Emory Law Journal

Strategic Immunity
Aaron L. Nielson,
Christopher J. Walker Aaron L. Nielson is Associate Professor of Law, J. Reuben Clark Law School, Brigham Young University.Christopher J. Walker is Associate Professor of Law, Michael E. Moritz College of Law, The Ohio State University. For helpful feedback on project design and framing, thanks are due to Kent Barnett, Tara Leigh Grove, John Harrison, Michael Heise, Daniel Hemel, Randy Kozel, Nancy Leong, James Lindgren, Janet Moore, Jeff Pojanowski, Richard Re, Dan Tokaji, and Amanda Tyler. For helpful comments on earlier drafts, thanks are due to Ian Farrell, Andy Hessick, Christine Hurt, RonNell Andersen Jones, Gillian Metzger, David Moore, Paul Stancil, Peter Strauss, Mark Seidenfeld, Lisa Grow Sun, and participants at the AALS New Voices in Administrative Law Workshop, BYU Law Work-in-Progress Series, J. Reuben Clark Law Society Faculty Conference, and Rocky Mountain Junior Scholars Forum. Thanks are also due to J. Paul Norton for assistance with statistical analysis and to our research assistants Andrew Bean, Jade Bollinger, Trevor Crowley, Roman Harper, James Heilpern, Bryant Hendriksen, Ellysa Hendriksen, Andrew Henley, Steven Hilton, Neal Hoopes, Jarom Jones, Austin Martineau, Perris Nelson, Robert Patterson, Rebecca Perez, Richard Storms, Mark Tyler, Morgen Weeks, and Lani Wilson for help with case analysis. This study was funded in part by The Ohio State University Center for Interdisciplinary Law and Policy Studies and an internal research grant from Brigham Young University.

Abstract

The first rule of administrative law is that discretion can be dangerous. Although discretion is often used for its intended purposes, scholars of the regulatory process understand from both theory and experience that unintended consequences sometimes result. This is one reason why the Supreme Court is cautious when it comes to agency discretion. After decades of preventing agencies from acting in arbitrary or even self-interested ways, the modern Court has developed a fairly sophisticated understanding of the risks and rewards of discretion and why it is essential to pay attention to incentives.

That is, unless the Supreme Court is addressing judicial discretion. Then, its sophistication all too often gets tossed aside. Qualified immunity is a perfect example. In Pearson v. Callahan, the Court granted judges confronting novel civil-rights claims maximalist discretion whether to decide constitutional questions for the public’s benefit. The intent behind this new discretion is sound: flexibility allows judges to balance constitutional avoidance versus constitutional stagnation in light of case-specific factors. What the Court forgot, however, is that discretion can also have unintended consequences.

This Article addresses perhaps the most serious of these unintended consequences: strategic behavior by judges. While the Court recognizes that federal agencies may have incentives to use discretion in strategic ways, neither the Justices nor scholars have considered the strategic considerations that can influence a judge’s discretionary decision to clearly establish constitutional rights. The potential for strategic behavior is especially sharp, moreover, when discretion to decide constitutional questions is combined with discretion to issue unpublished, nonprecedential opinions. To illustrate this danger, this Article examines real-world judicial decisionmaking. Reviewing over 800 published and unpublished circuit decisions, this Article identifies significant “panel effects”: politically unified panels are more likely to exercise discretion either to find no constitutional violation, for “all Republican” panels, or to recognize new constitutional rights, for “all Democratic” panels. The decision to publish also may be used strategically. For instance, one in five decisions recognizing new constitutional rights is unpublished. This potential for strategic behavior—as in the administrative law context—counsels in favor of reform.

