patently unsuccessful litigation can contribute to positive legal change.”220 Full process and consideration may hasten that innovation,

whereas perpetual second-tier treatment can stymie it. And, sometimes, even what may initially seem frivolous evolves too. All this to say, it’s easy to be casual with the labels “meritless” and “frivolous,” and there are likely few truly frivolous (in the sense of legally and factually silly) appeals (outside of, perhaps, claims about aliens, sovereign citizens, and the like).

Scholars and judges alike have often devalued and dismissed the output of the second-tier appellate system as meritless to justify the selective distribution of appellate resources. For example, in their major empirical and theoretical work examining the behavior of federal judges, Professors Lee Epstein, William Landes, and then-Judge Richard Posner decided not to include unpublished decisions in their study (a sizeable omission, of course).221 They did so because “[t]he vast majority of unpublished decisions are affirmances, and . . . this is a reflection of the signal lack of merit of most cases filed in the federal district courts.”222 At the same time, however, the authors argued that even though these “meritless” cases occupy a “majority” of judges’ time “in some of the busier circuits, . . . it is not their principal work.”223 The principal work, in their view, is the fun stuff—the cases of consequence in which legal rules are made and shaped. The authors compare judges to artists like T.S. Eliot and Lawrence Olivier, who may have spent most of their waking hours doing drudge work.224 So too with judges. Epstein, Landes, and Posner aren’t alone in thinking of the work of judges in these terms—that’s the attitude reflected in the candid assessment from the Eleventh Circuit and Fifth Circuit judges discussed above.225

But judges aren’t artists or actors (or competitive tennis players); they’re public servants. And doing the drudge work is a public service. It may be that courts rightly triaged their time and attention in a time of scarcity. That’s the point of triage: those with life-threatening injuries who can be treated go first; those with needs that exceed available resources and for whom efforts would be futile go last.226 But it need not be a time of scarcity anymore: Congress can respond to the persistent failure to distribute federal appellate resources more fairly, and, in so doing, it can incentivize

the courts to provide more and better process to more litigants. Before turning to one proposed solution—adding more judges—the next Section considers additional costs of the uneven distribution of federal appellate resources.

B. Disparate Impact of the Triage System Across Geographic Circuits

Not all circuits use the triage procedures described in Part I to the same extent. Some circuits continue to rely more heavily on oral argument than others, and some circuits issue substantially more published decisions than others. We might reasonably expect some variation across the circuits. But as Professors Richman and Reynolds observed, it is unlikely that “caseload differences, either in number or kind, account for the[se] differences.”227 Although some “experimentation” might be “tolerable,” courts aren’t really experimenting—they’re just doing things differently for “no apparent reason,” thus “mocking the notion of equal treatment under the law.”228

This Section argues that those inexplicable differences have a disparate impact based on race and, to a lesser extent, class. First, it evaluates the differences in how the courts distribute their available resources (that is, how they distribute oral argument and publication). Next, it considers the extent to which the different inputs (docket composition) may affect different procedural outputs (oral argument and published decisions). Finally, it examines how variations in the distribution of Article III resources correlate with race- and class-based differences across the circuits. Circuits with the largest per capita population of Black Americans rely most heavily on second-tier procedures, whereas circuit courts in the wealthiest regions rely the least on such processes. This is especially the case with the procedural shortcut of a nonargument track. As a result, the disparate distribution of judicial time and attention appears to impose a greater burden on our most vulnerable communities.

1. How the Courts of Appeals Distribute Their Resources Each circuit has an underbelly. Every circuit relies to some extent on second-tier appellate processes. To assess that underbelly, we can consider the extent to which the courts use the constituent parts of that process, such as unpublished decisions and summary or nonargument procedures.

Unpublication often, but not always, reflects a second-tier process. (Some appeals argued orally will end in unpublished decisions, but most will be published.) Nationwide, the unpublication rate for all merits terminations

in the geographic courts of appeals has hovered in the high-80% range since 2013.229 But unpublication is uneven across the circuits. At the high end, the Fourth Circuit has a five-year mean unpublication rate of 93.7%, followed closely by the Eleventh Circuit at 93.3% and the Ninth Circuit at 93.0%. At the low end, the D.C., First, and Seventh Circuits have five-year means below 67%. Table 3 provides unpublication rates for the last five years, as well as a five-year mean, for each geographic circuit.

60.1%

71.7%

88.3%

89.9%

93.1%

86.5%

87.6%

66.1%

78.2%

81.7%

91.5%

64.3%

69.2%

89.1%

90.5%

91.8%

87.7%

88.0%

64.7%

74.5%

93.2%

81.2%

93.5%

60.9%

63.6%

90.5%

90.8%

93.9%

91.4%

63.7%

79.1%

93.1%

92.8%

52.8%

63.6%

91.7%

89.1%

93.4%

91.0%

92.6%

65.2%

78.2%

78.5%

93.9%

61.8%

65.4%

90.4%

96.3%

90.4%

90.6%

69.8%

92.3%

85.1%

94.6%

89.3%

66.0%

Note. The mean of all circuits equals 82.6%, and the standard deviation equals 12.2%. Red circuits are above the five-year mean, and green circuits are below it.

The use of oral argument has similarly been low over this same stretch of time. Nationwide, the total percentage of merits terminations after oral argument has hovered around 20% annually since 2012.230 Table 4 identifies

the percentage of appeals terminated after oral argument over the last five years in each of the geographic circuits, as well as the five-year means. As with rates of unpublished decisions, there’s significant variation in the use of oral argument across the circuits; the chart highlights in red those circuits below the national five-year mean and highlights in green those circuits above the national five-year mean.

45.5%

23.7%

35.3%

11.4%

22.9%

13.5%

16.2%

19.0%

25.6%

45.1%

21.5%

34.7%

11.4%

22.0%

39.1%

21.0%

27.4%

11.0%

47.2%

33.3%

11.0%

18.7%

13.3%

41.8%

16.3%

23.0%

29.0%

51.6%

27.2%

32.4%

10.5%

16.4%

37.5%

15.5%

20.6%

27.9%

46.0%

27.6%

31.3%

5.9%

30.4%

23.2%

21.1%

7.7%

47.0%

25.2%

21.3%

Note. The mean of all circuits equals 22.6%, and the standard deviation equals 11.8%. Red circuits are below the five-year mean, and green circuits are above it.

Although more appeals receive oral argument nationwide than end in unpublished decisions, the two are generally related. That means that most circuits with the highest publication rates also have the highest oral-argument rates, and those courts with the highest unpublication rates have the lowest oral-argument rates. The two procedures go hand in hand: judges don’t

expect to invest much time in cases that will not make law, and these supposedly easier cases do not benefit from oral argument. Figure 1 demonstrates the strong inverse correlation232 between the unpublication rate and the rate of oral argument across the circuits: as the unpublication rate rises, the rate of oral argument generally falls. There are some exceptions— in particular, in the Second Circuit—in which a strong tradition for oral argument (even for pro se litigants) persists.233 But in other circuits, publication and oral argument are explicitly linked. For example, until the COVID-19 pandemic,234 the Fourth Circuit would only publish decisions in appeals that received oral argument.235 But overall, the strong inverse correlation between the rate of unpublished decisions and the use of oral argument is what we would expect given the efficiency goals that unpublication and reduced oral argument serve.

45.0%

55.0%60.0%65.0%70.0%75.0%80.0%85.0%90.0%95.0%100.0%

Rate of Unpublished Decisions

Thus, nationwide, over the past five years, only about 20% of appeals were argued orally and only about 13% ended in published decisions. Those are rough but reasonably fair proxies for the size of the first and second tiers of the federal appellate justice system. The first tier comprises roughly 20% of cases; the second tier is the bottom 80%. In any one circuit, the first tier might be smaller (perhaps 10% or less in the Fourth and Eleventh Circuits) or larger (more like 35% in the Seventh Circuit).

Finally, bringing data from Tables 3 and 4 together gives us a clearer sense of which circuits devote more judicial attention to a greater percentage of their dockets based on relative oral-argument and publication rates. Table 5 combines oral-argument and publication data and sorts each circuit from best to worst (highest oral-argument rate to lowest; lowest unpublication rate to highest) based on five-year means (and where the bold dividing line reflects the mean). Four circuits emerge as above average: the D.C., First, Seventh, and Tenth Circuits. Conversely, the Fourth and Eleventh Circuits are consistently below average.

Oral-Argument Rates

Unpublication Rates

47.1%

25.2%

21.4%

65.9%

89.3%

Whether these differences affect outcomes is a difficult—and perhaps impossible—question to answer. Is a prisoner who files a civil rights claim in the Fourth Circuit more likely to lose than a prisoner who files the same claim under similar circumstances in the Seventh Circuit? At the very least, we might expect the Seventh Circuit prisoner to be more likely to have his case heard by the court orally, and intuitively, we might expect a court that devotes more judicial attention to a greater percentage of appeals to have results that are more favorable to the appellant more of the time.237 And the data appear to support that intuition—at least in so far as there may be some relationship between the use of oral argument and favorable outcomes on appeal.238 Figure 2 demonstrates a moderately positive correlation across the circuits between favorable results (either reversal or remand) in cases terminated on the merits and the rate of oral argument. (Of course, if a court

thinks more cases raise sticky issues meriting oral argument, it follows that more such cases would end in favorable outcomes for the appellant, too.)

Rate of Favorable Outcomes in Merits Terminations (Reversal and Remand)

Underscoring that the amount of attention matters, Professor Copus’s recent work argues against the categorical treatment of certain appeals as presumptively not deserving of judicial attention.240 He uses predictive algorithms to identify likelihood of error and the need for lawmaking in casespecific contexts to help better deploy resources in the triage process.241 His case study (based on the Ninth Circuit) demonstrates that the courts may be underappreciating the risk of error in pro se appeals.242 Although many pro se appeals have low degrees of error, Professor Copus says approximately 10% of pro se appeals have error rates in the range of other lawyered civil appeals.243 Considering that pro se appeals account for approximately 50%

of the appellate workload,244 10% is a nontrivial amount. Based on his work, Professor Copus unsurprisingly concludes that he sees “little justification for assigning [all pro se appeals] to staff attorneys” based on an assumption that they are easy affirmances or meritless.245

The extent of process may indeed matter—and not just to the litigant who desires her day in court, but also to the outcome on appeal. Appellants appear more likely to obtain a favorable result where the court gives the appellant oral argument and, therefore, more Article III attention. How a court chooses to distribute Article III resources, therefore, may affect an appeal’s fate. If that’s true—if the extent of process matters instrumentally as well as intrinsically—then we should be especially attuned to the unevenness of how those procedures are being distributed. I turn to that question next after first explaining why docket composition alone cannot justify disparate circuit practices.

2. Docket Composition Across the Courts of Appeals Any uneven distribution of appellate resources might reflect different docket demands across the country. Perhaps one circuit court (such as, say, the D.C. Circuit) sees more complex cases than another circuit. Although we would not expect the identity of the appealing party alone to determine the extent of the processes afforded,246 it’s possible that some types of appeals, on average, demand more attention while other types of appeals, on average, demand less. At the district court level, for example, the courts weigh various filings based on expected resource demands and relative complexity.247 Although we have no similar metric for appellate cases, it’s reasonable to think docket composition might have something to do with how different courts distribute their resources differently.

But that is not what is happening—at least not for the most part. Although there are some notable exceptions (discussed below), most of the twelve geographic courts of appeals see roughly the same kinds of appeals in roughly the same percentages. Appendix A provides five-year averages

for docket composition in each circuit, and Table 6 details the nationwide five-year averages for docket composition:

Appeals Categories

Five-Year Mean of Means

20.3%

15.1%

23.2%

13.2%

7.8%

6.1%

3.3%

This Table shows that there is relatively high variability in some categories of appeals—administrative agency and criminal appeals have the largest standard deviations—and relatively consistent volumes of bankruptcy and U.S. prisoner appeals and original proceedings across the circuits.

As Appendix A demonstrates, only the D.C., First, and Second Circuits exceed two standard deviations from the mean in any one of these eight broad categories. The D.C. Circuit’s docket composition is the most unusual—and the clearest outlier among the geographic circuits. For administrative appeals and nonprisoner U.S. civil appeals, it exceeds two standard deviations on the high end, and for private (that is, nonfederal) prisoner appeals, it exceeds two standard deviations on the low end. The D.C. Circuit also exceeds (either above or below) one standard deviation (but is within two standard deviations) in every remaining appeal category except for U.S. prisoner

petitions, where it is close to the national average. For these reasons, I generally exclude the D.C. Circuit from my disparate impact analysis.249 Its docket composition is sufficiently different from its peer circuits that we might expect it to process appeals differently, irrespective of any demographic influences in its limited geographic footprint.

The First and Second Circuits are also outliers in one category each. The First Circuit sees a significantly greater percentage of bankruptcy appeals than its sister circuits, but the overall numbers of such appeals are slight (ranging from 28 to 44 bankruptcy appeals over the last five years).250 The Second Circuit—surely owing to its location in major business centers— sees a greater percentage of private civil cases than any of its sister circuits. These are distinctions to consider, but I do not think either justifies excluding these circuits from the analysis to follow in Section II.B.3.251

Except for these three circuits—the D.C., First, and Second Circuits— no other circuit exceeds two standard deviations with respect to any category of appeals based on five-year means. Most courts exceed one standard deviation in at least one category, however. Only two circuits are perfectly average and within a single standard deviation in each category: the Third Circuit and the Tenth Circuit.

Because there is a correlation between unpublished decisions and the volume of appeals filed by pro se litigants,252 the relative volume of pro se litigation across the circuits may affect the extent to which a circuit uses second-tier procedures. Table 7 identifies the average volume of pro se appeals filed over the last five years for each geographic circuit. Those highlighted in green are at least one standard deviation below the mean, while those highlighted in red are at least one standard deviation above the mean. Three circuits—the D.C., First, and Second Circuits, the same three

circuits highlighted above for unusual docket composition—all have a particularly low volume of pro se appeals.

34.6%

36.6%

39.5%

45.3%

45.4%

47.3%

52.4%

52.9%

57.4%

58.2%

59.1%

60.7%

49.1%

9.1%

Several circuits stand out for particularly large volumes of pro se filings: the Fourth and Seventh Circuits, in which the volume of pro se activity exceeds one standard deviation, and the Sixth and Eleventh Circuits, which are well above the mean of 49.1% (but within one standard deviation). If pro se litigation has driven the development of the appellate triage system, then one might expect the Seventh Circuit’s procedures to be on par with the Fourth Circuit’s procedures. But they are, in fact, opposites: the Seventh Circuit is routinely a publication and oral-argument leader, and the Fourth Circuit is not.

One might wonder whether the type of pro se litigation differs across the circuits, even when the volume remains high. We might expect that where certain types of pro se litigants—namely, pro se prisoners—are a large

percentage of a court’s docket that courts might rely more heavily on secondtier procedures. Not always. Table 8 identifies the percentage of pro se prisoner appeals as a percentage of all appeals filed in each geographic circuit.254 The three circuits with the lowest overall percentage of pro se litigation—the D.C., First, and Second Circuits—are all more than one standard deviation below the mean for overall percentage of pro se prisoner filings. The Fourth Circuit and the Seventh Circuit—the leaders in pro se filings, more generally—are one standard deviation above the mean.

8.6%

19.9%

22.0%

23.3%

23.4%

23.4%

25.1%

28.2%

29.2%

7.2%

All this to say: except for those circuits with a particularly low volume of pro se litigation (the D.C., First, and Second Circuits), most circuits see a relatively consistent256—and a relatively high—volume of pro se prisoner appeals as a percentage of their overall dockets. The volume of pro se prisoner appeals—just like the volume of pro se litigation more generally— cannot account for the disparities in the distribution of federal appellate resources across the circuits. Although the cause of these disparities may be

unknowable, the fact of the disparities is uncontroverted—and it creates a risk of disparate impact that I consider next.

3. Race, Class, and the Courts of Appeals Whatever accounts for the disparities in procedures across the circuits,257 it is also the case that these variations may be consequential to the communities the courts serve.258 Recall, again, that nearly half of all appeals involve individuals proceeding pro se; these are claimants seeking to vindicate important federal statutory and constitutional rights—and these procedures themselves developed in response to a perceived “flood” of these very kinds of claims. This Section argues that institutional design and substantial circuit variation have engendered processes that impose uneven burdens on individuals in our most vulnerable communities. As Professor Dorothy Brown’s recent book on the racial disparate impact of tax law concluded: “[T]here’s nothing in this country that race . . . [is not] a part of.”259 So too here. The variations in circuit procedures appear to have a disparate impact based on race and, to a lesser extent, class.

I make these observations with tentativeness. I do not know the race or ethnicity of any pro se appellant in any appeal in any court. The courts do not track these data. What I do know is that a great many litigants in the federal appellate system are individuals—often proceeding pro se. It follows that these litigants reflect their community at least in rough proportion; indeed, recent work from Professors Roger Michalski and Andrew Hammond suggests that pro se litigants in the district court are more likely to come from nonhomogeneously white communities.260 Thus, even without knowing the identity of any particular litigant, we can see some relationship between the disparities in how the circuits distribute procedures and the impact the courts may have on the communities they serve, because these courts serve communities in particular geographic regions about which we have demographic information.

The bottom line is this: relative circuit “whiteness” and “Blackness” based on census data for the population in a circuit correlate with differences

in appellate procedures across the country.261 That means that whiter circuits tend to publish more of their decisions, and Blacker circuits do not. And wealthier, whiter circuits tend to hold oral argument more often, and poorer, Blacker ones do not.

Table 9 ranks the circuits from least to most diverse based on circuit whiteness.262 Those highlighted in green are more than one standard deviation above the mean for whiteness (i.e., the least diverse); those highlighted in red are more than one standard deviation below the mean for whiteness (i.e., the most diverse). The Ninth and Fifth Circuits are the only majority-minority circuits.

Circuit Whiteness (2019)

79.1%

77.2%

75.7%

67.0%

61.5%

58.5%

57.6%

54.3%

46.7%

44.5%

62.9%

11.9%

Now consider circuit Blackness, i.e., those circuits with the largest percentage of individuals identifying as Black or African-American alone or in combination with other races according to the U.S. Census Bureau’s 2019 population estimates. Table 10 identifies those circuits from most to least Black. Those marked in red are more than one standard deviation above the

mean (the Fourth and Eleventh Circuits); those marked in green are more than one standard deviation below the mean (the Ninth and Tenth Circuits).

Circuit Blackness (2019)

24.1%

23.5%

18.1%

16.5%

12.4%

9.7%

8.0%

6.4%

5.6%

6.4%

All this to say: the Fifth, Ninth, and Eleventh Circuits are the most diverse, and in particular, the Fourth and Eleventh Circuits have the greatest concentration of Black Americans. Recall that these are also the circuits with the smallest first tier and the largest second tier of federal appellate procedure, as reflected in the overall volume of unpublished decisions and use of oral argument.263

It is no surprise, then, that circuit Blackness correlates moderately with the use of unpublished decisions across the circuits. Those circuits with higher percentages of Black Americans are generally more likely to use unpublished decisions; those circuits with lower percentages of Black Americans are less likely.264 Figure 3 demonstrates a moderate positive correlation between the two:

Proportion of Circuit Population that Is Black Alone (2019 Census Estimates)

Consistently with the results for circuit Blackness, circuit whiteness inversely correlates with the use of unpublished decisions. That means that the least diverse circuits tend to rely less on unpublished decisions than those circuits that have greater diversity. As the percentage of white Americans in a particular circuit increases, the rate of unpublished decisions falls. Figure 4 demonstrates a strong inverse correlation between the two:

40.0%45.0%50.0%55.0%60.0%65.0%70.0%75.0%80.0%

Proportion of Circuit Population that Is White, Not Hispanic (2019 Census Estimates)

Circuit Blackness also correlates inversely with the use of oral argument across the circuits, but circuit whiteness has no correlation with the use of oral argument. Circuit Blackness is a moderate predictor of the volume of oral argument—driven, no doubt, by the very low rates of oral argument in the circuits with the largest percentages of Black Americans (the Fourth and Eleventh Circuits). Figures 5 and 6 illustrate the lack of correlation between circuit whiteness and oral argument and the moderate inverse correlation between circuit Blackness and oral argument. That means that circuits with higher populations of Black Americans are less likely to use oral argument in their decision-making—and, potentially, that those communities are less likely to benefit from having their cases heard (assuming, of course, that Black Americans litigate where they reside and in numbers proportionate to their representation in the community). Again, this is a story of potential impact—and not of any intentional discrimination.

