Forum Crowding

Jurists and scholars have long debated (and often decried) the practice of forum shopping. Critics note, for example, that forum shopping both yields procedural gamesmanship and undermines the courts’ legitimacy, while supporters contend that competition for cases increases access to the courts and improves litigants’ experiences.

Such debates have overlooked the effects of forum shopping on an important constituency: litigants who have little choice over forum. When forum shopping causes a sudden influx of cases—when, that is, it crowds a forum—what happens to other cases that have nowhere else to go? Are busy judges—made busy by forum shopping—more prone to error?

In this Article, we conduct a novel analysis of the effects of such forum crowding. In particular, we draw on ongoing pathologies in patent litigation, which has been subject to extreme forum shopping for over a decade, to empirically examine the effects of these practices on other cases, including criminal and civil rights cases, which are subject to more stringent forum rules. We find that, in some crowded courts, these cases may get short shrift. And so we conclude by offering procedural reforms aimed at protecting these cases from being crowded out.

Table of Contents Show

Introduction

In 1952, Judge Irving Kaufman issued the first district court decision to use the phrase “forum shopping,” decrying a patentholder’s litigation tactics as “forum shopping with a vengeance.”[1] Specifically, beauty-products manufacturer Helene Curtis Industries filed suit in the Southern District of New York, seeking a declaratory judgment that a patent held by competitor Procter & Gamble was invalid. But Procter & Gamble preferred to litigate in Texas and so it filed a parallel patent infringement suit there.[2] In an extraordinary order, Judge Kaufman enjoined the parallel infringement suit, explaining that “there [wa]s nothing to be gained by a trial in Texas rather than here [in New York], and there is much which commends the conclusion that this forum will best serve the ends of justice.”[3]

It is remarkable how little seems to have changed, as patent-holding plaintiffs are still trying to get into Texas’s federal district courts. Judge Alan Albright of the Western District of Texas is known for his appetite for patent cases, having “openly solicited cases at lawyers’ meetings . . . and urged patent plaintiffs to file their infringement actions in his court.”[4] Judge Albright even adopted plaintiff-friendly procedural rules designed specifically to attract patent litigation.[5] And because of a quirk in the case assignment procedures in the Western District of Texas, Judge Albright could, until recently, guarantee patent plaintiffs that he—rather than any other judge in the Western District—would preside over their cases.[6] Hence, Judge Albright’s caseload—over 800 cases in 2021—encompassed nearly a quarter of all patent cases filed nationwide.[7] Indeed, Judge Albright remains the most popular patent judge in the nation.[8]

But what about the rest of Judge Albright’s docket? Some scholars have examined Judge Albright’s controversial methods for attracting patent plaintiffs—i.e., his “forum selling” conduct.[9] Others have looked at his record in certain patent appeals before the U.S. Court of Appeals for the Federal Circuit.[10] But no one has considered whether this deluge of patent cases affects Judge Albright’s consideration in, say, criminal or civil rights matters. Patent cases, after all, can be difficult, as they tend to be complicated and time intensive—perhaps leaving less time for these other sorts of cases (especially for those judges, like Judge Albright, who have expressed a strong preference for patent cases). Hard cases might thus make bad law in more than one way.[11]

Hence, while Judge Albright’s practices have renewed debates about forum shopping (by plaintiffs) and forum selling (by judges),[12] these exchanges have overlooked an important constituency—litigants who have far less control over the forum in which their case is heard. Scholars and policymakers have debated the merits of, say, allowing patent plaintiffs to choose a seemingly favorable federal forum, but they have largely missed the possible effects of such en masse case migration on the rest of that forum’s docket (criminal and civil rights cases, for example).[13] In all, many examinations of forum shopping consider such conduct from a perspective that is internal to the case or claims that give rise to the conduct—e.g., does forum shopping undermine fairness for patent defendants, or does forum shopping improve the local procedural rules governing patent claims?[14] But these studies often overlook the external effects of forum shopping—the effects, for example, on the civil rights plaintiffs or criminal defendants in that forum.

We examine such externalities of forum shopping in this Article. Our analysis includes a novel empirical study of several district courts subject to what we call “forum crowding”—substantial caseload increases resulting from forum shopping, forum selling, and responses to such behavior. Specifically, we draw on the well-documented forum-shopping-related pathologies of patent litigation to assess and measure the possibilities for forum crowding.

This study draws from and builds upon two strands in the legal literature. One strand is the scholarship considering the equities of forum shopping. This literature examines the harms that can flow from allowing plaintiffs to select favorable fora as well as the benefits that can result from interjurisdictional competition for cases.[15] Our Article adds to this literature by expanding the scope of the concerns implicated by forum shopping to encompass forum crowding. Specifically, we focus on the effects of court congestion caused by substantial forum shopping on other unrelated cases and parties (that is, on cases other than those subject to such forum shopping). By isolating forum shopping from other potential causes of court congestion, our Article can strengthen the link from one to the other, explaining that the downstream effects of congestion (e.g., workload effects) may also result from forum shopping and selling.[16] Forum crowding thus represents a specific sort of court congestion, namely, congestion caused by excessive forum shopping (including forum shopping that is induced by forum selling).

The second strand is the growing literature examining workload effects on legal decision-making. In one leading study, Bert Huang examined the effect of a sudden influx of agency appeals on the scrutiny given by federal appeals courts in other, unrelated cases.[17] He found that busy appellate judges are more likely to defer to their colleagues on the district courts: as appellate workloads increase, so do affirmance rates. Looking outside the Judiciary, Michael Frakes and Melissa Wasserman have estimated the consequences of workload effects on patent quality, finding that when patent examiners are expected to process more patent applications (and hence have less time to review each individual application), they are more likely to issue a low-quality patent—i.e., a patent more likely to be later deemed invalid.[18] Our study adds to, and builds upon, this literature by examining the effect, at the federal district courts, of a sudden influx of patent cases on non-patent cases.

Our analysis finds that this forum crowding has had important, significant effects on both process and substance in the district courts. We find that when a forum gets crowded, some judges move more quickly through their respective dockets. On reflection, this may seem obvious: judges face various institutional pressures to ensure that cases do not languish for long, such as the six-month list,[19] and so must process individual cases in a crowded docket more quickly to stay on schedule.[20] One possible consequence of this result is that district judges, when moving quickly, take less time and care with each case. But another possible consequence is that busy judges are no less careful, they are simply more efficient. In order to discern which is more likely, we turn to appeal rates and reversal rates.[21]

Here, the study’s research design takes advantage of the unique structure of patent appeals. Practically every appeal in a patent case is diverted to the U.S. Court of Appeals for the Federal Circuit (rather than the regional federal appeals court). Accordingly, a flood of patent cases in, say, the Western District of Texas will have no effect on the docket in the U.S. Court of Appeals for the Fifth Circuit.[22] We can thus compare appeal rates and reversal rates in non-patent cases without worrying that the relevant Court of Appeals’ work is itself affected by crowding.[23]

We find that decisions from crowded dockets are reversed more often.[24] And our qualitative examination of some of these district court procedures, orders, and decisions corroborates the view that cases on these crowded dockets can get short shrift.

Moreover, a closer look at the patterns in our results uncovers a nuanced story: are these reversal rates attributable to a judicial preference for commercial patent litigation over, say, criminal trials; or are they due to the squeeze on the most precious judicial resource—time—that accompanies these massive caseload increases? Our analysis suggests that both judicial preferences and resource constraints matter, particularly because preferences can drive how judges allocate their resources (bandwidth for cases, slots for law clerks, and so on). The upshot is that crowded dockets may yield lower-quality decisions when such crowding is associated with a judge’s preference for a certain kind of case.

This Article proceeds in three Parts.

First, we describe the debates regarding forum shopping and forum selling, both in general and as applied to patent cases in particular. While these developments are well trod, our telling focuses on the matters most pertinent to forum crowding, including the developments that inform our research design (most notably, the migratory patterns of patent plaintiffs: to the Eastern District of Texas, then north to the District of Delaware, and then (back) to the Western District of Texas).

Second, we describe the results of our study. We begin by describing our specific research design, including our methods for manually collecting and analyzing the data we present. We then present our findings regarding reversal rates, appeal rates, and overall time-to-termination alongside our interpretation of these findings—namely, that forum crowding seems to degrade decision quality, and that these effects are driven by judicial preferences for certain cases.

Third, we propose possible responses to these findings. We consider ways that courts should—or should be required to—force parties to account for the costs of their forum shopping conduct. Further, we examine procedural reforms to venue and transfer rules and to the Civil Justice Reform Act’s “six-month list” that address the externalities of forum crowding, both through the direct regulation of forum shopping and selling behavior and through changes that would cause forum shoppers to internalize these externalities. And, given our focus on patent cases, we consider the implications of our study to matters of particular importance to patent law, including the rise of agency adjudication as a substitute for litigation in the courts.

I.Forum Shopping, Forum Selling, and Forum Crowding

A.     Forum Shopping

In general, forum shopping refers to strategic venue selection: litigants may choose a forum because it offers favorable decisional rules or preferred procedural practices, among other possibilities.[25] While such behavior may well date to the beginnings of our legal system—choosing a venue is, after all, one of the first steps to filing any lawsuit—the phrase “forum shopping” itself seems to have originated in response to Erie Railroad v. Tompkins. In a foundational article, Harold Horowitz described “forum shopping” as an “evil” spawned by the Supreme Court’s decision in Swift v. Tyson,[26] among other federal decisions.[27] Erie, then, was an attempt to reckon with that evil.[28] By mandating that a federal court exercising its diversity jurisdiction apply substantive state law, Erie eliminated many incentives for forum shopping between state and federal venues (i.e., “vertical” forum shopping): both venues must apply the same decisional rules, and so neither system—state or federal—offers an advantage over the other.[29] Erie, however, did little to address “horizontal” forum shopping, in which litigants seek out favorable venues from among the state or federal courts, on the view that one state or federal district court will offer favorable rules or practices.[30]

Concerns about forum shopping thus persist.[31] Specifically, forum shopping continues to present concerns about gamesmanship, legitimacy, and court congestion.

Critics of forum shopping note that the practice promotes gamesmanship, as litigants seek out venues that treat claims differently for no reason other than the court in which they are brought.[32] Justice Kagan, for example, highlighted Texas’s strategy of filing its challenges to federal programs in selected divisions of selected districts, thereby assuring itself of an audience that seems more receptive to its claims.[33]

The seemingly arbitrary differences (particularly, perhaps, in outcomes) between fora that lead plaintiffs to choose one venue over another may also seem to undermine the legitimacy of our legal systems. Indeed, several notable jurists have suggested that forum shopping “feeds the growing perception that the courts are politicized.”[34] Such concerns about legitimacy extend to questions of judicial power and jurisdiction. Forum shopping that causes courts to hear cases that might otherwise seem beyond their purview—a case, for example, filed in a California state court about a “Pennsylvania-manufactured airplane with Ohio-made propellers [that] crashed in Scotland”—gives rise to concerns over aggrandizement and the scope of the judicial powers.[35]

Such forum shopping can also pressure courts’ internal procedures and docket management strategies. For example, the Supreme Court’s decision in Piper Aircraft explained that the domestic exercise of jurisdiction would make “American courts, which are already extremely attractive to foreign plaintiffs . . . even more attractive . . . and further congest already crowded courts.”[36]

We do not mean to say that the story is all bad. Pamela Bookman, for example, has identified several “unsung virtues” of forum shopping.[37] She explains that forum shopping can help preserve access to courts (thereby promoting private enforcement) and that forum shopping produces competition among courts for cases that can lead to positive legal reform.[38]

Hence, while the conversation on forum shopping is more complicated than simply decrying the practice as an “evil” tactic,[39] the general consensus among courts, scholars, and policymakers seems to strongly disfavor the practice.[40]

Notably, this literature largely ignores certain externalities of forum shopping—namely, its possible effects on the rest of a favored forum’s docket. As noted above, Bookman considers possible positive externalities of forum shopping, such as desirable legal reform. And others have described possible negative effects, such as diminished confidence in the courts.[41] These are certainly externalities of forum shopping: they are a cost or benefit imposed on the courts that is largely unrecognized by plaintiffs seeking favorable fora. But these are effects of forum shopping on the judicial system generally, and as far as we can tell, no study has examined the effects of forum shopping on other pending cases.[42]

B.     Forum Shopping and Forum Selling: A Study in Patent Litigation

1.     The Migrations of Patent Litigation

The debates over forum shopping have had special application to patent cases for the better part of three decades. Assessing caseloads from 1995 to 1999, then-Professor (now-Chief Judge) Kimberly Ann Moore found that much patent litigation was concentrated in only a few districts, and that some of these districts—the District of Delaware and the Northern District of California, for example—carried a disproportionate load of patent cases (as measured against their share of civil cases generally). The District of Delaware, for example, was home to 0.3% of all federal civil cases, but home to 3.2% of all patent cases.[43] Similarly, the Northern District of California was responsible for 2.3% of all federal cases, compared to 9.1% of all patent cases.[44]

Such figures, however, pale in comparison to the concentrations of patent cases that resulted from forum shopping in later years. See Figure 1, below.

Figure 1: Share of Patent Cases by District (D. Del, E.D. Tex., W.D. Tex.), 2011–2021

As Figure 1 shows,[45] the Eastern District of Texas accounted for nearly half of all patent cases in 2015.[46] In 2019, the District of Delaware housed nearly one-quarter of all patent cases.[47] And in 2021, a single judge in the Western District of Texas was responsible for nearly one-fourth of the nation’s patent litigation.[48]

Such unusual shifts of large concentrations of patent cases are the consequence of two interrelated features: (1) the law governing venue in patent litigation; and (2) the “substantive and procedural differences among district courts in resolving patent cases.”[49]

We begin with the law. The statute governing venue in patent cases, 28 U.S.C. § 1400(b), provides that “[a]ny civil action for patent infringement may be brought in the judicial district where the defendant resides, or where the defendant has committed acts of infringement and has a regular and established place of business.”[50] In 1990, the U.S. Court of Appeals for the Federal Circuit considered what it meant for a patent defendant to “reside” in a particular district.[51] Looking to a 1988 amendment to the general venue statutes,[52] the Federal Circuit reasoned that a corporate defendant could be understood to “reside” in “any judicial district in which it is subject to personal jurisdiction.”[53] And, given the relatively capacious boundaries of personal jurisdiction, this ruling meant that most companies operating in the streams of interstate commerce could be sued under § 1400(b) in almost any federal district court.[54] Hence, patent plaintiffs began to engage in so-called “horizontal” forum shopping,[55] seeking out district courts that seemed to offer a more favorable forum for their claims.

We thus turn next to interdistrict variation. In the late 1990s, for example, patent plaintiffs seemed to prefer districts that moved quickly (thereby minimizing their litigation costs), that resolved cases well before trial (often insulating their patents from being found invalid), and that generally seemed to offer the “greatest chances of success.”[56]

But modest differences across federal districts later metastasized into more significant variations, as some judges designed patent-specific procedures rules with the specific intent of luring patent plaintiffs to their districts.[57]

Consider the Eastern District of Texas: in 2007, that district was home to the largest concentration of patent litigation in the nation, responsible for about one-eighth of all such cases; in 2012, it housed about one-fourth of all patent litigation in the country; and in 2015, it accounted for nearly half.[58] How did the Eastern District of Texas come to be such a favored forum? Procedural innovation. Judge T. John Ward began by emulating some of the practices that led to some of the early, modest consolidation noted above: Judge Ward adopted, for example, rules mimicking the Northern District of California’s “patent local rules” that allowed such cases to move more quickly.[59] Other authors have suggested that juries in Marshall, Texas—familiar with disputes over oil and sensitive to property rights—were predisposed to ruling in favor of patentholders, thus leading patent plaintiffs to file claims there.[60]

In 2011, Judge Ward retired, and (now-Chief) Judge Rodney Gilstrap was appointed to his seat. As Jonas Anderson and Paul Gugliuzza explain, “[Chief] Judge Gilstrap adopted unique practices that made his courtroom even more appealing for patent plaintiffs, such as requiring defendants to seek his permission before filing a motion for summary judgment or a motion to invalidate a patent for lack of patent-eligible subject matter.”[61] Chief Judge Gilstrap also adopted special discovery rules—rules which expedited discovery, increasing settlement pressure on defendants—and he took further procedural steps to insulate patents from validity challenges.[62]

Most notably, the Eastern District of Texas employs a unique case assignment procedure. Six divisions comprise the Eastern District of Texas: the Beaumont Division, the Lufkin Division, the Marshall Division, the Sherman Division, the Texarkana Division, and the Tyler Division. While case assignment is ostensibly random, any randomness applies only within a division rather than across the entire district, subject to case assignment rules set out in standing orders by Chief Judge Gilstrap.[63] Under these orders, all patent cases filed in the Marshall Division were directed to Chief Judge Gilstrap.[64] And, notably, no venue rules govern the choice of a division within a district.[65] Hence, patent plaintiffs eligible to file in the Eastern District of Texas can simply select the Marshall Division in the court’s electronic filing systems, thus assuring themselves of a spot on Chief Judge Gilstrap’s docket.[66]

Such procedural innovations (together with this unique case assignment procedure) help to explain many plaintiffs’ decisions to file in the Eastern District of Texas. But what explains these judges’ decisions to create such patent-friendly rules? Greg Reilly and Daniel Klerman contend that the Eastern District of Texas—and Judges Ward and Gilstrap, in particular—engaged in a practice of “forum selling.”[67] In short, they created “pro-plaintiff law and procedures” in order to induce plaintiffs, who typically control questions of venue in non-contract cases (including many patent infringement cases),[68] to bring their litigation there.[69] And their hypotheses for this behavior, alongside those advanced by other scholars,[70] range from the innocuous to the worrying: Judge Gilstrap, for example, may do this in order to attract cases he finds interesting or  that he perceives to be more prestigious (perhaps because of their high stakes and complex nature); or Judge Ward might have done this to benefit the local economy (e.g., the hotels, restaurants, copy shops, cafés, and local law firms that benefit from increased local litigation) or perhaps for more direct pecuniary gain (as Judge Ward’s son is a local patent attorney).[71]

The Eastern District of Texas’s power over patent litigation caught the Supreme Court’s attention. As early as 2006, Justice Scalia referred to the district as a “renegade jurisdictio[n]” presenting a problem in need of redress.[72]

In 2017, over ten years after Justice Scalia’s initial critique, the Supreme Court finally offered a solution by way of its decision in TC Heartland v. Kraft Foods.[73] Specifically, the Court’s decision in TC Heartland revisited the Federal Circuit’s 1990 venue decision (which, recall, construed an amendment to the general venue statutes as applicable to patent venue statutes, too).[74] The Court upended the Federal Circuit’s prior approach and held unanimously that the amendment to the general venue statute did not apply to the patent-specific venue statute.[75] Hence, after TC Heartland, it is no longer the case that corporate defendants in patent litigation are suable anywhere they are subject to personal jurisdiction.[76] Rather, such defendants could be sued only in their state of incorporation or where they have “committed acts of infringement and ha[ve] a regular and established place of business.”[77] The Court granted review in TC Heartland to address a problem of forum shopping,[78] and its decision was effective at blunting the effect of the Eastern District of Texas’s procedural innovations.[79]

Because the Eastern District of Texas could no longer properly house so much patent litigation, the District of Delaware temporarily became the new venue of choice. Why Delaware? Because many corporate defendants are incorporated in Delaware, for reasons well-developed in the corporate law literature (namely, Delaware is an attractive forum in which to incorporate).[80] Colleen Chien and Michael Risch correctly predicted that, in TC Heartland’s wake, “cases would primarily move to the District of Delaware,” estimating that it would take responsibility for about one-quarter of all patent litigation.[81] Indeed, the District of Delaware’s share of new patent cases grew to slightly more than one-quarter of all filings after TC Heartland, while the Eastern District of Texas’s share shrank significantly.[82]

The District of Delaware’s prominence would, however, prove to be short-lived, as the Western District of Texas quickly took its place. As noted, after TC Heartland, patent defendants could be sued in their state of incorporation or wherever they have “committed acts of infringement and ha[ve] a regular and established place of business.”[83] And many corporate patent defendants have established places of business in San Antonio or Austin where patentholders may allege acts of infringement.[84] The Western District of Texas, which encompasses both cities,[85] thus offers one possible alternate venue for many patent cases.

Patentholders could thus file in the Western District of Texas, even after TC Heartland. But why would they? Again, the answer lies in procedural innovation and case assignment rules. Judge Albright, appointed to the Western District of Texas in 2018, has (like Judges Ward and Gilstrap) promulgated several standing orders that speed patent cases along faster, stay aspects of discovery in ways that seem to favor patent plaintiffs over defendants, and limit early motions to dismiss on certain questions of patent validity.[86] Such procedural practices strongly favor patent plaintiffs (though Judge Albright has defended his practices as “party agnostic”[87]).

Judge Albright not only takes steps to ensure that cases come to him; he also helps to make sure they stay with him. As noted, Judge Albright has limited how defendants can challenge a patent’s validity during a case’s early phases. Consequently, defendants might prefer to challenge these patents before the Patent Office, which offers a comparatively quick and cheap agency adjudication of a patent’s validity (known as inter partes review).[88] Most district judges stay patent litigation during the pendency of these agency reviews.[89] This is because an agency decision finding the patent invalid can quickly end the parallel litigation. But Judge Albright employs a different tack, instead often scheduling aggressively optimistic trial dates in order to dissuade the Patent Office from undertaking review at all.[90]

Judge Albright is similarly disinclined to grant motions to transfer cases to other venues, sometimes even in cases where the relevant factors evince a “striking imbalance favoring transfer.”[91] Indeed, the Federal Circuit has issued repeated writs of mandamus in many of these cases, sometimes finding Judge Albright’s decisions to deny transfer practically indefensible.[92]

Judge Albright’s popularity with patent plaintiffs has proved persistent. Until July 2022, the Western District of Texas employed a case assignment procedure akin to that of the Eastern District of Texas: case assignment was random only within a division, rather than across an entire district.[93] And because Judge Albright is the only member of the Western District’s Waco Division,[94] a patent plaintiff who filed in the Waco Division knew that Judge Albright would preside over her case. But in July 2022, the Chief Judge of the Western District, in an apparent response to critiques of Judge Albright,[95] unsuccessfully attempted to reform these assignment procedures.[96] And while Chief Judge Alia Moses, who recently succeeded Chief Judge Orlando Luis Garcia, has kept these reforms in place, Judge Albright remains the most popular patent judge in the nation.[97] In short, forum shopping in the Western District of Texas has proved sticky and resistant to efforts to dislodge it.

2.     Evaluating Patent Forum Shopping

One obvious question raised by these patterns of patent case migrations is whether such dramatic movements are normatively desirable. Some commentators have characterized this forum shopping as an “evil” to be contained.[98] As with the more general literature on forum shopping, the literature on patent forum shopping typically focuses either on the distortions that result from plaintiff-friendly practices (e.g., gamesmanship-related concerns) or on the systemic implications that forum shopping has for the judiciary generally (e.g., legitimacy-related concerns).[99] On the former, some scholars have explained that plaintiff-friendly procedural rules—designed, for instance, to solicit patent cases—lead to “inefficient distortions of substantive law.”[100] On the latter, some commentary notes the possibility that forum shopping will diminish the public’s confidence in the courts or in the patent system.[101] Others, however, have a more sanguine view, contending that such procedural innovation may offer some benefits to our systems of patent litigation.[102] Overall, the debates about forum shopping in patent cases reflect debates about forum shopping more generally.