Introduction

The “perennial question” in administrative law, now-Justice Elena Kagan reminds us, is “how to ensure appropriate control of agency discretion.” 1Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2254 (2001). See also Elizabeth Magill, Foreword: Agency Self-Regulation, 77 Geo. Wash. L. Rev. 859, 859 (2009) (“Discretion is at the center of most accounts of bureaucracy.”). The danger is that although discretion can be and, indeed, usually is used for the public’s benefit, 2See, e.g., Geoffrey C. Shaw, H. L. A. Hart’s Lost Essay: Discretion and the Legal Process School, 127 Harv. L. Rev. 666, 668 (2013). it can also serve self-interested ends 3See, e.g., Rachel E. Barkow, The Ascent of the Administrative State and the Demise of Mercy, 121 Harv. L. Rev. 1332, 1334 (2008) (“The expansion of the administrative state has showcased the dangers associated with the exercise of discretion.”). —for instance by allowing regulators to make their own lives easier. 4See, e.g., NetworkIP, LLC v. FCC, 548 F.3d 116, 127 (D.C. Cir. 2008) (“Complainants the agency ‘likes’ can be excused, while ‘difficult’ defendants can find themselves drawing the short straw.”); Lisa Schultz Bressman, Beyond Accountability: Arbitrariness and Legitimacy in the Administrative State, 78 N.Y.U. L. Rev. 461, 472–73 (2003) (“[A]gency discretion [may be used] to satisfy private or selfish interests at public expense.”). In other words, discretion carries with it the potential for strategic behavior by using flexibility intended for one purpose to achieve another, more self-serving purpose. The upshot of that realization, especially when combined with the reality that sometimes agencies behave arbitrarily, is that today’s Supreme Court often doubts whether agencies can police themselves. Invoking the presumption of judicial review, for example, Justice Kagan has recently explained that discretion is too important to leave in an agency’s “hands alone.” 5Mach Mining, LLC v. EEOC, 135 S. Ct. 1645, 1652 (2015). But see Nicholas Bagley, The Puzzling Presumption of Reviewability, 127 Harv. L. Rev. 1285, 1287 (2014) (questioning the presumption). This sort of sophisticated understanding of discretion is key to administrative law. 6See, e.g., Felix Frankfurter, The Task of Administrative Law, 75 U. Pa. L. Rev. 614, 619 (1927).

The Supreme Court, however, has a blind spot: judicial discretion. When it comes to agencies, the Justices frequently recognize both discretion’s benefits and costs and the risk that agencies will behave strategically. But when it comes to judges, the Court may put its skepticism aside. 7See, e.g., Thomas W. Merrill, The Mead Doctrine: Rules and Standards, Meta-Rules and Meta-Standards, 54 Admin. L. Rev. 807, 821 (2002) (observing the “danger of lower court discretion”); cf. Kathryn A. Watts, Constraining Certiorari Using Administrative Law Principles, 160 U. Pa. L. Rev. 1, 4 (2011) (noting the dangers of Supreme Court discretion). Both agencies and courts, however, exercise discretionary power, 8See Margaret H. Lemos, The Other Delegate: Judicially Administered Statutes and the Nondelegation Doctrine, 81 S. Cal. L. Rev. 405, 407–08 (2008). and both face complex incentives regarding how to use that power. 9See, e.g., Jennifer Nou, Agency Self-Insulation Under Presidential Review, 126 Harv. L. Rev. 1755, 1759–60 (2013) (discussing agency incentives); Lawrence B. Solum, The Positive Foundations of Formalism: False Necessity and American Legal Realism, 127 Harv. L. Rev. 2464, 2494–95 (2014) (book review) (discussing judicial incentives). Courts, moreover, like agencies, can issue decisions that profoundly affect the lives of individuals and the public, especially since “[a] certain degree of discretion, and thus of lawmaking, inheres in most executive or judicial action.” 10Whitman v. Am. Trucking Ass’ns, Inc., 531 U.S. 457, 475 (2001) (emphasis added).

Perhaps the best example of the Court’s blind spot is qualified immunity. In 2009, the Court in Pearson v. Callahan granted judges maximalist discretion in these cases: whenever someone seeks damages from many types of government officials based on a novel constitutional claim, the judges deciding the claim can choose—in their “sound discretion”—whether to decide the constitutional question or instead to simply dismiss it as barred by qualified immunity. 11Pearson v. Callahan, 555 U.S. 223, 236–37 (2009). The intended purpose of this discretion is to empower judges to consider case-specific factors, such as whether an argument is well briefed, 12See id. at 239 (noting that “woefully inadequate” briefs foster “bad decisionmaking”). before resolving a constitutional question. This new discretion can be, and no doubt often is, used in socially optimal ways. But it also has a dark side: such discretion empowers judges to act in “strategic” ways (i.e., ways that may benefit judges), especially when the discretionary power to recognize new constitutional rights is combined with the discretionary power to issue unpublished, nonprecedential opinions.

For instance, as explained in this Article, judges may choose to exercise their discretion to clarify constitutional doctrine for reasons that have nothing to do with, say, the objective quality of the briefing but rather because doing so is consistent with ideological preferences. Or judges may decline to clarify constitutional doctrine because the panel disagrees on how to do so, and it is unpleasant for judges to write and respond to dissents. These sorts of strategic uses of discretion may allow judges to avoid criticism, sidestep controversy, and advance ideological preferences, even though such outcomes are not the purpose of Pearson discretion. Needless to say, the Supreme Court did not intend to enable strategic behavior on the part of judges. But if administrative law teaches us anything, it teaches that when it comes to discretion, unintended consequences are real.