Proportion of Circuit Population that Is White, Non-Hispanic Alone

Proportion of Circuit Population that Is Black (or Black in Combination with Another Race)

This is a story of class, too. Circuit wealth appears to have some relationship to reliance on second-tier procedure, as well. Table 11 identifies the wealthiest to the poorest circuits per capita. The dividing line in the table reflects the mean ($35,807).

Per Capita Income

$43,202

$42,223

$37,887

$36,190

$36,080

$35,200

$34,078

$33,529

$32,259

$31,983

$31,243

Most notably, the rate of oral argument (a feature of first-tier appellate process) moderately correlates with circuit per capita income. Figure 7 demonstrates that as circuit per capita income rises, so too does the use of oral argument to resolve cases.

$44,000

$42,000

$40,000

$38,000

$36,000

$34,000

$32,000

$30,000

0.0%5.0%10.0%15.0%20.0%25.0%30.0%35.0%40.0%

Rate of Oral Argument

Circuit wealth is also suggestive of the use of unpublished decisions, but the correlation is significantly weaker than we see between wealth and oral argument. Figure 8 depicts the relationship between circuit wealth and unpublication rates.

$44,000

$42,000

$40,000

$38,000

$36,000

$34,000

$32,000

$30,000

60.0%65.0%70.0%75.0%80.0%85.0%90.0%95.0%

Rate of Unpublished Decisions

We should be mindful that correlation does not equal causation. But the fact remains: the most diverse circuits and, in particular, those with the largest percentages of Black Americans tend to rely on more procedural shortcuts than peer circuits. Similarly, the wealthiest circuits appear more likely to extend the procedural safeguard of oral argument.

Let me underscore again that this is not a narrative of disparate treatment. I do not claim that any single court has developed or deployed its procedures to discriminate intentionally against any particular group. But what is clear is that some circuits—the Fourth, Fifth, Ninth, and Eleventh Circuits, in particular—rely more heavily on second-tier procedures. And they do so in regions of the country that are more diverse than others. Their dockets are not notably different; each circuit sees a sufficient volume of pro se, administrative, and criminal appeals where the litigant on the other side of government power is an individual. And that individual may, more often than in other circuits, be a person of color. The disparate impact of the second tier of federal appellate procedure may be yet another way in which, as Professors Michalski and Hammond recently observed, “our civil legal system [has failed] to respond to the legal needs of minority communities.”270

III. HOW TO DISTRIBUTE ARTICLE III RESOURCES IN THE U.S. COURTS OF APPEALS

If the two-tiered system of federal appellate justice is one born of necessity, it is time to reduce its necessity. The number of appeals crested at nearly 70,000 in 2005271 and has fallen to—and held steady at—around 48,000 in recent years.272 Although that trend might suggest that the need for additional judges has receded—or, at least, is no longer at crisis-like levels— focusing on gross numbers oversimplifies the problem and underappreciates the distributive justice concerns of court scholars. The courts have limped along for the better part of the last century only because they have relied increasingly on shortcuts to handle their enormous workload.273 Those shortcuts have cost the most for the system’s least powerful.274 If there’s an opportunity now to rethink a good distribution of Article III resources, then Congress should take it. What follows is a starting place for that conversation.

Lower court reform can have many features, but my focus here is on the hows, whys, and wheres of Article III resource allocation at the federal

appellate level: how many judges the courts need, why, and where. This Part first considers the historical objections and roadblocks to adding judges. Then, it discusses the problems with the existing measures for judicial need. Finally, it offers three models that Congress could adopt to codify judicial expansion at the federal appellate level in periodic intervals.

A. Judicial Objections to Adding Judges—and Why Congress Shouldn’t Listen

Other than a slight readjustment,275 there have been no new judges added to the U.S. Courts of Appeals in over thirty years. That’s not for lack of trying—at least not exactly. Since 1997,276 the Judicial Conference of the U.S. Courts has made nearly biennial277 requests to Congress for more authorized judgeships in at least some of the U.S. Courts of Appeals.278 Only six of the eleven numbered circuits have made biennial requests over the last quarter century,279 and one court (the Fifth Circuit) promptly withdrew its only request the following year.280 One circuit—the Ninth Circuit—has made a request each time; only one other court (the Sixth Circuit) has asked for

any additional help since 2011, when the Second and Third Circuits dropped repeated requests for additional judges.281

The Judicial Conference’s process for making judgeship requests depends on a crucial first step: the circuit court itself must request more judges. As Judge Brian Miller recently explained in testimony before Congress: “If the court does not make a request, the Judicial Conference does not consider recommending a judgeship for that court.”282 And, just as important, for any court to make such a request, a majority of its sitting active judges must agree to ask the Judicial Conference to evaluate circuit need.283 If the majority doesn’t want new judges, the circuit will not get new judges— unless, that is, Congress overrides the Judicial Conference’s (lack of) recommendation.

Historically, there have been at least two major stumbling blocks to adding judges to the U.S. Courts of Appeals: (1) the courts themselves, to the extent that the courts are seen to initiate the process, and (2) Congress, as the ultimate decision-maker on judicial need. The latter is a complicated gatekeeper284—and the politics of judicial appointments and court expansion are beyond the scope of my work here. As difficult as those politics are, I suggest that judicial recalcitrance makes the politics harder. It might be easier for Congress (at least in some circumstances) to find the political will for modest expansion in the lower courts when the courts themselves give an imprimatur to the request. From that perspective, judicial recalcitrance has an effect; those on either side of the aisle can weaponize deference to judicial assessments about judicial need to block additions to the federal bench. What I’ve tried to show in Part II is that judicial recalcitrance and congressional acquiescence or inertia permits systemic inequities to propagate. Whatever

deference Congress owes to courts, then, is modest—and it should be sensitive to the entrenched interests that may drive courts not to ask Congress for more help.

Judicial opposition to significant expansion of the federal appellate bench has long been strong. The judges themselves have been the best lobbyists for the status quo—even over the objection of some dissenting colleagues. Dueling American Bar Association Journal articles from the early 1990s illustrate the conflict well. In 1992, a sitting Ninth Circuit judge wrote a letter (reprinted in the Journal) to then-Senator Joseph Biden, the Chair of the Senate Judiciary Committee, begging Congress to double the federal appellate courts: “There are simply far too few of us to do the job properly,” he told then-Chair Biden.285 In response, a sitting judge on the Eleventh Circuit, Gerald Tjoflat, weighed in with his own article in the Journal: More Judges, Less Justice. His West Coast colleague had fallen “prey to the sirens of size”; the real problem in “most circuits,” he said, “is not that [the circuits] have too few judges, but that they have too many.”286 That took some chutzpah, given that the Ninth Circuit, whence the judge advocating for doubling the federal appellate bench hailed, was then and still is the only “jumbo court” at the heart of the Eleventh Circuit judge’s critique.287 Surely a judge sitting on the Ninth Circuit might have a better perspective about what that court, in particular, needs.

The debate between these judges reflects a core debate over how we view the work of courts: either as elite institutions rightly selective in how they distribute their resources, or as fundamentally egalitarian institutions that should be committed to providing public services to every litigant. The Eleventh Circuit judge’s view is typical of the former; the Ninth Circuit judge’s view is typical of the latter. Ultimately, those advocating for the federal appellate bench to remain small, elite, and static have prevailed before Congress, whether because of political infeasibility, inertia, disinterest, or deference to the courts themselves.

Those judges concerned with expanding the judiciary would likely reject my suggestion that elitism has anything to do with their reluctance. Instead, they would argue that expansion threatens the “quality” of the federal judiciary, because the increase might reduce the attention paid to the appointments process (and open the pool to less qualified contenders).288 Relatedly, some have expressed concern that adding to the bench will reduce the prestige of the position, making it less desirable for the most qualified potential appointees (a significant worry given the relative low pay for the position).289 Other objections have focused less on the judges themselves than on the potential cost to the federal government and to the stability of circuit law that may follow from larger courts.290

Professors Richman and Reynolds persuasively argued more than a decade ago that most of these objections elevate the interests of the judge over the needs of justice.291 To be sure, the quality of judges—and their satisfaction with their job—matters and may ultimately affect the quality of justice those judges provide. But it’s difficult to see how a modest increase in judges of the sort I call for in Section III.C would lead to an erosion of judicial quality. Would taking a court from twelve judges to fifteen judges, for example, so dilute the prestige of the position as to make a lifetime appointment to the federal bench undesirable to qualified applicants? Will that loss of prestige (being one of fifteen instead of one of twelve) really affect the process—more so than, say, how grueling, contentious, and invasive judicial nominations and confirmations have become?

Ultimately, the appellate judiciary’s decades-long reluctance to make consistent and sustained requests for more Article III resources should not give Congress pause. There’s a limit to the deference Congress owes the federal appellate bench when making decisions that would shape the contours of judges’ power and influence. The procedures that are now firmly entrenched in the appellate courts have done much to insulate the bench from caseload pressures. Busy courts have outsourced their work to central staff attorneys, and that radically reshapes the daily work of appellate judges. To

the extent that more resources might change expectations (including hearing more oral argument or writing more published decisions), that only would add work to the judiciary’s plate or realign individual panel priorities. In a circuit where any one judge has the power to send a case to oral argument, the addition of a new judge with a different set of values, expectations, and normative priorities might have a radical effect. That is to say nothing of the potential for partisan preferences to play some role in how a judge perceives circuit need (and which president and which Senate fills it).

Outsiders to the courts, on the other hand, largely agree that the appellate courts need more Article III resources. There’s agreement among scholars of all stripes that the federal courts of appeals need more judges, and they need more judges in no small part because their work—and the quality of justice—has suffered.292 Professor Steven Calabresi and Shams Hirji cited the “significant[] deteriorat[ion]” of the work of the U.S. Courts of Appeals as a reason to support lower court packing.293 Although Calabresi and Hirji’s plan was criticized as partisan,294 it shouldn’t be the case that every addition to the courts need be seen through a partisan lens. Nor, for that matter, should the partisan implications of additional authorized judgeships be a reason to put needed (and even overdue) reforms on ice. The next Sections offer a modest, need-based framework for deciding when and where to add federal appellate judges that hopes to lessen its partisan implications.

B. How Courts and Congress Currently Think About Judicial Need—and What’s Wrong with It

Adding judges is tricky business. Even if we can jettison partisan concerns, one challenge is that the existing framework for assessing judicial need is deeply flawed. More than a decade ago, the United States Government Accountability Office observed that “there is no empirical basis” on which the Judicial Conference bases its recommendations for adding circuit court judgeships.295 And despite considering “a number of alternatives,” the Judicial Conference has never agreed on another approach.296

Under that problematic process, the Judicial Conference measures circuit busyness based on “a standard of 500 adjusted filings per panel.”297 That means that it assumes that each appellate court can handle about 500 “adjusted filings”—I’ll come back to that in a minute—per panel per year. It’s the Judicial Conference’s policy that where a court does not exceed 500 adjusted filings per panel, it will not consider a circuit for an additional judgeship.298 To calculate adjusted filings, the Judicial Conference starts with overall annual filings, “remov[es] reopened appeals[,] and count[s] original pro se appeals as one-third of a case.”299 Then, that new number is divided by the number of three-judge panels each court can compose.

So, for example, if a circuit has 1,000 filings per year, 50 reopened appeals, and 300 pro se filings, it would have 750 adjusted filings per year (1,000 – 350 + (300 / 3 . Then, if the court happened to have six authorized judgeships, it could constitute two panels of three, so the 750 adjusted filings would be divided by two to arrive at 375 adjusted filings per panel—a number that would make our hypothetical circuit ineligible for a Judicial Conference recommendation for additional judgeships.

Although this process “represents an effort to improve” earlier need assessments, which were based on nothing more than “the restraint of individual courts of appeals,” it suffers from a fatal flaw: “It is not based on empirical data regarding the judge time that different types of cases may

require.”300 And, conveniently, the 500 per-panel adjusted-filings benchmark just happens to coincide with what the courts of appeals were generally handling at the time the benchmark was developed (around 2003 and 2004).301 All this led the Government Accountability Office to issue the damning conclusion that the Judicial Conference had provided “no empirical basis” for its assessments.302

Making matters worse, the Judicial Conference neither publishes these adjusted-filings numbers nor makes public any part of its deliberative process. The numbers, at least, are easy enough to piece together, which I have done in Table 12 below. For this table, I removed reopened appeals and adjusted the number of filings based on each circuit’s pro se filings.303 I then divided that number by the number of panels available among the circuit’s authorized judgeships. Circuits exceeding 500 adjusted filings are in yellow (slight excess), orange (more considerable excess), and red (more than one standard deviation (176.6) above the mean (519 .

2019 Appeals Commenced (Original) 900

Current Authorized Judges 11

1,338

4,080

3,454

4,474

6,666

4,184

2,575

2,794

9,747

1,731

5,061

Existing Panels

Pro Se

Adjusted Filings

2.00

4.33

4.67

5.00

5.67

5.33

9.67

4.00

4.00

1,788

1,624

2,655

3,196

2,446

1,516

1,467

4,490

2,994

1,034

2,828

2,317

2,616

4,429

2,472

1,514

1,767

6,604

1,184

2,965

Existing perPanel Adjusted Filings 182

Recall that in recent years only the Ninth Circuit has asked Congress to add judges to its ranks through the Judicial Conference process. The Ninth Circuit, indeed, exceeds the 500 adjusted-filing benchmark, but it’s not the busiest court. That honor falls to the Fifth Circuit, followed by the Eleventh Circuit.

Here, I should note another outlier: the Tenth Circuit. Its volume of appellate activity is so low that the Judicial Conference has advocated for leaving an authorized judgeship on the Tenth Circuit vacant305 (a recommendation that neither President Trump nor Congress followed when Judge Allison Eid filled the open seat following the elevation of Justice Neil Gorsuch306).

Since 1997, when this wayward process began, requests for additional judgeships in the covered circuits have dwindled. That’s not terribly surprising, given that appellate activity has fallen since 2005 and stabilized in recent years.307 What is surprising, however, is that some circuits—in particular, those very busy courts that rely the most on second-tier procedures—have almost never requested additional help from Congress. As Table 13 demonstrates, the Fourth, Fifth, and Eleventh Circuits have not asked for more judges (save for a withdrawn request from the Fifth Circuit).

1997 1999 2002 2005 2007 2009 2011 2013 2015 2017 2019 2021

2P 1P, 1T

1P*

2P, 2T

2P, 1T

6P, 3T

2P, 3T

2P, 3T

5P, 2T

5P, 2T

2T

11th Cir. Note. * denotes that the Fifth Circuit’s request was withdrawn the following year, P denotes a request for a permanent judgeship, T denotes a request for a temporary judgeship, and –1 denotes a request that Congress not fill a vacancy.

Given the unreliability of the Judicial Conference’s efforts to quantify judicial busyness, it may be helpful to look instead at cases per judge—not adjusted filings or panel filings. Although I agree that not all appeals will involve the same amount of work, we don’t have any real way to measure the relative time investment that any potential appeal may take—including pro se appeals.309 There’s also a separate normative question about whether we should devalue from the outset the time investment we expect Article III judges to make on those appeals. If we instead treat every appeal the same and divide annual case filings by authorized judgeships, we see a slightly

different picture: how many appeals filed per judge in each circuit per year.310 We might then set as a baseline either the average across all circuits or the average in a circuit that relies less on the second-tier process (such as the Seventh Circuit or the Tenth Circuit, for example).

Table 14 considers per-judge filings across the circuits. It highlights in yellow the courts where filings per judge are one standard deviation (101) or more above the mean (273) and highlights in green those courts in which filings are more than one standard deviation below the mean. Based on these numbers alone, demands per judge are greatest in the Eleventh and Fifth Circuits, followed by (to a lesser extent) the Ninth and Second Circuits; needs are least in the D.C. and Tenth Circuits.

Appeals Filed

Authorized Judges Appeals Filed Per Judge

1,118

1,284

4,698

2,877

4,680

6,401

4,306

2,609

2,766

10,400

1,737

5,314

In 1990, when Congress last added authorized judgeships to the U.S. Courts of Appeals, filings per judgeship were at 237 (after the additional judges were added).312 Today, total filings per authorized judgeship in the

geographic courts of appeals stand at 289 (48,190 annual filings across 167 authorized judgeships).313 That is a nearly 22% increase in filings per judge per year from the last time Congress addressed judgeship needs at the federal appellate level.

One approach to judicial expansion might be to take the 1990 measure as the baseline: that means we treat around 237 appeals per judge as a reasonable expectation of judicial labor on an annual basis.314 That would have a judge decide a little less than an appeal per workday each year (assuming 261 workdays per year and a two-week vacation). Concededly, that’s an arbitrary guess at a reasonable workload. At other points in time, scholars have suggested that only 80 appeals per judge per year was an outer limit;315 when that limit was exceeded, others suggested around 130 appeals might be the maximum.316 Of course, those estimates dated back to the days before technology revolutionized legal research and word processing made drafting more efficient.

Today, courts are far more efficient in many respects—thanks not just to technology, but also to the amount of nonjudicial help they have. Case dispositions on the merits per judge across the circuits far exceed 1960s and 1970s expectations. Table 15 looks at judicial output across a variety of metrics. It identifies the number of judge-issued terminations per authorized judgeship during the 2020 reporting year (including procedural terminations issued by judges). It also identifies the number of published opinions per authorized judgeship and the number of oral arguments per panel (since judges hear appeals in panels). Today, the courts average close to 200 judgeissued terminations per year across the circuits—not wildly out of proportion to expectations from the baseline set in 1990.

Total Terminations by Judges per Judge

Merits Terminations per Judge

Published Opinions per Judge

Published Opinions per Panel

Oral Arguments per Panel

Oral Arguments per Judge

This Table also underscores exactly why some courts have not asked for additional resources despite having heavy caseloads. The triage system permits judges to focus on cases that are argued orally and that end in published opinions. A judge might be responsible for writing opinions in only 34 cases per year after argument, which may result, on average, in writing only 25 fully reasoned, published decisions (because some appeals from argument will not be published). (Of course, judges will have to prepare for oral argument in more cases.) Again, Article III attention goes to a select few in the federal appellate system, and the demands of the first tier—as reasonable as they are—drive the courts’ self-perception of judicial needs.

Staff attorneys help courts absorb all that extra work—that is, those cases that don’t receive first-tier treatment. Indeed, judges might well agree that their maximum today is only about, say, 130 (first-tier) appeals per year—that is, that they cannot give sustained attention to more than that number of appeals. The current distribution of resources permits judges to do a lot more—at least on paper—while insulating them from feeling any particular need for more Article III help.