Scholars have primarily regarded patent forum shopping as undesirable: most think that the practice is more harmful to the courts and to our patent system than it is helpful to improving the mechanics of patent litigation. Judge Moore, for example, has suggested that forum shopping is “normative[ly] evil” because it “thwarts the ideal of neutrality in a system whose objective is to create a level playing field” for dispute resolution, thereby “erod[ing] public confidence in the law.”[103] She also notes that forum shopping gives rise to increased litigation costs (for example, travel to inconvenient locales or costs associated with motions to transfer) and that patent forum shopping may well undermine the innovation-inducing purposes of the patent franchise.[104] In later work, other scholars have echoed these concerns. Jonas Anderson, Paul Gugliuzza, and Jason Rantanen, as well as Greg Reilly and Daniel Klerman, have all remarked on the tendency of such forum selling and shopping both to undermine fairness and legitimacy and to increase the costs associated with venue-related disputes.[105] And these scholars have emphasized that these concerns are especially pronounced where plaintiffs are engaged in judge shopping beyond mere forum shopping.[106] In all, these critiques echo the concerns, noted above, about gamesmanship and legitimacy.[107]

Some scholars, however, have emphasized the benefits that may accrue from such forum shopping. Specifically, such commentary reflects the possibility that forum shopping can lead to positive legal reform. Jeanne Fromer, for example, has advanced a nuanced view of patent forum shopping. Though she ultimately concluded that, on net, forum shopping seems to do more harm than good, she noted the possibility that the interdistrict competition occasioned by forum shopping could “foster[] a race to the top . . . [that] cause[s] district courts to develop useful rules.”[108] She also suggested that local case concentration can give rise to both legal and substantive expertise—i.e., familiarity with both patent law and the actual (patented) technologies underlying local industries.[109] And more favorably still, Xuan-Thao Nguyen concluded that patent forum shopping leads to adjudication by “reasonable and fair” judges who are “knowledgeable, welcoming, and organized,” particularly because they have adopted a series of efficiency-enhancing and cost-saving local rules that reflect a “customer-oriented approach.”[110]

But both these sets of scholars—those who think that forum shopping is, on net, good for our systems of patent litigation, as well as those who think it is deleterious—have overlooked the effects of forum shopping in patent cases on the non-patent cases, and non-patent litigants, in that forum. Xuan-Thao Nguyen, noted, for example, that forum shopping allows patent plaintiffs to avoid districts that are “burdened with criminal cases.”[111] But criminal defendants in, say, Waco, Texas can hardly avoid a district burdened with patent cases.[112] Habeas petitioners and civil rights plaintiffs are similarly stuck.[113]

Is there any reason to worry about the influx of patent cases on the non-patent cases in, for example, Judge Albright’s courtroom? So far, we have very little evidence to say. We have found no study, either in the patent literature or elsewhere, that has closely examined the effects of forum shopping on other unrelated pending cases. We turn to such an examination next.

II. The Effects of Forum Crowding

These migrations of patent litigation—into the Eastern District of Texas, then out of Texas to Delaware, and then back to the Western District of Texas—offer an opportunity to study the effects of forum crowding on other cases, such as criminal and civil rights cases. What happens when a district judge is suddenly inundated by an influx of technical and typically complex cases?

Judge Edwards has written about this problem from the perspective of the federal appeals courts: “The bigger the dockets, the less time we spend on the difficult cases and the more mistakes we make.”[114] Judge Wald has also written about how “time and docket pressures” may cause a judge to rely on doctrines such as waiver or forfeiture (rather than take a close look at a claim’s merits).[115]

Some scholars, moreover, have noted the effects that docket congestion can have on litigation delay[116] and on public access to the courts,[117] while more recent scholarly examinations of workload effects on legal decision-making echo Judge Edwards’ view that docket congestion can affect outcomes. As noted above, Bert Huang finds that docket pressures at appeals courts can cause appellate judges to apply only “lightened scrutiny” to the decisions of their trial court colleagues.[118] Melissa Wasserman and Michael Frakes similarly found that patent examiners with higher workload expectations review patent applications less carefully, thereby granting more low-quality patents.[119]

We add to that literature here by examining the effects of the substantial swings in patent cases among the Districts of Delaware, Eastern Texas, and Western Texas on the other cases in those venues. We begin by describing our study design, and then move on to presenting the results of that study.

A.     Measuring Forum Crowding

1.     Defining Crowded Fora

We began by identifying the districts and judges that faced relatively more crowded dockets (as well as when such crowding occurred) in order to identify our treatment and control study periods. Unsurprisingly, the series of legal and procedural shifts described above guided our approach. We emphasize that our measure of forum crowding is relative, not absolute—that is, we sought out (for example) time periods of comparatively higher and lower concentrations of patent cases to determine whether those changes corresponded to, say, changes in indicia of decision quality, such as reversal rates.

We focus on two major developments, giving rise to three comparison sets.

First, the Supreme Court’s decision in TC Heartland presents a unique opportunity to measure the effects of forum crowding on decision quality in the Eastern of District of Texas and the District of Delaware, whose crowding is the consequence of two very different sources (forum selling and changes in venue rules, respectively).[120] In order to concentrate attention on the effects of forum crowding and to reduce the risks that other developments confound our results, we focused our study to the two years immediately preceding and following TC Heartland.[121] Indeed, the District of Delaware was near the nadir of its popularity among patent plaintiffs in 2015 and 2016, just before the Court’s decision in TC Heartland, when it accounted for 9% and 10% of all patent cases,  respectively.[122] And the District of Delaware was the most popular patent venue in the two years immediately following TC Heartland, shouldering 24% and 28% of the national load in 2018 and 2019 respectively.[123] We thus treated the District of Delaware as non-crowded from January 1, 2015, to December 31, 2016, and as crowded from January 1, 2018, to December 31, 2019.[124]

By contrast, the Eastern District of Texas actively solicited patent cases in the two years immediately before TC Heartland, accounting for about 40 to 45% of all patent cases, and saw its share of patent litigation wane drastically in the two years immediately following that decision.[125] Hence, we treated the Eastern District of Texas as crowded from January 1, 2015, to December 31, 2016, and as non-crowded from January 1, 2018, to December 31, 2019.[126]

Moreover, because judicial districts can vary over time in ways that would affect our results—judges retire while new judges come on the bench, leading to changes in judicial experience and, perhaps, judicial ideology—we further narrowed our analysis to consistent sets of judges within each district. In particular, we used two criteria to select judges from the Eastern District of Texas and the District of Delaware. First, we identified judges whose service included the entirety of our selected timeframes, both to study the crowding effects on individual judges who faced both crowded and non-crowded dockets and to avoid confounding our results with variations in judicial ideology, temperament, or experience.[127] Second, we focused on judges who were responsible for a significant number of patent cases within their respective districts. Our study includes, for example, Chief Judge Gilstrap, who is among the judges primarily responsible for the Eastern District’s popularity among patent plaintiffs,[128] as well as Judge Stark, who was recently appointed to the Federal Circuit from the District of Delaware.[129]

In short, by selecting the busiest patent judges within the busiest patent districts, we focused our study on those courtrooms where forum crowding was the most pronounced and where we expected its potential effects, if any, to be most conspicuous. In all, our examinations of the Eastern District of Texas focused on Chief Judge Rodney Gilstrap and Judge Robert W. Schroeder III, and our examinations of the District of Delaware focused on Judges Richard G. Andrews and Leonard P. Stark.[130] Figure 2, below, visually depicts these crowded versus non-crowded comparisons.

Figure 2: Crowded and Non-Crowded Periods by Judge

Second, Judge Albright’s appointment to the Western District of Texas’s Article III bench offers an additional opportunity to examine the effects of forum crowding. As noted, the Western District of Texas saw a meteoric rise in patent filings beginning in 2019, and so we consider January 1, 2019, to June 1, 2022, the date we ended data collection, as crowded. This period coincides, unsurprisingly, with Judge Albright’s tenure on the bench, as he was confirmed in late 2018.

But because we treat nearly all of Judge Albright’s tenure on the bench as crowded, to whom or what should we compare his outcomes? We cannot compare a crowded Judge Albright docket to a non-crowded one. Instead, we compare Judge Albright’s metrics to those of his compatriots on the Western District of Texas who hear comparatively few patent cases. In particular, we selected two judges who were appointed at about the same time (late 2018 or early 2019) and by the same president (President Donald J. Trump) to help account for judicial experience and ideology: Judge Walter David Counts III and Judge Jason K. Pulliam.[131] Similar to Judge Albright, both Judge Counts and Judge Pulliam served as judges prior to joining the federal bench. Judge Counts, like Judge Albright, served as a Magistrate Judge for the Western District of Texas,[132] and Judge Pulliam served as a Justice on the Texas Fourth Court of Appeals.[133] Additionally, all three have served as active judges in the Western District of Texas for roughly the same amount of time, as each received their commissions in 2018 or 2019.[134]

Table 1: Crowded and Non-Crowded Judges, W.D. Tex.

Judge	Status
Albright	Crowded
Counts	Non-Crowded
Pulliam	Non-Crowded

Hence, of our three comparison sets, two are intrajudicial and intertemporal (that is, they compare metrics of a consistent set of judges over two different time periods) and one is interjudicial and intratemporal (that is, it compares different judges over a single time period). In all, these comparisons allow us to study the effects of forum crowding in two ways. First, we compare the metrics of a consistent set of judges on the Eastern District of Texas and the District of Delaware during periods of relative crowding to those of no crowding. Second, we compare the metrics of the judge experiencing crowding in the Western District of Texas to those who simultaneously experience no crowding in that same district. While this latter comparison offers less compelling evidence of the direct effects of forum crowding (as opposed to, say, the idiosyncratic behavior of one judge),[135] it allows us to examine some consequences of the specific phenomenon that is behind many contemporary complaints about forum shopping, and selling, in patent cases.

2.     Data and Metrics

Having described our comparison sets, we turn next to describing our dataset, including our data sources and metrics for analysis.

Our data draws from three primary sources: the Federal Judicial Center, Lex Machina, and Westlaw. While we might have preferred to rely on a single source for all of our data and analysis, thus ensuring consistent definitions for, and treatments of, cases and their categorization,[136] we found that no single source provided a comprehensive set of docket-level information across all case types for a given judge. For example, Lex Machina provides a sophisticated set of tools for civil cases but contains no criminal case data. Conversely, the Federal Judicial Center’s Criminal Integrated Database provides comprehensive data on criminal dockets but no civil case data. And it offers only division-level information, rather than a judge-specific breakdown of cases. Moreover, while Westlaw included some features that proved particularly useful for tracking appeal outcomes and disposition dates, its methods of case categorization lacked the level of detail and consistency we required for other aspects of our analysis.

To measure the effects of forum crowding on decision quality, we collected information on non-patent cases. Our reason for excluding patent cases is simple: we want to distinguish the causes of crowding (i.e., patent cases) from their possible effects.[137] Consider, for example, the possibility that Judge Albright (or Judge Gilstrap) sought to attract patent litigation because of a passion for patent law and a belief that a particular application of patent doctrine is best for innovation policy. It may be, though, that Judge Albright’s vision for patent law is not shared by his colleagues on the Federal Circuit. And so the effect of Judge Albright’s forum crowding behavior (high reversal rates—due, perhaps, to an idiosyncratic view of patent law; or due, perhaps, to crowding effects) may be, in this hypothetical example, entangled with its cause (i.e., a desire to implement that idiosyncratic view). But this is not so for other cases, which provide a more reliable measure of the effects of this forum crowding. Hence, we collected information on non-patent civil and, where feasible, criminal cases. Moreover, within the civil case category, we also collected information about selected civil rights and habeas cases, defined to include claims arising under Section 1983, Bivens, or federal habeas corpus statutes.[138]

We focused our data collection efforts on three case-specific metrics: time-to-termination (i.e., the amount of time between a case’s filing and its final disposition); appeal rates (i.e., whether a given case was appealed, yielding an overall appeal rate calculation); and reversal rates (i.e., the outcomes of each of those appeals, yielding an overall reversal rate calculation).[139]

In selecting time-to-termination, appeal rates, and reversal rates as our metrics for analysis, we do not mean to suggest that these are the only relevant measures of forum crowding’s effects. For example, while time-to-termination measures may be suggestive proxies for judicial attention or procedural fairness, other more difficult-to-capture metrics, such as “bench presence,” may offer a richer metric.[140] And so we emphasize that our examination does not rely on bare statistics alone but also encompasses a qualitative examination of the practices of the judges in our dataset—Judge Albright’s practice, for example, of immediately referring nearly all of his non-patent cases to a magistrate judge in the first instance[141]—as well as a review of the underlying decisions and opinions themselves.[142] Indeed, we consider other metrics of decision quality, such as opinion length and detail, in our qualitative reviews of these decisions.

Likewise, we acknowledge some uncertainty over whether reversal rates are a perfect proxy for measuring decision quality, as relying on reversals may imply an assumption that, in a given reversal, the appeals court was correct and the trial court was not. But we mean no such implication and agree that correctness might be assessed on terms outside a court’s place in the judicial hierarchy. Hence, while we acknowledge these important epistemic questions, we rely on appellate outcomes for several other reasons: first, even if correctness might be measured on other terms, a reversal does suggest that the trial court’s decision was deficient—“improper or inadequate”—in at least one way; second, that deficiency gave rise to additional investments in the litigation process, and “a goal of the justice system is to reduce the need for appeals;” and third, reversal rates are consequently widely used in the literature as a proxy for decision quality.[143]

We turn next to a more detailed description of each of these three metrics.

3.     Time-to-Termination

We collected time-to-termination information in part to test our hypothesis that busy judges would have to move through individual cases on a crowded docket more quickly.[144]

Lex Machina automatically calculates time-to-termination, computing it as the difference between the filing date of the case and the termination date of the case as recorded on PACER.[145] Hence, for civil cases (including civil rights and habeas cases), we used Lex Machina’s time-to-termination metric.[146] For criminal cases, we used the Federal Judicial Center’s Criminal Integrated Database. While this database does not offer judge-specific information, it does offer division-specific information—and some divisions, recall, house only one judge. Hence, we filtered the database by district and by office code (i.e., division[147]), in order to obtain time-to-termination information for the respective criminal dockets of Judges Gilstrap, Schroeder, and Albright (each of whom is the only judge in their division).[148] Like Lex Machina, we calculated the difference between the filing date and the disposition date.[149] And because we were unable to obtain comprehensive judge-level criminal case data for Judges Andrews and Stark in the District of Delaware and for Judges Counts and Pulliam in the Western District of Texas, we cannot report their time-to-termination metrics in criminal cases.[150]

4.     Appeal Rates

As noted, we also collected information on how many cases are appealed. Some literature has used appeal rate data as a proxy for a litigant’s confidence in a district court decision.[151] Appellants may be likely to file only those appeals that they think they can win, or, at least, where they calculate its expected value to exceed its expected costs. Hence, as a decision’s quality decreases, the likelihood of appeal increases. And as decision quality in general decreases, appeal rates should increase in aggregate.[152]

While we collect and report on this metric, we share other scholars’ trepidation about its utility.[153] Appeal rates are the function of myriad variables, including confidence in the appealed decision, the ability to pay for the appeal, or the desire for finality. So it is not obvious that appeal rates will uniquely reflect litigant confidence in the district court’s opinion, as opposed to these other concerns.

Appeal rate information can also serve a distinct, but related, function: it can help alert for possible selection effects. For example, one might view an increase in reversal rates accompanied by a drop in appeal rates as suggestive of appeal selection and not decision quality. That is, the new reversal rate might be attributable to better appellant decision-making—i.e., better selection of the decisions that merit an appeal—rather than an overall effect on decision quality.

We calculate appeal rate metrics straightforwardly, dividing the total number of appeals taken by the number of district court dispositions during the study period.[154] We collect the total number of district court dispositions as described above (for calculating time-to-termination rates), using either Lex Machina or the Criminal Integrated Database. We collect the number of appeals taken using Westlaw’s appeal analytics tools.[155]

5.     Reversal Rates

Finally, we calculated reversal rates to help assess decision quality. These measures are based on appeal outcomes for each judge during both crowded and non-crowded periods, relying, again, on Westlaw’s database.[156] More specifically, we retrieved all the appeals Westlaw identified as originating with the judges during our study period and divided those into three sets: criminal cases, non-patent civil cases, and patent cases.[157] As noted above, we excluded the patent appeals from our review.[158]

Indeed, our research design, emphasizing patent-related crowding, is uniquely suited to study reversal rate effects (as opposed to a design emphasizing crowding caused by some other sorts of cases). This is because patent cases have their own appellate pathway—patent cases are appealed to the Federal Circuit, while other cases are typically appealed to the relevant regional circuit (the Fifth Circuit in the case of Texas; the Third Circuit in the case of Delaware).[159] Hence, because patent cases are filtered out at the appellate level, we need not worry about any possible effects of appellate crowding; that is, an influx of patent cases will not yield an influx of patent appeals that may affect appellate outcome measures.[160]

We readily acknowledge that our approach leads, in some instances, to smaller sample sizes. Our analysis is limited to appealed cases from a select set of judges over a limited time period. But we emphasize that our study’s sample encompasses nearly the entire universe of opportunities to study this phenomenon. Because of their unique appellate pathway, patent cases present a rare opportunity to measure district court effects by reference to appellate outcomes; and TC Heartland is a unique “policy shock” that caused a substantial change to patent plaintiffs’ litigating behavior. On balance, we think the benefits of limiting our view to a consistent set of judges, and to the time period most immediately adjacent to TC Heartland (namely, reducing the likelihood that our results are affected by confounding factors), offset concerns about the smaller sample size. Furthermore, we accompany our quantitative analysis with a corroborating qualitative examination of these courts’ decisions and procedures.

After identifying and categorizing the relevant appeals, we recorded the outcome (affirmed, or not) for each case. Affirmances include any decision affirmed in its entirety, any appeal that was dismissed, and any writs of mandamus that were denied.[161] Reversals include decisions that were reversed, vacated, or remanded, whether in whole or in part,[162] as well as petitions for writs of mandamus that were granted. And outcome measures are reported as a percentage of all appeals. For example, twenty-four of the non-patent cases that Judge Stark decided between January 1, 2015, and December 31, 2016, were appealed to the Third Circuit. Of those, twenty-two were affirmed. Hence, Judge Stark’s affirmance rate during this non-crowded period was 91.7% (or twenty-two out of twenty-four).

B.     Results and Analysis

We present our results and analysis below. Our analysis of crowded and non-crowded dockets in the Eastern District of Texas, the District of Delaware, and the Western District of Texas suggests a stark pattern: when judges crowd their dockets, decision quality suffers. We emphasize, moreover, that our results are founded on the entire population of relevant cases and satisfy other measures of statistical significance.[163]

We begin by presenting our reversal rate and appeals rate data, alongside our qualitative review of decisions in these dockets, in order to describe the relationship between crowding and decision quality. We find that quality is lower when crowding is higher. We then turn to the question of mechanism: why, exactly, are crowding and decision quality related? Here, we find that our more granular results, alongside our time-to-termination statistics, help to reveal the extent to which judicial preferences matter.

1.     Decision Quality

Our examination of the quality effects on crowding starts with our reversal rate metric. These measures suggest that decisions rendered on crowded dockets have been reversed more frequently than those on non-crowded dockets.

We begin with the Eastern District of Texas and the District of Delaware. Recall that these comparisons examine the behavior of a consistent set of judges over time, namely, before and after TC Heartland, which caused a vast number of cases to shift from the Eastern District of Texas to the District of Delaware.[164] But the sources of crowding in each of these districts is very different: in the Eastern District of Texas, it is forum selling; in the District of Delaware, it is the change in venue rules occasioned by TC Heartland.

Figure 3: Reversal Rates, Eastern District of Texas (All Cases, n=18)

Figure 4: Reversal Rates, District of Delaware (All Cases, n=158)

In the Eastern District of Texas, reversal rates during periods of crowding are substantially higher than during periods of less crowding.[165] The District of Delaware evinces a similar, if less pronounced, trend. While we revisit the differences in the strength of the trend in the next Section,[166] it is worth remarking on how this effect is consistent across various classes of cases.[167] For example, in civil rights and habeas cases, reversal rates are starkly higher in both the Eastern District of Texas and the District of Delaware during periods of forum crowding.

Figure 5: Reversal Rates, Eastern District of Texas (Civil Cases, n=17)

Figure 6: Reversal Rates, District of Delaware (Civil Cases, n=123)

Figure 7: Reversal Rates, Eastern District of Texas (Civil Rights & Habeas Cases, n=6

Figure 8: Reversal Rates, District of Delaware (Civil Rights & Habeas Cases, n=20)

Moreover, we do not believe that the results illustrated above result from changes in the body of appealed cases. As noted above, reversal rates, if accompanied by changes to the appeal rate, might be ascribed to a more careful selection of cases for appeals—i.e., putative appellants may improve their selection of cases for appeal, bringing only the one they are likely to win, and forgoing others. But here, the appeal rates appear rather steady over our study periods, as demonstrated in Tables 2 and 3. Hence, changes in the appeal rate cannot account for these differences in the reversal rate.

Table 2: Appeal Rates, Eastern District of Texas (Civil Cases)

District	Crowded Rate(%)	Non-Crowded Rate(%)
E.D. Tex.	1.5	1.4

Table 3: Appeal Rates, District of Delaware (Civil Cases)

District	Crowded Rate(%)	Non-Crowded Rate(%)
D. Del.	3.4	3.3

In all, the overall trends at the district level across both the Eastern District of Texas and the District of Delaware are consistent with the view that forum crowding affects decision quality. Indeed, the difference between reversal rates during crowded and non-crowded periods is significant.[168] This result suggests that forum crowding adversely affects decision quality. These findings are further corroborated by our qualitative examination of the orders and decisions issued by these judges during these periods.[169]

We might, for example, look at two similar cases—two appeals of a bankruptcy court ruling in a proceeding regarding individual debtors—before Judge Gilstrap (of the Eastern District of Texas), one decided during a crowded period and one not.[170] Judge Gilstrap presided over both cases without the assistance of a magistrate judge, and he affirmed the Bankruptcy Court’s rulings in both cases. However, the order rendered during the crowded period cites no caselaw and is barely a page in length, while the order from the non-crowded period is more than five times longer and features more thorough background and discussion sections (including citations to caselaw, the docket, and relevant statutes).[171]

Two of Judge Gilstrap’s employment cases—again, one before TC Heartland and another after—further highlight this phenomenon.[172] Both cases involved lawsuits brought by employees against their employers under the Fair Labor Standards Act. In both cases, Judge Gilstrap granted the defendant employers’ motions to transfer venue against the plaintiff employees’ wishes. The order granting the motion to transfer venue from the crowded period, however, is a little under two pages and offers no analysis of the parties’ arguments. The order filed during the non-crowded period is almost four times as long and provides an in-depth discussion of the applicable legal standard, the relevant public and private interest factors, and the parties’ claims.