To appreciate the danger of discretion in qualified immunity cases, one must step back and understand qualified immunity and the procedural puzzle it creates. It is no secret that the Supreme Court pays close attention to suits against police officers and other government actors—especially when officers may be on the hook for damages. Nor is the reason secret: the Court’s worry “is essentially the same as that in First Amendment overbreadth cases—that the prospect of civil liability will induce timidity and caution in the exercise of government powers that generally operate to the public good.” 13John C. Jeffries, Jr., The Liability Rule for Constitutional Torts, 99 Va. L. Rev. 207, 244 (2013). But see Joanna C. Schwartz, Police Indemnification, 89 N.Y.U. L. Rev. 885, 952–60 (2014) (questioning this “chill” argument regarding whether the risk of civil liability makes government officials too cautious).

Perhaps nowhere is the Justices’ wariness about chilling government conduct more pronounced than with qualified immunity. Few disagree that officers who knowingly violate the Constitution should be punished—including by paying damages. Yet the threat of liability for unknowing violations may cause officers to flinch from the “discharge of their duties,” to the detriment of innocent third parties who depend on state protection. 14Harlow v. Fitzgerald, 457 U.S. 800, 814 (1982). Hence, it is now blackletter law that “officials [need] breathing room to make reasonable but mistaken judgments.” 15Messerschmidt v. Millender, 132 S. Ct. 1235, 1244 (2012). And this “reasonable mistake” principle is no platitude. By their own account, “[b]ecause of the importance of qualified immunity ‘to society as a whole,’” the Justices give special weight to these cases in deciding whether to grant certiorari. 16City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1774 n.3 (2015) (quoting Harlow, 457 U.S. at 814).

Qualified immunity, however, is also controversial. 17See, e.g., John Paul Stevens, Six Amendments: How and Why We Should Change the Constitution 106 (2014) (urging that qualified immunity be banned by constitutional amendment); Stephen R. Reinhardt, The Demise of Habeas Corpus and the Rise of Qualified Immunity: The Court’s Ever Increasing Limitations on the Development and Enforcement of Constitutional Rights and Some Particularly Unfortunate Consequences, 113 Mich. L. Rev. 1219, 1244 (2015) (denouncing qualified immunity for “limiting the ability of all persons to receive the protections of the Constitution”). Even if an officer has violated the Constitution, she cannot be personally liable unless the rights were “clearly established” at the time. For example, even though strip-searching students can be unconstitutional, the officials who conducted a strip-search in Safford Unified School District v. Redding did not have to pay damages because the law was not “sufficiently clear.” 18557 U.S. 364, 379 (2009). Likewise, last year in San Francisco v. Sheehan, the Court awarded immunity to police officers who shot a mentally-ill woman; the Court reasoned, that even assuming the officers should have accommodated her illness, her rights were not clearly established. 19Sheehan, 135 S. Ct. at 1775. Nor are these cases outliers, 20See id. at 1774 n.3 (citing five separate cases from the preceding three years). especially when the Court’s “shadow docket” is considered. 21William Baude, Foreword: The Supreme Court’s Shadow Docket, 9 N.Y.U. J.L. & Liberty 1, 1 (2015); see also Scott Michelman, Taylor v. Barkes: Summary Reversal Is Part of a Qualified Immunity Trend, SCOTUSblog.com (June 2, 2015, 11:17 AM), http://www.scotusblog.com/2015/06/taylor-v-barkes-summary-reversal-is-part-of-a-qualified-immunity-trend/. Citing Sheehan, for instance, the Court recently summarily reversed the Third Circuit in a case involving “suicide screening” of those in custody. 22Taylor v. Barkes, 135 S. Ct. 2042, 2044 (2015) (per curiam). And even more recently, the Court summarily awarded immunity to an officer who, while standing on an overpass, shot and killed a driver below. 23See Mullenix v. Luna, 136 S. Ct. 305, 312 (2015) (per curiam). At the same time, the Court summarily vacated and remanded another case in light of Mullenix. Pickens v. Aldaba, 136 S. Ct. 479 (2015) (mem.). Notably writing only for herself, Justice Sonia Sotomayor criticized “sanctioning a ‘shoot first, think later’ approach to policing.” 24Mullenix, 136 S. Ct. at 316 (Sotomayor, J., dissenting).

The controversy, however, goes beyond substance; it also includes procedure. 25See James E. Pfander, Resolving the Qualified Immunity Dilemma: Constitutional Tort Claims for Nominal Damages, 111 Colum. L. Rev. 1601, 1604 (2011). Qualified immunity shields an officer from liability whenever the asserted right was not clearly established. This means that a plaintiff seeking money damages must clear two hurdles. She must prove not only that the officer violated her rights, but also that any competent officer would have known at the time that he was doing so. This creates a dilemma. If the “right” was not clearly established, the plaintiff cannot prevail. But then why would a court ever reach the constitutional merits of novel theories? 26See Pearson v. Callahan, 555 U.S. 223, 241 (2009) (discussing “the ‘older, wiser . . . counsel “not to pass . . . constitutionality . . . unless such adjudication is unavoidable”’” (quoting Scott v. Harris, 550 U.S. 372, 388 (2007) (Breyer, J., concurring))). The principle of constitutional avoidance suggests that courts confronting new constitutional claims—such as those involving novel technologies or situations—should simply dismiss them. Yet avoidance has a problem of its own: if courts always do that, novel claims may never be clearly established. Constitutional avoidance may devolve into “constitutional stagnation.” 27Pearson, 555 U.S. at 232.