Circuits with more staff-attorney help do a lot more work with fewer Article III resources. Based on data that Judge Posner compiled in 2017,318

Table 16 identifies the number of full-time-equivalent staff attorneys across the geographic circuits and compares those numbers to the number of authorized judgeships in each circuit. Only one circuit’s ratio of staff attorneys to authorized judgeships exceeds two standard deviations above the mean: the Eleventh Circuit, which is also the court with the highest number of terminations per judge per year.319

Staff Attorneys

Ratio to Authorized Judgeships

19.5

35.5

26.5

18.5

18.5

77.5

36.4

19.8

3.3

1.9

3.0

2.5

2.0

5.4

2.6

1.1

By almost any measure—adjusted filings per panel, filings per judge, terminations per judge, and reliance on non-judicial staff—a court like the Eleventh Circuit needs more Article III resources.320 Yet it has chosen, instead, to invest in nonjudicial staff, and it has continued to distribute Article III resources sparingly.321 Any reasonable assessment of judicial need should ensure that a circuit like the Eleventh Circuit does not escape scrutiny simply because a majority of its judges have not asked Congress for more help.

Finally, none of these data take into account the work of senior or visiting judges. Across the circuits today, senior and visiting judges participate in approximately 25% of merits terminations annually322—a

significant volume of work that greatly reduces the burden on the active federal appellate judges. The help that senior and visiting judges provide varies widely across the circuits—ranging from a high of 41% of cases in the Ninth Circuit to a low of around 12% in the Fourth and Eighth Circuits.323 This labor is essentially volunteer help, and it could vanish or be reduced without warning. To be sure, senior and visiting judges are essential to the efficient operation of the federal appellate courts, but how to assess and measure their help, and how to account for it on a going-forward basis, is especially difficult.324 Were the supply of senior judges, in particular, to become much lower in some circuits, undoubtedly those courts would see an immediate need for more Article III judicial resources—the question this Article turns to next.

C. How to Add Judges to the Courts of Appeals

I propose three different models for thinking about how to add judges to the federal appellate courts. These are starting places. They are high-level roadmaps for legislation that could trigger congressional inquiry into judgeship needs automatically and at fixed intervals. That structure might lessen the partisan implications of authorizing additional judgeships; were expansion procedures codified and standardized, they would become predictable and routine. These models are also starting places in another sense: I focus on only one-half of the equation. These models describe what would trigger a congressional inquiry into the need for additional judgeships on a particular court. No matter which model (or models—these could be enacted in combination) Congress might choose, I also envision the courts providing significant feedback during the process.

1. The Average-Circuit Model The first possibility is the average-circuit model. It identifies circuit norms for the distribution of Article III resources and then seeks to bring all other circuits in line with the existing average distribution of resources. It would make modest, but routine, adjustments in authorized judgeships whenever a court falls out of step with its peers over a five-year period. This model has both a mandatory and an optional (or court-initiated) trigger.

Every five years, Congress would initiate a mandatory review for at least one new authorized judgeship in any geographic circuit in which,325 based on five-year means, any one of the following conditions is met:

• A circuit’s unpublication rate is one standard deviation higher

than the national geographic-circuit mean;

• A circuit’s rate of oral argument is one standard deviation lower

than the national geographic-circuit mean;

• A circuit’s ratio of staff attorneys to authorized judgeships is one standard deviation higher than the national geographiccircuit mean;

• A circuit’s ratio of total filings per judge is one standard deviation higher than the national geographic-circuit mean; • A circuit’s number of judge-issued terminations per judge is one standard deviation higher than the national geographiccircuit mean; or

• A circuit’s reversal rate of district court decisions is one standard deviation lower than the national geographic-circuit mean.

Whenever one of these mandatory “triggering” events occurs, the circuit would undergo review for at least one new authorized judgeship. Where more than one of these factors is present, that would create a strong presumption in favor of at least one additional judgeship that the circuit could only overcome with a compelling reason (e.g., an unusual spike in caseload or a stable and consistent bench of senior judges). A court could request more than one additional judgeship, but given the periodic (five-year) review period, that may not be necessary.

Under the optional-review alternative, any court that exceeds (or falls below, depending on the metric) the national mean in any two categories could request an additional judgeship based on a vote of a majority of the judges on the court (subject to the transparency procedures outlined below). Congress would then evaluate that request along with its mandatory-review process. Congress should ordinarily grant such court-initiated requests.

Importantly, the first five of these factors are within each court’s control. Courts that desire to keep their benches smaller could work to ensure that their decisional practices do not fall significantly out of step with peer courts. Even where the mandatory-review procedures apply, a court could

oppose additional judgeships through the court- and public-review process described below.

If the average-circuit model were in place today, Congress would be required to conduct a review of judgeship needs in the Fourth, Fifth, Eighth, and Eleventh Circuits. A majority of judges in the Second, Third, Sixth, and Ninth Circuits could initiate a request for additional judges, thereby triggering the same congressional inquiry process. Table 17 demonstrates how this model operates with current data.

OralArgument Rate 25.2%

Staff Attorney Ratio 3.3

Judge-Issued Terminations per Judge 160.7

Unpublication Rate

91.3%

65.9%

84.8%

10.5%

21.4%

20.4%

95.3%

Mean Standard Deviation Triggering Amount

1.9

3.0

2.5

2.0

2.3

5.4

2.6

1.1

3.7

303.1

94.6

397.7

215.8

164.5

245.0

282.6

221.0

167.1

230.7

271.3

120.9

356.7

221.5

67.2

288.7

Reversal Rate

9.0%

8.5%

8.2%

5.9%

9.0%

4.9%

6.9%

7.4%

Note. Circuits in red would presumptively need at least one additional judge; circuits in orange would be entitled to one judge and would have triggered the automatic-review process; circuits in green would be able to request an additional judge if the circuit majority agreed.

2. The Population Model The second model is the population model, which would entitle any circuit to additional authorized judgeships whenever, based on census data, (1) either its population-to-authorized-judgeship ratio exceeds the national geographic-circuit average or the circuit has experienced population growth that exceeds the national geographic-circuit average, and (2) the circuit is also above average in filings per judge (based on a five-year mean). Under this model, Congress would take into account population growth in a particular region, including how population shifts over time might indicate future judicial need. Congress would add a sufficient number of authorized judgeships to bring the circuit within approximately 10% of the national geographic-circuit mean for filings or terminations per judge (whichever is highest: currently, 333 for filings or 244 for terminations). One advantage of this model is that it has an easier mechanism for identifying how many judges a court may need. Thus, it may be a good resource to use in combination with the average-circuit model. A circuit would remain free to lobby for an adjustment (up or down).

Using 2020 census data for circuit population, Table 18 identifies which circuits would be eligible to receive additional authorized judgeships now and in what numbers under the population model:

Total Population (2020) 10,867,184

24,450,270

26,567,516

32,160,146

36,764,541

33,293,455

25,491,754

21,690,565

66,848,869

18,636,936

37,274,374

Percent Growth (from 2010 Census) 6.1%

Authorized Judges per Person (1:value) 1,811,197.33

Judge-Issued Terminations per Judge 160.7

8.0%

2.0%

8.6%

9.5%

6.7%

1,880,790.00

1,897,679.71

2,144,009.73

2,162,620.06

2,080,840.94

2,317,432.18

1,971,869.55

2,305,133.41

1,553,078.00

3,106,197.83

2,170,268.68

303.1

215.8

164.5

245.0

282.6

221.0

167.1

230.7

271.3

120.9

356.7

221.5

Judges to Add

Note that because this model balances both population metrics with current demands on judge time, it would not entitle circuits like the Seventh Circuit or the Tenth Circuit—which meet the population threshold—to any new judgeships (because they do not currently experience more demand than average).

3. The Gross-Filings-Benchmark Model The third model is the gross-filings-benchmark model, which would entitle any circuit to additional judges whenever it substantially exceeds the last authorized additional filings-to-judge ratio set when Congress judgeships to any federal appellate court. At present, that model would use the 1990 ratio—the number of filings per judge after the addition of authorized judges in 1990—as an acceptable benchmark. Accordingly, any circuit that exceeds 237 filings per judge would be eligible for additional judgeships. The court would be eligible for whatever number of judges would bring the court to within approximately 20% of the previous benchmark (so, around 284 filings per judge).328 The court would not be eligible, however, if, with the addition of a new judgeship, the court’s new per-judge filings rate would fall below the previous benchmark. That carveout ensures that a court that only slightly exceeds the previous benchmark does not receive a new judge. Table 19 identifies where judges might be added under this model based on the 1990 benchmark.

Current Authorized Judgeships 6

Judges to Add 0

New Total Filings per Judge N/A

0*

286.0

285.0

282.0

263.0

249.0

288.0

281.0

Note. Even though the Seventh Circuit exceeds the 237-per-judge filing benchmark, because the addition of a new authorized judgeship would reduce per-judge filings below the benchmark, the Seventh Circuit would not receive a judge.

The gross-filings benchmark appears to be the most aggressive of these various approaches. That’s partly because it plays catch-up for lost time (meaning, the thirty years of stagnant benches). The other two models anticipate regular recalibration over time (every five or ten years, depending on the model), thus resulting in continual, but exceedingly modest, adjustments to the bench. Regardless, the gross-filings benchmark is consistent with the pace of growth that the courts of appeals have seen in the past, and it would reflect an overall expansion of 18.6%, which is commensurate with the percentage growth in filings since 1990. Appendix B identifies the pace of judgeship growth in the courts of appeals over time. The last large omnibus judge bill in 1984329 grew the federal bench by 18.2%, and the largest ever (in 1978)330 grew the bench by 36.1%.

* * *

Under any of these models, whenever a court is eligible for additional judgeships, that eligibility would trigger an internal court process for each

court to develop a recommendation on judgeship needs to Congress. That process should engage those with dissenting views and the public, ensuring that Congress hears from anyone with an interest in judicial need.331 If a majority of the court agrees that additional judges are needed, the court should provide a recommendation to Congress to that effect (including how many judges and in what duty stations or states). If a majority of a court disagrees, however, it should provide Congress with a detailed, specific explanation as to why additional Article III resources are not needed. No matter the ultimate recommendation, any judge who disagrees with the majority’s recommendation should have an opportunity to submit a minority report to Congress. The circuit should also prepare a public record of any comments it receives during the review process.

No matter which model is adopted, measures should be put in place to disincentivize overreliance on court staff. To discourage overdelegation, Congress should adopt a measure that would discourage courts from relying on nonjudicial staff instead of asking for Article III resources to meet actual needs. As part of any judicial-need model, Congress should freeze that circuit’s ratio of staff attorneys to authorized judgeships at the then-existing national average whenever (1) a circuit opposes the addition of judges despite qualifying for an additional judge and (2) Congress determines not to add judges based on that opposition. That limitation will let courts know in advance that if they oppose additional judgeships, they may not be eligible for more nonjudicial resources to absorb their need.

Finally, I should say a word about the political implications of these proposals. I favor a periodic and standardized process that might lessen the political stakes of any increase in the federal appellate bench. If Congress were to revisit this question every five or ten years, then sometimes Republicans might be in charge, sometimes Democrats, and sometimes we’d have a divided government.332 It’s also conceivable that, to further lessen the

partisan implications of any reform, Congress could set new authorized judgeships on a time delay, adding one per year over time, thus ensuring that different political arrangements participate in the process. Although that’s not been done before with judgeship bills, these are no longer ordinary political times.

This Section has offered a glimpse at how Congress might address judicial need at the federal appellate level now and into the future. The next and final Section offers a brief word on why adding judges matters and how doing so might ameliorate some of the inequities wrought by the judiciary’s seeming path dependence to its procedural shortcuts.

D. Why Adding Authorized Judgeships Matters

The systemic deficits and potential disparate impact those deficits may impose on vulnerable communities are not easy problems to solve. By focusing here on adding authorized judgeships, I am suggesting that such additions are at least the beginning of an answer to a complex problem. Even as judicial workload has eased, the courts have not begun publishing more cases or hearing more oral argument. That suggests that having more judicial help may not offer any more procedural opportunities for marginalized litigants, which is one of the concerns animating this project. Judges might just take more leisure time instead.

If that’s the case, how will adding judges make any difference? First, in the aggregate, it appears that there is a correlation between appeals per judge—at least one measure of court busyness—and the percentage of cases argued orally and the circuit publication rate. As Figures 9 and 10 demonstrate,333 as the number of appeals filed per judge go up, the rate of oral argument declines and the rate of unpublication increases. Put differently, the busier the court, the more likely it is to rely on procedural shortcuts. That’s not always the case, of course, but the overall trendline suggests a relationship between the volume of work before each judge and the opportunities for oral argument and reasoned, precedential decisionmaking.

45.0%

Appeals Filed per Judge

Appeals Filed per Judge

Finally, it may matter whom the President adds to the federal bench. If judges have grown to like the status quo because it permits them to pay selective attention to certain cases, it is conceivable that efforts to diversify the bench through additional judgeships might diversify and expand circuit priorities at the same time—especially if judges with different professional backgrounds and experiences are added to the courts.335 Those who come to the federal bench from private practice may be comfortable with a large degree of delegation; those who come to the bench from nonprofit backgrounds, who were public defenders, or who were academics may, instead, prefer to do the work themselves. It may be no accident, for example, that the Seventh Circuit has both a high volume of oral argument and a (relatively) low volume of unpublished decisions given the academic bent of that court.336 Moreover, circuit cultures may shift—the so-called “Posner effect,” for example, on the Seventh Circuit—propelling individual judges to do more judicial work themselves.337 Congress should not just add authorized judgeships, but the President should appoint judges who will bring a strong service-minded work ethic to the federal bench.

Finally, I recognize that any growth in the federal appellate bench— even modest, incremental growth like I suggest here—may disrupt the courts themselves. Not all change may be seen as good change by existing members of a court. I do not mean my proposals to minimize the real risk that some courts grow too large, either. But, I suggest, fears over where exactly that line may be—is a court of twenty-nine too big but a court of seventeen okay?—should not stop us from considering modest, incremental reform. When and if any one court grows too large, then Congress can revisit that question—just as it did when it split the old Fifth Circuit in two.

CONCLUSION

Congress bears responsibility for ensuring the even-handed distribution of federal appellate resources throughout the nation. In an effort to spur innovation and to reimagine the appellate courts in a time of plenty, not scarcity, this Article offers several models for doing so. Foremost, it urges Congress to distribute additional Article III resources even over the objection of reluctant courts. That reluctance may reflect the values of slim majorities or entrenched attitudes that have done more harm than good—especially to vulnerable litigants who more often find their cases resolved through secondtier processes. The federal courts, and the litigants who rely on them, deserve more than making do with little.

It is entirely possible that the addition of more judges to the federal bench will produce more appeals—the familiar problem of adding more lanes to a highway only to create more, rather than less, traffic. I don’t conceive of that as a “problem” when it comes to administering a judicial system, however. (Or at least I don’t think it should be added to the “con” column.) The opportunity to petition a neutral decision-maker for redress for a constitutional violation or a private injury—including the opportunity to appeal—is a cornerstone of civilized society. If adding more judges produces more work for courts, then that suggests the possibility of more access to justice (as well as the possibility that there had been pent-up demand). This Article’s central premise is that such a possibility is a far better objective for the federal appellate courts than maintenance of the status quo.338

APPENDIX A: CIRCUIT-COMPOSITION DATA

This Appendix contains docket-composition data averaged over the last five years for each of the geographic circuit courts. The rightmost column notes the standard deviation for each category in each circuit to capture the variation during that five-year period for each circuit. Cells marked yellow contain figures that exceed one standard deviation from the nationwide fiveyear means that appear in Table 6. Cells marked red contain figures that exceed two standard deviations of the nationwide five-year means. All circuit composition data derive from each table B-1 of the 2016–2020 Judicial Business reports, averaged over the five-year period.339

D.C. Circuit Appeals

8.2%

24.3%

14.9%

35.4%

9.7%

0.1%

First Circuit Appeals

30.9%

7.6%

5.4%

27.9%

7.4%

11.3%

2.6%

3.2%

Second Circuit Appeals

5.1%

5.2%

36.1%

1.9%

16.6%

1.6%

1.9%

Third Circuit Appeals

13.6%

4.0%

17.7%

28.7%

2.5%

6.5%

Fourth Circuit Appeals

24.7%

14.7%

16.1%

16.8%

2.7%

Fifth Circuit Appeals

31.5%

7.8%

2.5%

18.0%

19.3%

7.5%

12.5%

0.4%

2.7%

4.0%

Sixth Circuit Appeals

9.8%

3.9%

19.8%

21.1%

0.9%

17.3%

4.3%

Seventh Circuit Appeals

17.8%

10.1%

22.1%

25.8%

13.2%

3.1%

Eighth Circuit Appeals

28.2%

11.3%

4.4%

17.5%

18.0%

3.15%

7.3%

Ninth Circuit Appeals

11.7%

4.9%

18.4%

21.2%

27.7%

8.7%

0.9%

Tenth Circuit Appeals

20.7%

11.7%

15.6%

27.8%

3.0%

5.8%

Eleventh Circuit Appeals

21.2%

3.6%

17.2%

0.4%

3.0%

9.4%

APPENDIX B: AUTHORIZED JUDGESHIPS OVER TIME

Table B1 reflects the regular and periodic growth in the geographic U.S. Courts of Appeals over time. Where a negative number appears, it reflects a redistribution of a judge from one circuit to another (ordinarily when the latter circuit was created). The percentage of growth reflected in each new authorized-judgeship bill appears in the rightmost column.

1st Cir. 2

2d Cir. 3

3d Cir. 2

4th Cir. 2

5th Cir. 2

6th Cir. 2

7th Cir. 2

8th Cir. 2

9th Cir. 2

Total Added 19

Agg. Total 19

Pct. Growth

Year

–12

15.8%

4.5%

8.7%

3.6%

6.7%

3.1%

6.1%

2.9%

4.3%

12.5%

5.6%

4.6%

14.7%

7.7%

15.5%

36.1%

18.2%

7.1%

–0.6%

--- FOOTNOTES ---

Copyright 2022 by Merritt E. McAlister

Printed in U.S.A. Vol. 116, No. 5

III. HOW TO DISTRIBUTE ARTICLE III RESOURCES IN THE U.S. COURTS OF

1 See, e.g., Daniel Epps & Ganesh Sitaraman, How to Save the Supreme Court, 129 YALE L.J. 148, 169 (2019) (arguing that significant reform is required to save the Supreme Court); Opinion, How to Fix the Supreme Court, N.Y. TIMES (Oct. 27, 2020), https://www.nytimes.com/interactive/2020/10/27/ opinion/supreme-court-reform.html (last visited Nov. 11, 2021) (compiling arguments for and against reforming the federal courts).

2 See, e.g., Leah Litman, Expand the Lower Courts, N.Y. TIMES (Oct. 27, 2020), https://www. nytimes.com/interactive/2020/10/27/opinion/expand-federal-courts-judges.html#litman [https://perma. cc/XA3H-M53Q] (stating that the vast majority of cases do not reach the Supreme Court but are instead resolved in the lower courts, and advocating for lower federal court reform); Xiao Wang, In Defense of (Circuit) Court-Packing, 119 MICH. L. REV. ONLINE 32, 33 (2020), https://repository.law.umich.edu/cgi/ viewcontent.cgi?article=1035&context=mlr_online [https://perma.cc/L3JU-LWUC] (explaining that federal “circuit courts are actually the last resort for most litigants” and advocating for lower federal court reform).