Indeed, a more general review of Judge Gilstrap’s case dockets before and after TC Heartland suggests that, during periods of crowding, Judge Gilstrap’s orders are far less frequently accompanied by explanatory memoranda, and even when such memoranda are included, they tend to be less comprehensive. Such markers of “judicial effort” per case reflect not only on decision quality, but also have important implications for public reasoning and procedural fairness values in judicial decision-making.

We can find similar examples from the District of Delaware. During Judge Stark’s non-crowded period, he issued a twelve-page order granting in part and denying in part cross-motions for summary judgment on a plaintiff’s appeal of her denial of disability insurance benefits under the Social Security Act (SSA).[173] Judge Stark spent four pages of his opinion in Cannon v. Colvin on the factual background, describing each of the plaintiff’s injuries in detail.[174] And he spent another four pages on his discussion applying the relevant law to the facts, largely regarding the appropriate weight to be given to expert testimony.[175] In contrast, during his crowded period, Judge Stark considered a similar case in an analogous posture: cross-motions for summary judgment on a plaintiff’s appeal of her denial of disability insurance benefits under the SSA.[176] This eight-page opinion, which reached the same legal conclusion based on the same issues as Cannon, spent only two of those eight pages applying the relevant law to the weight of the expert’s testimony.[177] Notably, the order spends only one short paragraph discussing the justifications for the administrative law judge’s choice not to credit the treating physician’s records.[178] But the same issue in the non-crowded opinion spans almost three pages, and includes copious citations to the administrative record.[179]

Similar results attend to our quantitative and qualitative assessments of the Western District of Texas, lending further support to the inference that crowding affects quality. Judge Albright’s track record before the Fifth Circuit features a substantially higher reversal rate compared to that of his contemporaries in the Western District of Texas. Figure 9, for example, implies that Judge Albright was reversed (or remanded, etc., in whole or in part) more than twice as often as Judge Counts and 1.4 times as often as Judge Pulliam.

Figure 9: Reversal Rates, Western District of Texas (All Cases, n=232)

Figure 10: Reversal Rates, Western District of Texas (Civil Cases, n=62)

Moreover, Judge Albright’s governing standing orders assign nearly all civil matters to magistrate judges, save for only a few exceptions: patent cases, as well as habeas corpus, Section 1983, and Bivens cases (i.e., our category of “civil rights and habeas” cases[180]).[181] The results illustrated in Figures 9 and 10 are further exaggerated when we narrow focus to only those cases that do not automatically benefit from an initial review by a magistrate judge.[182]

Figure 11: Reversal Rates, Western District of Texas (Civil Rights & Habeas Cases, n=24)

As Figure 11 demonstrates, Judge Counts and Judge Pulliam have a perfect record in habeas, Section 1983, and Bivens cases, whereas Judge Albright—who reserves only these cases for his initial first review (alongside his extensive patent docket, of course)—is reversed in more than one-quarter of these cases, suggesting an effect on decision quality in his comparatively crowded docket.[183]

These results, moreover, are reflected in the Western District of Texas’s appeal rates data. As noted above, we are somewhat unsure that appeal rate serves as a clear signal of decision quality, given that the decisions to appeal are informed by a wide range of factors beyond the mere likelihood of error.[184] But these appeal rate measures offer some additional soft support for a view that crowding affects decision quality, at least in the Western District of Texas, where Judge Albright’s non-patent civil cases were both appealed more frequently and reversed more frequently than those of his judicial colleagues. These measures thus also strongly undercut the possibility that differences in the reversal rate are attributable to differences in the selection of cases for appeal.

 Table 4: Appeal Rates, Western District of Texas (Civil Cases)

Judge	Status	Appeal Rate (%)
Albright	Crowded	5.4
Counts	Non-Crowded	2.2
Pulliam	Non-Crowded	2.1

Again, these results are even more pronounced when narrowed to only those cases for which Judge Albright bears primary responsibility (i.e., excluding those cases referred to a magistrate for a first review).

Table 5: Appeal Rates, Western District of Texas (Civil Rights & Habeas Cases)

Judge	Status	Appeal Rate (%)
Albright	Crowded	42.9
Counts	Non-Crowded	5.7
Pulliam	Non-Crowded	7.3

 Judge Albright’s performance in criminal cases (cases which are initially referred to a magistrate for pre-trial matters, among other things[185]) similarly lags behind his peers. As Figure 12 illustrates, the Fifth Circuit reversed Judge Albright’s criminal decisions more than twice as often as it reversed Judge Counts, while Judge Pulliam has not been reversed in any criminal matter.[186]

Figure 12: Reversal Rates, Western District of Texas (Criminal Cases, n=170)

And, as in the Eastern District of Texas and the District of Delaware, a more qualitative review of some of the opinions in these sets of cases reinforces the finding that forum crowding affects decision quality.

In Judge Albright’s first year as the nation’s busiest patent judge, he presided over Antonio Gardner’s federal criminal proceedings.[187] One day after Gardner filed a motion to withdraw his guilty plea, Judge Albright, “in a one-word order, denied his request to withdraw the plea, without an evidentiary hearing,” later sentencing Gardner to a twenty-year prison term (followed by six years of supervised release).[188] On appeal, the Fifth Circuit chided Judge Albright for failing to adequately analyze Gardner’s arguments and for offering no reasoning beyond “a single-word order—‘DENIED.’”[189] The terseness of Judge Albright’s consideration in this case (and others) offers further evidence of a possible effect on decision quality.[190]

Consider, too, the Fifth Circuit’s decision in Alvarez v. Akwitti.[191] There, Alvarez asked Judge Albright to hear his pro se complaint, alleging that prison officials were putting him at risk of retaliation from a “sexually violent predator inmate.”[192] Judge Albright dismissed the complaint sua sponte, even before the defendant warden had filed a response. The Fifth Circuit reversed and remanded, explaining that Judge Albright’s “failure to consider the entirety of Alvarez’s allegations” required a closer look and reminding Judge Albright of his obligations to pro se applicants in particular.[193]

We cannot, of course, say conclusively that these results are not the byproduct of some idiosyncrasy in Judge Albright’s approach to civil rights and criminal cases. It may well be that Judge Albright simply approaches these cases differently from his compatriots on the Western District’s bench.[194] Nevertheless, our appeals outcome results in the Eastern District of Texas and District of Delaware (which, recall, assess trends over a consistent set of judges in each venue) offer stronger evidence for the view that forum crowding matters. And our qualitative reviews of decisions from those districts reinforce the finding that forum crowding matters—and that it matters for values that sound not only in “correctness,”[195] but also in such concerns as public reasoning,[196] procedural fairness, and judicial legitimacy.

2.     Explaining Forum Crowding’s Effects

There are at least two possible explanations for why forum crowding—and forum crowding resulting from the sort of patent-related forum-shopping and forum-selling conduct described above—may adversely affect decision quality.

One possibility is that forum-selling conduct reflects judicial preferences: Judge Albright and Chief Judge Gilstrap seem to strongly prefer patent cases.[197] And so, given the opportunity to choose to work on one among a wide range of patent cases or something else, we might expect them to prefer patent cases (and, perhaps, to neglect the rest of their dockets). If this were so, then we would expect a stronger effect among the judges who seem to prefer patent cases (a preference evinced, perhaps, by forum selling).

A second, alternate (and potentially complementary) possibility is encapsulated in Judge Edwards’s view that bigger dockets give judges less time to work on cases, thereby increasing the likelihood of error.[198] This hypothesis is more indifferent to judicial preferences, focusing instead on a court’s finite bandwidth or capacity for cases. Under this view, forum-crowding-related capacity constraints affect decision quality no matter whether the cause of such crowding is internal to the forum (e.g., a consequence of forum-selling conduct) or external (e.g., a byproduct of the Court’s decision in TC Heartland).[199]

Our results are suggestive of both possibilities. On the one hand, we see decision-quality effects across both sets of districts—those whose active solicitations of patent litigation seem to suggest a preference for such cases (i.e., the Eastern and Western Districts of Texas), as well as those where forum crowding was the mere byproduct of TC Heartland’s revised interpretation of the governing venue statute (i.e., the District of Delaware). But, on the other hand, the strength of the trend is more pronounced in those districts where there is an apparent preference for patent litigation (evidenced by forum selling). And so, while we ultimately conclude that both judicial preferences and resources seem to have a role to play in this story, it is worth further unpacking our results and their relationship to these possible mechanisms.

We begin by trying to discern the source of the disparity in the strength of the forum-crowding effects between the District of Delaware and the Eastern District of Texas. Specifically, we disaggregate the district-level results into more judge-specific results. Here, we uncover something possibly surprising. While Judge Stark’s reversal rate more than doubled when his docket was crowded, Judge Andrews, interestingly, did better on appeal to the Third Circuit during the more crowded years immediately following TC Heartland.

Figure 13: Reversal Rates, Judge Stark (All Cases, n=72)

Figure 14: Reversal Rates, Judge Andrews (All Cases, n=87)

We might conclude from these results that we have resolved our mechanism puzzle. After all, Judge Stark was, as noted, elevated to the Federal Circuit after our study period. This appointment might signal a substantial interest in patent cases. Judge Stark might have, for example, developed a preference for patent cases after having been exposed to so many of them as a consequence of TC Heartland (or, perhaps, he always preferred patent cases, but not so much as to engage in forum selling). These individuated results cleanly cleave those judges who have evinced some preference for patent cases (either in their forum-selling conduct or in their subsequent career paths) from those who have not, with negative effects on decision quality in only the former group.[200] But Judge Stark’s promotion, while suggestive of a preference for patent litigation, is less dispositive of such an inclination (as compared to, say, Judge Albright’s open solicitations to patent plaintiffs).[201]

Hence, we turn to our time-to-termination results in order to assess the extent to which capacity matters. Here, we find that both Judge Stark and Judge Andrews took less time to resolve non-patent civil cases during the crowded study period than during the non-crowded study period.[202] Stated similarly, forum crowding constrained the time they could dedicate to each case, especially given that judges often hesitate from—and are admonished against—letting cases languish for too long.[203] These effects were far more pronounced for Judge Stark, whose median time to termination went from 460 days before TC Heartland to only 181 days afterward—a decrease of over 60%. By contrast, Judge Andrews, who was working rather quickly to begin with, saw his median time-to-termination reduced by a more modest amount.

Figure 15: Time to Termination, Judge Stark (Civil Cases, n=939)

Figure 16: Time to Termination, Judge Andrews (Civil Cases, n=869)

Judge Andrews, moreover, appears to have been more strategic in allocating his time to cases. Our closer look at some of the orders and cases underlying these results suggests that Judge Andrews is the only judge in the District of Delaware to invariably include an option to refer a case to alternative dispute resolution on his scheduling order form, and he seems to have used that procedure with some regularity.[204] Hence, some cases settled quickly, and Judge Andrews accordingly spent less time on them; other cases, however, still required more attention. Judge Stark, by contrast, reduced the time he spent on all non-patent civil cases, no matter whether they ended with a settlement or some other judgment.[205] In all, these time-to-termination results—which can be understood to estimate available judicial resources or capacity—mirror our decision-quality results: where judicial capacity is more tightly constrained (relative to a non-crowded baseline), effects on decision quality are stronger.[206]

Our Western District of Texas results are similar: Judge Albright moved through his civil docket the most quickly and had the highest reversal rate of these judges, while Judge Counts took the most time with each civil case and had the lowest reversal rate.[207]

Figure 17: Time to Termination, Western District of Texas (Civil Cases, n=1,552)

But our time-to-termination results in the Eastern District of Texas are more surprising: both Chief Judge Gilstrap and Judge Schroeder take slightly more time per case during periods of crowding, as Figure 18 (and Appendix Figures 6–9) illustrate.

Figure 18: Time to Termination, Chief Judge Gilstrap (Civil Cases, n=636)

How could this be? As noted above, judges face significant institutional pressure against drawing cases out; but here, the Eastern District appears to be doing exactly that—taking longer to resolve cases, even as its docket grows.[208] Our hypothesis is that the Eastern District of Texas was simply too busy. Recall that, during years of crowding, the Eastern District was responsible for nearly half of all the nation’s patent cases.[209] Facing such an incredible workload, Judges Gilstrap and Schroeder may have had no choice but to let some cases languish for longer. We return to this hypothesis in the next Part. For now, it suffices to note that these results make it harder to discern whether the Eastern District of Texas in fact dedicated less capacity to each nonpatent civil case during years of crowding (compared to years of less crowding).[210]

Hence, the relationship between judicial capacity and decision quality does not align quite as neatly as the relationship between judicial preferences and decision quality. Constraints on judicial capacity—reflected in less time dedicated to each case—correspond roughly to increases in the reversal rate in the District of Delaware and the Western District of Texas, while our time-to-termination results from the Eastern District of Texas are harder to interpret. Meanwhile, every judge who has evinced a preference for patent cases seems to neglect the rest of their docket during periods of crowding (measured by reversal rate), and every judge who has evinced no such preference seems to be affirmed at the same—or better—rates during periods of crowding.

In short, these results corroborate the view that judicial preferences matter: we see negative effects on decision quality only among those judges who have evinced a preference for patent litigation, either through their forum selling conduct or through other choices.

But we are also mindful of the fact that our time-to-termination results in both the District of Delaware and the Western District of Texas also correlate, roughly, to our decision-quality results, while the Eastern District’s time-to-termination statistics are somewhat perplexing. And so we do not discount the oft-repeated view, voiced by judges with relevant experience, that resource constraints affect the care and attention that judges can give to the cases before them.

This is particularly so because we might understand these two hypotheses—one about judicial preferences and another regarding capacity and bandwidth—as complementary. Judicial preferences matter. And they might matter, especially, to how judges allocate their time and other resources. We can uncover some evidence for this hypothesis in the way these judges allocate other resources, such as in clerk hiring practices, or in the assignment of matters to magistrate judges. Judge Gilstrap, for example, prefers to hire law clerks with expertise relevant to patent litigation.[211] But such hiring practices may mean that these chambers lack familiarity with other bodies of doctrine. Similarly, Judge Albright has explained that the magistrate judges in the Western Division of Texas help him stay “afloat” with his patent cases by “taking on more duties from [his] docket” such as “felony pleas of guilt,”[212] while Judge Albright keeps other matters—such as all patent cases—for himself. In short, Judge Albright and Judge Gilstrap, perhaps among others, allocate their resources to focus on the cases they prefer—patent cases—to the seeming exclusion of other matters.

III. Reforming Forum Crowding

When judges crowd their dockets, the quality of their judicial decisions seems to decline. We consider these consequences of forum crowding to be among the externalities of forum shopping and, especially, forum selling, given our view that judicial preferences are a primary driver of the decision-quality effects noted above.[213] Such preferences—and the forum crowding that they give rise to—have real effects on real litigants. Consider, again, Antonio Gardner, whose criminal case sat on Judge Albright’s docket in 2020. In cases like his, the difference between a crowded forum (crowded because of forum selling) and a non-crowded one—one, perhaps, with more capacity or criminal law expertise—could be the difference not only between a terse one-word summary denial and a reasoned decision on the merits of a motion but also between a twenty-year prison sentence and a different outcome altogether. And so, in the following sections, we consider procedural reforms that can help mitigate these pernicious effects.

Externality regulation often takes one of (at least) two forms: first, policymakers might directly address the underlying behavior giving rise to the consequences of concern; or, second, policymakers might impose rules—taxes, say—that require the actors causing these negative effects to internalize the costs of those effects.[214] Consider, for example, a chemical factory using a production process that creates a toxic byproduct. Policymakers may respond in a few ways. First, they might decide to directly ban the dangerous process, thus forcing the factory owners to use a different process or enter a different business. Alternatively, they might decide to tax the toxic byproduct to create a fund for mitigation and clean-up measures, letting the factory owners decide whether the cost of the tax is worth the pecuniary benefits of the toxic process.

We consider similar approaches to the problems of forum crowding. First, we consider procedural reforms that directly address forum selling and forum shopping.[215] Second, we consider ways in which the courts might cause these plaintiffs to internalize the costs of selecting a crowded venue, so that they may decide whether the benefits of selecting a crowded forum outweigh the costs of doing so. Though we stop short of suggesting that plaintiffs with venue flexibility pay a toll to file in crowded venues, we consider reforms to the six-month list that may cause these plaintiffs to cede time, in the form of priority, to other litigants.

A.     Addressing Forum Selling and Forum Shopping

We begin by considering a range of reforms that would directly address forum selling and forum shopping, thereby mitigating the decision-quality consequences of forum crowding. For one, we join other scholars in advocating for reform to the procedural flexibility that allows judges to engage in forum selling. We also echo calls for reform to the venue rules that allow plaintiffs to file in practically any district (and any division of that district). And we highlight related reforms, such as transferring cases out of crowded venues into less crowded ones.

1.     Reforming Forum Selling and Procedural Innovation

We begin by advocating for limits on forum selling, particularly in view of our findings that judicial preferences may matter most for decision quality in crowded fora. Scholars studying the problems of forum selling have advocated for limits on local procedural flexibility. In patent contexts, for example, some scholars have focused on the possibility of standardizing local rules so as to limit possibilities for forum selling through patent-specific (and plaintiff-friendly) procedure.[216] Indeed, the Administrative Office of the U.S. Courts canceled its “Patent Pilot Program,” which created patent-specific procedures in selected districts, in part to “ensure that all district judges remain generalists.”[217] We concur with such calls for more regularity across districts and division in procedure because such standardization can help reduce forum selling and the pernicious effects of forum crowding that follow while also addressing concerns about judicial legitimacy.

One target for procedural reform is especially ripe: the rules governing district court stays while parallel Patent Office proceedings are pending. In many patent cases, an alleged infringer will challenge the validity of the patent giving rise to the dispute. Such challenges were traditionally brought in district courts as a counterclaim.[218] But such challenges are now often also brought directly to the Patent Office’s Patent Trial and Appeal Board in a specialized proceeding known as inter partes review (or IPR).[219] This dual-track system—one in the courts, one in the agency—gives rise to a difficult coordination problem. When an alleged infringer responds to a lawsuit by filing both a counterclaim and an agency challenge (as patent defendants typically do[220]) regarding the patent’s validity, who goes first—the court or the agency?[221]

The doctrinal analysis is somewhat circular: on one hand, many courts are more likely to stay litigation in view of an active agency challenge;[222] but, on the other, the Patent Office is less likely to agree to hear the challenge without a judicial stay, or if a trial date is looming.[223] That helps to explain, at least in part, why Judge Albright declines to stay litigation pending Patent Office review and instead often schedules unrealistically optimistic trial dates.[224] Such tactics help ensure that Judge Albright retains control over these disputes, rather than ceding them to the Patent Office.

Practitioners, policymakers, and scholars have all invoked considerations such as cost, efficiency, and fairness in advocating for resolutions for this venue conflict. One alternative to the current paradigm emphasizes agency substitution. A substitution—replacing court cases with an agency process—is significantly cheaper.[225] Agency adjudication also tends to be faster (except, perhaps, where the parallel litigation is at a significantly advanced stage).[226] But some have argued that litigation better honors the patent-holding plaintiff’s choice of forum and that the standards applied in district court are more likely to vindicate the patentholder’s expectation interests.[227] Indeed, some have even suggested, rather speciously, that choosing litigation over agency adjudication upholds patentholders’ “[c]onstitution[al] . . . rights.”[228]

Our forum crowding results favor agency adjudication over litigation. Moving these cases out of the courts can help address crowding-related concerns, including by improving decision quality across a wide range of cases. Hence, as we elaborate below, both the Patent Office and the courts should account for these possible docket effects when deciding whether to institute agency review or stay litigation pending an agency decision. Specifically, the standards applied in both the agency and the district courts invite consideration of the public interest, and such considerations should encompass the effects of forum crowding.

We begin with the Patent Office. As noted, the Patent Office hosts an agency-based adjudicatory process for challenging a patent’s validity. This process, known as inter partes review (IPR), proceeds in two stages: first, institution; second, merits. At the first stage, the Patent Office has total discretion to grant or deny any petition for review that evinces a “reasonable likelihood of success” on the merits (and that complies with various other statutory requirements).[229] If the Patent Office decides to grant a petition for review and therefore institutes an IPR, then it proceeds to a determination of the patent’s merits, reviewing whether the patent satisfies certain statutory requirements of patentability.

The standards governing whether the agency should grant or deny a petition are set out in a short series of decisions, known collectively as NHK-Fintiv, regarding the Patent Office’s institution-stage IPR decisions. The factors emphasized by NHK-Fintiv include whether the competing district court has stayed its case pending the Patent Office’s decision, whether the district court is likely to begin a trial on the merits before the Patent Office can issue its final decision, and “other circumstances” that may affect the agency’s decision to exercise its discretion.[230] Agency officials have explained that such “other circumstances” include an evaluation of the public’s interest in instituting IPR.[231]

Our findings suggest that instituting IPR can serve the public’s general interest in better, higher-quality adjudication in the federal district courts, particularly where any parallel litigation takes place in a crowded forum.[232] This is because the Patent Office may only institute review where a petitioner evinces a “reasonable likelihood” of proving a patent to be invalid.[233] And an IPR that successfully voids an incorrectly issued patent is likely to dramatically simplify any litigation based on that patent.[234] Stated simply, IPR will likely end the patent litigation. And our results suggest that some district courts—particularly those that are especially popular with patent plaintiffs—issue higher-quality decisions as their patent litigation burden wanes.[235] In short, the public has an interest in high-quality decisions across a crowded forum’s docket; IPR institution decisions should expressly reflect this concern.

Likewise, district courts should stay litigation pending agency review with more regularity. So far, district courts “have been nearly uniform in granting motions to stay” trial proceedings pending the agency’s merits decision, particularly because doing so will, as noted, typically “reduce the burden of litigation on . . . the court” (one “consideration that courts often take into account in determining whether to grant a stay pending inter partes review”).[236] But district courts have been somewhat more reluctant to issue stays while the Patent Office is still deciding whether to institute review at all.[237] That is misguided. Such a stay improves the likelihood of institution by ensuring that the parallel litigation does not advance too fast (one factor, as noted, that the agency considers when ruling on an institution petition) during the six months in which the Patent Office may review and consider the petition.[238] And if other aspects of the petition counsel in favor of IPR, then the agency’s decision to institute review may simplify the litigation down the line, giving rise to similar effects.[239]

The interworkings of these procedural provisions across two branches of the federal government are somewhat complicated. But the bottom lines for the Patent Office, for the district courts, and even for Congress, are straightforward.

First, the Patent Office should consider forum crowding as among the other public interest circumstances that counsel in favor of instituting review.[240] If a petition seems reasonably likely to succeed and if it relates to litigation in a crowded forum (especially one engaged in forum selling), the agency should consider the reduced docket pressure on that district court and related benefits as among the considerations weighing in favor of review.

Second, district courts with crowded dockets should stay litigation pending IPR. This is true across both stages of IPR—at institution and on the merits. Indeed, district courts should even encourage parties to take their validity disputes to the Patent Office, much as Judge Andrews directed his cases to means of alternative dispute resolution.[241] Doing so may help to both reduce docket pressure and give the court time to focus on other cases, including those that have comparatively less venue flexibility.[242] And such a presumption in favor of a stay is consistent with the traditional principle of staying litigation while sorting out the availability of mechanisms of alternative dispute resolution.[243]

Finally, we note two “legislative recommendations” to address concerns flowing from forum shopping, forum selling, and forum crowding.[244] One, Congress can simplify the complicated interactions between pending litigation and inter partes review by enacting a mandatory stay provision. Any litigation based on a patent that is the subject of petition for IPR should be stayed automatically for six months,[245] pending the agency’s resolution of the petition. And if the petition is granted, then litigation should be stayed for twelve more months pending the agency’s merits decision (as the Patent Office is required, by statute, to reach a merits decision within twelve months).[246] Two, inter partes review is presently limited in scope, but Congress can help by expanding IPR’s scope to encompass all available validity challenges (and, of course, by expanding the concomitant estoppel provisions that prevent petitioners from bringing later redundant challenges to the courts[247]).[248] For example, petitioners can currently allege that a patent is invalid because it is a mere obvious improvement over the state of the art, but they cannot challenge a patent for insufficiently disclosing and describing the new advance it purports to cover.[249] As a consequence, agency review does not always simplify future litigation, as a patent may survive IPR but be subject to further validity challenges in the district courts.[250]

Stated simply, our results counsel in favor of more agency substitution—i.e., more agency review—particularly where agency process will reduce docket pressure in crowded fora.

2.     Addressing Forum Shopping and Venue Rules

In addition to addressing forum-selling conduct by, say, improving procedural regularity across the district courts (especially as pertaining to parallel agency proceedings), policymakers can also address forum-shopping conduct by litigants through reforms to the venue rules.

Several scholars have already advanced a range of proposed reforms. Jeanne Fromer, for example, proposed a rule that largely restricts venue in infringement cases to the district of any defendant’s principal place of business. She reasoned that such a rule would improve district court decision-making in patent cases by ensuring that local courts can build—and build upon—their familiarity with local industry (alongside substantive patent law).[251] While that proposal focuses on district-level case assignment, Jonas Anderson and Paul Gugliuzza have gone further, proposing a rule requiring that divisional assignments also reflect the facts of the underlying dispute.[252] In their view, if a case alleged that infringing conduct occurred in San Antonio, then the case should be heard in the San Antonio Division, not Waco, even if both San Antonio and Waco fall within the Western District of Texas. And they would add further rules to promote random case assignment when filing in single-judge divisions.[253]

We concur with proposals such as these, primarily because we think they may address the pernicious effects of forum crowding. Specifically, these sorts of proposals seem likely to give rise to a more even distribution of patent cases, thus reducing crowding in any given venue. Jeanne Fromer, for example, demonstrated that her proposed venue rule would give rise to distinct geographic clusters of certain patent cases and, importantly, that these clusters would be more geographically dispersed than under current conditions (which, as noted, tend to cluster cases in Texas and Delaware).[254] Similarly, the proposals to avoid concentrations of cases within a single division of a district—through both more restrictive divisional venue rules and procedures promoting a wider distribution of cases among more judges—seem likely to help disperse crowds of cases.[255] In all, we favor modifications to venue rules—including the ones described here—that tend to reduce crowding in any given courtroom by creating a more even distribution of (patent) cases.[256] Such rules will blunt the effect of forum selling and forum shopping.

Our focus on venue rules has further implications for transfer motions, too. Crowded venues should grant motions to transfer cases to courts that are less busy, provided that doing so is consistent with any applicable venue rules and does not raise concerns about defensive gamesmanship. In short, crowding should factor into decisions about whether to grant venue transfer motions.

In criminal contexts, courts already occasionally consider docket load and capacity when deciding whether to relocate a criminal trial. Federal Rule of Criminal Procedure 18 gives courts discretion to set the “place of trial within the district” after giving “due regard [to] the convenience of the defendant . . . and the prompt administration of justice” (among other considerations).[257] And in approving intradistrict (interdivision) transfers, some courts have explained that “the prompt administration of justice includes more than the case at bar; the phrase includes the state of the court’s docket generally, [and so the] court . . . must weigh the impact the trial location will have on the timely disposition of the instant case and other cases.”[258]

Capacity and load could similarly influence decisions on transfer motions in civil patent cases. Indeed, the Fifth Circuit’s test for evaluating a transfer motion, which is the same test that the Federal Circuit ostensibly applies when reviewing transfer decisions out of the Eastern and Western Districts of Texas, directs district courts to consider “difficulties flowing from court congestion,” alongside seven other factors.[259] In practice, however, congestion-related considerations have been largely (if inconsistently) minimized.[260] For example, Judge Albright has reasoned, with the Federal Circuit’s apparent approval, that the “most important factor in the transfer analysis is the convenience of the witnesses.”[261] By contrast, immediately after TC Heartland, some (visiting) judges in the District of Delaware cited court congestion, among other factors, in support of decisions to transfer litigation out of the district.[262] Our results suggest that court congestion deserves more attention in transfer contexts than it currently receives.

One further clarification is in order. Some readers may object that our focus on the distribution of cases does not truly address the problems of docket crowding. That is, a court’s decision to transfer a case to another venue (or a plaintiff’s decision to file in that other venue to begin with) does not fully resolve the crowding effect—it simply moves the problem elsewhere.[263] Each case, per this objection, contributes to docket congestion, no matter where it is filed. Perhaps. But there are at least three responses to such an objection.

First, recent work by Christian Helmers and Brian Love indicates that venue flexibility and forum selling induce additional patent litigation.[264] Stated otherwise, they find that some patent plaintiffs sue only because of the availability of favorable fora. Hence, more consistent procedures across fora and more restrictive venue rules (including on motions to transfer) seem likely to reduce the overall volume of patent litigation.

Second, as we suggested above, we suspect that judicial capacity for cases tracks something approximating an s-curve: some courts may have ample capacity and can easily accommodate an additional case with little effect; other courts may be so congested such that one additional case may make little practical difference. Indeed, the unusual time-to-termination results from the Eastern District of Texas seem consistent with a view that, past a certain point, judges cannot keep up and must let cases sit on the docket for longer. Hence, we emphasize venue and transfer reforms that seem likely to move cases primarily to those venues that have excess docket capacity and away from those venues that have none. We cannot definitively say, of course, that any one of these venue reforms will solve the problem in this way. These reforms may simply move cases from a crowded venue to one that is at or near capacity. And, indeed, we acknowledge uncertainty over whether any other reform proposals will move cases to a venue with capacity or to one with none. While it may be possible to predict where some changes will send cases,[265] it is difficult to say whether that influx of cases will give rise to crowding-related effects that ascend the s-curve that we suspect characterizes crowding externalities. Stated simply, we do not know—and do not here purport to measure—how many cases are too many. Our view is that the answer likely varies across venues and depends on a wide range of factors. But, having said this, it also seems most probable that venue and transfer reforms that distribute cases more evenly are likely to reduce crowding effects to at least some degree.[266]

Third, we not only expect that these reforms will decrease crowding effects overall, we also expect that they will ensure that crowding effects, to the extent that they persist, are more evenly applied. Until very recently, litigants (such as criminal defendants or civil rights plaintiffs) in Waco were subject to an arbitrary disadvantage by virtue of geography and Judge Albright’s peculiar appetite for patent cases. Such arbitrary disuniformity in the courts has long been cause for concern. Indeed, one primary motive for the Supreme Court’s grants of certiorari is to address disuniformity in the application of federal law.[267] And a more uniform distribution of cases seems likely to reduce both crowding effects and the extent to which those effects are arbitrarily applied.

In all, policy-makers should directly address the problems of forum crowding by taking on their causes—forum shopping and, especially, forum selling. Improving procedural regularity (including as to inter partes review at the Patent Office), and reforming venue and transfer rules offer some promising avenues for doing so.

B.     Internalizing Crowding-Related Externalities

One mode of addressing the problems of forum crowding is to eliminate its causes, namely, forum shopping and selling. This could be accomplished through venue and transfer reform or through greater agency substitution. Another mode of addressing forum crowding is by taxing it, so that the crowders (say, patent plaintiffs) internalize the effects of being crowded out. We consider such a proposal here.

In their examination of the externality-regulating effects of procedure rules, Ronan Avraham and William Hubbard “imagine[d] . . . a cap-and-trade system” for judicial resources, explaining that “parties who ‘hog’ court time [c]ould be forced to pay for it while parties who use less court time [c]ould be monetarily rewarded.”[268] And they elaborated on this idea in the particular context of multidistrict litigation (MDL), suggesting that courts replace a Lone Pine examination—a “figurative toll” to participate in an MDL—with a “literal toll.” In so doing, they expect that cases will sort themselves into strong (those willing to pay) and weak (those unwilling to do so), all while improving the allocation of scarce judicial resources.[269]

In our specific context of patent litigation, we are less sure that a literal toll offers an ideal response to the problems of forum crowding. For one, our examination ranges beyond a single class of civil cases to encompass a court’s entire docket, including criminal and civil rights cases. Imposing a toll on any plaintiff or petitioner to a crowded forum presents substantial questions of fairness, administrability, and arbitrariness. On fairness, we need go no further than to acknowledge the substantial equity considerations raised by allowing a patent plaintiff to purchase priority over, say, a civil rights plaintiff or a criminal defendant.[270] It is, of course, possible to imagine a more nuanced regime, one which ensures criminal defendants access to the judicial resources required to comply with the Speedy Trial Act and associated constitutional guarantees, and which protects other policy priorities, such as a desire to vindicate constitutional rights in civil litigation. But such a complicated regime is a far cry from an easily administered fee system. Moreover, any such code would have to account for local variation, since docket conditions vary widely from district to district and division to division. But these varying local schemes awaken a specter of arbitrariness, as different plaintiffs bringing similar claims may be subject to different requirements, solely because venue rules require that those plaintiffs file their case in different districts, one more popular than the other.

While we differ on the specific implementation of this mode of externality regulation, we agree on the basic premise. Traditional modes of resolving concerns regarding docket crowding and court congestion—“proactive judicial oversight and case management”—can seem “paradoxical.”[271] “If the problem is that the excessive devotion of court time and attention to problematic cases leads to cost and delay, how can the solution be for the court to devote more time and attention to those cases?”[272]

Hence, we offer an alternate proposal that addresses the concern of using additional judicial resources to manage an already-taxed system. This approach may also cause crowding-causing plaintiffs to internalize the costs of filing in favored fora. As noted, many judges feel pressure to move cases along, and they do so at a faster clip during periods of crowding. This is in part because various institutional measures, including, most notably, the so-called six-month list, pressure judges into staying on top of their docket.[273] In short, the list is a shaming mechanism intended to address concerns about litigation delay by requiring that judges report all matters that have been pending for more than six months.[274]

Congress could amend the Civil Justice Reform Act (which instantiated the six-month list) to forgive delays in cases with venue flexibility by exempting such cases from the reporting obligation. Doing so would ease crowding and case management concerns by placing cases with venue flexibility on the back burner. Plaintiffs, anticipating this, would be put to a choice: they could wait at the back of a long line in a crowded but seemingly favorable forum, or they could choose another, less crowded venue. Stated in more general terms, courts should prioritize cases with little or no venue flexibility over those with venue choice. Doing so would cause plaintiffs with venue flexibility to internalize the costs of forum crowding in the form of deprioritization.

We concede that this proposal is not without risks. First, it risks generally deprioritizing certain cases, including patent cases, across all district courts for no reason other than their governing venue rules. Second, given our view that preferences are a leading cause of crowding effects, it is possible that this solution—trained more squarely on capacity constraints—will prove less effective than other options. Judges motivated to work on, for example, patent cases will simply resolve all their other work even more quickly to get to the patent cases waiting at the back of the line. Therefore, we emphasize the reforms outlined above—improved procedural regularity, stays pending inter partes review, and tighter venue and transfer rules. Indeed, we especially emphasize those reforms that begin with Congress and the Executive Branch.[275] If judicial preferences matter, as our findings suggest, then it might be impractical to expect judges who prefer patent cases to voluntarily transfer those cases away.

Conclusion

Although forum shopping has long been criticized as an “evil” tactic,[276] it remains a mainstay of patent litigation. For nearly as long as the term has been a part of legal vernacular, it has been an integral aspect of a patent plaintiff’s arsenal.[277] In recent years, patent plaintiffs have sought out certain federal venues in Texas—first, the Marshall Division of its Eastern District, and now, the Waco Division of its Western District—for their favorable procedures.

Critics of forum shopping tend to focus on the possibilities for gamesmanship by plaintiffs, suggesting that the practice undermines fairness in litigation. Others note that forum shopping reflects poorly on the Judiciary, as forum shopping implies that the same legal claim will earn different treatment, depending on whether it is filed in Texas or California. But these sharp swings in patent filings—from the Eastern District of Texas, to the District of Delaware, and then to the Western District of Texas—offer an opportunity to examine another aspect of the practice: when forum shopping leads to court congestion—when there is forum crowding—what happens to the rest of the docket?

Our results suggest that when patent plaintiffs have crowded a forum, decision quality in other cases has tended to suffer. Decisions rendered on crowded dockets tend to be reversed more frequently and reasoned more thinly than those rendered on non-crowded dockets. And a closer look at our results suggests that judicial preferences are a primary driver of these effects. Every judge in our study that has evinced a preference for patent cases, either through forum selling conduct or otherwise, is reversed more frequently in the rest of their docket during periods of crowding. Meanwhile, judges with no such preference are affirmed at the same (or better) rates during analogous periods of crowding.

Courts, Congress, and the Executive Branch should thus do more to limit forum shopping and forum crowding, especially in patent cases. For one, district courts should stop exploiting local procedural flexibility to solicit crowds of cases that consume vast judicial resources. Instead, these courts should be more willing to defer to the Patent Office, to transfer cases out of crowded dockets, and to deprioritize cases with venue flexibility (thus forcing plaintiffs to internalize some of the consequences of crowding). The Patent Office should likewise conduct patent reviews that may eliminate or simplify litigation in crowded dockets. Congress can help by enacting these reforms into law. Even more importantly, the President and Congress should affirm the Judiciary’s commitment to “ensure that all district judges remain generalists”[278] by nominating and confirming jurists who can—and will endeavor to—treat all cases, patent or not, alike.

Appendix

Appendix Figure 1: Share of Patent Cases by Judge (D. Del, E.D. Tex., W.D. Tex.), 2011–2021

Appendix Figure 2: Reversal Rates, Eastern District of Texas (Criminal Cases, n=1)

Note: Neither Chief Judge Gilstrap nor Judge Schroeder had any criminal cases appealed during our crowded study periods.

Appendix Figure 3: Reversal Rates, District of Delaware (Criminal Cases, n=16)

Appendix Figure 4: Reversal Rates, Judge Gilstrap (All Cases, n=10)

Appendix Figure 5: Reversal Rates, Judge Schroeder (All Cases, n=8)

Appendix Figure 6: Time to Termination, Judge Schroeder (Civil Cases, n=692)

Appendix Figure 7: Time to Termination, Chief Judge Gilstrap (Criminal Cases, n=76)

Appendix Figure 8: Time to Termination, Judge Schroeder (Criminal Cases, n=143)

Appendix Figure 9: Time to Termination, Western District of Texas (Criminal Cases, n=699)

As noted supra notes 148 and 151 and accompanying text, time-to-termination data for the criminal cases before Judges Counts and Pulliam is not readily available.

DOI: https://doi.org/10.15779/Z38J38KJ8G

                    Copyright © 2024 Tejas N. Narechania, Tian Kisch, and Delia Scoville.

             *     Professor of Law, University of California, Berkeley, School of Law.

**     J.D., University of California, Berkeley, School of Law (2022).

***     J.D., University of California, Berkeley, School of Law (2022).

                    For valuable comments and suggestions, we thank Abhay Aneja, Andrew Baker, Andrew Crespo, Pam Bookman, Andrew Bradt, Colleen Chien, Katherine Mims Crocker, Courtney Cox, Nikolas Datzov, Tabrez Ebrahim, Judge Jeremy Fogel (ret.), Jonah Gelbach, Paul Gugliuzza, David Hausman, Nick Hoppe, Bert Huang, Sonia Katyal, Aniket Kesari, Anya Ku, Christa Laser, Mark Lemley, Marin Levy, Brian Love, Michael Mattioli, Peter Menell, Khushali Narechania, C.E. Petit, Greg Reilly, Rachel Sachs, Amy Semet, Sepehr Shahshahani, Erik Stallman, Neel Sukhatme, Steve Vladeck, Rebecca Wexler, Ryan Whalen, Elliott Wyse, John Yoo, and Emily Zhang, as well as audiences at Duke Law School, the Indiana University (Bloomington) Maurer School of Law, Stanford Law School, and the University of California, Berkeley, School of Law. We thank Jennifer Sun for excellent research assistance and the outstanding law librarians at the University of California, Berkeley, School of Law for expert help. Our thanks also to Jung Kim and the California Law Review for their careful edits and helpful suggestions.

           [1].     Helene Curtis Indus. v. Sales Affiliates, 105 F. Supp. 886, 902 (S.D.N.Y. 1952).

                    Searches on Westlaw and LexisNexis suggest that the first mention of the phrase was in Covey Gas & Oil v. Checketts, 187 F.2d 561, 563 (9th Cir. 1951) (quoting Harold W. Horowitz, Erie R.R. Co. v. Tompkins—A Test to Determine Those Rules of State Law to Which Its Doctrine Applies, 23 S. Cal. L. Rev. 204, 215 (1950)) (describing a “clear case of what is aptly called ‘forum shopping’”), making Helene Curtis the second judicial decision (but first district court decision, per those databases) to use the phrase. The phrase has a slightly longer pedigree in the legal literature, dating to at least 1946. See Note, How a Federal Court Determines State Law, 59 Harv. L. Rev. 1299, 1301 (1946) (regarding the Erie doctrine). But see Forum-Shopping, Black’s Law Dictionary (11th ed. 2019) (dating the phrase to 1954).

           [2].     For readers interested in the details, the patent at issue is No. 2,577,710 (regarding a cold hair perm solution), which Sales Affiliates immediately assigned to Procter & Gamble. Helene Curtis Industries filed its declaratory judgment suit “within a matter of hours after the patent issued.” Helene Curtis, 105 F. Supp. at 891. Sales Affiliates later filed several infringement lawsuits, including the parallel suits in Texas, against a variety of customers and distributors of Helene Curtis. Id.

           [3].     Helene Curtis, 105 F. Supp. at 902–03, aff’d, 199 F.2d 732 (2d Cir. 1952) (noting that, in general, “a simple case pending in diverse courts may be allowed to go forward simultaneously in each tribunal until one reaches final judgment, and prior judicial control or direction is unnecessary if not undesirable,” but that the particular circumstances of this case warranted an unusual intervention).

           [4].     Letter from Sens. Thom Tillis & Patrick Leahy to Chief Justice John Roberts 1 (Nov. 2, 2021) [hereinafter Roberts Letter]. For more on the Chief Justice’s response to the letter, see Sup. Ct. of the U.S., 2021 Year-End Report on the Federal Judiciary 5 (2021).

           [5].     See Order Governing Proceedings—Patent Case at 1–4 (W.D. Tex. Feb. 26, 2020).

           [6].     Compare, e.g., Amended Order Assigning the Business of the Court, at 3 (W.D. Tex. Nov. 15, 2022) (assigning to Judge Albright “[a]ll cases and proceedings in the Waco Division”), with Amended Order Assigning the Business of the Court, at 4 (W.D. Tex. Dec. 16, 2022) (citing Amended Order Assigning the Business of the Court (W.D. Tex. July 25, 2022)) (explaining that patent cases filed in the Waco Division will be distributed among eleven judges in the Western District of Texas).

           [7].     Ryan Davis, Albright Transfer Drama Will Keep Eyes on Texas in 2022, Law360 (Dec. 17, 2021), https://www.law360.com/articles/1448846/albrighttransfer-drama-will-keep-eyes-on-texas-in-2022 [https://perma.cc/UX5H-YJ7Q] (noting that, as of December 15, 2021, 875 out of 3,843 total patent cases had been assigned to Judge Albright of the Western District of Texas). Eight hundred cases a year is, notably, about three new cases per business day.

           [8].     Ryan Davis, After Rules Shake-Up, Albright Remains the Top Patent Judge, Law360 (Feb. 15, 2023), https://www.law360.com/articles/1573848/after-rules-shake-up-albright-remains-the-top-patent-judge [https://perma.cc/FG7L-9PR8] (noting that Judge Albright has retained responsibility for over 20% of the nation’s patent caseload); Ryan Davis, New WDTX Top Judge Keeps Random Patent Suit Distribution, Law360 (Dec. 21, 2022), https://www.law360.com/articles/1560535?e_id=932e3440-ed91-4123-9d0e-fdfc5ed81f75 [https://perma.cc/D5S3-3YLL] (noting that Judge Albright “get[s] more patent cases than any other U.S. judge”).

           [9].     Daniel Klerman & Greg Reilly, Forum Selling, 89 S. Cal. L. Rev. 241, 262 (2016) (defining forum selling); see also J. Jonas Anderson & Paul R. Gugliuzza, Federal Judge Seeks Patent Cases, 71 Duke L.J. 419, 421–22 (2021) (regarding Judge Albright’s forum selling efforts).

         [10].     See J. Jonas Anderson, Paul R. Gugliuzza & Jason A. Rantanen, Extraordinary Writ or Ordinary Remedy? Mandamus at the Federal Circuit, 100 Wash. U. L. Rev. 327, 374–83 (2022) (examining the Federal Circuit’s recent mandamus orders, with particular emphasis on those directed to Judge Albright).

                    Indeed, Judge Albright’s practices have sparked interest from policymakers across all three branches of the government. Senator Thom Tillis, for example, sent letters to Chief Justice Roberts seeking “legislative recommendations” to address forum shopping concerns and to officials in the U.S. Patent and Trademark Office, seeking them to “undertake a study and review of [forum shopping]” and consider whether procedural changes induced by case law “should be modified to account for unrealistic trial scheduling.” See Roberts Letter, supra note 4, at 2; Letter from Sen. Thom Tillis to Acting Dir. Andrew Hirshfeld, Comm’r for Patent, U.S. Patent & Trademark Off. 3 (Nov. 2, 2021) [hereinafter Hirshfeld Letter]. The Roberts Letter elicited a flurry of responses. For one, the Director of the Administrative Office wrote that random case assignment “operates to safeguard the Judiciary’s autonomy while deterring judge-shopping and the assignment of cases based on the perceived merits or abilities of a particular judge.” Letter from Judge Roslynn R. Mauskopf to Sens. Thom Tillis & Patrick J. Leahy 1 (Dec. 15, 2021) [hereinafter Mauskopf Letter]. The response also noted that the Administrative Office recently counseled against the extension of the Patent Pilot Program, a program that had allocated patent cases to designated judges in certain district courts, in part “to help ensure that all district judges remain generalists.” Id. Shortly thereafter, Chief Justice Roberts’s 2021 Year-End Report promised to investigate the “case assignment procedures [that], in effect, enable [a] plaintiff to select a particular judge.” Sup. Ct. of the U.S., supra note 4, at 5. In July 2022, Chief Judge Garcia of the Western District of Texas singled out the Waco Division for new rules to promote random case assignment in a responsive, if unsuccessful, effort to address Judge Albright’s attempts to lure patent cases to his courtroom. See Order Assigning the Business of the Court as it Relates to Patent Cases (W.D. Tex. July 25, 2022) (order made “upon consideration of the volume of new patent cases assigned to the Waco Division, in an effort to equitably distribute those cases,” randomly assigning “all civil cases involving patents” to eleven other judges in the district); see also infra Part I.B.1 (describing this reform alongside data suggesting that it has failed to address forum shopping concerns in the Western District of Texas).