3 The same could be said of the study of federal courts more generally. See, e.g., Justin WeinsteinTull, The Structures of Local Courts, 106 VA. L. REV. 1031, 1039–40 (2020) (observing that litigants filed 86.2 million cases in nonfederal courts in 2015 and only 343,176 cases in federal courts).

4 Professors Daniel Epps and William Ortman have argued that the “legal system would benefit if, every Term, the Supreme Court were forced to decide some unquestionably unimportant cases—some run-of-the-mill appeals dealing with the kinds of ordinary and seemingly inconsequential legal questions that the lower courts resolve every day.” Daniel Epps & William Ortman, The Lottery Docket, 116 MICH. L. REV. 705, 707 (2018). The very ordinariness of many of these cases, combined with the Supreme Court’s obsessive focus on circuit splits, has created an accountability gap insulating many—if not most— court of appeals decisions from the possibility of Supreme Court review. Id. at 721–23 (explaining how “sharply circumscribed” the Supreme Court’s ability to monitor output of the circuit courts is under current certiorari process).

5 See infra Appendix A (identifying the docket composition of the U.S. Courts of Appeals). Social Security and immigration appeals fall within the broad category of administrative appeals; habeas corpus and prisoner civil rights claims are within the categories of U.S. and private prisoner claims.

6 See ADMIN. OFF. OF THE U.S. CTS., JUDICIAL BUSINESS 2020, at tbl.B-9 (2021) [hereinafter JUDICIAL BUSINESS 2020], https://www.uscourts.gov/sites/default/files/data_tables/jb_b9_0930.2020. pdf [https://perma.cc/W2PC-9BUJ]. I rely heavily on data from several installments of Judicial Business in this Article, which for convenience are cited simply as “JUDICIAL BUSINESS [year],” with a hyperlink to each table the first time it appears. For all reports, see Judicial Business of the United States Courts, U.S. CTS., https://www.uscourts.gov/statistics-reports/analysis-reports/judicial-business-united-statescourts [https://perma.cc/8XLG-AMY4]. Note that the reporting period for the Administrative Office’s Judicial Business publication runs from October 1 to September 30 each year. Id.

7 See infra Section I.A (discussing the distribution of federal appellate resources); see also Judith Resnik, Tiers, 57 S. CAL. L. REV. 837, 840–44 (1984) (exploring procedure’s “value-expressive functions” and identifying methods of process that preserve lower court decisions out of concerns about resource conservation for habeas corpus petitions, civil rights claims, and employment law claims).

8 I have previously suggested that there may be three tiers of appellate process. See Merritt E. McAlister, Missing Decisions, 169 U. PA. L. REV. 1101, 1118 (2021). Further, some circuits (such as the Ninth Circuit) have erected tiers within tiers by identifying cases within each tier that need a harder look than others. See, e.g., J. Clifford Wallace, Improving the Appellate Process Worldwide Through Maximizing Judicial Resources, 38 VAND. J. TRANSNAT’L L. 187, 197 (2005) (discussing how nonjudicial personnel screen and weigh cases—at least in the Ninth Circuit—according to relative difficulty, “to apportion [the court’s] workload more equally among judges”).

9 Part I will describe these tiers, and the triage system that sorts cases into tiers, in detail. Although the courts themselves acknowledge having argument and nonargument tracks for cases, see, e.g., 2D CIR. LOC. R. 34.2 (listing the classes of cases subject to the court’s nonargument calendar), my use of the term “tiers” reflects the scholarly consensus as to the effect of the courts’ procedural shifts over the last halfcentury at the federal appellate level. See generally infra Section I.A (describing procedures and tiers of federal appellate process).

10 For an extended examination of the staff-attorney position across circuits, see RICHARD A. POSNER, REFORMING THE FEDERAL JUDICIARY: MY FORMER COURT NEEDS TO OVERHAUL ITS STAFF ATTORNEY PROGRAM AND BEGIN TELEVISING ITS ORAL ARGUMENTS 49–61 (2017).

11 Marin K. Levy, Judicial Attention as a Scarce Resource: A Preliminary Defense of How Judges Allocate Time Across Cases in the Federal Courts of Appeals, 81 GEO. WASH. L. REV. 401, 416 (2013) [hereinafter Levy, Judicial Attention] (“[C]ases that are not scheduled for oral argument tend to be prepared by staff attorneys, and those same cases almost always result in unpublished dispositions.”).

12 See id. at 416–19 (discussing categorical presumptions that drive tracking decisions). 13 See POSNER, supra note 10, at 164–65 (identifying similarities and differences across circuits in

the use of staff attorneys).

14 See WILLIAM M. RICHMAN & WILLIAM L. REYNOLDS, INJUSTICE ON APPEAL: THE UNITED STATES COURTS OF APPEALS IN CRISIS, at xii, 115 (2013) (describing changes in appellate court structure and case processing as “judicial activism of the highest order” and concluding that this “unilateral change in [their] function . . . is deeply subversive of the entire constitutional scheme”).

15 For a discussion of the history of this caseload growth, see Penelope Pether, Inequitable Injunctions: The Scandal of Private Judging in the U.S. Courts, 56 STAN. L. REV. 1435, 1444 & n.30 (2004).

16 See, e.g., Bert I. Huang, Lightened Scrutiny, 124 HARV. L. REV. 1109, 1112 & n.9 (2011) (tracing origins of the “crisis of volume” narrative (citing Henry J. Friendly, Averting the Flood by Lessening the Flow, 59 CORNELL L. REV. 634, 642 (1974 .

17 Brown v. Allen, 344 U.S. 443, 536–37 (1953) (Jackson, J., concurring in the judgment). 18 See Boyce F. Martin, Jr., In Defense of Unpublished Opinions, 60 OHIO ST. L.J. 177, 181–82

(1999).

19 See, e.g., Resnik, supra note 7, at 843, 939–64 (documenting procedural shifts in district courts for resolving large volumes of habeas corpus petitions, civil rights claims, and employment law matters). 20 The magistrate system began in 1968 with the Federal Magistrates Act, Pub. L. No. 90-578, § 82 Stat. 1107 (1968) (codified as amended in scattered sections of 18 U.S.C. and 28 U.S.C.). See generally Peter G. McCabe, The Federal Magistrate Act of 1979, 16 HARV. J. ON LEGIS. 343, 344 (1979) (describing the history of the federal magistrates).

21 Resnik, supra note 7, at 840. 22 Id. at 841.

23 My use of the term “distributive justice” borrows from Professor Matthew Shapiro’s recent work to identify “which specific goods access to justice actually comprises.” Matthew A. Shapiro, Distributing Civil Justice, 109 GEO. L.J. 1473, 1477 (2021). Professor Shapiro categorizes access to “judicial resources” as one of the goods to be distributed in a civil justice system and observes that judicial decisions to allocate time and attention differentially may be seen both “as a form of disparate treatment” and “disparate impact,” depending on the circumstances. See id. at 1491–92 (discussing how “from an access-to-justice perspective, the worry is that courts are giving certain categories of cases short shrift so that they can focus on the ones that, from many judges’ perspectives, really matter”).

24 See, e.g., Thomas E. Baker, Intramural Reforms: How the U.S. Courts of Appeals Have Helped Themselves, 22 FLA. ST. U. L. REV. 913, 949 (1995) (“Judges have gone to great lengths to defend the level of delegation that already has taken place.”).

25 Penelope Pether, Sorcerers, Not Apprentices: How Judicial Clerks and Staff Attorneys Impoverish

U.S. Law, 39 ARIZ. ST. L.J. 1, 29 (2007).

26 RICHMAN & REYNOLDS, supra note 14; see also THOMAS E. BAKER, RATIONING JUSTICE ON APPEAL: THE PROBLEMS OF THE U.S. COURTS OF APPEALS 14–30 (1994) (making similar observations about the erosion of the “federal appellate tradition” and warning that such tradition may “disappear altogether” without intervention).

27 See Merritt E. McAlister, “Downright Indifference”: Examining Unpublished Decisions in the Federal Courts of Appeals, 118 MICH. L. REV. 533, 554 & fig.5 (2020) (discussing the lack of correlation between use of unpublished decisions and overall caseload volume).

28 In the 2005 reporting year, in which the U.S. Courts of Appeals received 68,473 total filings (their tbl.B-1, https://www.uscourts.gov/sites/default/files/ most ever), JUDICIAL BUSINESS 2005, at statistics_import_dir/b1_1.pdf [https://perma.cc/C52K-XDYN], the nationwide unpublication rate was

81.6%, id. tbl.S-3, https://www.uscourts.gov/sites/default/files/statistics_import_dir/s3_1.pdf [https:// perma.cc/F572-72JH], and the nationwide oral-argument rate was 30.1%, id. tbl.S-1, https://www. uscourts.gov/sites/default/files/statistics_import_dir/s1_1.pdf [https://perma.cc/9FQQ-N8CZ]. In the 2020 reporting year, with more than 20,000 fewer filings, JUDICIAL BUSINESS 2020, at tbl.B-1, https:// www.uscourts.gov/sites/default/files/data_tables/jb_b1_0930.2020.pdf [https://perma.cc/3S6H-2UGW] (48,190 total filings), the nationwide unpublication rate was 87%, id. tbl.B-12, https://www. uscourts.gov/sites/default/files/data_tables/jb_b12_0930.2020.pdf [https://perma.cc/252V-ED5L], and the oral-argument rate was 19.3%, id. tbl.B-10, https://www.uscourts.gov/sites/default/files/data_tables/ jb_b10_0930.2020.pdf [https://perma.cc/7C69-GXZ2].

29 See McAlister, supra note 27, at 583–94 (arguing that minimally reasoned unpublished

decisions—a product of second-tier review—threaten procedural justice values).

30 See infra Figure 2 (identifying strong correlation between rates of oral argument and favorable

outcomes on appeal).

31 See infra Section II.A (discussing the disparate use of shortcuts across appellate courts).

32 Cf. SUP. CT. R. 10 (identifying considerations for Supreme Court review). At bottom, these disparities defy Supreme Court review because that model of adjudication is case specific, whereas the trends we see in how the courts deploy procedures operate systemically. The Court could influence trends through its oversight of the rulemaking process or through the Judicial Conference.

33 See Pether, supra note 15, at 1445–46 (tracing the Fourth Circuit’s role in institutionalizing nonpublication and that innovation’s connection to pro se prisoner litigation). Although some judges on the old Fifth Circuit were hailed as heroes in the post-Brown Civil Rights Era, that court also innovated summary procedures for many criminal and habeas corpus claims. See Charles R. Haworth, Screening and Summary Procedures in the United States Courts of Appeals, 1973 WASH. U. L.Q. 257, 277–78 (discussing screening procedures for different types of cases).

34 See generally Jon O. Newman, The Second Circuit’s Expedited Adjudication of Asylum Cases: A Case Study of a Judicial Response to an Unprecedented Problem of Caseload Management, 74 BROOK. L. REV. 429, 430–34 (2008) (describing the development of the nonargument track for immigration appeals).

35 See infra Section III.A (discussing disagreement among the federal bench).

36 See Marin K. Levy, The Promise of Senior Judges, 115 NW. U. L. REV. 1227, 1232–33 (2021) [hereinafter Levy, Senior Judges] (arguing that there are “no easy fixes” to make federal appellate courts bigger and advocating for lowering institutional barriers to taking senior status to increase judicial resources); see also Marin K. Levy, Judging Justice on Appeal, 123 YALE L.J. 2386, 2403–07 (2014) [hereinafter Levy, Judging] (arguing that expanding the federal appellate bench is an impractical solution to the appellate courts’ workload problem).

37 Levy, Senior Judges, supra note 36, at 1235; see also David R. Stras & Ryan W. Scott, Retaining Life Tenure: The Case for a “Golden Parachute,” 83 WASH. U. L.Q. 1397, 1400–02 (2005) (advocating for retirement-benefit and other institutional reforms that might incentivize federal judges—in particular, Supreme Court Justices—to retire at an appropriate age or should they become mentally or physically disabled).

38 Levy, Senior Judges, supra note 36, at 1232 (stating that courts need to be bigger but recognizing the political difficulty of adding judges and seeking to avoid partisanship that could impair the judiciary). 39 I also recognize that it’s just as possible that the courts have been reluctant to ask Congress for more judges because they recognize the political difficulty of that task. But the debate over judicial expansion discussed in Section III.B appears to run deeper than that and reflects concerns about the nature of the federal judiciary that extend well beyond how it intersects with the political realm.

40 Compare JUD. CONF. OF THE U.S., JUDGESHIP RECOMMENDATIONS tbl.1 (2021), https://www. uscourts.gov/sites/default/files/2021_judicial_conference_recommendations_0.pdf [https://perma.cc/ 3R7P-F3PQ] (requesting two permanent judgeships for the Ninth Circuit), with JUD. CONF. OF THE U.S., (2019), https://www.uscourts.gov/sites/default/files/2019_ JUDGESHIP RECOMMENDATIONS judicial_conference_judgeship_recommendations_0.pdf [https://perma.cc/SLQ5-P72C] (requesting five permanent judgeships for the Ninth Circuit).

tbl.1

41 See infra Appendix B (charting the history of requests for additional authorized judgeships). Congress added three judges to the Fifth Circuit through two bills in 1984 and 1990. See Bankruptcy Amendments and Federal Judgeship Act of 1984, Pub. L. No. 98-353, § 201(a)(1), 98 Stat. 333, 346; Judicial Improvements Act of 1990, Pub. L. No. 101-650, § 202(a)(3), 104 Stat. 5089, 5099.

42 Compare ADMIN. OFF. OF THE U.S. CTS., U.S. COURTS OF APPEALS - JUDICIAL CASELOAD PROFILE, https://www.uscourts.gov/sites/default/files/data_tables/fcms_na_appprofile0331.2021.pdf [https://perma.cc/PMT9-UGR7] (detailing that the Eleventh Circuit had 4,955 total filings in the twelvemonth period ending on March 31, 2021), with Robert A. Ainsworth, Jr., Fifth Circuit Court of Appeals Reorganization Act of 1980, 1981 B.Y.U. L. REV. 523, 523 n.2 (detailing that the old Fifth Circuit had 4,404 total filings in the twelve-month period ending on September 30, 1980 before its October 1, 1981 reorganization).

43 Gerald Bard Tjoflat, More Judges, Less Justice, 79 A.B.A. J. 70, 72 (1993) (describing the old

Fifth Circuit, the author’s former court, as a “jumbo court”).

44 Ainsworth, supra note 42, at 523. 45 See infra notes 282–283 and accompanying text (discussing the current process for evaluating

judicial need).

46 A Judiciary Diminished Is Justice Denied: The Constitution, the Senate, and the Vacancy Crisis in the Federal Judiciary: Hearing Before the Subcomm. on the Const. of the Comm. on the Judiciary, H.R., 107th Cong. 90 (2002) (statement of J.L. Edmondson, C.J., U.S. Court of Appeals for the Eleventh Circuit).

47 Compare ADMIN. OFF. OF THE U.S. CTS., FEDERAL JUDICIAL CASELOAD STATISTICS 2002 tbl.B (2002), https://www.uscourts.gov/sites/default/files/statistics_import_dir/b00dec02.pdf [https://perma. cc/46W4-93EF] (detailing that the Eleventh Circuit had 7,431 filings in the twelve-month period ending on December 31, 2002), with Ainsworth, supra note 42, at 523 n.2 (detailing that the old Fifth Circuit had 4,404 filings in the twelve-month period before its October 1, 1981 reorganization).

48 See Tjoflat, supra note 43, at 70 (arguing for a small federal appellate judiciary and emphasizing

difficulties encountered on the jumbo old Fifth Circuit).

49 See generally Arthur D. Hellman, Assessing Judgeship Needs in the Federal Courts of Appeals: Policy Choices and Process Concerns, 5 J. APP. PRAC. & PROCESS 239, 264–67 (2003) (making recommendations to reform the Judicial Conference process related to assessment of judicial need).

50 This history has been recounted in greater detail by others. See, e.g., RICHMAN & REYNOLDS, supra note 14, at 1–16 (discussing the caseload crisis leading to internal procedural reforms); see also Haworth, supra note 33, at 258–64 (explaining the need for courts to develop more efficient procedures in response to caseload pressures).

51 The current structure of the U.S. Courts of Appeals is not the same, either. For a succinct history of the evolution of the federal appellate system, see Paul D. Carrington, The Function of the Civil Appeal: A Late-Century View, 38 S.C. L. REV. 411, 411‒31 (1987).

52 See William M. Richman & William L. Reynolds, Elitism, Expediency, and the New Certiorari: Requiem for the Learned Hand Tradition, 81 CORNELL L. REV. 273, 278 (1996) (describing “traditional appellate procedure”).

53 See id. 54 See RICHMAN & REYNOLDS, supra note 14, at 115 (describing reforms as being “unilateral”). 55 See DANIEL J. MEADOR, APPELLATE COURTS: STAFF AND PROCESS IN THE CRISIS OF VOLUME 72‒ 75 (1974) (examining how two state appellate courts responded to a “crisis of volume” as models for broader institutional reforms); see also PAUL D. CARRINGTON, DANIEL J. MEADOR & MAURICE ROSENBERG, JUSTICE ON APPEAL, at V (1976) (“A cause for deep concern about appellate justice is the runaway inflation in the volume of appeals.”); Irving R. Kaufman, The Pre-Argument Conference: An Appellate Procedural Reform, 74 COLUM. L. REV. 1094, 1094 (1974) (recognizing the “caseload crisis affecting” the Second Circuit and “other federal appellate courts”).

56 DONALD R. SONGER, REGINALD S. SHEEHAN & SUSAN B. HAIRE, CONTINUITY AND CHANGE ON

THE UNITED STATES COURTS OF APPEALS 15 tbl.1.2 (2000).

57 ADMIN. OFF. OF THE U.S. CTS., U.S. COURTS OF APPEALS ADDITIONAL AUTHORIZED JUDGESHIPS [hereinafter ADDITIONAL AUTHORIZED JUDGESHIPS], http://www.uscourts.gov/sites/default/files/ appealsauth.pdf [https://perma.cc/KEW7-GYL8].

58 Notably, these statistics are not weighted by circuit, and some circuits see greater volume than others. See Peter S. Menell & Ryan Vacca, Revisiting and Confronting the Federal Judiciary Capacity “Crisis”: Charting a Path for Federal Judiciary Reform, 108 CALIF. L. REV. 789, 859 fig.21 (2020) (illustrating that in 2017 the Fifth Circuit and Eleventh Circuit had the heaviest workload per judge). 59 See McAlister, supra note 27, at 552 fig.3 (identifying a downward trend in caseload volume). 60 Thomas E. Baker & Denis J. Hauptly, Taking Another Measure of the “Crisis of Volume” in the U.S. Courts of Appeals, 51 WASH. & LEE L. REV. 97, 103 (1994); see ADDITIONAL AUTHORIZED JUDGESHIPS, supra note 57.

61 JUDICIAL BUSINESS 2020, at tbl.B, https://www.uscourts.gov/sites/default/files/data_tables/jb_b_

0930.2020.pdf [https://perma.cc/7E8E-Y2CW].

62 Lawrence M. Friedman, Legal Rules and the Process of Social Change, 19 STAN. L. REV. 786,

798 (1967).

63 See generally William L. Reynolds & William M. Richman, The Non-Precedential Precedent— Limited Publication and No-Citation Rules in the United States Courts of Appeals, 78 COLUM. L. REV. 1167, 1168 (1978) (tracing the early history of unpublished decisions).