         [11].     See N. Secs. Co. v. United States, 193 U.S. 197, 400 (1904) (Holmes, J., dissenting).

         [12].     See supra note 9.

         [13].     See Ronen Avraham & William H.J. Hubbard, Civil Procedure as the Regulation of Externalities: Toward a New Theory of Civil Litigation, 89 U. Chi. L. Rev. 1, 30, 31, 60 (2022) (“Parties have an incentive to file motions when they expect to gain from a favorable ruling, notwithstanding any negative effect this has on the court’s ability to give justice to other parties in other cases.”).

         [14].     Infra Part I (describing such literature, both as to forum shopping in general and as to forum shopping in patent cases specifically).

         [15].     Compare, e.g., Erie R.R. Co. v. Tompkins, 304 U.S. 64, 74–78 (1938) (regarding federal-state forum shopping), and Lynn M. LoPucki & William C. Whitford, Venue Choice and Forum Shopping in the Bankruptcy Reorganization of Large, Publicly Held Companies, 1991 Wis. L. Rev. 11, 13 (regarding federal forum shopping in the bankruptcy context in particular), with Pamela K. Bookman, The Unsung Virtues of Global Forum Shopping, 92 Notre Dame L. Rev. 579, 583, 591 (2016) (highlighting “three unappreciated virtues of global forum shopping: forum shopping’s importance in protecting access to justice, promoting regulatory enforcement, and propelling substantive and procedural reform” (emphasis added)). See also Brief of Professor Stephen I. Vladeck as Amicus Curiae in Support of Petitioners at ii, United States v. Texas, 599 U.S. 670 (No. 22-58) (2023) (describing “how judge-shopping occurs, impairs the public interest, and . . . can damage the credibility of the federal judiciary”).

                    There is a closely related literature that examines these matters in the specific context of patent cases, given patent litigation’s unique, recent relationship with forum shopping. See, e.g., Colleen V. Chien & Michael Risch, Recalibrating Patent Venue, 77 Md. L. Rev. 47, 59 (2017) (noting that, “while permissive venue is not unique to patent law, . . . forum shopping in patents has attracted the most attention”); see also Jonas Anderson, Court Competition for Patent Cases, 163 U. Pa. L. Rev. 631, 637 (2015); Kimberly A. Moore, Forum Shopping in Patent Cases: Does Geographic Choice Affect Innovation?, 79 N.C. L. Rev. 889, 892 (2001); Klerman & Reilly, supra note 9, at 247–70. We canvass this literature more completely infra Part I.

         [16].     See, e.g., Christopher A. Whytock, The Evolving Forum Shopping System, 96 Cornell L. Rev. 481, 500 (2011) (citing Piper Aircraft v. Reyno, 454 U.S. 235 (1981)) (explaining that some of the Supreme Court’s forum non conveniens cases, arising in contexts of transnational litigation, expressly link forum shopping concerns with concerns for domestic docket congestion). Indeed, the docket congestion effects of forum shopping have been a target of forum non conveniens doctrine since its earliest days. See Paxton Blair, The Doctrine of Forum Non Conveniens in Anglo-American Law, 29 Colum. L. Rev. 1, 1 (1929).

         [17].     See Bert I. Huang, Lightened Scrutiny, 124 Harv. L. Rev. 1109, 1115 (2011).

         [18].     See Michael D. Frakes & Melissa F. Wasserman, Irrational Ignorance at the Patent Office, 72 Vand. L. Rev. 975, 987 (2019) [hereinafter Irrational Ignorance]; see also Michael D. Frakes & Melissa F. Wasserman, Is the Time Allocated to Review Patent Applications Inducing Examiners to Grant Invalid Patents?: Evidence from Microlevel Application Data, 99 Rev. Econ. & Stat. 550, 551 (2017) [hereinafter Time Allocated].

         [19].     See Civil Justice Reform Act of 1990, Pub. L. No. 101-650, tit. I, 104 Stat. 5089; Miguel F.P. de Figueiredo, Alexandra D. Lahav & Peter Siegelman, The Six-Month List and the Unintended Consequences of Judicial Accountability, 105 Cornell L. Rev. 363, 367 (2020); Jonathan Petkun, Nudges for Judges: The Effects of the “Six-Month” List on Federal Civil Justice 2 (August 2021) (unpublished manuscript). Several scholars have studied the six-month list’s substantial role both in ensuring that cases do not drag on for too long and in organizing judicial priorities. As an example, pending motions approaching the six-month mark tend to be decided more quickly—and perhaps more sloppily—in the weeks before the list’s release than at other times. See de Figueiredo, Lahav & Siegelman, supra, at 434–35. And this is so across categories of civil cases: patent, civil rights, antitrust, and so on.

         [20].     We say that this may seem obvious upon reflection because some readers might reasonably have a different intuition: namely, that as a forum gets crowded, every case takes longer and longer to resolve. Indeed, though the majority of judges we examined in our study took less time to terminate cases when their dockets were crowded, judges in one district we studied (namely, the Eastern District of Texas) took slightly more time. See infra Part II. The Eastern District of Texas aside, the institutional pressures against letting cases drag on for too long, such as the so-called six-month list, seem generally effective (informal though they may be) at guarding against such an outcome. See, e.g., de Figueiredo, Lahav & Siegelman, supra note 19, at 369–70 & fig.1; Petkun, supra note 19, at 2 (finding that “social pressure can be a key driver of workplace behavior”). We offer an explanation for the Eastern District of Texas infra Part II.B.2, and we revisit the institutional pressures occasioned by the six-month list infra Part III.B.

         [21].     See, e.g., David F. Levi & Mitu Gulati, Judging Measures, 77 UMKC L. Rev. 381, 410–12 (2008) (describing the utility of such measures); Huang, supra note 17, at 1130–33 (similarly turning to reversal rates to study decision quality); Hon. Harry T. Edwards, The Role of a Judge in Modern Society: Some Reflections on Current Practice in Federal Appellate Adjudication, 32 Clev. St. L. Rev. 385, 403 (1983) (similarly turning to reversal rates to study decision quality).

         [22].     See 28 U.S.C. §§ 41, 1295 (explaining that the Fifth Circuit has jurisdiction over appeals from Texas district courts, except in patent cases (which flow, instead, to the Federal Circuit)).

         [23].     Indeed, the unique structure of patent appeals suggests that patent cases are one of the few contexts in which we can reliably study workload effects on district court decision-making by reference to appellate outcomes.

         [24].     See infra Part II.B.1. Specifically, we find that decisions from crowded dockets are appealed at the same (or even higher) rates, and that, among appealed decisions, reversal rates are significantly higher in crowded dockets.

         [25].     See, e.g., Forum-Shopping, Black’s Law Dictionary (11th ed. 2019); see also Mary Garvey Algero, In Defense of Forum Shopping: A Realistic Look at Selecting a Venue, 78 Neb. L. Rev. 79, 80 (1999); Kimberly Jade Norwood, Shopping for a Venue: The Need for More Limits on Choice, 50 U. Mia. L. Rev. 267, 268 (1996).

         [26].     41 U.S. 1 (16 Pet.) (1842).

         [27].     Horowitz, supra note 1, at 208.

         [28].     Id.

         [29].     A series of subsequent federal cases contributed to the elimination of the incentives for “vertical” forum shopping by limiting the availability of federal law in diversity actions; simultaneously, these cases may have given rise to additional incentives for “horizontal” forum shopping by expanding the categories of state law that counted as “substantive” law. See, e.g., Cities Serv. Oil Co. v. Dunlap, 308 U.S. 208 (1939); Klaxon Co. v. Stentor Elec. Mfg. Co., 313 U.S. 487 (1941); Palmer v. Hoffman, 318 U.S. 109 (1943); Ferens v. John Deere Co., 494 U.S. 516 (1990).

         [30].     Algero, supra note 25, at 85 (referring to the distinction between vertical and horizontal forum shopping).

         [31].     Indeed, this is so even in the context of vertical forum shopping. For example, in Stewart Organization, Inc. v. Ricoh Corp., the Supreme Court reasoned that questions over the validity of forum selection clauses were to be resolved by reference to federal venue rules, such as 28 U.S.C. § 1404, rather than state law. 487 U.S. 22, 28–29 (1988). But Justice Scalia’s dissent contends that the Court’s decision causes plaintiffs to shop for a forum—state or federal—that is more likely to offer a favorable interpretation of a contract’s forum selection clause. Cf. Lapides v. Bd. of Regents of the Univ. Sys. of Ga., 535 U.S. 613, 621 (2002) (discussing Georgia’s potential “motive for removal” of case from state to federal court).

         [32].     See, e.g., Friedrich K. Juenger, Jurisdiction, Choice of Law and the Elusive Goal of Decisional Harmony, in Law And Reality: Essays On National And International Procedural Law 137 (1992).

         [33].     See Transcript of Oral Argument at 94, United States v. Texas, 599 U.S. 670 (2022) (No. 22-58) (statement of Justice Elena Kagan) (“In Texas, there are divisions within districts, you can pick your trial court judge . . . [and that] judge stops a federal immigration policy in its tracks, because you have a kind of sort of speculative argument that your budget is going to be affected.”); see also Brief of Professor Stephen I. Vladeck as Amicus Curiae, supra note 15, at ii (describing “how judge shopping occurs, impairs the public interest, and . . . can damage the credibility of the federal judiciary”); Steve Vladeck, Texas Judge’s COVID Mandate Ruling Exposes Federal “Judge-Shopping” Problem, MSNBC (Jan. 11, 2022),  https://www.msnbc.com/opinion/texas-judge-s-covid-mandate-ruling-exposes-federal-judge-shopping-n1287324 [https://perma.cc/4FNA-DAQL].

         [34].     In addition to the statement highlighted supra note 33 and accompanying text, see, e.g., Gregg Costa, An Old Solution to the Nationwide Injunction Problem, Harv. L. Rev. Blog (Jan 25, 2018), https://harvardlawreview.org/blog/2018/01/an-old-solution-to-the-nationwide-injunction-problem/ [https://perma.cc/42NC-FYNZ]; see also Alex Botoman, Divisional Judge-Shopping, 49 Colum. Hum. Rts. L. Rev. 297, 323–24 (2018).

         [35].     Bookman, supra note 15, at 591; see also Hannah L. Buxbaum, Foreign Governments as Plaintiffs in U.S. Courts and the Case Against “Judicial Imperialism,” 73 Wash. & Lee L. Rev. 653, 699–703 (2016); Kevin M. Clermont & John R.B. Palmer, Exorbitant Jurisdiction, 58 Me. L. Rev. 474, 477–82 (2006).

         [36].     454 U.S. 235, 252 (1981); see also Whytock, supra note 16, at 500 (explaining that Piper Aircraft expressly links concerns about forum shopping with concerns about domestic docket congestion); Marin K. Levy, Judging the Flood of Litigation, 80 U. Chi. L. Rev. 1007, 1028, 1036 (2013) (similarly noting that the Supreme Court deploys congestion concerns to avoid “taking too many cases that they believe belong in state court” or to avoid creating a “flood” of other sorts of federal litigation). But cf. Levy, supra, at 1066–73 (expressing skepticism about such congestion arguments, especially to the extent they are directed at addressing concerns about judicial workload).

         [37].     Bookman, supra note 15, at 583.

         [38].     Id.

         [39].     Compare, e.g., Horowitz, supra note 1, at 208, with Theodore Eisenberg & Lynn M. LoPucki, Shopping for Judges, 84 Cornell L. Rev. 967, 971 (1999) (noting that forum shopping is “far from universally condemned”).

         [40].     Eisenberg & LoPucki, supra note 39, at 971 n.12 (quoting George D. Brown, The Ideologies of Forum Shopping—Why Doesn’t a Conservative Court Protect Defendants?, 71 N.C. L. Rev. 649, 666–68 (1993)) (collecting sources that “describ[e] ‘anti-forum-shopping’ as ‘the classical position’”).

                    Indeed, Congress has legislated to limit forum shopping in various specific contexts. The Parental Kidnapping Prevention Act, the Uniformed Services Former Spouses’ Protection Act, and the Uniform Child Custody Jurisdiction and Enforcement Act, for example, set venue rules and jurisdictional requirements to “remove [a] parent’s legal incentive to . . . search [for] a friendly forum.” Patricia M. Hoff, U.S. Dep’t of Just., The Uniform Child-Custody Jurisdiction and Enforcement Act 2 (2001); see also Algero, supra note 25, at 99 (similar, for the PKPA). Similarly, Congress has designated the D.C. Circuit as the only court with jurisdiction to hear disputes arising under some complex or highly technical areas of the law, such as energy and telecommunications law. See 15 U.S.C. § 717r(d) (stating that the D.C. Circuit has “original and exclusive jurisdiction” over natural gas permitting disputes); see also 47 U.S.C. § 402(b) (enumerating the specific telecommunications-related orders, acts, and disputes that must be appealed at the D.C. Circuit).

         [41].     On fairness-related concerns, see, e.g., Anderson & Gugliuzza, supra note 9, at 476–77 (noting that Judge Albright’s conduct both “undermine[s] the judiciary’s integrity and public confidence in its impartiality” and implicates judicial “ethics rules”); Anderson, supra note 15, at 679 (noting the risk that courts competing for cases would “bend the adjudicative process to favor one group of litigants”); Reilly & Klerman, supra note 9, at 308 (noting that forum selling conduct will tend to “favor those with the power to choose where the case will be brought”).

         [42].     Cf. Avraham & Hubbard, supra note 13, at 17. In Avraham & Hubbard’s terms, prior examinations of forum shopping have considered the “strategic externalities” (e.g., gamesmanship concerns that patent plaintiffs choose fora to take advantage of favorable discovery rules, or to avoid patent validity challenges) and the “public-goods externalities” (e.g., concerns about eroding the courts’ reputation as a fair and neutral venue for resolving patent disputes) of the practice. In this Article, we shift focus to the “system externalities” of forum shopping and selling. Id. For the only two examples of brief scholarly discussions of such effects we have found, see Matthew A. Shapiro, Procedural Wrongdoing, 48 B.Y.U. L. Rev. 197, 240–41 (2022) (explaining that one purpose of Civil Rule 11 is to deter “protracting the proceedings so as to dissipate the court’s . . . resources,” because “with such a dilatory motive, the party seeks to divert the civil justice system from its core public functions, compromising the court’s ability to accurately dispose of her own case as well as others on its docket”) and Whytock, supra note 16, at 500 (citing Piper Aircraft v. Reyno, 454 U.S. 235 (1981)) (explaining that some of the Supreme Court’s forum non conveniens cases that arise from transnational litigation expressly link forum shopping concerns with concerns for domestic docket congestion).

         [43].     See Moore, supra note 15, at 903.

         [44].     Id.

         [45].     The data in Figure 1 is derived from Lex Machina. See infra Part II.A.2 (describing our data collection methods).

         [46].     See, e.g., Ofer Eldar & Neel U. Sukhatme, Will Delaware Be Different? An Empirical Study of TC Heartland and the Shift to Defendant Choice of Venue, 104 Cornell L. Rev. 101, 112 (2019); Anderson & Gugliuzza, supra note 9, at 443.

         [47].     See, e.g., Shawn P. Miller, Venue One Year After TC Heartland: An Early Empirical Assessment of the Major Changes in Patent Filing, 52 Akron L. Rev. 763, 782 tbl.1 (2019) (calculating that the District of Delaware was responsible for 24% of the nation’s patent cases in the year immediately following the Court’s May 2017 decision in TC Heartland); cf. Delaware Was 2019’s Top Patent District, but West Texas Has Been Rising in 2020, RPX Corp. (Mar. 11, 2020), https://www.rpxcorp.com/data-byte/delaware-was-2019s-top-patent-district-but-west-texas-has-been-rising-in-2020/ [https://perma.cc/3ERX-S5AQ] (finding that the District of Delaware was responsible for 27% of the nation’s patent cases in the fourth quarter of 2019).

         [48].     Davis, Albright Transfer Drama, supra note 7 (noting that, as of December 15, 2021, 875 out of 3,843 total national patent cases had been assigned to Judge Albright of the Western District of Texas).

         [49].     Moore, supra note 14, at 924.

         [50].     28 U.S.C. § 1400(b).

         [51].     VE Holding Corp. v. Johnson Gas Appliance Co., 917 F.2d 1574, 1575 (1990).

         [52].     Judicial Improvements and Access to Justice Act, Pub. L. No. 100-702, § 1013(a), 102 Stat. 4642, 4669 (1988).

         [53].     VE Holding Corp., 917 F. 2d at 1577.

                    While we focus on more recent developments, the story of VE Holding begins in 1948 with the enactment of the patent venue statute. That statute, recall, establishes that either “the judicial district where the defendant resides, or where the defendant has committed acts of infringement and has a regular and established place of business,” offers an appropriate venue for patent litigation. In 1957, the Supreme Court held that the statutory term “resides” referred to a corporate defendant’s place of incorporation (rejecting an argument that an intervening statutory definition of “reside” in a general venue statute—one extending venue to any district where that defendant could be found—also modified the patent-specific venue statute). See Fourco Glass Co. v. Transmirra Prods. Corp., 353 U.S. 222, 229 (1957). In 1988, Congress again modified the general venue statute, explaining that its definition of a corporate defendant’s residence applied to any provision “under this chapter.” 28 U.S.C. § 1391(d) (1988). And, as noted, the Federal Circuit, in VE Holding, construed this amendment as effectively abrogating the Supreme Court’s decision in Fourco Glass, applying to the patent-specific provision too (given that it fell under the same “chapter”). However, as we explain infra Part II.B.1, the Supreme Court would again reject that reasoning and conclude, in TC Heartland LLC v. Kraft Foods Group Brands LLC, 581 U.S. 258, 263 (2017), that the amendment to the general venue statute did not apply to the patent-specific provisions, thereby restricting venue choice in patent litigation.

         [54].     See, e.g., Megan M. La Belle, Patent Litigation, Personal Jurisdiction, and the Public Good, 18 Geo. Mason L. Rev. 43, 69–70 (2010); Moore, supra note 15, at 895 (“[A]ny company that operates in national commerce is likely subject to personal jurisdiction in many possible districts.”).

                    We readily concede that the personal jurisdiction and procedural due process analysis is more complicated than we have set out here. See, e.g., Kathy McCarroll, Note, Civil Procedure—Reassessing Personal Jurisdiction in Arkansas and the Eighth Circuit After Goodyear Dunlop Tires Operations v. Brown, 131 S. Ct. 2846 (2011) and J. Mcintyre Machinery, Ltd. v. Nicastro, 131 S. Ct. 2780 (2011), 36 U. Ark. Little Rock L. Rev. 229, 229 & nn.1–4 (collecting sources). And it is also true that, since the Federal Circuit’s decision in VE Holding, the Supreme Court has narrowed general personal jurisdiction in some important respects. See, e.g., Daimler AG v. Bauman, 571 U.S. 117 (2014); Goodyear Dunlop Tires Operations, S.A. v. Brown, 564 U.S. 915 (2011). But what we have said is more than sufficient for our present purposes, and we can safely sidestep the Supreme Court’s more recent pronouncements because, as we note infra Part II.B.1, the Court has since rejected the reasoning in the Federal Circuit’s VE Holding opinion.

         [55].     See supra note 29 and accompanying text (defining “horizontal” forum shopping).

         [56].     See Moore, supra note 15, at 908–11, 917.

         [57].     See, e.g., Klerman & Reilly, supra note 9, at 242; see also Anderson, supra note 15, at 677–78 (describing a similar phenomenon under a rubric of court competition (rather than “forum selling”)).

         [58].     See, e.g., Eldar & Sukhatme, supra note 46, at 112 fig.1; see also supra Figure 1.

         [59].     See, e.g., Anderson & Gugliuzza, supra note 10, at 438.

         [60].     Yan Leychkis, Of Fire Ants and Claim Construction: An Empirical Study of the Meteoric Rise of the Eastern District of Texas as a Preeminent Forum for Patent Litigation, 9 Yale J.L. & Tech. 193, 213–14 (2007); see also Julie Creswell, So Small a Town, So Many Patent Suits, N.Y. Times (Sept. 24, 2006), https://www.nytimes.com/2006/09/24/business/24ward.html [https://perma.cc/KQ97-E7GW] (noting that patent plaintiffs in Marshall, Texas, won at trial in nearly 80% of cases).

         [61].     Anderson & Gugliuzza, supra note 10, at 439.

         [62].     Id.; Klerman & Reilly, supra note 9, at 268–70; see also Moore, supra note 15, at 908–11, 917.

         [63].     See, e.g., Anderson & Gugliuzza, supra note 9, at 440; see also Klerman & Reilly, supra note 9, at 255–56 (describing a similar system prior to Judge Ward’s retirement).

         [64].     See, e.g., General Order No. 21-08, Assigning Civil and Criminal Actions (E.D. Tex. Apr. 30, 2021). Chief Judge Gilstrap recently amended these orders. Under the most recent version, Chief Judge Gilstrap no longer guarantees a patent plaintiff that he will preside over their case, though it remains overwhelmingly likely. See General Order No. 21-19, Assigning Civil and Criminal Actions (E.D. Tex. Apr. 30, 2021) (assigning 90% of the civil cases filed in the Marshall Division to Chief Judge Gilstrap).

         [65].     See, e.g., Anderson & Gugliuzza, supra note 10, at 439–40, 476–82 (noting that no venue rules govern the choice of division within a district and arguing in favor of such rules).

         [66].     Similar prior orders offered comparable guarantees in earlier years. See Klerman & Reilly, supra note 9, at 255–56 (2016) (describing a similar system prior to Judge Ward’s retirement).

         [67].     Klerman & Reilly, supra note 9, at 242–43.

         [68].     We emphasize non-contract cases because a valid forum selection clause negotiated among parties to a contract may undermine the view that plaintiffs alone control questions of venue in contract cases. See, e.g., Stewart Org., Inc. v. Ricoh Corp., 487 U.S. 22, 29 (1988).

         [69].     Klerman & Reilly, supra note 9, at 242; see also Anderson, supra note 15, at 677–78 (describing a similar phenomenon under a rubric of court competition (rather than “forum selling”)).

         [70].     See, e.g., Anderson, supra note 14, at 661–65 (2015) (identifying a range of individual and institutional incentives for forum selling conduct); Paul R. Gugliuzza & J. Jonas Anderson, Why Do Judges Compete for (Patent) Cases?, 65 Wm. & Mary L. Rev. (forthcoming 2024).

         [71].     Klerman & Reilly, supra note 9, at 270–77.

         [72].     Transcript of Oral Argument at 10, eBay v. MercExchange, 547 U.S. 388 (2006) (No. 05-130) (suggesting that the Court should “remedy th[e] problem” with Marshall, Texas).

         [73].     581 U.S. 258 (2017).

         [74].     Id. at 267–69.

         [75].     Id.

         [76].     See VE Holding Corp. v. Johnson Gas Appliance Co., 917 F.2d 1574 (1990), abrogated by TC Heartland, 581 U.S. at 267–69.

         [77].     28 U.S.C. § 1400(b).

         [78].     See Tejas N. Narechania, Certiorari, Universality, and a Patent Puzzle, 116 Mich. L. Rev. 1345, 1361 n.78 (2018) (“Consider the Court’s distaste for forum shopping. The Court’s decision to grant the petition in TC Heartland, which alleged ‘rampant forum shopping’ in patent cases, may be informed by this underlying interest . . . .” (citation omitted)).

         [79].     See supra Figure 1. The Eastern District of Texas’s share of patent cases continued to decrease after 2017, the year that TC Heartland was decided. Whether the Court’s decision offers the best interpretation of the various interacting provisions and precedents is a matter of some debate. See, e.g., Paul R. Gugliuzza & Megan M. La Belle, The Patently Unexceptional Venue Statute, 66 Am. U. L. Rev. 1027, 1046–53 (2017) (contending that VE Holding was correctly decided and remained doctrinally sound (at least until the Court’s decision in TC Heartland)). But we can safely set that debate aside for our present purposes.

         [80].     See, e.g., Stephen M. Bainbridge, Introduction, in Can Delaware Be Dethroned?: Evaluating Delaware’s Dominance of Corporate Law 3–6 (Stephen M. Bainbridge, Iman Anabtawi, Sung Hui Kim & James Park eds., 2018). We can safely set aside the particulars of why Delaware may be so attractive (including the debate about whether Delaware is popular because of its substantive law or because of a network effect in incorporation decisions). See, e.g., Sarath Sanga, Network Effects in Corporate Governance, 63 J.L. & Econ. 1, 1–3 (2020).

         [81].     Chien & Risch, supra note 15, at 91 (estimating that the District of Delaware would receive 19% of operating company cases and 25% of non-practicing entity cases).

         [82].     See supra Figure 1.

         [83].     28 U.S.C. § 1400(b).

         [84].     See, e.g., SITO Mobile R&D IP v. Hulu, 2021 WL 1166772 (W.D. Tex. 2021) (describing Hulu’s presence in San Antonio in the context of a decision to deny Hulu’s transfer motion), vacated sub nom In re Hulu, 2021 WL 3278194 (Fed. Cir. 2021); Uniloc 2017 LLC Non-Confidential Response to Apple Inc.’s Petition for Writ of Mandamus at 6–10, 15, 17, In re Apple, 979 F.3d 1332 (Fed. Cir. 2020) (No. 2020-135) (describing Apple’s presence in Austin, Texas).

         [85].     The United States Attorney’s Office: Western District of Texas, U.S. Dep’t of Just., https://www.justice.gov/usao-wdtx [https://perma.cc/RWT9-E3PJ] (“The United States Attorney’s Office for the Western District of Texas proudly serves more than 7 million residents, many of whom make their home in San Antonio, Austin, or El Paso.”).

         [86].     Anderson & Gugliuzza, supra note 10, at 456–58.

         [87].     Miriam Rozen, How Waco Became a Patent Litigation Hotspot, Fin. Times (June 15, 2022), https://www.ft.com/content/22cf012e-2f89-412c-bfdf-4bbdd07018be [https://perma.cc/266A-3F6P].

         [88].     See Cuozzo Speed Techs. v. Lee, 579 U.S. 261, 279–80 (2016) (quoting H.R. Rep. No. 112-98, pt. 1, at 39–40 (2011)) (“Inter partes review is an ‘efficient system for challenging patents that should not have issued.’”).

                    To be sure, inter partes review is limited in scope, meaning that some patent validity questions can only be addressed in the district courts. See, e.g., 35 U.S.C. § 311(b) (defining the scope of inter partes review). We set these concerns aside for now, though we return to them in Part III.A.1.

         [89].     Jonathan Stroud, Linda Thayer & Jeffrey C. Totten, Stay Awhile: The Evolving Law of District Court Stays in Light of Inter Partes Review, Post-Grant Review, and Covered Business Method Post-Grant Review, 11 Buff. Intell. Prop. L.J. 226, 238 fig.1 (2015).

         [90].     Hirshfeld Letter, supra note 10, at 1–2; see also Dennis Crouch, Fintiv: Dir. Vidal Calls for Fewer Discretionary Denials of Inter Partes Review Petitions, Patently-O Blog (June 22, 2022), https://patentlyo.com/patent/2022/06/discretionary-denials-petitions.html [https://perma.cc/TYC9-E6C6] (noting that Judge Albright “has been strategically setting early trial dates in order to trigger discretionary denial”); Pauline M. Pelletier, Deborah Sterling & Anna G. Phillips, How West Texas Patent Trial Speed Affects PTAB Denials, Law360 (Feb. 16, 2021), https://www.law360.com/articles/1355139/how-west-texas-patent-trial-speed-affects-ptab-denials [https://perma.cc/46HJ-P5B2] (discussing how Judge Albright’s schedule puts pressure on defendants to pursue inter partes review quickly); see also infra Part III.A.1.

         [91].     In re Apple, No. 2022-128, 2022 WL 1196768, at *1–2 (Fed. Cir. Apr. 22, 2022); see also Anderson, Gugliuzza & Ranteaen, supra note 10, at 331 (describing the Federal Circuit’s growing willingness to reverse, via mandamus, Judge Albright’s decisions to deny transfer motions).

         [92].     See, e.g., In re Apple, 2022 WL 1196768 (Fed. Cir. Apr. 22, 2022); In re Dish Networks, No. 2021-182, 2021 WL 4911981 (Oct. 21, 2021); In re Apple, 979 F.3d 1332 (Fed. Cir. 2020); see also Anderson, Gugliuzza, & Rantanen, supra note 10, at 331.

         [93].     See Anderson, Gugliuzza, & Rantanen, supra note 10, at 453.

         [94].     Id.

         [95].     See Sup. Ct. of the U.S., supra note 4, at 5 (promising to investigate the “case assignment procedures [that], in effect, enable [a] plaintiff to select a particular judge”).

         [96].     Specifically, Chief Judge Garcia attempted to assign cases more randomly, assigning new patent filings in the Waco Division to any one of twelve different judges in the Western District.

                    But this tactic proved unsuccessful. If case assignment were random, then each judge would have assumed responsibility for only 8.5% of these new patent filings. But Judge Albright had managed to assume responsibility for nearly half of the new patent case filings. More precisely, Judge Albright was assigned 112 out of the 233 (i.e., 48% of the) new patent cases filed in the Western District of Texas between July 25, 2022 (the date that Chief Judge Garcia’s new assignment order issued) and October 31, 2022. Western District of Texas CM/ECF Civil Case Reports, PACER, https://pacer.uscourts.gov [https://perma.cc/X4V3-V57G]. One likely explanation for the limited effect of Chief Judge Garcia’s intervention may lie in the Western District’s case assignment rules for related cases. Under those rules, cases related to previously filed cases are assigned to the judge who presided over that prior, related case. See May 2003 Amended Plan for Random and Direct Assignment of Cases in Multi-Judge Divisions at 1 (May 28, 2003); Order at 1, Bassfield IP LLC v. Paradies Lagardere @ AUS, LLC, No. 6:22-cv-890-KC (W.D. Tex. Aug. 30, 2022), ECF No. 6 (explaining that, “[t]o facilitate efficient docket management, all related cases are ordinarily allocated to the Judge that was randomly assigned the earliest-numbered related case,” and citing to that May 2003 order). Of the 112 cases assigned to Judge Albright since the July 25 Order, 103 fall within the ambit of these related-case rules. See PACER, supra. It is possible that this effect will wane in the future, as the number of new cases related to Judge Albright’s prior docket diminishes over time. But it is not guaranteed, particularly in view of the relatively capacious understanding of “related case” applied by judges in the district. See Order Granting Motion to Relate and Reassign Case at 3–4, Sonrai v. Micron, No. 6:22-CV-00855 (W.D. Tex. Mar. 16, 2023) (reassigning to Judge Albright a case that was initially randomly assigned to Judge Biery, on the grounds that the two cases shared one patent in common, notwithstanding other dissimilarities, including differences in the allegedly infringing products and claims regarding other patents not before Judge Albright).

         [97].     Davis, New WDTX Top Judge, supra note 8; Jasmin Jackson, Even Without Promise of Albright, NPEs Still Prefer WDTX, Law360 (Oct. 12, 2022), https://www.law360.com/articles/1539247/even-without-promise-of-albright-npes-still-prefer-wdtx [https://perma.cc/2BLS-GGWJ].

                    More specifically, in November 2022, Chief Judge Garcia issued a new superseding case assignment order that unwound his July order, giving Judge Albright responsibility for any case—patent or not—filed in the Western District’s Waco Division. Amended Order Assigning the Business of the Court at 3 (W.D. Tex. Nov. 15, 2022) (assigning to Judge Albright “[a]ll cases and proceedings in the Waco Division”). Once Chief Judge Moses succeeded (former) Chief Judge Garcia, she restored, in substantial part, the July order. See Amended Order Assigning the Business of the Court at 4 (W.D. Tex. Dec. 16, 2022) (explaining that patent cases filed in the Waco Division will be distributed among eleven judges in the Western District of Texas, citing Amended Order Assigning the Business of the Court (W.D. Tex. July 25, 2022)). But, as noted, such reforms have proved largely unsuccessful so far, see supra note 95, and their future success remains uncertain.

         [98].     See, e.g., supra notes 26–28 and accompanying text; see also Moore, supra note 15, at 924.

         [99].     See supra Part 1.A.

      [100].     Klerman & Reilly, supra note 10, at 247; see also Amy Semet, An Empirical Examination of Venue in Patent Law (unpublished manuscript); LoPucki & Whitford, supra note 15, at 13 (finding that “[t]he prevalence of forum shopping may be influencing the content of bankruptcy law”).

      [101].     See Moore, supra note 15, at 924–31.

      [102].     See Xuan-Thao Nguyen, Justice Scalia’s “Renegade Jurisdiction”: Lessons for Patent Law Reform, 83 Tul. L. Rev. 111, 136, 138, 141 (2008).

      [103].     Moore, supra note 15, at 924.

      [104].     Id. at 924–31.

      [105].     On fairness-related concerns, see, e.g., Anderson & Gugliuzza, supra note 9, at 477 (noting that Judge Albright’s conduct both “undermine[s] the judiciary’s integrity and public confidence in its impartiality” and implicates judicial “ethics rules”); Anderson, supra note 15, at 679 (noting the risk that courts competing for cases will “bend the adjudicative process to favor one group of litigants”); Reilly & Klerman, supra note 9, at 308 (noting that forum selling conduct will tend to “favor those with the power to choose where the case will be brought”). On cost-related concerns, see, e.g., Anderson, Gugliuzza, & Rantenen, supra note 10, at 384 (noting that “reduc[ing] the incentives for and availability of court competition” could save “litigants time and money” associated with transfer proceedings (and the mandamus appeals that follow)).

      [106].     See, e.g., Anderson & Gugliuzza, supra note 9, at 439; Klerman & Reilly, supra note 9, at 308; see also Eisenberg, supra note 39, at 971 (“Observers seem to agree that judge shopping ‘breeds disrespect for and threatens the integrity of our judicial system’ and undermines the aphorism that ‘ours is a government of laws, not men.’” (citation omitted)).

      [107].     See supra notes 32–35 and accompanying text.

      [108].     Jeanne C. Fromer, Patentography, 85 N.Y.U. L. Rev. 1444, 1466 n.139 (2010).

      [109].     See id. at 1466. Jeanne Fromer further describes how the cost savings from more efficient litigation are symmetric, accruing to both plaintiffs and defendants alike (though, of course, infringement defendants might be willing to spend more on litigation if it allowed them to litigate in a forum they perceive to be fairer and less likely to give rise to billion-dollar verdicts).

      [110].     Nguyen, supra note 102, at 136, 138, 141 (concluding that patent forum shopping leads to adjudication by “reasonable and fair” judges who are “knowledgeable, welcoming, and organized,” particularly because they have adopted a series of efficiency-enhancing and cost-saving local rules that reflect a “customer-oriented approach”); see also Beth Thornburg, Michael Smith, Robert Conklin & Andrei Iancu, The Jury in the EDTX: Unsophisticated American Peers or Idealists of Property Rights in Patents, 14 SMU Sci. & Tech. L. Rev. 203, 203–04 (2011) (transcript of Xuan-Thao Nguyen’s opening remarks arriving at a similar conclusion); Andrei Iancu & Jay Chung, Real Reasons the Eastern District of Texas Draws Patent Cases—Beyond Lore and Anecdote, 14 SMU Sci. & Tech. L. Rev. 299, 320 (2011) (contending that “the case assignment scheme” in the Eastern District of Texas “is designed to get patent cases to those judges” “that have an affinity for, and deep experience with, patent cases,” thereby providing “more efficient” and “more accurate” “resolution of patent cases”).

      [111].     Nguyen, supra note 110, at 141.

      [112].     See U.S. Const. art. III, § 2, cl. 3 (requiring that criminal trials take place in “the State where the [alleged] Crimes shall have been committed”); id. amend. VI (requiring that criminal trials take place with jurors from “the state and district wherein the crime shall have been committed”); see also Fed. R. Crim. P. 18 (explaining that the government must generally prosecute defendants in “the state and district wherein the crime shall have been committed”).

                    Earlier versions of Rule 18 were even more strict, requiring that “[a]ll prosecutions for crimes or offenses . . . be had within the division of such districts where the same were committed.” Fed. R. Crim. P. 18 advisory committee’s notes (1944) (citing 28 U.S.C. § 114); see also Salinger v. Loisel, 265 U.S. 224, 237 (1924) (interpreting Rule 18 to require that trial be had in the division where the offense was committed, where a district contains more than one division). The Rule was amended in 1966 to “eliminat[e] the requirement that the prosecution shall be in a division in which the offense was committed[, vesting] discretion in the court to fix the place of trial at any place within the district with due regard to the convenience of the defendant and his witnesses.” Fed. R. Crim. P. 18 advisory committee’s notes (1966). In 1979, the set of considerations bearing on the venue question was expanded to include “the prompt administration of justice.” Those considerations were expanded again in 2008 to include the convenience of the victims. Fed. R. Crim. P. 18. The reference to the “prompt administration of justice” is meant to primarily capture the requirements of the Speedy Trial Act. See Fed. R. Crim. P. 18 advisory committee’s notes (1979). But as we explain in more detail infra Part III.A.2, it has also been interpreted to encompass the state of the court’s docket generally, thereby enabling some courts to transfer cases to other divisions in view of docket congestion concerns.

                    Nevertheless, prevailing doctrine still requires “the government [to] prosecute an offense in a district where the offense was committed,” and there is little evidence to suggest that courts routinely transfer criminal cases out of crowded divisional dockets. Fed. R. Crim. P. 18. Indeed, our study found no outgoing intradistrict (interdivision) transfers of criminal cases for the crowded divisions we examined. See infra Part III.A.2. Hence, criminal defendants in, say, Waco generally have little opportunity to avoid a forum burdened by patent cases.

      [113].     See, e.g., Rumsfeld v. Padilla, 542 U.S. 426, 428, 443 (2004) (explaining “the general rule that for core habeas petitions challenging present physical confinement, jurisdiction lies in only one district: the district of confinement” and noting that this “general rule” “serves the important purpose of preventing forum shopping by habeas petitioners”); see also 28 U.S.C. § 1391(e) (regarding venue in actions “against an officer or employee of the United States or any agency thereof acting in his official capacity or under color of legal authority”).

      [114].     Edwards, supra note 21, at 403; see also Moore v. City of E. Cleveland, 431 U.S. 494, 523 (1977) (Burger, C.J., dissenting) (describing the “devastating impact overcrowded dockets have on the quality of justice received by all litigants”); United States v. Jacobs, 429 U.S. 909, 910 (1976) (Stewart, J., dissenting).

      [115].     Hon. Patricia M. Wald, Thoughts on Decisionmaking, 87 W. Va. L. Rev. 1, 10 (1984).

      [116].     See, e.g., Hans Zeisel, Harry Kalven, Jr., & Bernard Buchholz, Delay in the Court 12 (1959) (likening congestion and delay to the problems of a literal “logjam”); see also George L. Priest, Private Litigants and the Court Congestion Problem, 69 Boston Univ. L. Rev. 527, 527 (1989) (“Much of our understanding of litigation delay has been influenced by the early important work of Zeisel, Kalven, and Buchholz. The Zeisel team approach derives from their view of the litigation delay problem in terms of the metaphor, drawn from the lumber industry, of a logjam.” (footnote omitted)).

      [117].     Victor Marrero, The Cost of Rules, the Rule of Costs, 37 Cardozo L. Rev. 1599, 1624–25 (2016) (noting that litigants who “command the best legal talent and inordinately inflate the price of lawsuits” tend to “unduly conges[t] the court dockets with costly complex disputes,” thus causing other cases to “become economically borderline or too financially unappealing for many more attorneys to consider accepting. In consequence, people who need counsel no less—and often more so—than those able to pay the going rate for lawyers, are priced out of the legal services market”); see also Marc Galanter, Why the Haves Have Come Out Ahead, 9 Law & Soc’y Rev. 95, 121 (1974) (describing some distributional effects of “overload”).

      [118].     Huang, supra note 17, at 1115.

      [119].     Frakes & Wasserman, Irrational Ignorance, supra note 18, at 987; see also Frakes & Wasserman, Time Allocated, supra note 18, at 551.

      [120].     See supra Figure 1; see also Gloria Huang, TC Heartland, Legal Trends, One Year Later, Lex Machina Blog (May 23, 2018), https://lexmachina.com/blog/tc-heartland-legal-trends-one-year-later/ [https://perma.cc/AM8H-FAVY] (noting that the District of Delaware became “the leading court for patent litigation” after the TC Heartland decision).

      [121].     Cf. Bert I. Huang & Tejas N. Narechania, Judicial Priorities, 163 U. Pa. L. Rev. 1719, 1737 (2015) (measuring effects three years before and after a policy shock).

      [122].     See supra Figure 1.

      [123].     See id.

      [124].     We chose December 31, 2016, as our cutoff in order to account for any strategic behavior during the pendency of TC Heartland. The Supreme Court granted the petition for a writ of certiorari in TC Heartland in December 2016 and it decided the case later that Term in May 2017. We chose January 1, 2018, as the start date for any post-TC Heartland analysis for similar reasons and to streamline data collection.

      [125].     See supra Figure 1.

      [126].     See supra note 124.

      [127].     See Levi & Gulati, supra note 20, at 410–11 (noting the need to account for such differences); see also Huang & Narechania, supra note 117, at 1737 (similarly measuring effects for a “stable sample” of judges who were on the court for the entire period of the study).

      [128].     See supra notes 67–69 and accompanying text.

      [129].     Judge Biographies, U.S. Ct. of Appeals for the Fed. Cir., https://cafc.uscourts.gov/home/the-court/judges/judge-biographies/ [https://perma.cc/AC54-YQWY].

      [130].     The only other active judges who sat on the federal bench in either the District of Delaware or the Eastern District of Texas during the entire duration of our study period were Judge Marcia Crone and Judge Amos Mazzant, both of whom heard far fewer patent cases than either Chief Judge Gilstrap (3,306 patent cases during our study period) or Judge Schroeder (1,482 patent cases) (deriving data from Lex Machina) (data on file with the authors). Judge Crone heard six patent cases during her entire tenure, and Judge Mazzant heard 177 during our study period (deriving data from Lex Machina) (data on file with authors).

      [131].     From January 1, 2019, to June 1, 2022, 2,327 patent cases were assigned to Judge Albright, three to Judge Counts, and one to Judge Pulliam. See Appendix Figure 1.

      [132].     See Press Release, The White House, Off. of the Press Sec’y, President Obama Nominates Six to Serve on the United States District Courts (Mar. 15, 2016), https://obamawhitehouse.archives.gov/the-press-office/2016/03/15/president-obama-nominates-six-serve-united-states-district-courts [https://perma.cc/MMF7-79UY].

      [133].     President Donald J. Trump Announces Judicial Nominees, a United States Attorney Nominee, and United States Marshal Nominee (Mar. 1, 2019), https://trumpwhitehouse.archives.gov/presidential-actions/president-donald-j-trump-announces-judicial-nominees-united-states-attorney-nominee-united-states-marshal-nominees/ [https://perma.cc/2LPJ-W68T].

      [134].     See Counts, Walter David III, Fed. Jud. Ctr., https://www.fjc.gov/node/4028221 [https://perma.cc/PN8S-HULJ] (stating Judge Counts received his commission on January 17, 2018); Albright, Alan D., Fed. Jud. Ctr., https://www.fjc.gov/node/5198506 [https://perma.cc/7KUV-PJUZ] (stating Judge Albright received his commission on September 10, 2018); Pulliam, Jason Kenneth, Fed. Jud. Ctr., https://www.fjc.gov/node/6840711 [https://perma.cc/9NQ7-LQLR] (stating Judge Pulliam received his commission on August 5, 2019).

      [135].     See, e.g., infra Figure 12; Order Regarding Court Docket Management for Waco Division (W.D. Tex. May 23, 2023).

      [136].     Consider, for example, the various ways a research database might define a case. A case might refer to a given opinion (of which there might be many in one proceeding); or it might refer to a defined set of parties within a proceeding; or it might refer to various consolidating proceedings. We discovered that our data sources used slightly different definitions (though not in any substantial way that gives us concern for our results). Lex Machina states that “[a] District Court ‘case’ in Lex Machina is akin to a ‘case’ in PACER. It represents the entire docket, which may contain hundreds or even thousands of docket entries and documents. A case is represented by a civil action number in a particular District Court. This notion of a case contrasts with traditional legal research software, which sometimes refers to an individual opinion or judgment as a ‘case.’ A Lex Machina ‘case’ may contain many opinions or judgments, as well as all the other pleadings, motions, and various documents filed throughout the life of a case.” District Court Overview, Lex Machina, https://law.lexmachina.com/help/documentation/district-court/overview [https://perma.cc/6TN5-EAHD]. By contrast, Westlaw provides no definition on its site, and when we contacted a representative, they responded that they “do not have a definition of ‘cases.’” The FJC’s Integrated Database (“IDB”), meanwhile, presents data at the criminal defendant level, rather than the case level. Fed. Jud. Ctr., The Integrated Database: A Research Guide 2, https://www.fjc.gov/sites/default/files/IDB-Research-Guide.pdf [https://perma.cc/LA2L-SUS4]. This is because criminal cases tend to involve multiple defendants. Id.

      [137].     See, e.g., Huang, supra note 17, at 1127–29 (employing a similar research design that separates the increase of caseloads from its effects).

      [138].     The reasons for this limited definition of civil rights cases are elaborated infra Part II.B.1 (describing Judge Albright’s standing order). Such cases incidentally also have relatively more strict venue rules. See supra note 113 and accompanying text.