64 See William L. Reynolds & William M. Richman, An Evaluation of Limited Publication in the United States Courts of Appeals: The Price of Reform, 48 U. CHI. L. REV. 573, 577 n.12, 581 (1981) (“West publishes only opinions designated for publication by the several circuits.”).

65 We thought of precedent differently too, but that’s beyond the scope of this Article. See, e.g., Hart v. Massanari, 266 F.3d 1155, 1168 (9th Cir. 2001) (“The modern concept of binding precedent—where a single opinion sets the course on a particular point of law and must be followed by courts at the same level and lower within a pyramidal judicial hierarchy—came about only gradually over the nineteenth and early twentieth centuries.”).

66 See Anastasoff v. United States, 223 F.3d 898, 903 (8th Cir. 2000) (discussing the history of

publication), vacated as moot, 235 F.3d 1054 (8th Cir.) (en banc).

67 See RICHMAN & REYNOLDS, supra note 14, at 11 (“Limited publication has been actively

considered by the federal judicial establishment since the end of World War II.”).

68 There was another, even more practical reason to reduce publication: the growing volume of lawbooks. See Reynolds & Richman, supra note 63, at 1168–69 (describing concerns over the growing volume of decisional law); see also John B. Winslow, The Courts and the Papermills, 10 ILL. L. REV. 157, 158 (1915) (“The law library of the future staggers the imagination as one thinks of multitudes of shelves . . . all loaded with their many volumes of precious precedents.”).

69 See McAlister, supra note 27, at 538–41 (discussing and developing an efficiency rationale for

unpublished decisions that are poorly and lightly reasoned).

70 See Reynolds & Richman, supra note 64, at 601–02, 606 (observing that the “lower quality of

unpublished opinions may be the most important of the costs of limited publication”).

71 See Pether, supra note 25, at 6, 17 (discussing perceptions of staff attorneys’ work). 72 See McAlister, supra note 8, at 1101, 1111–14, 1111 n.48, 1146 (2021) (discussing no-citation

rules and objections to those rules, and also collecting sources).

73 FED. R. APP. P. 32.1. 74 See Reynolds & Richman, supra note 63, at 1180–81 (discussing different circuit approaches to

so-called “no citation rules”).

75 See McAlister, supra note 8, at 1116–17 (describing categories of appellate terminations on the

merits).

76 JUDICIAL BUSINESS 2020, at tbl.B-12 (indicating that 87% of merits terminations nationwide are

unpublished).

77 See Brian Soucek, Copy-Paste Precedent, 13 J. APP. PRAC. & PROCESS 153, 153–54 (2012) (examining how unpublished decisions are “repeatedly copied and pasted into other unpublished opinions,” infecting such decisions with errors that are hard to fix “because no one thinks to do so”).

78 See McAlister, supra note 27, at 550–51 (discussing disparities in publication rates across circuits). 79 See Eric M. Fraser, David K. Kessler, Matthew J.B. Lawrence & Stephen A. Calhoun, The Jurisdiction of the D.C. Circuit, 23 CORNELL J.L. & PUB. POL’Y 131, 132 & n.2 (2013) (discussing “the accepted wisdom that the D.C. Circuit is ‘a special court’ that tackles a disproportionate share of thorny administrative and regulatory cases affecting a wide range of important national public policy issues”).

80 See Pether, supra note 25, at 7 (describing the scholarly critique of “institutionalized

unpublication” as “multiplicitous and damning”).

81 The literature on unpublished decisions is vast and varied; it challenges every aspect of the practice. See, e.g., Pether, supra note 15, at 1483, 1511 & nn.441–42 (identifying damaging “rule of law” effects of unpublication, including that it enables power in that repeat litigants can rig the system in their favor); Reynolds & Richman, supra note 64, at 574 (evaluating and critiquing the substance of unpublished decisions); David Dunn, Note, Unreported Decisions in the United States Courts of Appeals, 63 CORNELL L. REV. 128, 141–45 (1977) (discussing cases that examined the no-citation rule under principles of due process and equal protection); Salem M. Katsh & Alex V. Chachkes, Constitutionality of “No-Citation” Rules, 3 J. APP. PRAC. & PROCESS 287, 297–315 (2001) (arguing that the no-citation rule violates the First Amendment); see also David S. Caudill, Parades of Horribles, Circles of Hell: Ethical Dimensions of the Publication Controversy, 62 WASH. & LEE L. REV. 1653, 1657 (2005) (raising ethical concerns with unpublished decisions); Shenoa L. Payne, The Ethical Conundrums of Unpublished Opinions, 44 WILLAMETTE L. REV. 723, 725 (2008) (similar); J. Lyn Entrikin Goering, Legal Fiction of the “Unpublished” Kind: The Surreal Paradox of No-Citation Rules and the Ethical Duty of Candor, 1 SETON HALL CIR. REV. 27, 31 (2005) (similar).

82 Preeminent Eighth Circuit Judge Richard Arnold and now-disgraced but once-preeminent Ninth Circuit Judge Alex Kozinski disagreed over the constitutionality of unpublished decisions in Anastasoff v. United States, 223 F.3d 898, 899 (8th Cir. 2000), vacated as moot, 235 F.3d 1054 (8th Cir.) (en banc), and Hart v. Massanari, 266 F.3d 1155, 1175 (9th Cir. 2001).

83 For a discussion of the contentious and lengthy debates over FED. R. APP. P. 32.1, see the work of the Reporter to the Advisory Committee, Patrick J. Schiltz, Much Ado About Little: Explaining the Sturm Und Drang over the Citation of Unpublished Opinions, 62 WASH. & LEE L. REV. 1429, 1458–90 (2005). 84 See infra Table 3 and accompanying text (describing the current use of unpublication across

circuits and five-year averages).

85 Compare JUDICIAL BUSINESS 2005, at tbl.B (68,473 appeals commenced), with JUDICIAL BUSINESS 2020, at tbl.B (48,190 appeals commenced); compare JUDICIAL BUSINESS 2005, at tbl.S-3 (81.6% unpublication rate), with JUDICIAL BUSINESS 2020, at tbl.B-12 (87% unpublication rate).

86 See RICHMAN & REYNOLDS, supra note 14, at 92 (discussing reduction in oral argument and observing that in Administrative Office Reporting Year 2010, “only 26.4 percent of all appeals were terminated following an ‘oral hearing,’” whereas “[i]n 1985, 56 percent of terminations followed oral argument”).

87 FED. R. APP. P. 34(a)(2) (“Oral argument must be allowed in every case unless a panel of three judges who have examined the briefs and record unanimously agrees that oral argument is unnecessary for any of the following reasons: (A) the appeal is frivolous; (B) the dispositive issue or issues have been

authoritatively decided; or (C) the facts and legal arguments are adequately presented in the briefs and record, and the decisional process would not be significantly aided by oral argument.”).

88 Id. 34(a)(1). 89 See David R. Cleveland & Steven Wisotsky, The Decline of Oral Argument in the Federal Courts of Appeals: A Modest Proposal for Reform, 13 J. APP. PRAC. & PROCESS 119, 138–39 (2012) (discussing transparency issues related to the denial of oral argument).

90 Id. at 119. 91 Id. at 119–20 & nn.2–3; see also Marin K. Levy, The Mechanics of Federal Appeals: Uniformity and Case Management in the Circuit Courts, 61 DUKE L.J. 315, 345–47 (2011) [hereinafter Levy, Mechanics] (discussing the decline of oral argument and the related increased reliance on centralized judicial staff attorneys).

92 JUDICIAL BUSINESS 2020, at tbl.B-10. 93 RICHMAN & REYNOLDS, supra note 14, at 93. 94 See Levy, Mechanics, supra note 91, at 355–56 (discussing various approaches to oral argument). 95 See, e.g., Robert J. Martineau, The Value of Appellate Oral Argument: A Challenge to the Conventional Wisdom, 72 IOWA L. REV. 1, 3 (1986) (recognizing that “extent to which many appellate courts have eliminated or severely restricted oral argument” has “caused more concern among both appellate judges and appellate lawyers” than other procedural reforms); Cleveland & Wisotsky, supra note 89, at 121 (arguing that reduction in oral argument has “weakened or diluted” “values implicit in appellate review” and noting that there has been a concomitant decline in reversal rates); see also RICHMAN & REYNOLDS, supra note 14, at 94 (expressing concern that courts deny oral argument inconsistently and “for no apparent reason”).

96 See, e.g., John B. Oakley, The Screening of Appeals: The Ninth Circuit’s Experience in the Eighties and Innovations for the Nineties, 1991 B.Y.U. L. REV. 859, 860 (observing that case-screening procedures that lead to nonargument involve “a significantly lesser degree of personal attention by judges”).

97 Cleveland & Wisotsky, supra note 89, at 121. 98 RICHMAN & REYNOLDS, supra note 14, at 97. 99 See also Pether, supra note 25, at 7 (arguing that “[t]he institution of the staff attorney/habeas

clerk/pro se clerk has its shared origins with modern institutionalized unpublication”).

100 For a history of the law clerk, see JOHN BILYEU OAKLEY & ROBERT S. THOMPSON, LAW CLERKS AND THE JUDICIAL PROCESS: PERCEPTIONS OF THE QUALITIES AND FUNCTIONS OF LAW CLERKS IN AMERICAN COURTS (1980).

101 Charles H. Sheldon, The Evolution of Law Clerking with the Washington Supreme Court: From

“Elbow Clerks” to “Puisne Judges,” 24 GONZ. L. REV. 45, 48 (1988).

102 Act of June 17, 1930, ch. 509, 46 Stat. 774. 103 OAKLEY & THOMPSON, supra note 100, at 19–20. 104 RICHMAN & REYNOLDS, supra note 14 at 97–98; see also Todd C. Peppers, Michael W. Giles & Bridget Tainer-Parkins, Surgeons or Scribes? The Role of United States Court of Appeals Law Clerks in “Appellate Triage,” 98 MARQ. L. REV. 313, 315 (2014) (noting that federal appellate judges may hire four law clerks “if they forgo a second secretarial position”). I know of at least one federal appellate judge who hired five clerks (one in a position that was designated as a staff attorney) to work in chambers.

105 See Todd C. Peppers, Birth of an Institution: Horace Gray and the Lost Law Clerks, 32 J. SUP.

CT. HIST. 229, 231 (2007).

106 OAKLEY & THOMPSON, supra note 100, at 10–11 (discussing the origins of the law clerk role and

Justice Gray’s paternity of the practice).

107 Id. at 11. 108 Id. In 1886, Congress funded an “assistant” position, which usually performed secretarial work. Id. at 15. In 1919, Congress funded both a “law clerk” and a “stenographic clerk” position for each Justice. Id. at 16–17.

109 See Tony Mauro, Diversity and Supreme Court Law Clerks, 98 MARQ. L. REV. 361, 361, 363–65 (2014) (discussing the fact that overwhelmingly Supreme Court clerks have been white men). The first female law clerk was not hired until 1944; the next was not hired until 1966. See Mark R. Brown, Gender Discrimination in the Supreme Court’s Clerkship Selection Process, 75 OR. L. REV. 359, 362 n.19 (1996). In the early days, the Justices preferred law students tapped by law professors and deans at Harvard, Yale, and Columbia. See OAKLEY & THOMPSON, supra note 100, at 17 (tracing the early history of Supreme Court law clerks and connections to these schools). They still do. See Jason Iuliano & Avery Stewart, The New Diversity Crisis in the Federal Judiciary, 84 TENN. L. REV. 247, 296 (2016) (gathering data to show that approximately 70% of Supreme Court clerks hired between October Term 1950 and October Term 2015 came from one of just six schools: Harvard (24.8%), Yale (19.0%), the University of Chicago (8.2%), Stanford (6.7%), the University of Virginia (5.6%), or Columbia (5.5% .

110 OAKLEY & THOMPSON, supra note 100, at 14. 111 MEADOR, supra note 55, at 73 (describing law clerks’ responsibilities).

112 See OAKLEY & THOMPSON, supra note 100, at 23–24 (“While isolated examples of utilization of a centralized pool of legal assistants to a court have existed for some years, it was the creation of a central staff at the Michigan Court of Appeals a decade ago [in 1968] that ‘gave impetus and focus to the contemporary development’ of the process.” (quoting MEADOR, supra note 55, at 17 ; see also DANIEL JOHN MEADOR & JORDANA SIMONE BERNSTEIN, APPELLATE COURTS IN THE UNITED STATES 108–09 (1994) (similar).

113 MEADOR & BERNSTEIN, supra note 112, at 108 (“Central staff attorneys are lawyers employed by an appellate court to work for the court as an entity. . . . [T]hey have no close relationship to any particular judge. They are organized centrally under the supervision of a lawyer who is the head of the staff and who in turn is answerable to the court.”).

114 Id. at 109 (observing that the use of a central legal staff “spread gradually, meeting with resistance

from appellate judges”).

115 Id. 116 Id. 117 Id. (“The other development motivating courts to employ central staffs was the continuing rise in their caseloads. Each year during the 1970s brought an increase in appellate filings in many courts all across the country.”).

118 See id. at 110 (“In effect, the ABA standards provided a formal legitimation of the concept by the

American legal-judicial community.”).

119 RICHMAN & REYNOLDS, supra note 14, at 103. Professor Meador traces the federal lineage of the central judicial staff to the United States Court of Military Appeals, which had employed research attorneys, called commissioners, since its creation in 1951. See FED. JUD. CTR., CENTRAL LEGAL STAFFS IN THE UNITED STATES COURTS OF APPEALS 1 (1978) [hereinafter CENTRAL LEGAL STAFFS], (citing https://www.ojp.gov/pdffiles1/Digitization/49649NCJRS.pdf MEADOR, supra note 55, at 218).

[https://perma.cc/RT52-YU8A]

120 CENTRAL LEGAL STAFFS, supra note 119, at 1–2. 121 Id. 122 Pether, supra note 25, at 53–54 (observing that judges view “‘have-not’ litigants . . . as

troublesome or vexatious [and] their appeals [as] unmeritorious”).

123 On this point, I disagree in some respects with Professor Pether, who suggests that work by elbow clerks and staff attorneys is almost interchangeable—indeed, even her title suggests as much. See id. at 1 (titling the work “Sorcerers, Not Apprentices: How Judicial Clerks and Staff Attorneys Impoverish U.S. Law”). Judge Posner’s recent work on staff attorneys is more consistent with my personal experience: judges and their clerks work on the “important” cases; staff attorneys generally handle the rest. See POSNER, supra note 10, at 50–59 (discussing the role of staff attorneys across the circuits); see id. at 162– 63 (noting that very few staff-attorney cases result in published decisions across the circuits, except in the Seventh Circuit, where approximately 13% are published).

124 RICHMAN & REYNOLDS, supra note 14, at 104. 125 CENTRAL LEGAL STAFFS, supra note 119, at 2. 126 Act of Oct. 5, 1974, Pub. L. No. 93-433, 88 Stat. 1187, 1202 (“[T]he chief judge of each circuit may appoint a senior law clerk to the court at not more than $30,000 per annum.”); see also CENTRAL LEGAL STAFFS, supra note 119, at 2 (identifying 1975 as the first year for congressional recognition of central staff).

127 See CENTRAL LEGAL STAFFS, supra note 119, at 6–35 (surveying the state of central-staff

programs across the circuits by the late 1970s).

128 Federal Courts Improvement Act of 1982, Pub. L. No. 97-164, § 120(c)(1), 96 Stat. 25, 34

(codified at 28 U.S.C. § 715). 129 28 U.S.C. § 715. 130 Congress created the Judicial Conference in 1922 to generate policy guidelines for administration

of the federal courts. 28 U.S.C. § 331.

131 See Baker, supra note 24, at 949 (discussing the history of staff-attorney program and Judicial

Conference recommendations regarding staffing ratios).

132 Id. 133 RICHMAN & REYNOLDS, supra note 14, at 103. 134 See MEADOR & BERNSTEIN, supra note 112, at 112 (“The staff director or chief staff attorney, as this officer is usually called, is typically the only member of the staff who has contact with the court’s judges, and that contact is often limited to dealings with the chief judge.”). But see POSNER, supra note 10, at 59 (noting limited interactions between judges and staff attorneys in most circuits).

135 See POSNER, supra note 10, at 57 (“Line attorneys [in the Eleventh Circuit] must complete a quota

of cases by the end of every two months.”).

136 See Pether, supra note 25, at 53 (describing the “work that produces unpublished opinions and

the work of producing them” as being “perceived as low status”).

137 Id. at 54 (noting that staff attorneys “generally do not have the blue chip credentials of those in

elite elbow clerkships”).

138 See, e.g., POSNER, supra note 10, at 59 (noting that Seventh Circuit staff attorneys reported to Judge Posner that staff attorneys in other circuits were “reluctant to recommend reversal of a district judge”).

139 See id. (noting that “many of the staff attorneys in the other circuits [outside of the Seventh

Circuit] were reluctant to recommend reversal of a district court judge”).

140 See Pether, supra note 25, at 18–19 (arguing that increased reliance on staff attorneys has impoverished the federal appellate system); POSNER, supra note 10, at 62–64 (arguing for reform of the staff-attorney system across the circuits).

141 That is not to say that the courts did not ask Congress for any Article III help, nor to suggest that some courts (mainly, the Ninth Circuit) have not continued to do so. See infra Table 13 and accompanying text (discussing recent requests to add authorized judgeships). But those requests did not keep pace with caseload demands, and judges routinely expressed concern over growing the federal bench too much. See infra notes 285–287 and accompanying text (discussing judicial opposition to growing the federal bench). 142 See POSNER, supra note 10, at 49–61, 162–65 (discussing the use of staff-attorneys’ offices across circuits); Levy, Mechanics, supra note 91, at 333–39 (discussing screening procedures in the D.C., First,

Second, Third, and Fourth Circuits and noting “complexity, variation, and importance” of courts’ casemanagement procedures); see also Newman, supra note 34, at 432 (“In most other circuits [outside of the Second Circuit], all cases are reviewed at an early stage to determine whether they warrant oral argument or should be decided on the briefs without argument.”).

143 See Levy, Mechanics, supra note 91, at 333–40 (noting that “[d]epending on the circuit, this screening is performed by counsel in the clerk’s office, by staff attorneys, or by judges,” and describing the nature of the initial review across the circuits by different decision-makers).

144 Id. at 339. This is, to be sure, a meaningful and necessary check in an effective triage system, but the extent to which individual judges exercise that discretion and oversight is unknowable. Moreover, the data on certiorari grants discussed below suggest that, at least in some circuits, the check may be less effective. See infra Section II.A (discussing the extent to which the Supreme Court has reviewed unpublished decisions through its certiorari process). Further, to the extent that circuits have norms against staff-attorney recommendations of reversal—a norm that Judge Posner’s work suggests—that too would undermine the meaningfulness of this systemic check, given the staff attorneys’ primacy in this process. See POSNER, supra note 10, at 135.

145 POSNER, supra note 10, at 164; see also Levy, Mechanics, supra note 91, at 351 (“[T]he Third Circuit does not have its staff attorneys screen cases for oral argument. Rather, only particular kinds of cases—pro se cases that do not involve direct criminal appeals and most immigration cases—are sent to panels that do not hear argument.”).

146 See Levy, Mechanics, supra note 91, at 336–37 (discussing Second Circuit procedures). 147 See Newman, supra note 34, at 429; Levy, Mechanics, supra note 91, at 349–50 (describing

Second Circuit Non-Argument Calendar procedures).