                    We also used Lex Machina to identify these cases in particular. Besides setting filters consistent with our judicial selections and study time periods, we narrowed results using specific search terms (“§ 1983” or “Section 1983” or “Bivens” or “habeas” or “2255” or “2241” or “2254”) and by excluding particular “Case Types” (antitrust, bankruptcy, consumer protection, contracts, copyright, ERISA, insurance, patent, product liability, securities, tax, trade secret, and trademark). We then manually confirmed that the remaining cases did indeed include either a Section 1983, habeas, or Bivens claim by checking the text of the decision on Westlaw or the claims stated in the plaintiff’s complaint.

      [139].     Other studies of judicial efficiency and quality have used these three metrics as measures of judicial decision-making. See, e.g., Admin. Off. of the U.S. Cts., Final Report to Congress Pursuant to Section (1)(e) of the Pilot Program in Certain District Courts Act, Pub. L. No. 111-349 (2011), at 4 (2021) [hereinafter Patent Pilot Program Final Report] (relying on termination time, appeal rates, and reversal rates, among other metrics, to evaluate the success of the Patent Pilot Program).

      [140].     See, e.g., Hon. William G. Young & Jordan M. Singer, Bench Presence: Toward a More Complete Model of Federal District Court Productivity, 118 Penn. St. L. Rev. 55 (2013) (proposing a bench presence measure, defined as “a measure of the time that a federal district judge spends on the bench, presiding over the adjudication of issues in an open forum” thereby offering a “rough but meaningful proxy for procedural fairness”); see also Mirya R. Holman, Measuring Merit in Rhode Island’s Natural Experiment in Judicial Selection, 15 Roger Williams U. L. Rev. 705, 710–11 (2010) (describing the debate over using opinion length to assess decision quality and concluding that “[r]egardless of who is correct, both sides—and many scholars—believe that page length is an important measure of judicial effort”).

      [141].     See Order Regarding Court Docket Management for Waco Division, supra note 135.

      [142].     See infra Part II.B.1; cf. Tejas N. Narechania, Certiorari in Important Cases, 122 Colum. L. Rev. 923, 956 (2022) (“[I]t is worth emphasizing that the approach here does not rely solely on the results generated by this method, but also on a complementary analysis of the Court’s underlying opinions.”).

      [143].     Petkun, supra note 19, at 42. See, e.g., Levi & Gulati, supra note 21, at 410–11; Frank B. Cross & Stefanie Lindquist, Judging the Judges, 58 Duke L.J. 1383, 1403 (2009); Kimberly A. Moore, Are District Court Judges Equipped to Resolve Patent Cases?, 15 Harv. J.L. & Tech. 1, 3–4 (2001); Richard S. Higgins & Paul H. Rubin, Judicial Discretion, 9 J. Legal Stud. 129, 134 (1980).

      [144].     See Edwards, supra note 20, at 403; see also Moore v. City of E. Cleveland, 431 U.S. 494, 523 (1977) (Burger, C.J., dissenting) (describing the “devastating impact overcrowded dockets have on the quality of justice received by all litigants”); cf. Wasserman & Frakes, Irrational Ignorance, supra note 18, at 982–87 (noting similar time restraints for patent examiners).

      [145].     See Timing and Box Plots, Lex Machina, https://law.lexmachina.com/help/documentation/features/timing-events [https://perma.cc/4E64-D6KF].

      [146].     Specifically, we used Lex Machina’s filters to exclude all cases categorized as “Patent” and to include all other civil cases. (Lex Machina, recall, does not include criminal case data.) We limited our dataset to terminated cases whose filing date fell within a corresponding study period. That is, we counted cases that were filed during the relevant period of crowding (or not).

                    For all non-patent cases, we reported Lex Machina’s median time-to-termination. For civil rights and habeas cases, we hand-tabulated the time-to-termination metrics because narrowing the case category required some manual filtering. See supra note 138. We report the median time-to-termination.

      [147].     See Facts about FJC’s Integrated Database, Free L. Project, https://free.law/idb-facts [https://perma.cc/DL9N-KTB2] (explaining office codes from the FJC IDB); see also Fed. Jud. Ctr., Criminal Integrated Database (IDB): 1996 to Present, Codebook 21 (2016) (listing district codes from the FJC IDB).

      [148].     We used the most recent update of the FJC Criminal Integrated Database (as of our data collection efforts), which is current through March 31, 2022. See IDB Criminal 1996–present, Fed. Jud. Ctr., https://www.fjc.gov/research/idb/interactive/21/IDB-criminal-since-1996 [https://perma.cc/8F5F-WLHP]. Accordingly, the time-to-termination information for criminal dockets spans from January 1, 2019, to March 31, 2022.

      [149].     See Fed. Jud. Ctr., supra note 147, at 9, 12 (defining filing date as “when a case was first docketed in the district court” and disposition date as the “date upon which judicial proceedings before the court concluded”). In cases where a criminal defendant committed multiple offenses with multiple disposition dates, we used the first—and therefore earliest—disposition date to remain consistent with cases where the defendant committed only one offense. We report median time-to-termination values, consistent with Lex Machina.

      [150].     These omissions are not problematic for our study. As we explain infra, we rely primarily on time-to-termination data in civil cases, rather than criminal cases, because provisions such as the Speedy Trial Act affect the timelines for processing various criminal matters. Indeed, where we can measure time-to-termination in criminal cases, we find that they tend to max out at about 200 days, crowding or not. Some judges, however, move such cases along substantially more quickly. See infra Appendix Figure 9 (noting that Judge Albright typically resolves criminal matters in an astounding fifty-five days).

      [151].     See, e.g., Petkun, supra note 19, at 54 (studying appeal rates but finding no effect in this metric).

      [152].     See, e.g., id. at 42.

      [153].     See, e.g., Levi & Gulati, supra note 21, at 410–12 (considering the use of both appeal rates and reversal rates and explaining that “[r]eversals . . . should be a finer measure of a judge’s error rate,” while both noting qualms about both measures and concluding “that there is an enormous amount of information available to be gleaned from [both] such measures”).

      [154].     Both here and in the context of reversal rates, we exclude appeals that were dismissed for jurisdictional reasons, as well as those denied certificates of appealability. Where a certificate of appealability was granted, we count the case as appealed.

                    Moreover, to ensure fidelity to our study time periods, we relied on the date of the appealed decision, rather than the date of the appellate decision. After all, our hypothesis is that crowding might affect district court decision-making. To do this, we read each appellate court opinion to identify the underlying issue or disposition that was appealed, and then used Westlaw’s “History” function to locate the corresponding district court decision at issue, manually recording the disposition date of that decision. Where the underlying district court decision was not available from the “History” section on Westlaw, we employed several different strategies to find the correct date, including by looking at Westlaw’s docket tracker (under its “Filings” tab), which often reported the termination date of the lower court decision or otherwise included the appellant’s brief usually describes the district court disposition at issue. If we were unable to locate the district court disposition date in Westlaw, we turned to Lex Machina, or, for criminal cases, to Bloomberg Law’s docket searching capabilities. Where district court cases were consolidated, we used the date of disposition of the lead case after consolidation.

      [155].     We initially hesitated to employ a design relying on two different databases (Westlaw’s appeal analytics for the numerator and another data source, either Lex Machina or the Criminal Integrated Database, for the denominator) to calculate one metric. However, because Lex Machina does not report criminal data, we would be forced to turn to two data sources no matter what: Lex Machina for civil cases, and a combination of the Criminal Integrated Database and Westlaw for criminal cases. We chose, instead, to use a consistent source (Westlaw) for all numerators across case categories and to adhere to our methods for calculating other metrics, such as time-to-termination, for our denominators (i.e., total district court dispositions). Our decision to do so was reinforced by some capabilities absent from Westlaw, which offers no comprehensive method for finding cases by termination or filing date. Moreover, because Westlaw and Lex Machina have significant overlap with respect to each judge’s docket-wide dataset, we are confident that, for purposes of this calculation, any differences between the two databases would not significantly affect our results. And any effect is unlikely to be systematically biased in favor of either crowded or non-crowded time periods.

      [156].     We rely on Westlaw for several reasons, some of which we have already elaborated above. See supra note 155. For one, when we began our data collection, Lex Machina was not reporting appeals data. For another, Westlaw’s functionality allowed us to trace the history of an appealed district court decision. And as noted above, Lex Machina does not report criminal case data. Therefore, to maintain consistency across appeal outcomes, we proceeded with a single database. Moreover, discrepancies in the categorization approaches between Westlaw and Lex Machina for certain appeals data informed our decision to use only one of the two databases for the numerators in our reversal rate calculations (as opposed to relying on one data source for civil cases and another for criminal cases).

      [157].     As suggested, our primary method for separating patent appeals from non-patent appeals is by looking to the Court of Appeals. We treat appeals to the Federal Circuit as patent appeals and appeals to a regional circuit as non-patent appeals. It is possible, in very limited circumstances, for a case involving a patent claim to be appealed to a regional circuit (such as through a permissive counterclaim, see ABS Glob. v. Inguran LLC, 914 F.3d 1054 (7th Cir. 2019)). But we found no such cases in our dataset.

                    To further categorize the non-patent appeals, we ensured that each appealed decision fell within our study period, see supra note 146, and we manually categorized each judge’s appealed cases by reading the text of the appellate court’s decision and classifying each appeal as civil or criminal. Within the civil category, we also identified if the appeal was a civil rights or habeas case. See supra note 138 and accompanying text (describing our methods for doing so). We did not rely on Westlaw’s auto-generated tags, which we found to be inaccurate and underinclusive, particularly for criminal cases.

      [158].     See supra note 154 and accompanying text.

      [159].     See 28 U.S.C. § 41.

      [160].     As noted above, our study focuses on a consistent set of district judges throughout our study period to address effects related to ideology, temperament, or experience. Some readers might suggest extending such controls to the appellate courts, limiting our analysis of appeals to only those appeals decided by a consistent set of judges over time. Limiting panel selection in this way, however, severely restricts the number of appeals we can study—there may be, for example, only one appeal decided by any given panel. Moreover, the dynamics of panel decision-making can help to mitigate the changes brought on by the appointment of any one new judge. Hence, we use the entire body of appeals to measure reversal rates.

      [161].     As noted above, we disregard appeals dismissed for a lack of appellate jurisdiction (including cases in which a certificate of appealability was denied). See supra note 154. Moreover, there is one affirmance that the Supreme Court later vacated and remanded. See Henry Schein, Inc. v. Archer & White Sales, 139 S. Ct. 524 (2019). We treat this as an affirmance for our purposes, as our method looks only to decisions rendered by the regional circuit, for reasons elaborated above.

      [162].     See, e.g., Huang & Narechania, supra note 121, at 1740 n.103 (similarly treating reversals-in-part as reversals for measurement purposes).

      [163].     The relevant cases do not include criminal cases. See infra Part II.B.1. See also Sean Tu & Mark A. Lemley, What Litigators Can Teach the Patent Office About Pharmaceutical Patents, 99 Wash. U. L. Rev. 1673, 1688 (2022) (“Because this is a population study that includes every litigated patent, by definition the results are statistically significant.”). But cf. Christina L. Boyd, Pauline T. Kim, & Margo Schlanger, Mapping the Iceberg: The Impact of Data Sources on the Study of District Courts, 17 J. Empirical Legal Stud. 466, 466 (2020) (analyzing “how the work of federal district courts looks different depending on whether research relies on published opinions, on opinions available on Westlaw or Lexis (both ‘published’ and ‘unpublished’), or on more comprehensive data available on PACER”). As noted above, our research strategy collects information across various sources, including Westlaw and sources derived from PACER, in an effort to avoid problems related to such “missing decisions.” Cf. Merritt E. McAlister, Missing Decisions, 169 U. Pa. L. Rev. 1101, 1104–05 (2021).

      [164].     See supra note 81 and accompanying text.

      [165].     We can also compare these changes to the Fifth Circuit’s overall reversal rate during comparatively crowded and non-crowded years. Across these years, the Fifth Circuit’s reversal rate remained rather steady, reversing 6.3% of appeals in 2016 and 6.2% in 2018. See Admin. Off. for the U.S. Cts., Table B-5: U.S. Courts of Appeals—Decisions in Cases Terminated on the Merits, by Circuit and Nature of Proceeding, During the 12-Month Period Ending December 31, 2016, at 3 (2016); Admin. Off. for the U.S. Cts., Table B-5: U.S. Courts of Appeals—Decisions in Cases Terminated on the Merits, by Circuit and Nature of Proceeding, During the 12-Month Period Ending December 31, 2018, at 3 (2018). Our concentrated sample of judges in the Eastern District of Texas does worse against this benchmark during comparatively crowded years than during non-crowded years.

                    Again, we emphasize that our measure of crowding is comparative, because, as noted, the Eastern District heard a substantial number of patent cases even during periods designated as not crowded (which may, but need not, help to explain why these reversal rate figures are consistently higher than the Fifth Circuit’s average).

      [166].     See infra Part II.B.2.

      [167].     We could not calculate appeal rates for criminal cases because, as noted above, the FJC Criminal Integrated Database provides data only at the division level, so we were unable to determine the number of criminal filings for a given judge in multi-judge divisions. See supra note 136.

      [168].     See supra note 165 and accompanying text. As noted there, we mean significant in a statistical, rather than normative, sense. But see Ronald L. Wasserstein, Allen L. Schirm & Nicole A. Lazar, Moving to a World Beyond “p < 0.05,” 73 Am. Statistician 1, 2 (2019) [hereinafter Beyond 0.05] (“[I]t is time to stop using the term ‘statistically significant’ entirely.”). To elaborate further, we performed a Welch’s t-test to compare the difference in reversal rates for each judge with data for both crowded and non-crowded periods (i.e., Judges Stark, Andrews, Gilstrap, and Schroeder). We interpret the difference in reversal rates between the two periods, across all included judges, to be significant (p = 0.062), at least to the p < 0.10 level (indeed, even more so), particularly in view of the applied context of our research and the magnitude of our effect sizes (especially after accounting for the likely mechanism, see infra Part II.B.2). See Ronald L. Wasserstein & Nicole A. Lazar, The ASA’s Statement on Statistical Significance and p-Values: Context, Process, and Purpose, 70 Am. Statistician 129, 129–33 (2016) (concluding that the interpretation of p-values should turn not only on numerical thresholds, but also context and common sense, and should be reported on a continuous basis); Wasserstein, Schirm & Lazar, Beyond 0.05, supra, at 3–4. See also Matthew S. Thiese, Brenden Ronna & Ulrike Ott, P Value Interpretations and Considerations, 8 J. Thoracic Disease 928, 929 (2016) (distinguishing between “clinical significance” and “statistical significance” and explaining that “low P values” traditionally considered as “‘trending toward statistical significance’ may be clinically relevant for improving practice, particularly in smaller studies”). Cf. Randall Munroe, P-Values, xkcd, https://xkcd.com/1478 [https://perma.cc/4XQ5-DF58] (pithily characterizing various p-values in terms of significance). Consistent with practice in other disciplines, we report our p-value as a two-tailed p-value. Notably, this method of analysis assumes that neither the identity of the judge nor the type of case contributes to changes in reversal rates. As suggested above, this is consistent with our approach, as we can discern no bias in the types of non-patent cases assigned to these judges, nor any confounding change in the judges’ ideology over time.

      [169].     See supra note 142 and accompanying text.

      [170].     Compare Caldwell-Blow v. Wells Fargo Bank, N.A., 687 Fed. App’x 380 (5th Cir. 2017), with Lohri v. CSAB Mortgage-Backed Pass Through Certificate Series 2007-1 U.S. Bank, No. 4:18-CV-00143-JRG, 2019 WL 1239608 (E.D. Tex. March 18, 2019).

      [171].     See Holman, supra note 140, at 710–11 (describing the debate over using opinion length to assess decision quality and concluding that “[r]egardless of who is correct, both sides—and many scholars—believe that page length is an important measure of judicial effort”).

      [172].     Compare Adams v. Stripe-A-Zone, Inc., No. 2:15-cv-532-JRG, 2015 WL 12806516 (E. D. Tex. Dec. 7, 2015), with Potter v. Cardinal Health 200, LLC, 381 F. Supp. 3d 729 (E.D. Tex. 2019).

      [173].     Cannon v. Colvin, Civ. No. 15-29-LPS, 2016 WL 5660392 (D. Del. Sept. 29, 2016).

      [174].     Id. at *1–5.

      [175].     Id. at *8–12.

      [176].     Ransom v. Berryhill, Civ. No. 17-939-LPS, 2018 WL 3617944 (D. Del. July 30, 2018).

      [177].     Id. at *6–7.

      [178].     Id. at *7.

      [179].     Cannon, 2016 WL 5660392, at *9–11.

      [180].     See supra note 138.

      [181].     See Order Regarding Court Docket Management for Waco Division, supra note 135. This is the case, notwithstanding Judge Albright’s claims to “treat every civil case the same way.” Blake Brittain, Texas’ Busiest Patent Judge Shows No Sign of Slowing Down, Reuters (June 28, 2021) https://www.reuters.com/legal/transactional/texas-busiest-patent-judge-shows-no-signs-slowing-down-2021-06-28/ [https://perma.cc/3K3A-FAYP].

      [182].     We note that Judge Albright issued an earlier iteration of this Order a few months into his tenure on the bench, and so it is possible that some of these cases were subject to a slightly different process (i.e., Judge Albright’s pre-standing order process). Moreover, Judge Albright might have decided in any individual case to refer certain matters to a magistrate judge. Nevertheless, this set of cases seems most likely to represent those that receive Judge Albright’s primary attention—beyond patent cases, of course.

      [183].     See, e.g., Levi & Gulati, supra note 21, at 411 (“Of course, a higher reversal rate might just demonstrate a less careful . . . district judge.”).

      [184].     See supra Part II.A.2.

      [185].     See Order Regarding Court Docket Management for Waco Division, supra note 135.

      [186].     The Fifth Circuit’s average reversal rate in criminal cases is about 3%. See Admin. Off. for the U.S. Cts., Table B-5: U.S. Courts of Appeals—Decisions in Cases Terminated on the Merits, by Circuit and Nature of Proceeding, During the 12-Month Period Ending September 30, 2021, at 4 (2021) (noting a criminal reversal rate of 2.8%).

      [187].     See United States v. Gardner, No. 20-50481, 2022 WL 422167 (5th Cir. Feb. 11, 2022); see also Appellant’s Initial Brief at 13, United States v. Gardner, No. 20-50481 (5th Cir. Sept. 15, 2020) (stating that the district court denied Mr. Gardner’s motion to withdraw plea of guilty on February 27, 2020, and conducted his sentencing hearing on June 10, 2020). In 2020, Judge Albright led the country in number of patent cases filed, with 20% of all such cases nationwide. See Appendix Figure 1.

      [188].     Gardner, 2022 WL 422167, at *1.

      [189].     See id. at *2.

      [190].     See Holman, supra note 140, at 710–11.

      [191].     997 F.3d 211 (5th Cir. 2021).

      [192].     Id. at 213.

      [193].     See id. at 214 n.1.

      [194].     We emphasize, however, that these results cannot easily be attributed to differences such as experience or ideology. As described above, we have implemented some rough controls for such concerns by limiting our analysis to judges with similar backgrounds, experience, and ideology (all were appointed by President Trump in close proximity to one another after similar prior judicial experience). See supra note 131 and accompanying text.

      [195].     See supra note 143 and accompanying text.

      [196].     Cf. Huang & Narechania, supra note 121 (examining which court cases are prioritized during busy times and how that process may be affected by “higher-level judicial priorities” such as public image).

      [197].     See supra Part II.

      [198].     Edwards, supra note 21, at 403.

      [199].     Cf. Huang, supra note 17, at 1137–38 (describing the relationship between judicial capacity and decision quality).

      [200].     Indeed, these findings extend through Judge Schroeder of the Eastern District of Texas. Judge Schroeder played comparatively little role in that district’s forum-selling conduct, perhaps suggesting that Judge Schroeder did not share Chief Judge Gilstrap’s zeal for patent litigation. Indeed, the size of Judge Schroeder’s patent docket seems to have been a consequence of procedural decisions that largely predated his appointment to the bench. See Matthew Bultman, Beyond Gilstrap: 5 Judges To Watch In Patent Litigation, Law360 (Feb. 7, 2018) https://www.law360.com/articles/1007206/beyond-gilstrap-5-judges-to-watch-in-patent-litigation, [https://perma.cc/LVA8-22LW] (explaining that Judge Schroeder’s “patent docket has been substantial, in part because he inherited a large share of the cases filed in nearby Tyler after his appointment”). Judge Schroeder’s reversal rate remained steady during periods of crowding and non-crowding (though we note that the sample size here is especially small—only eight of Judge Schroeder’s non-patent decisions across our study periods were appealed, and he was affirmed in all eight cases). See infra Appendix Figure 5.

      [201].     We cannot say whether Judge Stark was appointed to the Federal Circuit because he actively prefers patent cases, because of his extensive (if accidental) experience with them, or because of some other reason entirely.

      [202].     Here and throughout, we focus on civil cases in our time-to-termination results both because various constitutional and statutory measures, such as the Constitution’s Speedy Trial Clause under the Sixth Amendment and the Speedy Trial Act, affect how long judges may take to process such cases and because of the limited availability of certain criminal data. Cf. supra note 86 and accompanying text (noting that Judge Albright reviews criminal cases much faster than these provisions require).

      [203].     See de Figueiredo, Lahav, & Siegelman, supra note 19, at 369–70; Petkun, supra note 19, at 2.

      [204].     See, e.g., Wooten v. City of Wilmington, No. 19-2133-RGA, 2021 WL 411707 (D. Del. Feb. 5, 2021); Twin Spans Bus. Park v. Cincinnati Ins. Co., No. 1:19-cv-00476-SB-SRF, 2021 WL 1226497 (D. Del. Apr. 21, 2021); VoterLabs, Inc. v. Ethos Grp. Consulting Servs., LLC, No. 19-524-RGA, 2020 WL 10622570 (D. Del. Dec. 4, 2020).

      [205].     More specifically, for Judge Andrews, the median time to termination for non-patent cases denoted as a “Likely Settlement” in Lex Machina dropped while the median time to termination for other non-patent civil cases increased. By contrast, the median time to termination for Judge Stark reduced across all non-patent civil cases. Lex Machina defines a “Likely Settlement” as a case in which “one or both parties stipulated to dismiss the case.”

      [206].     Such results, moreover, might have implications for other crowded courts, including state courts, which are notoriously backlogged. See, e.g., Gina Jurva, Thomson Reuters Inst., The Impacts of the COVID-19 Pandemic on State & Local Courts Study 2021: A Look at Remote Hearings, Legal Technology, Case Backlogs, and Access to Justice 4 (2021). While many scholars and commentators regard these backlogs and capacity constraints as problematic, we note that some innovative scholars have described these limits as offering an opportunity to subvert the criminal legal system. See Andrew Manuel Crespo, No Justice, No Pleas: Subverting Mass Incarceration Through Defendant Collective Action, 90 Fordham L. Rev. 1999, 2002–04 (2022).

      [207].     We note that these results may also have implications beyond their utility in assessing possible mechanisms. Comparatively terse treatments of non-patent cases may also have implications for a more qualitative review of a judge’s decisions and case management procedures, giving rise to further procedural fairness concerns. Cf. Young & Singer, supra note 140, at 55 (proposing a measure of procedural fairness that considers “the time that a federal district judge spends on the bench, presiding over the adjudication of issues in an open forum”). Indeed, turning away from civil cases and to criminal cases, we noted above that Judge Albright resolves criminal matters in only fifty-five days. See supra note 150.