148 See POSNER, supra note 10, at 164–65. 149 See id. at 56 (“In the Tenth Circuit, judges perform all screening of cases. Nonargument cases are then moved to the staff attorney’s office, whose director assigns line staff attorneys to a mix of cases.”).

150 See infra note 305 and accompanying text (identifying the volume of appellate activity in the Tenth Circuit and discussing the recommendation that a vacancy on that court go unfilled); infra Table 12.

151 Haworth, supra note 33, at 264. 152 Id. at 317 (“The thesis of the statistical test was simply to determine if a relation existed between screening and summary procedures and affirmances of the lower court decision. The test showed that the relation apparently exists, but it is not proof that the screening procedures were the cause of the relation.”); see id. at 314–16 (discussing comparison with the Third Circuit).

153 Id. at 319. 154 Id. 155 JUDICIAL BUSINESS 2020, at tbl.B-5, https://www.uscourts.gov/sites/default/files/data_tables/

jb_b5_0930.2020.pdf [https://perma.cc/7DH2-CA63].

156 See Haworth, supra note 33, at 268–69 (describing screening procedures in the circuits as of the

early 1970s).

157 See id. at 317–19 (discussing concerns over deciding cases without oral argument). 158 See McAlister, supra note 8, at 1134 tbl.3 (identifying the percentage of decisions issued by each

circuit that have been described as unreasoned by courts).

159 See McAlister, supra note 27, at 575–82 (describing decisional practices of courts that purport to issue only reasoned decisions and describing a body of unpublished decisions that are “Kafkaesque” in nature).

160 See RICHMAN & REYNOLDS, supra note 14, at 179–80 (seeming to advocate for adding forty new judges to the federal appellate bench); Pether, supra note 15, at 1535 (noting that solutions “to stopping the practices of private judging would be to effectively publish all opinions, even if some have a designated inferior precedential status,” to use a panel of interested persons to make decisions about publication decisions, or both).

161 JUDITH A. MCKENNA, FED. JUD. CTR., STRUCTURAL AND OTHER ALTERNATIVES FOR THE FEDERAL COURTS OF APPEALS: REPORT TO THE UNITED STATES CONGRESS AND THE JUDICIAL CONFERENCE OF THE UNITED STATES 37 (1993).

162 See, e.g., RICHARD A. POSNER, THE FEDERAL COURTS: CHALLENGE AND REFORM 168–69 (1996) (“Given the workload of the federal courts of appeals today, the realistic choice is not between limited publication, on the one hand, and, on the other, improving and then publishing all the opinions that are not published today; it is . . . . between giving the parties reasons for the decision of their appeal and not giving them reasons even though the appeal is not frivolous.”); Newman, supra note 34, at 437 (explaining and defending the Second Circuit’s use of a nonargument calendar as a “fair[], effective[],

and efficient[]” response to “an extraordinary challenge”); Martin, supra note 18, at 178–79 (“Whereas academicians tend to see unpublished opinions as causing a variety of systemic problems, judges tend to see them as a necessary, and not necessarily evil, part of the job.”).

163 See Levy, Judicial Attention, supra note 11, at 440, 443–46 (concluding “that many of the courts’ current case management practices are loosely consistent with . . . the twin goals of error correction and law development” and also suggesting reforms). Professor Levy also appears to be of the view that the courts should have more judges too. “The Need for New Lower Court Judgeships, 30 Years in the Making”: Hearing Before the Subcomm. on Cts., Intell. Prop., and the Internet of the H. Comm. on the Judiciary, 117th Cong. 9 (2021) (testimony of Marin K. Levy, Professor of Law, Duke University School of Law) [hereinafter Levy Testimony], https://docs.house.gov/meetings/JU/JU03/20210224/111237/ HHRG-117-JU03-Wstate-LevyM-20210224-U1.pdf [https://perma.cc/K6DT-88HN].

164 See Ryan W. Copus, Statistical Precedent: Allocating Judicial Attention, 73 VAND. L. REV. 605, 611 (2020) (advocating for the use of artificial intelligence to help judges identify and sort cases that require more and less judicial attention).

165 Id. 166 See FED. R. APP. P. 32.1 (authorizing citation of any unpublished decision by any U.S. court of

appeals issued after January 1, 2007).

167 Even this modest reform was contentious. See Pether, supra note 25, at 2–3 (observing that Rule 32.1 involved “a bitter dispute over the apparently trivial issue of a proposed and eventually enacted uniform citation rule” that “split[] the ranks of the federal judiciary”).

168 Thomas E. Baker, Applied Freakonomics: Explaining the “Crisis of Volume,” 8 J. APP. PRAC. &

PROCESS 101, 113 (2006).

169 Id. at 114 (“[W]e now take for granted what were once characterized as ‘emergency’ procedures. We have lowered our expectations for appellate procedure. . . . We all have internalized the postmodern norms of the minimalist procedural paradigm.”).

170 In so doing, I distinguish between purposeful racism that intentionally harms people of color and government acts that otherwise appear facially race neutral yet impose a disproportionate harm on racial minorities and maintain white supremacy. See, e.g., Derrick Bell, Racial Realism, 24 CONN. L. REV. 363, 369 (1992) (“[D]espite . . . belief in the ‘rule of law’ . . . abstract principles lead to legal results that harm blacks and perpetuate their inferior status.”); I. Bennett Capers, Critical Race Theory and Criminal Justice, 12 OHIO ST. J. CRIM. L. 1, 2 (2014) (“Critical Race Theorists have not placed their faith in neutral procedures and the substantive doctrines of formal equality; . . . both the procedure and the substances of American law, including American antidiscrimination law, are structured to maintain white privilege.” (quoting Francisco Valdes, Jerome McCristal Culp & Angela P. Harris, Battles Waged, Won, and Lost: Critical Race Theory at the Turn of the Millennium, in CROSSROADS, DIRECTIONS, AND A NEW CRITICAL RACE THEORY 1, 1 (Francisco Valdes, Jerome McCristal Culp & Angela P. Harris eds., 2002 .

171 See, e.g., Washington v. Davis, 426 U.S. 229, 239 (1976) (drawing a distinction between facially neutral laws that have a racially disproportionate impact and those that “reflect[] a racially discriminatory purpose”).

172 This problem was foreseen long ago by Professors Reynolds and Richman. See Reynolds & Richman, supra note 63, at 1191–94 (arguing that judges may not be able to predict when decisions should be precedential).

173 Judith Resnik, Statement for the Record for the Public Hearing, Presidential Commission on the Supreme Court of the United States at 19 (June 30, 2021) [hereinafter Resnik Testimony], https://www. whitehouse.gov/wp-content/uploads/2021/06/Resnik-PDF-Presidential-Commission.pdf [https://perma. cc/B722-FF7J].

174 These numbers count consolidated cases as one opinion; I excluded opinions from within the Supreme Court’s original jurisdiction or decisions arising from three-judge district court panels within the Court’s mandatory jurisdiction. I coded the decisions based on lower court and publication status of the decision from which certiorari was sought. All data are on file with author.

TABLE 1: CERTIORARI GRANTS BY CIRCUIT

175 See JUDICIAL BUSINESS 2020, at tbl.B-12.

TABLE 2: OUTCOMES AT THE U.S. SUPREME COURT

176 140 S. Ct. 1731 (2020). 177 See Bostock v. Clayton Cnty. Bd. of Comm’rs, 723 F. App’x 964, 965 (2018) (noting that the decision in an earlier published case foreclosed the argument that Title VII protected plaintiff from discrimination because of sexual orientation), rev’d sub nom. Bostock v. Clayton County, 140 S. Ct. 1731. 178 See, e.g., Uzuegbunam v. Preczewski, 781 F. App’x 824, 832 (11th Cir. 2019) (holding that a claim for nominal damages in a religious-freedom case did not avoid mootness concerns), rev’d, 141 S. Ct. 792 (2021); United States v. Manrique, 618 F. App’x 579, 582–83 (11th Cir. 2015) (clarifying jurisdictional rules related to deferred order of restitution and timely filing of notice of appeal), aff’d, 137 S. Ct. 1266 (2017).

179 Data are derived from En Banc Poll Orders, U.S. CT. OF APPEALS FOR THE 11TH CIR., https:// www.ca11.uscourts.gov/enbanc-poll-orders [https://perma.cc/A9FN-MUEL], and are on file with author.

180 Young v. Borders, 620 F. App’x 889, 890 (11th Cir. 2015). 181 Bostock v. Clayton Cnty. Bd. of Comm’rs, 894 F.3d 1335 (11th Cir. 2018) (denying rehearing en

banc).

182 See Young v. Borders, 850 F.3d 1274 (11th Cir. 2017) (denying rehearing en banc). 183 Resnik Testimony, supra note 173, at 20. 184 Id. 185 See Melissa M. Serfass & Jessie Wallace Cranford, Federal and State Court Rules Governing Publication and Citation of Opinions: An Update, 6 J. APP. PRAC. & PROCESS 349, 351–56 (2004) (culling circuit-specific publication rules).

186 See generally Ben Grunwald, Strategic Publication, 92 TUL. L. REV. 745, 759–60 (2018)

(demonstrating that panels that disagree may bargain for publication or unpublication strategically).

187 See SUP. CT. R. 10 (explaining that a petition for a writ of certiorari “will be granted only for compelling reasons,” which include where (1) a U.S. court of appeals “has entered a decision in conflict with the decision of another” U.S. court of appeals “on the same important matter,” (2) a U.S. court of appeals is in conflict with a state court of last resort, (3) a U.S. court of appeals “has so far departed from the accepted and usual course of judicial proceedings, or sanctioned such a departure by a lower court, as to call for an exercise of this Court’s supervisory power,” or (4) a U.S. court of appeals “has decided an important question of federal law that has not been, but should be, settled by” the Supreme Court).

188 Id. 189 See McAlister, supra note 8, at 1126–32 (explaining that unpublished decisions remain “missing” from commercial legal databases despite technological innovations and legal reforms providing for access).

190 William H. Pryor, Jr., Conservatives Should Oppose Expanding the Federal Courts, N.Y. TIMES (Nov. 29, 2017), https://www.nytimes.com/2017/11/29/opinion/conservatives-expanding-federal-courts. html [https://perma.cc/PM23-Y8SU].

191 Edith H. Jones, Back to the Future for Federal Appeals Courts: Rationing Federal Justice by Recovering Limited Jurisdiction, 73 TEX. L. REV. 1485, 1493 (1995) (reviewing BAKER, supra note 26).

192 Id. 193 See McAlister, supra note 27, at 554 fig.5 (showing a lack of correlation between caseload

volume and reliance on unpublished decisions).

194 Id. at 550 fig.1. 195 See supra notes 65–76 and accompanying text (discussing the need for unpublished decisions). 196 See McAlister, supra note 27, at 555 fig.6. 197 See id. at 558 fig.9; see also id. at 555–58 (discussing the rise in pro se litigation and the positive

correlation between that increase and the continued and increasing reliance on unpublished decisions).

198 See JUDICIAL BUSINESS 2020, at tbl.B-1 (48,190 appeals commenced); JUDICIAL BUSINESS 2019, at tbl.B-1 (2019), https://www.uscourts.gov/sites/default/files/data_tables/jb_b1_0930.2019.pdf [https:// perma.cc/ET2T-DWMB] (48,486 appeals commenced).

199 In 1992, there were 47,013 new appeals filed in the U.S. Courts of Appeals. Michael C. Gizzi, Examining the Crisis of Volume in the U.S. Courts of Appeals, 77 JUDICATURE 96, 96 (1993). That year, the nationwide rate of unpublished opinions was 71.7%. Michael Hannon, A Closer Look at Unpublished Opinions in the United States Courts of Appeals, 3 J. APP. PRAC. & PROCESS 199, 204 (2001).

200 JUDICIAL BUSINESS 2020, at tbl.B-12. 201 RICHMAN & REYNOLDS, supra note 14, at 118 (“[M]andatory review [in the U.S. Courts of Appeals] has . . . come to mean something quite different [as a result of the triage procedures]. Now the trial court’s decision is subject to review at the discretion of the judges . . . .”). But see Levy, Judging, supra note 36, at 2402 (observing, in response, that this argument may contain some “hyperbole” and that appellate judges are still involved—albeit to a lesser extent—in resolving second-tier cases on the merits). 202 See LEE EPSTEIN, WILLIAM M. LANDES & RICHARD A. POSNER, THE BEHAVIOR OF FEDERAL JUDGES: A THEORETICAL AND EMPIRICAL STUDY OF RATIONAL CHOICE 36–37 (2013) (“Especially as judges age, and either their energy wanes or boredom sets in, or they realize that further effort will do little to alter their career achievements or reputation, we can expect them to delegate more of their work to law clerks and other staff . . . .”); see also Thomas Y. Davies, Gresham’s Law Revisited: Expedited Processing Techniques and the Allocation of Appellate Resources, 6 JUST. SYS. J. 372, 372 (1981) (finding, based on an analysis of one state intermediate appellate court, that appellate resources generally will be allocated for civil appeals, rather than criminal appeals, for a host of reasons and irrespective of the incidence of “frivolous” criminal appeals).

203 See EPSTEIN ET AL., supra note 202, at 25–26 (theorizing judicial behavior through lens of “the judge as a participant in a labor market—that is, as a worker”—and arguing that “the most fruitful theory to guide empirical study of judicial behavior is one of self-interested behavior, broadly understood, in a labor-market setting”). See generally RICHARD A. POSNER, HOW JUDGES THINK 7 (2008) (advancing a labor-market hypothesis for understanding judges as workers).

204 See Daniel Klerman, Nonpromotion and Judicial Independence, 72 S. CAL. L. REV. 455, 456 (1999) (“[B]ecause of the pyramidal structure of the federal judiciary, the typical judge’s chance of promotion is so low that it is unlikely that desire for promotion affects the decisions of more than a handful of judges.”).

205 In the trial court context, Professor Resnik has explored how the label “housekeeping” (and that term’s use) has had a similar expressive effect to identify low-value (and gendered) work. See Judith Resnik, Housekeeping: The Nature and Allocation of Work in Federal Trial Courts, 24 GA. L. REV. 909, 916, 933 (1990).

206 See Martin, supra note 18, at 181–82 (identifying little-merit appeals as a major problem affecting

the federal appeals courts).

207 Surely there were meritless appeals filed well before the 1960s—and possibly many of them— but they did not enter the judicial and scholarly conscience in the way we think of them today. See, e.g., Pether, supra note 15, at 1449–51, 1450 n.65 (tracing the early history of unpublished decisions as a response to the onslaught of criminal and civil rights claims perceived to be meritless).

208 See McAlister, supra note 27, at 543–44 (discussing the relationship between unpublished

decisions, appellate caseload growth, and the expansion of federally created rights).

209 To obtain this number, I searched Westlaw’s federal appellate courts’ database for the word

“meritless” using a date range for decisions issued before 1960.

210 To obtain this number, I searched Westlaw’s federal appellate courts’ database for the word “meritless” using a date range for each decade after 1960 (because searching multiple decades sometimes exceeded Westlaw’s limit on 10,000 returned database hits).

211 JUDICIAL BUSINESS 2020, at tbl.B-5. 212 Id. 213 One can also see a chicken-and-egg problem: is the reversal rate so low because the courts are

misers with their time, or have the courts been appropriately shrewd given the paucity of meaty issues?

214 Alexander A. Reinert, Screening Out Innovation: The Merits of Meritless Litigation, 89 IND. L.J. 1191, 1198–1201 (2014) (explaining that frivolous cases are those “based on a conception of the law that

is both unsupported and unsupportable under any existing principle” or “based on allegations of outlandish facts”).

215 See id. at 1226–30 (arguing that meritless litigation has value because meritless cases (1) can be “necessary to the development of a doctrine,” (2) can “prompt more direct change in the law,” including from congressional responses, (3) can inform “proper institutional conduct and behavior,” and (4) can “produce stability” by reducing the risk that parties “resort to disruptive self-help”).

216 Estelle v. Gamble, 429 U.S. 97, 106 (1976). 217 See, e.g., Youmans v. Gagnon, 626 F.3d 557, 564 (11th Cir. 2010) (“A serious medical need is one that has been diagnosed by a physician as mandating treatment or one that is so obvious that even a lay person would easily recognize the necessity for a doctor’s attention.” (quoting Mann v. Taser Int’l, Inc., 588 F.3d 1291, 1306–07 (11th Cir. 2009 .

218 The diagnosis of “transsexualism” first appeared in the Diagnostic and Statistical Manual of Mental Disorders in 1980; in 1994, that diagnosis was replaced with “gender identity disorder in adults and adolescence” to reduce stigma. See Gender Dysphoria Diagnosis, AM. PSYCHIATRIC ASS’N (Nov. 2017), https://www.psychiatry.org/psychiatrists/cultural-competency/education/transgender-and-gendernonconforming-patients/gender-dysphoria-diagnosis [https://perma.cc/8L2P-PN4Y].

219 Compare Edmo v. Corizon, Inc., 935 F.3d 757, 803 (9th Cir. 2019) (holding that a prisoner established deliberate indifference to medical needs for the failure of the prison to provide gender confirmation surgery), with Gibson v. Collier, 920 F.3d 212, 221–28 (5th Cir. 2019) (rejecting similar claim on the basis that there is no medical consensus on proper treatment for gender dysphoria).

220 Reinert, supra note 214, at 1193.

221 EPSTEIN ET AL., supra note 202, at 55. That choice led to some criticism of the work. See Renée Cohn Jubelirer, The Behavior of Federal Judges: The “Careerist” in Robes, 97 JUDICATURE 98, 103–04 (2013).

222 EPSTEIN ET AL., supra note 202, at 55. 223 Id. at 56. 224 Id. at 55–56. 225 See supra notes 190–191 and accompanying text (discussing candid observations from federal

appellate judges about the nature of unpublished decisions).

226 See, e.g., Katharyn Kennedy, Richard V. Aghababian, Lucille Gans & C. Phuli Lewis, Triage: Techniques and Applications in Decisionmaking, 28 ANNALS EMERGENCY MED. 136, 138 (1996) (describing triage systems in the historical, military, and mass-casualty contexts).

227 RICHMAN & REYNOLDS, supra note 14, at 94. 228 Id.

TABLE 3: UNPUBLICATION RATES OVER THE LAST FIVE YEARS

tbl.B-12, https://www.uscourts.gov/sites/default/files/data_tables/jb_b12_0930.2017.pdf

229 See JUDICIAL BUSINESS 2020, at tbl.B-12 (87.0%); JUDICIAL BUSINESS 2019, at tbl.B-12, https://www.uscourts.gov/sites/default/files/data_tables/jb_b12_0930.2019.pdf [https://perma.cc/8TYE4JMW] (87.0%); JUDICIAL BUSINESS 2018, at tbl.B-12, https://www.uscourts.gov/sites/default/files/ data_tables/jb_b12_0930.2018.pdf [https://perma.cc/XW26-E9C5] (88.2%); JUDICIAL BUSINESS 2017, at [https:// perma.cc/9R7D-ER67] (88.2%); JUDICIAL BUSINESS 2016, at tbl.B-12, https://www.uscourts.gov/sites/ default/files/data_tables/jb_b12_0930.2016.pdf (88.7%); JUDICIAL BUSINESS 2015, at tbl.B-12, https://www.uscourts.gov/sites/default/files/data_tables/B12Sep15.pdf [https://perma.cc/7CNB-5RQH] (87.0%); JUDICIAL BUSINESS 2014, at tbl.B-12, https://www.uscourts. gov/sites/default/files/statistics_import_dir/B12Sep14.pdf (87.7%); JUDICIAL BUSINESS 2013, at tbl.B-12, https://www.uscourts.gov/sites/default/files/statistics_import_dir/ S03Sep13.pdf [https://perma.cc/5A2E-Y6YN] (88.2%).

[https://perma.cc/4Q92-ZGAR]

[https://perma.cc/A7SW-8P2F]

230 JUDICIAL BUSINESS 2020, at tbl.B-10 (19.3%); JUDICIAL BUSINESS 2019, at tbl.B-10, https:// www.uscourts.gov/sites/default/files/data_tables/jb_b10_0930.2019.pdf [https://perma.cc/83FN-HJHS]

TABLE 4: ORAL-ARGUMENT RATES OVER THE LAST FIVE YEARS231

(20.4%); JUDICIAL BUSINESS 2018, at tbl.B-10, https://www.uscourts.gov/sites/default/files/data_tables/ jb_b10_0930.2018.pdf [https://perma.cc/2ZVY-CENX] (20.1%); JUDICIAL BUSINESS 2017, at tbl.B-10, https://www.uscourts.gov/sites/default/files/data_tables/jb_b10_0930.2017.pdf [https://perma.cc/NW9X -F57X] (18.8%); JUDICIAL BUSINESS 2016, at tbl.B-10, https://www.uscourts.gov/sites/default/files/ data_tables/jb_b10_0930.2016.pdf [https://perma.cc/798A-9SHB] (17.5%); JUDICIAL BUSINESS 2015, at tbl.B-10, https://www.uscourts.gov/sites/default/files/data_tables/B10Sep15.pdf [https://perma.cc/FS48NPG3] (20.5%); JUDICIAL BUSINESS 2014, at tbl.B-10, https://www.uscourts.gov/sites/default/files/ statistics_import_dir/B10Sep14.pdf [https://perma.cc/S7CG-B6RR] (19.5%); JUDICIAL BUSINESS 2013, at [https:// perma.cc/DE5V-UNVV] (19.6%); JUDICIAL BUSINESS 2012, at tbl.S-1, https://www.uscourts.gov/sites/ default/files/statistics_import_dir/S01Sep12.pdf [https://perma.cc/HWB2-U8VV] (18.8%).

tbl.S-1, https://www.uscourts.gov/sites/default/files/statistics_import_dir/S01Sep13.pdf

231 Data are derived from JUDICIAL BUSINESS 2020, at tbl.B-10. For a discussion of the Judicial Business reports as the only data the Administrative Office of the U.S. Courts publishes on publication status and which “focuses on ‘merits’ terminations,” see McAlister, supra note 8, at 1120.

232 Figures 1–8 present the results of linear regression fits using the Excel LINEST function, which returns a slope and intercept along with the square of the Pearson correlation coefficient R. The values of R are given in footnotes for each of the figures. The qualitative adjectival degrees of the correlation of strong, moderate, weak, etc., are rather arbitrary. That said, this Article follows the approach of JACOB COHEN, STATISTICAL POWER ANALYSIS FOR THE BEHAVIORAL SCIENCES 82 (2d ed. 1988), which defines the broad boundaries of 0.1 < |R| < 0.3, 0.3 < |R| <0.5, and 0.5 < |R| to imply weak, moderate, and strong correlation, respectively.

233 See Levy, Mechanics, supra note 91, at 369 (“Judge Newman of the Second Circuit has noted that his court gives particular weight to oral argument, explaining that until the creation of [a special track to handle influx of immigration appeals], the circuit ‘prided itself as the last remaining circuit to afford oral argument to all litigants, with the exception of prisoners whose cases [had] been deemed of insufficient merit to warrant the appointment of counsel.’” (quoting Newman, supra note 34, at 433 .

234 Notice of Further Amendment to Standing Order 20-01 (4th Cir. Aug. 24, 2020), https://www. ca4.uscourts.gov/docs/pdfs/secondamendedstandingorder20-01.pdf?sfvrsn=7a34ba09 [https://perma.cc/ 8FPK-R4SF].

235 4TH CIR. LOC. R. 36(a) (“The Court will publish opinions only in cases that have been fully

briefed and presented at oral argument.”).

FIGURE 1: RELATIONSHIP BETWEEN ORAL-ARGUMENT AND UNPUBLICATION RATES (2020)236

236 The Pearson correlation value for the data presented in Figure 1 is 0.62, implying strong

TABLE 5: ORAL-ARGUMENT AND UNPUBLICATION RATES BY CIRCUIT (USING FIVE-YEAR MEANS)

237 See Haworth, supra note 33, at 320 (“Probably no appellate judge believes that sixty percent of the cases on his court’s docket can be decided without oral argument without some mistakes being made . . . .”). See generally Huang, supra note 16, at 1114–15 (finding that reversal rates fell in nonprisoner civil cases as Second and Ninth Circuits diverted attention to handle an overwhelming volume of immigration appeals).

238 The correlation between favorable outcomes and publication rate is weak; the Pearson correlation value is 0.2, compared with 0.42 for the relationship between oral argument and favorable outcomes. See infra note 239 (discussing correlation between oral argument and favorable outcomes). That fact— coupled with the positive correlation in Figure 2—may suggest that oral argument is a more significant factor in litigant success than the decision to publish a particular outcome. For reasons discussed above, cases that go to oral argument are more likely to receive significant judicial attention while preparing for oral argument, even if the decision is not ultimately published. See supra Figure 1; infra notes 242–244 and accompanying text.

FIGURE 2: RATES OF ORAL ARGUMENT AGAINST FAVORABLE OUTCOMES ON APPEAL239

239 Figure 2 compares data on oral argument from JUDICIAL BUSINESS 2020, at tbl.B-10, with data on the number of reversals and vacaturs in merits terminations from id. tbl.B-5. To obtain the percentage of “favorable outcomes” described in this Figure, I divided the number of reversals and vacaturs by the total merits terminations for the reporting year. See supra note 232 and accompanying text (discussing the description of correlation). The Pearson correlation value for the data presented in Figure 2 is 0.42, implying moderate correlation. See supra note 232.

240 Copus, supra note 164, at 654 (arguing against the categorical treatment of pro se appeals as

presumptively meritless). 241 See id. at 618. 242 See id. at 611 (developing “a system of statistical precedent” that can help judges better predict when to deploy their error-correcting and lawmaking functions “based on finely tuned patterns automatically mined from large-scale datasets of previous decisions”).

243 Id. at 654.

244 JUDICIAL BUSINESS 2020, at tbl.B-9. 245 Copus, supra note 161, at 654. 246 See Donald R. Songer, Criteria for Publication of Opinions in the U.S. Courts of Appeals: Formal Rules Versus Empirical Reality, 73 JUDICATURE 307, 310 (1990) (“While some groups of appellants may be statistically more likely to bring appeals which have less merit than those brought by other types of appellants, the identification of the appellant per se should make no difference.”).

247 See CAROL L. KRAFKA & PATRICIA A. LOMBARD, FED. JUD. CTR., 2003–2004 DISTRICT COURT CASE-WEIGHTING STUDY: FINAL REPORT TO THE SUBCOMMITTEE ON JUDICIAL STATISTICS OF THE COMMITTEE ON JUDICIAL RESOURCES OF THE JUDICIAL CONFERENCE OF THE UNITED STATES 1, 2 (2005), https://www.fjc.gov/sites/default/files/2012/CaseWts0.pdf [https://perma.cc/5ZFN-KJFP] (explaining case weights).

TABLE 6: FIVE-YEAR MEAN DOCKET COMPOSITION ACROSS ALL GEOGRAPHIC CIRCUITS248

248 For this Table, I averaged the docket-composition data for the last five consecutive reporting years. See JUDICIAL BUSINESS 2020, at tbl.B-1; JUDICIAL BUSINESS 2019, at tbl.B-1; JUDICIAL BUSINESS 2018, at tbl.B-1, https://www.uscourts.gov/sites/default/files/data_tables/jb_b1_0930.2018.pdf [https:// perma.cc/9QWB-FWHP]; JUDICIAL BUSINESS 2017, at tbl.B-1, https://www.uscourts.gov/sites/default/ files/data_tables/jb_b1_0930.2017.pdf [https://perma.cc/99FG-3X4W]; JUDICIAL BUSINESS 2016, at tbl.B-1, https://www.uscourts.gov/sites/default/files/data_tables/jb_b1_0930.2016.pdf [https://perma.cc/ M334-LHUB]. I calculated the percentage composition for each geographic circuit court in each category for each year, and I averaged those results for each court over the five-year period. I then averaged all twelve geographic courts to create mean nationwide docket-composition numbers from which I calculated the standard deviation. Note that “private” simply refers to prisoners not held by federal authority (and thus captures filings by those in custody under state and local law). See, e.g., James C. Martin & Susan M. Freeman, Wither Oral Argument? The American Academy of Appellate Lawyers Says Let’s Resurrect It!, 19 J. APP. PRAC. & PROCESS 89, 114 (2018) (observing that term “private” in Administrative Office statistics “refers to prisoners held by state and local, not U.S., authorities”).

249 Excluding the D.C. Circuit in studies of judicial behavior and the effects of triage decisions is not uncommon. See, e.g., Songer, supra note 246, at 310 (“The District of Columbia Circuit was not included in the comparison because the nature of the issues litigated before it are not comparable to those litigated in the numbered circuits.”). Further, I have excluded data from the U.S. Court of Appeals for the Federal Circuit, which has a distinct jurisdiction and a nationwide scope. See 28 U.S.C. § 1295. It does not hear appeals in criminal, prisoner, or immigration matters—categories of appeals that dominate the second tier of federal appellate process.

250 For data from each respective reporting year, see table B-1 of each of the 2016–2020 installments

of Judicial Business cited supra note 248.

251 Because the volume of bankruptcy appellate activity is, overall, slight, it is hard to justify excluding the First Circuit on this basis. It may be more tempting to see the Second Circuit as an outlier, given the volume of private civil cases it sees, but it also has a large volume of immigration litigation that offsets that large amount of first-tier appellate work. See Newman, supra note 34, at 429 (discussing demands on the Second Circuit from a “flood” of immigration appeals).

252 See supra notes 193–200 and accompanying text (discussing the correlation between pro se

appeals and use of unpublished decisions).

TABLE 7: PRO SE AT FILING253

253 Data in this Table derive from table B-9 of Judicial Business averaged over the last five reporting years. See JUDICIAL BUSINESS 2020, at tbl.B-9, https://www.uscourts.gov/sites/default/files/data_tables/ jb_b9_0930.2020.pdf [https://perma.cc/XJA8-V6LA]; JUDICIAL BUSINESS 2019, at tbl.B-9, https:// www.uscourts.gov/sites/default/files/data_tables/jb_b9_0930.2019.pdf [https://perma.cc/XSU8-PDVG]; JUDICIAL BUSINESS 2018, at tbl.B-9, https://www.uscourts.gov/sites/default/files/data_tables/jb_b9_ 0930.2018.pdf tbl.B-9, https:// www.uscourts.gov/sites/default/files/data_tables/jb_b9_0930.2017.pdf [https://perma.cc/TSA5-ZZBB]; JUDICIAL BUSINESS 2016, at tbl.B-9, https://www.uscourts.gov/sites/default/files/data_tables/jb_b9_ 0930.2016.pdf [https://perma.cc/5YZG-7P2V].

[https://perma.cc/R4P8-UJCQ]; JUDICIAL BUSINESS 2017, at

TABLE 8: 2020 OVERALL DOCKET PERCENTAGE OF PRO SE PRISONER APPEALS (EXCLUDING ORIGINAL APPLICATIONS)255

254 This category combines the volume of appellate activity in two Administrative Office reporting categories: U.S. Prisoner Petitions and Private Prisoner Petitions, which includes nonfederal or state petitions.

255 Data in this Table derive from table B-9 of Judicial Business averaged over the last five reporting

years. See sources cited supra note 253.

256 See infra Appendix A.

257 Professor Levy has suggested that circuit “culture” should be taken into account, an assessment

with which I agree. Levy, Mechanics, supra note 91, at 368.

258 McAlister, supra note 27, at 559. 259 DOROTHY A. BROWN, THE WHITENESS OF WEALTH: HOW THE TAX SYSTEM IMPOVERISHES

BLACK AMERICANS—AND HOW WE CAN FIX IT 223 (2021).

260 See Roger Michalski & Andrew Hammond, Mapping the Justice Gap in Federal Court, 57 WAKE FOREST L. REV. (forthcoming May 2022) (manuscript at 4), https://papers.ssrn.com/a=3931568 [https://perma.cc/GB5Z-CC4J] (indicating that, even after accounting for both population and income, pro se litigants in the federal trial courts are more likely to reside in communities that are not homogeneously white).

TABLE 9: CIRCUIT WHITENESS

261 Circuit whiteness refers to the percentage of individuals living within a circuit who identify as “White alone, not Hispanic or Latino,” according to the U.S. Census Bureau. Circuit “Blackness” refers to the percentage of individuals living within a circuit who identify as “Black or African American alone” or in combination with one or more other races in the 2019 American Community Survey estimates (table B02009). See QuickFacts United States, U.S. CENSUS BUREAU, https://www.census.gov/quickfacts/ fact/table/US/PST045219 [https://perma.cc/54GP-G9VA]. I combined the relevant state-level data into circuit-level data. For all Census Bureau information, see Explore Census Data, U.S. CENSUS BUREAU, https://data.census.gov/cedsci/ [https://perma.cc/FC7G-CPXT]. All data are on file with the author.

262 Note that I omit the D.C. Circuit, which is very diverse but a docket-composition outlier. See

supra note 249 and accompanying text.

TABLE 10: CIRCUIT BLACKNESS

263 See supra notes 232–235 and accompanying text (discussing relative rates of oral argument and

use of unpublished decisions); supra Figure 1; supra Table 5.

264 Some may argue that the disparities we see have resulted from population growth in the sun belt. See infra Table 17 and accompanying text (identifying patterns of population growth across the circuits). But these disparities predate modern population trends; indeed, the old Fifth Circuit originated the triage regime in the 1960s. See generally Haworth, supra note 33 (describing the origins of the screening and nonargument track).

FIGURE 3: PERCENTAGE OF UNPUBLISHED DECISIONS AGAINST CIRCUIT BLACKNESS (2019–2020)265

s n o

i s i c e D d e h s i l

265 The Pearson correlation value for the data presented in Figure 3 is 0.47, implying moderate

FIGURE 4: PERCENTAGE OF UNPUBLISHED DECISIONS AGAINST CIRCUIT WHITENESS (2019–2020)266

s n o

i s i c e D d e h s i l

266 The Pearson correlation value for the data presented in Figure 4 is 0.61, implying strong

FIGURE 5: PERCENTAGE OF CASES WITH ORAL ARGUMENT AGAINST CIRCUIT WHITENESS (2019–2020)267

FIGURE 6: PERCENTAGE OF CASES WITH ORAL ARGUMENT AGAINST CIRCUIT BLACKNESS (2019–2020)

267 The Pearson correlation values for the data presented in Figures 5 and 6 are 0.04 and 0.39,

respectively, implying weak and moderate correlation. See supra note 232.

TABLE 11: CIRCUIT PER CAPITA INCOME (2019)268

268 To derive circuit per capita income, I used 2019 per capita income estimates from the U.S. Census Bureau. I multiplied each state’s per capita income by the state’s population to generate the total income per state, and I added each state’s total income to generate total income per circuit. I then divided that circuit-wide income by overall circuit population. See QuickFacts United States, supra note 261.

FIGURE 7: CIRCUIT PER CAPITA INCOME AGAINST RATE OF ORAL ARGUMENT (2019–2020)269

e

m o c n I

a t i

p a C

r e p

t i

u c r i

C

FIGURE 8: CIRCUIT PER CAPITA INCOME AGAINST RATE OF UNPUBLISHED DECISIONS (2019–2020)

e

m o c n I

a t i

p a C

r e p

t i

u c r i

C

269 The Pearson correlation values for the data presented in Figures 7 and 8 are 0.39 and 0.18,

respectively, implying moderate and weak correlation. See supra note 232.

270 Michalski & Hammond, supra note 260, at 4. 271 JUDICIAL BUSINESS 2005, at tbl.B-1. 272 JUDICIAL BUSINESS 2020, at tbl.B-1; JUDICIAL BUSINESS 2019, at tbl.B-1. 273 See supra Section I.A (discussing how the courts of appeals have adapted to increased caseloads). 274 See supra Section II.B (discussing the disparate impact of reforms across race and class).

275 Although no judges have been added to the U.S. Courts of Appeals since 1990, one authorized judgeship was reassigned from the D.C. Circuit to the Ninth Circuit in 2009. Levy, Senior Judges, supra note 36, at 1236 n.50.

276 This history postdates the last major omnibus judge bill in 1990, which added eleven judges to the U.S. Courts of Appeals. Judicial Improvements Act of 1990, Pub. L. No. 101-650, § 202, 104 Stat. 5098, 5098–99.

277 For all Judicial Conference reports, see Reports of the Proceedings - Judicial Conference of the U.S., U.S. CTS., https://www.uscourts.gov/about-federal-courts/reports-proceedings-judicial-conferenceus [https://perma.cc/6BQ9-BZMY]; see also infra Table 13 (identifying relevant requests per year).

278 The modern process began in 1993, when the Judicial Conference’s Committee on Long Range Planning reaffirmed its 1990 “position favoring ‘a relatively small Article III judiciary,’” while also requesting that it collaborate with other judicial institutions on “an updated, comprehensive method for evaluating needs for permanent circuit judgeships, including reexamination of caseload formulas and standards.” Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 51 (Sept. 20, 1993) (quoting Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 93 (Sept. 12, 1990), https://www.uscourts.gov/sites/default/files/ reports_of_the_proceedings_1990-09_0.pdf [https://perma.cc/M4LY-TMEK]), https://www.uscourts. gov/sites/default/files/1993-09_0.pdf [https://perma.cc/XAK7-8HSV]. In its March 1997 report, the Judicial Conference made its first recommendation based on that updated process. See Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 27 (Mar. 11, 1997), https://www.uscourts.gov/sites/default/files/1997-03_0.pdf [https://perma.cc/LM8K-8PWH].

279 See infra Table 12 (identifying relevant requests since 1997). 280 See Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 19 (Mar. 10, 1998) [hereinafter Judicial Conference Report 1998], https://www.uscourts.gov/sites/ default/files/1998-03_0.pdf [https://perma.cc/K4PN-SLU2] (“At the request of the Fifth Circuit Court of Appeals and on recommendation of the Committee on Judicial Resources, the Judicial Conference agreed at this session to rescind its recommendation for the additional Fifth Circuit judgeship.”).

281 The Sixth Circuit stopped repeated requests in 2013. See, e.g., Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 21 (Mar. 15, 2011), https://www.uscourts. gov/sites/default/files/2011-03.pdf [https://perma.cc/3HZP-MMN6] (noting requests from the Second Circuit for two permanent judgeships, the Third Circuit for one permanent judgeship, and the Sixth Circuit for one permanent judgeship); see also infra Table 13 (identifying the timeline of relevant requests).

282 The Judicial Conference’s Recommendation for More Judgeships: Hearing Before the S. Comm. on the Judiciary, 116th Cong. 3 (2020) (statement of Hon. Miller, J., U.S. District Court for the Eastern District of Arkansas, Chair, Judicial Resources Committee, Subcommittee on Judicial Statistics) [hereinafter Miller Testimony].