      [208].     See de Figueiredo, Lahav, & Siegelman, supra note 19, at 369–70; Petkun, supra note 19, at 2.

      [209].     See supra Figure 2.

      [210].     Stated similarly, it is possible that the Eastern District dedicated less capacity to each case, but each case still took longer to resolve because there were so many. Our longer time-to-termination results might alternately reflect the possibility that the Eastern District dedicated the same, or even more, judicial capacity to each case.

      [211].     See Jonas Anderson, Judge Shopping in the Eastern District of Texas, 48 Loyola U. Chi. L.J. 539, 540–41 n.8 (2016) (citing an invitation for clerkship applications that explains that “Judge Gilstrap has the largest patent docket in the country and appreciates applicants who have a science or engineering background”); see also Tommy Witherspoon, Waco Becoming Hotbed for Intellectual Property Cases with New Federal Judge, Waco Trib.-Herald (Jan. 18, 2020), https://www.wacotrib.com/news/local/waco-becoming-hotbed-for-intellectual-property-cases-with-new-federal-judge/article_0bcd75b0-07c5-5e70-b371-b20e059a3717.html [https://perma.cc/XS7Q-JU6L] (suggesting that Judge Albright hires “law clerks . . . [with] knowledge and expertise to handle patent cases”).

      [212].     Witherspoon, supra note 211.

      [213].     See Avraham & Hubbard, supra note 13, at 17.

      [214].     See Stephen G. Breyer, Regulation and Its Reform 23–26 (1982) (suggesting that assigning liability to one party forces that party to internalize the costs of their actions).

      [215].     See Levy, supra note 36, at 1070–72 (explaining that courts should favor procedural, rather than substantive, solutions to the problems of docket congestion); cf. Matthew A. Shapiro, Procedural Wrongdoing, 48 B.Y.U. L. Rev. 197, 240–41 (2022) (describing one procedural rule—namely, Civil Rule 11—as partially aimed at addressing “dilatory” crowding or congestion tactics).

      [216].     See, e.g., Megan M. La Belle, The Local Rules of Patent Procedure, 47 Ariz. St. L.J. 63, 111 (2015) (proposing that “a uniform set of federal procedural rules” apply to all patent cases nationwide).

      [217].     Mauskopf Letter, supra note 10, at 1; see also Patent Pilot Program Final Report, supra note 139, at 6–7.

      [218].     See, e.g., Lipocine Inc. v. Clarus Therapeutics, No. 19-622 (WCB), 2020 WL 4794576, at *2 (D. Del. Aug. 18, 2020) (Bryson, J.) (“The general fact pattern in each of these cases is essentially the same: the plaintiff alleges infringement; the defendant responds with an answer and counterclaims denying infringement and asserting patent invalidity. . . .”); see also Trico Prods. v. Anderson Co., 147 F.2d 721, 722 (7th Cir. 1945) (Defendant interposed a counterclaim seeking a declaratory judgment of invalidity and noninfringement in response to plaintiff’s motion.).

      [219].     See Return Mail Inc. v. U.S. Postal Serv., 139 S. Ct. 1853, 1868 (2019) (Breyer, J., dissenting) (“When A sues B for patent infringement, B may defend against the lawsuit by claiming that A’s patent is invalid. . . . Congress, however, has also established a variety of administrative procedures that B may use to challenge the validity of A’s patent.”); see also Saurabh Vishnubhakat, Arti K. Rai & Jay P. Kesan, Strategic Decision Making in Dual PTAB and District Court Proceedings, 31 Berkeley Tech. L.J. 45, 81 (2016) (“Most patents challenged at the PTAB are also in Article III litigation . . . .”).

      [220].     See supra notes 218–219 and accompanying text.

      [221].     See Vishnubhakat, Rai, & Kesan, supra note 219, at 80–81 (evaluating data on stays pending agency challenges).

      [222].     See, e.g., NFC Tech. LLC v. HTC Am., Inc., No. 2:13-CV-1058-WCB, 2015 WL 1069111, at *5–6 (E.D. Tex. Mar. 11, 2015) (granting a party’s motion to stay proceedings pending the completion of inter partes review before the Patent Trial and Appeal Board).

      [223].     See NHK Spring Co., Ltd. v. Intri-Plex Techs., Inc., No. IPR2018-00752, 2018 WL 4373643, at *4–7 (P.T.A.B. Sept. 12, 2018); Apple Inc. v. Fintiv, Inc., No. IPR2020-00019, 2020 WL 2126495, at *6, *9 (P.T.A.B. Mar. 20, 2020) [hereinafter, collectively, NHK-Fintiv] (noting that whether the agency will institute review depends, in part, on “whether the court [in parallel patent proceedings] granted a stay or evidence exists that one may be granted if a proceeding is instituted,” among other factors).

      [224].     See supra note 90.

      [225].     See, e.g., Saurabh Vishnubhakat, Patent Inconsistency, 97 Ind. L.J. 59, 64, 70 (2022) (describing agency substitution as efficient, but lamenting that, in practice, much agency process has been duplicative rather than substitutionary).

      [226].     See id. at 120.

      [227].     In our view, both of these asserted interests are thin reeds upon which to rest an absolute preference for litigation. See also supra note 219 and accompanying text. For one, it is not at all obvious why our legal system should vindicate a plaintiff’s choice of forum, other than blind adherence to the maxim that “the plaintiff is the master of the complaint.” See Holmes Grp. v. Vornado Air Circulation Sys., Inc., 535 U.S. 826, 831 (2002). This is particularly so when, as elaborated above, honoring the plaintiff’s choice of forum gives rise to gamesmanship, imposes externalities on other litigants, and undermines the courts more generally. See supra Parts I.A, I.B.2. Moreover, the patentholder’s legitimate expectation interests must be balanced against the public’s “paramount interest in seeing that patent monopolies . . . are kept within their legitimate scope.” Precision Instrument Mfg. Co. v. Auto. Maint. Mach. Co., 324 U.S. 806, 816 (1945). And the Patent Office, which has both legal and technical expertise and which may be held to account for its decisions, is uniquely suited to conduct such balancing. Cf. Tejas N. Narechania, Arthrex and the Politics of Patents, 12 Calif. L. Rev. Online 65, 71–73 (2022) (describing the Patent Office’s comparative institutional advantages with respect to conducting such a balancing). A more complete examination of these questions is better left to future work.

      [228].     Compare, e.g., Tex. Intell. Prop. L.J., Judge Alan Albright and Kat Li Speak About IP Litigation, Youtube, at 4:53–5:11 (Mar. 8, 2022), https://www.youtube.com/watch?v=h8fHZZOZqvI [https://perma.cc/PTQ8-CBVE] (statements of Judge Alan Albright) [hereinafter Albright Speech]  (“I’m a very strong believer that patents are in the Constitution—there’s a Seventh Amendment in the Constitution that gives people rights to jury trials—and that if someone has gotten a patent and they believe that their patent is being infringed . . . they ought to get a jury trial.”), with Oil States Energy Servs. v. Greene’s Energy Grp., 138 S. Ct. 1365, 1370, 1373 (2018) (explaining that patents, as public franchises, may be reviewed and revoked in agency proceedings without implicating the Seventh Amendment’s jury trial right).

                    To the extent this “constitutional rights” argument is premised not on the Seventh Amendment jury trial right, but rather on a constitutional entitlement to the patent itself, see Albright Speech, supra, the argument is even more specious. For one, the Constitution only authorizes Congress to issue (or establish an agency for issuing) patents, but it nowhere requires that Congress do so. And even if that constitutional authorization could be understood as a mandate, the Constitution is silent as to the standard that applies to patentable advances. This suggests that the government could, in some cases and after review, decline to issue—or (as Oil States holds) revoke previously issued—patents over inventions that are insufficiently inventive.

      [229].     See Cuozzo Speed Techs. v. Lee, 579 U.S. 261, 268 (2016).

      [230].     See NHK Spring Co., Ltd. v. Intri-Plex Techs., Inc., No. IPR2018-00752, 2018 WL 4373643, at *4–7 (P.T.A.B. Sept. 12, 2018); Apple Inc. v. Fintiv, Inc., No. IPR2020-00019, 2020 WL 2126495, at *6, *9 (P.T.A.B. Mar. 20, 2020).

      [231].     Tex. Intell. Prop. L.J., Professor Tejas Narechania and USPTO Panel, Youtube, at 23:07–24:40 (Mar. 8, 2022), https://www.youtube.com/watch?v=x__mjrJBFzU [https://perma.cc/Z8HR-XYA7] (statements of Administrative Patent Judge Brian Range); see also Clearwire Corp. v. Mobile Telecomms. Techs., No. IPR2013-00306, 2013 WL 8563943, at *3 (P.T.A.B. Oct. 22, 2013) (considering the public’s interest in avoiding past infringement liability when deciding to institute review of a soon-to-expire patent).

      [232].     Jeremy S. Graboyes, Kali Murray, Kazia Nowacki, Arti K. Rai, Alexandra Sybo, & Melissa F. Wasserman, Admin. Conf. of the U.S., Off. of the Chairman, Patent Small Claims: Report to the U.S. Patent & Trademark Office 32 (2023), https://www.acus.gov/sites/default/files/documents/ACUS%20Report%20on%20Patent%20Small%20Claims%202023.pdf [https://perma.cc/TMT7-ZWM2] (noting that “patent suits compete for limited judicial capacity on crowded district court dockets, including against criminal cases,” and so suggesting greater use of “non-Article III judicial officers” “to hear or help manage patent cases”).

      [233].     35 U.S.C. § 314(a).

      [234].     See, e.g., Christian Helmers & Brian J. Love, Patent Validity and Litigation: Evidence from U.S. Inter Partes Review, 66 J.L. & Econ. 53, 56 (2023).

      [235].     See supra Part II.B.

      [236].     NFC Tech. LLC v. HTC Am., Inc., No. 2:13-CV-1058-WCB, 2015 WL 1069111, at *4–6 (E.D. Tex. Mar. 11, 2015).

      [237].     See, e.g., Sierra Pac. Indus. v. Kolbe & Kolbe Millwork Co., No. 18-CV-853-WMC, 2019 WL 1924836, at *3 (W.D. Wis. Apr. 30, 2019) (denying stay pending institution decision, but without prejudice to a renewed motion for stay if the agency institutes review); see also Stroud, supra note 89, at 244.

      [238].     Under NHK-Fintiv, the Patent Office’s decision to institute the patent depends in part on “whether the court [in parallel patent proceedings] granted a stay,” the “proximity of the court’s trial date to the Board’s projected statutory deadline” as well as the “amount and type of work already completed in the parallel litigation.” Apple Inc. v. Fintiv, Inc., No. IPR2020-00019, 2020 WL 2126495, at *5, *8 (P.T.A.B. Mar. 20, 2020). And litigation can progress a substantial amount in six months (which is the Patent Office’s internal deadline for resolving a petition for inter partes review), particularly in venues that emphasize speed in order to lure patent plaintiffs. See 37 C.F.R. § 42.107(b) (giving patent owners three months to file a preliminary response to a petition for review); 35 U.S.C. § 314(b) (giving the agency three months from the preliminary response’s due date to act on a petition for review).

      [239].     See Helmers & Love, supra note 234, at 53 (finding that agency review can have “positive effects on the settlement of parallel court proceedings”).

      [240].     Cf. Apple Inc. v. Fintiv, Inc., No. IPR2020-00019, 2020 WL 2126495, at *5–6 (P.T.A.B. Mar. 20, 2020) (explaining that whether to institute review is guided by many factors, including “other circumstances that impact the Board’s exercise of discretion”).

      [241].     See supra notes 204–205 and accompanying text.

      [242].     See supra notes 112–113 and accompanying text; see also infra Parts III.A.2–III.B.

      [243].     Cf. Coinbase, Inc. v. Bielski, 599 U.S. 736, 744 (2023) (describing the traditional rule that “requires that a district court stay its proceedings while [proceedings resolving] the question of [alternative dispute resolution] is ongoing”); In re TikTok, Inc., 85 F.4th 352, 364 n.13 (5th Cir. 2023) (chastising the district court for failing to stay litigation while sitting on a transfer motion, and then justifying a decision to deny transfer by reference, in part, to the case’s progress).

      [244].     Roberts Letter, supra note 4, at 2 (describing concerns related to patent case assignment procedures).

      [245].     We choose six months because 35 U.S.C. § 314(b) and 34 C.F.R. § 42.107(b), read together, suggest that the Patent Office must decide whether to institute inter partes review within six months of receiving a petition.

      [246].     See, e.g., Eric W. Schweibenz, Robert C. Mattson & Lisa M. Mandrusiak, Automatic Stay of Litigation Pending Inter Partes Review?: A Simple Proposal for Solving the Patent Troll Riddle, 7 Landslide 41–42 (2014) (proposing similar legislation and analogizing it to the automatic stay provision that applies to parallel proceedings at the International Trade Commission regarding infringing imports).

      [247].     See 35 U.S.C. § 315(e)(2) (providing that, in civil actions or proceedings before the International Trade Commission, the petitioner is barred from asserting the claim’s invalidity based on any ground that was raised or reasonably could have been raised during the IPR); cf. Ryan Davis, PTAB Judges Discuss Their Approach To New Fintiv Guidance, Law360 (July 21, 2022) https://www.law360.com/articles/1510565/ptab-judges-discuss-their-approach-to-new-fintiv-guidance [https://perma.cc/GE2T-RHSZ] (noting Administrative Patent Judges’ concern for “duplicative work” across the agency and the courts).

      [248].     See Paul R. Gugliuzza, (In)valid Patents, 92 Notre Dame L. Rev. 271, 283, 327 (2016) (similarly advocating for expanded review in IPR); cf. Julia Schönbohm, Bolko Ehlgen & Natalie Ackermann-Blome, Germany, in The Patent Litigation Law Review 108–27 (Trevor Cook ed., 3d ed. 2019) (describing Germany’s bifurcated system of patent adjudication, which features a separate, initial validity determination conducted by experts and a subsequent infringement trial); Sapna Kumar, Judging Patents, 62 Wm. & Mary L. Rev. 871, 903–04 (2021) (similar).

      [249].     See 35 U.S.C. § 311(b) (limiting the scope of permissible challenges to “only . . . ground[s] that could be raised under section 102 or 103 and only on the basis of prior art consisting of patents or printed publications”); see also 35 U.S.C. §§ 102–103.

      [250].     See, e.g., Norgren Automation Sol., LLC v. PHD, Inc., No. 14-cv-13400, 2015 WL 1245942, at *2 (E.D. Mich. Mar. 18, 2015) (denying a stay motion in part because an inter partes review would not resolve certain other issues, including “validity under 35 U.S.C. § 101”).

      [251].     Fromer, supra note 108, at 1478–79 (proposing to “constrain venue to require suit in the district of the principal place of business of any of the defendants,” subject to certain limited exceptions); see also Klerman & Reilley, supra note 9, at 304 (similarly proposing that Congress “amend the patent venue statute to require patent owners to sue in the defendant’s principal place of business or largest market”). We note that not everyone agrees with proposals such as these. As Jeanne Fromer explains, both Kimberly Ann Moore and Rochelle Dreyfuss contend that such proposals will undermine the expertise in technical aspects of patent law that has accrued in some courts as a consequence of concentrations of patent filings. See Fromer, supra note 108, at 1478; Moore, supra note 15, at 934 (noting how limiting venue to defendant’s principal place of business might disperse patent infringement cases across judicial districts, reducing the current efficiency gained from consolidated cases in frequently targeted districts); Rochelle Cooper Dreyfuss, In Search of Institutional Identity: The Federal Circuit Comes of Age, 23 Berkeley Tech. L.J. 787, 805 (2008) (endorsing Moore’s suggested approach along with a proposed experiment designating specific judges in large districts for all patent cases); but see Hon. Diane P. Wood, Is It Time to Abolish the Federal Circuit’s Exclusive Jurisdiction in Patent Cases?, 13 Chi.-Kent J. Intell. Prop. 1, 6–8 (2013) (questioning the expertise rationale for concentrating patent cases in the context of the Federal Circuit’s exclusive jurisdiction over patent appeals).

      [252].     See Anderson & Gugliuzza, supra note 9, at 480 (advocating “for Congress to revise the patent venue statute to require the case to be connected to not just the district in which it is filed, but to the division in which it is filed, if the district is divided into divisions”) (emphasis in original).

                    Jonas Anderson and Paul Gugliuzza’s proposal has a back-to-the-future feel about it, as there was once a requirement that criminal cases be tried within the division (not only the district) in which the underlying crime was committed. See, e.g., supra note 112 (citing, inter alia, Salinger v. Loisel, 265 U.S. 224, 237 (1924)). As we explain there, the relevant statutes and procedural rules have since been amended to permit criminal trials anywhere in the district of the offense.

      [253].     See Anderson & Gugliuzza, supra note 9, at 478. Specifically, Jonas Anderson and Paul Gugliuzza would modify the statute governing case assignment, 28 U.S.C. § 137, to ensure that “no judge in a district court having more than one judge shall have greater than a 50 percent probability of being assigned a given case.”

      [254].     Compare Fromer, supra note 108, at 1496 (finding that, under current rules, patent cases regarding communications and computing technology tend to be filed in four main districts), with id. at 1497 (estimating that venue reform would cause such cases to be filed in six main districts (instead of four)). See also id. at 1500 (demonstrating similar distributions among jurisdictions for patent cases regarding pharmaceuticals and medical devices).

      [255].     We agree, of course, that the effectiveness of these proposals depends on avoiding the sorts of gaming that have undermined the reforms attempted in the Western District of Texas. See supra note 96 and accompanying text.

      [256].     We note, however, that not all proposed venue modifications would have this effect. A rule that, say, limited venue to a defendant’s district of incorporation would crowd the District of Delaware even more than before. And so we limit our endorsement to venue rules that address not only concerns about substantive distortion, gamesmanship, and judicial neutrality, but also concerns about crowding.

      [257].     Fed. R. Crim. P. 18.

      [258].     In re Chesson, 897 F.2d 156, 159 (5th Cir. 1990); see also United States v. Kaufman, 858 F.2d 94, 106 (5th Cir. 1988); United States v. Dickie, 775 F.2d 607, 610 (5th Cir. 1985) (similarly considering docket crowding in the context of a transfer motion).

                    We clarify that Rule 18 still requires “the government [to] prosecute an offense in a district where the offense was committed” (though not the division, as these cases regard intradistrict, interdivision transfers). See supra note 112. Moreover, we have found little evidence to suggest that courts now routinely transfer criminal cases out of crowded divisional dockets, as all the cases cited above predate the forum crowding analyzed above. See supra notes 124, 126, 131, and accompanying text (describing our study periods). Indeed, we have found no outgoing intradistrict transfers of criminal cases for the three divisions from our study for which we have data. But cf. United States v. Christensen, No. 17-cr-20037-JES-JEH, 2018 WL 6382050 (C.D. Ill. Dec. 6, 2018) (suggesting one counterexample from an entirely different district). In general, it is correct to say that criminal defendants in, say, Waco have little opportunity to avoid a forum that is overrun with patent cases.

      [259].     See In re Volkswagen of Am., Inc. 545 F.3d 304, 315 (5th Cir. 2008). The seven-factor test for venue transfer is divided into “private interest” and “public interest” factors. Id. Yet of these potential causes to grant or deny a motion for transfer, only one (court congestion) appears to involve a consideration of the effects of that transfer on other litigants in the transferee district. This list thus also reflects the longstanding tendency in our thinking about forum shopping and venue transfer to overlook the externalities of such choices on other litigants in a given district.

      [260].     See Mark Liang, The Aftermath of TS Tech: The End of Forum Shopping in Patent Litigation and Implications for Non-Practicing Entities, 19 Tex. Intell. Prop. L.J. 29, 48, 61–63 (2010) (noting that this factor has generally been treated as “inconclusive” or “speculative” in courts’ transfer motion analysis).

      [261].     See In re Apple Inc., No. 2022-128, 2022 WL 1196768, at *2 (Fed. Cir. Apr. 22, 2022); see also In re Genentech, Inc., 566 F.3d 1338, 1341–42 (Fed. Cir. 2009).

      [262].     See, e.g., MEC Res., LLC v. Apple, Inc., 269 F. Supp. 3d 218, 224, 228 (D. Del. 2017); see also Malathi Nayak, Swelling Docket Pushing Delaware Judges to Transfer Patent Cases, Bloomberg L. (Sept. 20, 2017), https://news.bloomberglaw.com/us-law-week/swelling-docket-pushing-delaware-judges-to-transfer-patent-cases-1 [https://perma.cc/ZT8E-NC9K] (“U.S. District Court for the District of Delaware judges have cited court congestion as the reason for transferring cases to other courts . . . .”). And, as our Judge Andrews results suggest, better case management practices may help to address forum crowding concerns.

      [263].     Avraham & Hubbard, supra note 13, at 30–31.

      [264].     See Brian J. Love & Christian Helmers, Welcome to Waco! The Impact of Judge Shopping on Litigation, 00 J.L. Econ. & Org. 1, 1 (2023).

      [265].     See, e.g., Chien & Risch, supra note 15, at 90–93 (predicting the distribution of patent cases under two distinct proposed or anticipated venue reforms); Fromer, supra note 108, at 1496–506 (predicting where patent cases will cluster under a hypothetical venue regime).

      [266].     See, e.g., supra note 254 and accompanying text (describing Jeanne Fromer’s study of a proposed venue reform that gives rise to a more even distribution of cases).

      [267].     See, e.g., Nichols v. U.S., 578 U.S. 104, 108 (2016).

      [268].     Avraham & Hubbard, supra note 13, at 30.

      [269].     Id. at 60.

      [270].     Even if instituting a system of “pay[ments]” for patent plaintiffs and “rewards” for civil rights plaintiffs and criminal defendants was a good idea (which it may or may not be), arriving at such a conclusion requires assessing and balancing a set of seemingly incommensurable values—a task well beyond our present scope. Indeed, there have long been concerns around economically quantifying the value of access to justice. See, e.g., Bail Reform, Am. C.L. Union, https://www.aclu.org/issues/smart-justice/bail-reform [https://perma.cc/2CNV-4R3Y]. And we suspect that implementing such a scheme would be highly controversial, thus implicating additional administrability concerns beyond those discussed in the remainder of this paragraph.

      [271].     Avraham & Hubbard, supra note 13, at 31.

      [272].     Id.

      [273].     See, e.g., de Figueiredo, Lahav, & Siegelman, supra note 19, at 369–70; Petkun, supra note 19, at 2.

      [274].     See supra note 19.

      [275].     See Levy, supra note 36, at 1070–72 (explaining that Congress can play a leading role in addressing problems of docket congestion).

      [276].     See, e.g., Moore, supra note 15, at 924; Norwood, supra note 25, at 268; see also Henry J. Friendly, Averting the Flood By Lessening the Flow, 59 Cornell L. Rev. 634, 641 (1974).

      [277].     See supra note 1 and accompanying text.

      [278].     Mauskopf Letter, supra note 10, at 1; Patent Pilot Program Final Report, supra note 139, at 6–7.