283 See BARRY J. MCMILLION, CONG. RSCH. SERV., R45899, RECENT RECOMMENDATIONS BY THE JUDICIAL CONFERENCE FOR NEW U.S. CIRCUIT AND DISTRICT COURT JUDGESHIPS: OVERVIEW AND ANALYSIS 10 n.40 (2019) (“U.S. circuit courts are asked to submit requests for additional judgeships only if a majority of the court’s active judges approve of the request.”).

284 Congress has constitutional control over the existence and structure of the lower federal courts, including the number of authorized judgeships on each court. See, e.g., U.S. CONST. art. III, § 1 (“The judicial Power of the United States, shall be vested in one supreme Court, and in such inferior Courts as the Congress may from time to time ordain and establish.”); U.S. CONST. art. I, § 8 (“The Congress shall have Power . . . [t]o constitute Tribunals inferior to the supreme Court.”).

285 See Stephen Reinhardt, Too Few Judges, Too Many Cases, 79 A.B.A. J. 52, 52 (1993). The late Judge Reinhardt has been credibly accused of sexually harassing his then-law clerk. See Protecting Federal Judiciary Employees from Sexual Harassment, Discrimination, and Other Workplace Misconduct: Hearing Before the H. Subcomm. on Cts., Intell. Prop., & the Internet, 116th Cong. 1 (2020) (statement of Olivia Warren). For this reason, I have not referred to him by name above the line.

286 Tjoflat, supra note 43, at 70. 287 Id. Judge Tjoflat had previously served on the old Fifth Circuit which, at twenty-six judges, had been the largest court before it was split in 1980. At the time of the article, however, the Ninth Circuit had twenty-eight judges and the Fifth Circuit, the second largest court, had seventeen. See Chronological History of Authorized Judgeships in U.S. Courts of Appeals, U.S. CTS., https://www.uscourts.gov/sites/ default/files/appealschronol.pdf [https://perma.cc/ZVA3-EGFZ]; see also Timothy J. Corrigan, The “Duke” of Federal Court: Celebrating Gerald B. Tjoflat’s 50 Years as a Federal Judge, 104 JUDICATURE 6, 6 (2020).

288 Tjoflat, supra note 43, at 70. 289 See Irving R. Kaufman, New Remedies for the Next Century of Judicial Reform: Time as the

Greatest Innovator, 57 FORDHAM L. REV. 253, 260–61 (1988).

290 See RICHMAN & REYNOLDS, supra note 14, at 179–84 (discussing common objections to

substantial increases in the size of the federal appellate bench).

291 See id. at 177 (“All of the foregoing quality-of-the-bench arguments suffer from an additional and very serious flaw. They all focus on the quality of the active circuit judges—not on the total average quality of the appellate justice dispensed.”). I would put concerns over diminished “collegiality” on big courts in this bucket, too. See, e.g., id. at 204–06 (discussing collegiality concerns with adding too many judges to the courts). Diminished collegiality is surely a real concern, but it’s a nebulous one. It’s one thing to add one or two judges over time, and it’s another thing to add ten at once. My proposals, thus, focus on incremental additions.

292 See, e.g., id. at 115; Litman, supra note 2 (“Any court reform—indeed, any democracy reform— requires more lower federal courts.”); see also Menell & Vacca, supra note 58, at 879 (“Unless Congress can overcome the impediments to judiciary reform, the American justice system will continue to experience the erosion of its capacity to fulfill its essential role.”). But see Ronald A. Cass, Adding Judges: Issues in Federal Courts’ Governance 17 (C. Boyden Gray Ctr. for the Study of the Admin. State, Working Paper No. 21-25, 2021), https://ssrn.com/abstract=3836045 [https://perma.cc/3EXZ-V462] (opposing the addition of judges to the U.S. Courts of Appeals as based on “[t]he most glaringly flawed arguments”). Cass generally would have Congress add judges to the appellate courts only when they struggle to clear their dockets year over year. Id. at 29. He does not engage meaningfully with the kinds of distributive justice concerns I raise here to support judgeship additions.

293 Steven G. Calabresi & Shams Hirji, Proposed Judgeship Bill 10 (Nw. Univ. Pritzker Sch. of L., Working Paper No. 17-24, 2017), https://archive.thinkprogress.org/uploads/2017/11/calabresi-courtpacking-memo.pdf [https://perma.cc/PP9C-4BJV].

294 See, e.g., Ilya Somin, Opinion, The Case Against Court-Packing, WASH. POST: THE VOLOKH CONSPIRACY (Nov. 27, 2017), https://www.washingtonpost.com/news/volokh-conspiracy/wp/2017/11/ 27/the-case-against-court-packing/ [https://perma.cc/S4K7-HZKR] (“[I]t is not unfair to conclude that court-packing is a major objective of their proposal, even if it is not the only one.”); see also Richard Primus, Rulebooks, Playgrounds, and Endgames: A Constitutional Analysis of the Calabresi-Hirji Judgeship Proposal, HARV. L. REV. BLOG (Nov. 24, 2017), https://blog.harvardlawreview.org/rulebooksplaygrounds-and-endgames-a-constitutional-analysis-of-the-calabresi-hirji-judgeship-proposal/ [https:// perma.cc/V7U6-NDUT] (“The express motive for the proposal is to change the ideological and/or partisan composition of the judiciary.”).

295 Federal Judgeships: The General Accuracy of District and Appellate Judgeship Case-Related Workload Measures: Hearing Before the Subcomm. On Admin. Oversight & the Cts. of the S. Comm. on the Judiciary, 111th Cong. 7–9 (2009) (statement of William O. Jenkins, Jr., Director, Homeland Security & Justice, U.S. Government Accountability Office) [hereinafter GAO Testimony].

296 Id. at 4. 297 Miller Testimony, supra note 282, at 6. 298 See GAO Testimony, supra note 295, at 7. 299 Miller Testimony, supra note 282, at 6. “Reopened appeals” account for a relatively slight volume of appellate activity. In 2020, for example, there were 1,350 reopened appeals out of 48,190 commenced (less than 3%). JUDICIAL BUSINESS 2020, at tbl.B-1.

TABLE 12: GENERAL ADJUSTED WORKLOAD PER PANEL BY CIRCUIT304

300 GAO Testimony, supra note 295, at 8–9. 301 Id. at 8. 302 Id. at 9. 303 I should note that neither the D.C. Circuit nor the Federal Circuit is part of the Judicial Conference process for evaluating judgeship needs, but I have included the D.C. Circuit here for the sake of comparison. See Judicial Conference Report 1998, supra note 280, at 19 (observing that D.C. and Federal Circuits are “excluded from the process and standards used by the Conference for determining judgeship needs”).

304 Data derive from JUDICIAL BUSINESS 2019, at tbls.B-1, B-9.

305 Since 2017, the Judicial Conference has “recommend[ed] to the President and the Senate [that they] not fill[] the next judgeship vacancy in the Court of Appeals for the Tenth Circuit.” Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 17 (Mar. 14, 2017), https://www.uscourts.gov/sites/default/files/2017-03_0.pdf [https://perma.cc/24QE-4VYG]. Beginning in 1996, the Judicial Conference began recommending not only adding authorized judgeships, but also leaving them vacant or reducing them. Jud. Conf. of the U.S., Report of the Proceedings of the Judicial Conference of the United States 22 (Mar. 16, 1999), https://www.uscourts.gov/sites/default/files/199903_0.pdf [https://perma.cc/8LGD-Y48V].

306 Mark K. Matthews, President Trump Nominates Allison Eid to Fill Neil Gorsuch’s Seat on 10th U.S. Circuit Court of Appeals, DENVER POST (June 7, 2017, 11:47 AM), https://www.denverpost. com/2017/06/07/donald-trump-allison-eid-nomination-10th-circuit-court-of-appeals/ [https://perma.cc/ 5YDX-Z39H].

307 See supra notes 271–272.

TABLE 13: JUDICIAL CONFERENCE REQUESTS TO CONGRESS FOR AUTHORIZED JUDGESHIPS BY CIRCUIT AND YEAR308

308 This Table captures the biennial requests from the Judicial Conference available in the Judicial Conference’s Report of Proceedings. See Reports of the Proceedings—Judicial Conference of the U.S., supra note 277. Generally, the requests are made in odd years and included in its March report. A “[t]emporary judgeship[] [is] created by statutory language stipulating that the first vacancy occurring in a judicial district after a specified period of time shall not be filled.” MCMILLION, supra note 283, at 5 n.24. The handful of temporary judgeships in U.S. Courts of Appeals history have all been converted to permanent judgeships. See U.S. Courts of Appeals Authorized Temporary Judgeships, U.S. CTS., https://www.uscourts.gov/sites/default/files/appealstemp.pdf [https://perma.cc/26VZ-RBM5].

309 Indeed, it’s quite possible that in some circuits pro se appeals take more judicial time than a counseled appeal that involves a less complex legal issue (or that may possibly settle on appeal). See Resnik, supra note 205, at 950 (describing the burden that pro se habeas filings may impose on judges and law clerks). Regardless, a blanket discount of pro se appeals is, at the very least, empirically unsupported, as the Government Accountability Office has observed. See GAO Testimony, supra note 295, at 4.

TABLE 14: PER-JUDGE FILINGS BY CIRCUIT (2020)311

310 Gross filings per judge is not a perfect measure either because some circuits rely heavily on senior judges (and visiting judges) to help with workload, but that labor is essentially “free” and could vanish (or be reduced) at any point. That said, in some circuits senior judges provide significant labor to help busy courts; they average between 40% and 50% of full caseloads. See Levy, Senior Judges, supra note 36, at 1232 (discussing the vital service of senior judges).

311 See JUDICIAL BUSINESS 2020, at tbl.B. Note, these are original numbers that do not exclude “reopened” cases. Because a “reopened” case captures work that the court must now do—that is, it reflects cases where a default has been lifted—it seems these appeals should count towards the workload before a judge, even if they are not “new” work. That does mean that they are potentially counted twice—once at filing, and once at “reopening”—but they reflect actual work before the court only upon being reopened. Regardless, it’s a small number of cases per year.

312 COMM’N ON STRUCTURAL ALTS. FOR THE FED. CTS. OF APPEALS, FINAL REPORT 14 (1998); Levy

Testimony, supra note 163, at 5.

313 Note that I exclude the Federal Circuit from this calculation, and I use data from JUDICIAL

BUSINESS 2020, at tbl.B.

314 Professor Levy’s recent testimony before a Subcommittee of the House Committee on the Judiciary suggests this approach. Levy Testimony, supra note 163, at 6 (observing that courts have seen an “increase of approximately 20%” in caseload since 1990).

315 See Charles Alan Wright, The Overloaded Fifth Circuit: A Crisis in Judicial Administration, 42 TEX. L. REV. 949, 957 (1964) (“Eighty dispositions per judge is visionary for some judges, and far above what most courts regularly achieve.”).

316 See Haworth, supra note 33, at 259 (questioning whether judges “can carefully consider and

dispose of” 128 appeals per year).

TABLE 15: TERMINATIONS, PUBLISHED OPINIONS, AND ORAL ARGUMENTS PER JUDGE AND PANEL317

317 See JUDICIAL BUSINESS 2020, at tbls.B-1, B-9, B-12. 318 POSNER, supra note 10, at 162–63.

TABLE 16: CIRCUIT USE OF STAFF ATTORNEYS

319 Supra Table 15. 320 See supra Table 12 (741 adjusted filings per panel); Table 14 (443 filings per judge); Table 15

(310 terminations per judge); Table 16 (5.4 ratio of staff attorneys to judges).

321 For example, the Eleventh Circuit is well below average in publication and oral argument. See

supra Table 5 (oral-argument rate of 10.24% and unpublication rate of 93.26%).

322 JUDICIAL BUSINESS 2020, at tbl.B-11, https://www.uscourts.gov/sites/default/files/data_tables/

jb_b11_0930.2020.pdf [https://perma.cc/A8MR-6X53].

323 Id. 324 Accounting for and adjusting for the impact of senior judges, in particular, is beyond the scope of this Article. It is possible that the work of senior judges mitigates some of the concerns I raise here about overreliance on nonjudicial decision-makers.

325 Because of their unusual docket compositions, I would exclude both the Federal Circuit and the D.C. Circuit from this process and these statistics—just as the Judicial Conference currently does. See supra note 303 (discussing the exclusion of the Federal and D.C. Circuits from the Judicial Conference process).

TABLE 17: APPLICATION OF THE AVERAGE-CIRCUIT MODEL TO CURRENT METRICS326

326 Total filings per judge in Table 17 differ from Table 14, supra, because Table 17 reflects fiveyear averages, whereas Table 14 uses 2020 data only. Data from this table are annualized over a five-year period, and they appear in Judicial Business tables. See supra note 248 (tables B-1); supra note 229 (tables B-12), JUDICIAL BUSINESS 2020, at tbl.B-5; JUDICIAL BUSINESS 2019, at tbl.B-5, https:// www.uscourts.gov/sites/default/files/data_tables/jb_b5_0930.2019.pdf [https://perma.cc/UH4T-CTAE]; JUDICIAL BUSINESS 2018, at tbl.B-5, https://www.uscourts.gov/sites/default/files/data_tables/jb_b5_ 0930.2018.pdf tbl.B-5, https:// www.uscourts.gov/sites/default/files/data_tables/jb_b5_0930.2017.pdf [https://perma.cc/73RZ-DNYF]; JUDICIAL BUSINESS 2016, at tbl.B-5, https://www.uscourts.gov/sites/default/files/data_tables/jb_b5_ 0930.2016.pdf [https://perma.cc/C59Z-LJEU].

[https://perma.cc/6BDF-6LXB]; JUDICIAL BUSINESS 2017, at

TABLE 18: APPLICATION OF THE POPULATION MODEL327

327 Population figures derive from 2020 census data, which are available at 2020 Population and Housing State Data, U.S. CENSUS BUREAU, https://www.census.gov/library/visualizations/interactive/ 2020-population-and-housing-state-data.html [https://perma.cc/9VLS-HBMP].

328 Because the last benchmark was set in 1990—before technology made at least some of the courts’ work more efficient—I suggest a higher benchmark than for the population model, where I would bring the courts within 10% of peer circuits in either filings or terminations per judge (whichever is higher). Once a new benchmark is set, it may be that a 10% range would be more appropriate.

TABLE 19: APPLICATION OF THE GROSS-FILINGS-BENCHMARK MODEL

329 Bankruptcy Amendments and Federal Judgeships Act of 1984, Pub. L. No. 98-353, § 201(a),

98 Stat. 333, 346.

330 Act of Oct. 20, 1978, Pub. L. No. 95-486, § 3(a), 92 Stat. 1629, 1632. Overall judgeship additions appear in infra Appendix B, and percentage increases have been calculated using the growth rate year over year.

331 Cf. Arthur D. Hellman, Assessing Judgeship Needs in the Federal Courts of Appeals: Policy Choices and Process Concerns, 5 J. APP. PRAC. & PROCESS 239, 262, 266–67 (2003) (describing a “particularly unfortunate” aspect of the Judicial Conference process as its relative secrecy about when a majority of judges are opposed to adding judges to a circuit, and recommending a public-comment process and the airing of dissenting views).

332 This argument is consistent with how others have conceptualized the benefits of Supreme Court term limits. See Adam Chilton, Daniel Epps, Kyle Rozema & Maya Sen, Designing Supreme Court Term Limits, 95 S. CAL. L. REV. (forthcoming 2021) (manuscript at 9), https://papers.ssrn.com/a=3788497 [https://perma.cc/KP7S-LSCH] (recognizing that term limits for Supreme Court Justices would produce regular and predictable turnover at the Court, thereby potentially “reduc[ing]” “the political stakes over each appointment”); see also Steven G. Calabresi & James Lindgren, Term Limits for the Supreme Court: Life Tenure Reconsidered, 29 HARV. J.L. & PUB. POL’Y 769, 832–33 (2006) (arguing that a Supreme Court term-limit proposal “would reduce the stakes of the nomination process and eliminate the uncertainty that now exists regarding when vacancies will occur”).

333 Data here derive from tables B-1 (cases filed), B-10 (oral argument), and B-12 (publication) in the 2016–2020 editions of Judicial Business. The oral-argument and publication rates are five-year means, for the five-year period ending on September 30, 2020; the number of appeals filed per judge reflects total circuit filings divided by authorized judgeships.

FIGURE 9: ORAL-ARGUMENT RATE AGAINST APPEALS PER JUDGE334

n o

i t a c i l

FIGURE 10: UNPUBLICATION RATE AGAINST APPEALS PER JUDGE

334 The Pearson correlation values for the data presented in Figure 9 and 10 are 0.39 and 0.59,

respectively, implying moderate and strong correlation. See supra note 232.

335 See John P. Collins, Jr., Judging Biden, 75 SMU L. REV. F. 150, 165–70 (2022) (discussing the professional diversity of recent nominees to federal bench by President Joe Biden). On the value of diversity more generally, see Written Testimony on the Importance of Judicial Diversity: Hearing Before the H. Subcomm. on Cts., Intell. Prop., & the Internet, 117th Cong. 4–9 (2021) (statement of Maya Sen, Professor of Public Policy, John F. Kennedy School of Government, Harvard University).

336 Notably, Judge Posner famously used only two of his allotted four clerks and did most of the work himself. See, e.g., Stephen J. Choi & G. Mitu Gulati, Mr. Justice Posner? Unpacking the Statistics, 61 N.Y.U. ANN. SURV. AM. L. 19, 25 & n.18 (2005) (describing Judges Posner’s and Easterbrook’s delegation habits—or lack thereof—and Judge Posner’s failure to hire the full number of allotted clerks). Judge Diane Wood, likewise, “almost always write[s] . . . [her] own opinions in full.” William Domnarski, Opinion, Judges Should Write Their Own Opinions, N.Y. TIMES (May 31, 2012), https:// www.nytimes.com/2012/06/01/opinion/judges-should-write-their-own-opinions.html [https://perma.cc/ 2JD5-V69J].

337 See Choi & Gulati, supra note 336, at 25–27 (discussing influences that Judge Posner’s work

ethic may have had on the Seventh Circuit).

338 That premise is also consistent with the stated goals of the federal judiciary itself. One of seven strategic issues identified by the Judicial Conference in a 2020 strategic plan is “enhancing access to justice and the judicial process,” and another is “providing justice” by “pursu[ing] improvements in the delivery of fair and impartial justice on a nationwide basis” and “secur[ing] resources that are sufficient to enable the judiciary to accomplish its mission.” JUD. CONF. OF THE U.S., STRATEGIC PLAN FOR THE FEDERAL JUDICIARY 4, 21 (2020), https://www.uscourts.gov/sites/default/files/federaljudiciary_ strategicplan2020.pdf [https://perma.cc/9X8W-7FTY].

TABLE A1: D.C. CIRCUIT

TABLE A2: FIRST CIRCUIT

339 See supra note 248 (explaining docket-composition statistics).

TABLE A3: SECOND CIRCUIT

TABLE A4: THIRD CIRCUIT

TABLE A5: FOURTH CIRCUIT

TABLE A6: FIFTH CIRCUIT

TABLE A7: SIXTH CIRCUIT

TABLE A8: SEVENTH CIRCUIT

TABLE A9: EIGHTH CIRCUIT

TABLE A10: NINTH CIRCUIT

TABLE A11: TENTH CIRCUIT

TABLE A12: ELEVENTH CIRCUIT

TABLE B1: GROWTH OF THE GEOGRAPHIC CIRCUITS OVER TIME