The Supreme Court has struggled with this procedural puzzle. The solution the Justices have settled on is judicial discretion. In Pearson, the Court unanimously held that judges, exercising their “sound discretion,” are free to decide constitutional questions for the benefit of future litigants or, if they wish, to jump directly to the clearly established prong—for instance where the constitutional briefing is “woefully inadequate.” 28Id. at 236, 239. Since 2009, the Court has exercised this discretion over ten times, 29See, e.g., Wesby v. District of Columbia, 816 F.3d 96, 102 (D.C. Cir. 2016) (Kavanaugh, J., dissenting from the denial of rehearing en banc) (“Indeed, in just the past five years, the Supreme Court has issued 11 decisions reversing federal courts of appeals in qualified immunity cases, including five strongly worded summary reversals.”) (collecting authorities). and lower courts have exercised it thousands of times. Because of the critical role that qualified immunity cases play in explicating constitutional rights, it is unsurprising that much has been written about whether this new discretion strikes the right balance between constitutional avoidance and stagnation. 30See, e.g., Nancy Leong, Improving Rights, 100 Va. L. Rev. 377, 396–409 (2014); Nancy Leong, Rethinking the Order of Battle in Constitutional Torts, 105 Nw. U. L. Rev. 969, 973–77 (2011) [hereinafter Leong, Rethinking the Order of Battle]; Reinhardt, supra note 17, at 1250. Judge Reinhardt, for instance, argues that the Justices badly erred by allowing courts to ever sidestep constitutional questions. 31Reinhardt, supra note 17, at 1249–50 (criticizing Pearson). Justices Kennedy and Scalia, by contrast, have hinted that courts should never decide constitutional claims in these cases. 32Camreta v. Greene, 563 U.S. 692, 726 (2011) (Kennedy, J., dissenting); id. at 714 (Scalia, J., concurring).

Whether Pearson has achieved its intended consequence, in large part, is an empirical question. In the real world, is constitutional stagnation a valid concern? Elsewhere, we have shown that although stagnation fears are overstated (since judges resolve constitutional questions most of the time), it is true that this discretion may diminish recognition of new constitutional rights—at least at the margins. 33Aaron L. Nielson & Christopher J. Walker, The New Qualified Immunity, 89 S. Cal. L. Rev. 1, 6 (2015). Part IV.B provides an overview of the findings from The New Qualified Immunity, which examines post-Pearson constitutional stagnation concerns, disparities across circuits, and a preliminary inquiry into the role of ideology. See also Nancy Leong, A Fresh Look at Qualified Immunity, Jotwell (Dec. 3, 2015), http://courtslaw.jotwell.com/?s=a+fresh+look+at+qualified+immunity (reviewing Nielson & Walker, supra). This Article leverages that data to look beyond how Pearson discretion has been used in the circuit courts to explore theoretically and empirically the more troubling unintended consequence of this discretion: how it encourages strategic judicial behavior. The upshot of this empirical finding is that even on the Supreme Court’s own “avoidance balanced against stagnation” terms, it is debatable whether the Justices’ approach to judicial discretion in qualified immunity cases is sound.

This Article, however, addresses an unintended consequence of this new discretion: strategic behavior by judges. 34See, e.g., Jack M. Beermann, Qualified Immunity and Constitutional Avoidance, 2009 Sup. Ct. Rev. 139, 143 (noting the risk of strategic behavior). At bottom, the Court has empowered judges to decide whether to resolve constitutional questions without imposing safeguards on how judges use that discretionary power. 35See id. at 142. Given the nation’s experience with administrative law, this development should raise red flags. In creating this discretion, after all, the Justices appear to believe that lower-court judges will clarify constitutional law only in certain classes of cases—such as when “there would be little if any conservation of judicial resources” by not resolving the constitutional merits 36Pearson v. Callahan, 555 U.S. 223, 236 (2009). —while jumping to the clearly established prong only in other classes of cases, including when “the constitutional question is so factbound that the decision provides little guidance for future cases.” 37Id. at 237. No doubt such legitimate concerns play a major role in how judges opt to exercise discretion. But are they all that go through a judge’s mind?

Almost certainly not. Discretion, after all, carries with it the opportunity for strategic behavior. For instance, confronted with a Fourth Amendment suit for damages, judges may choose not to reach the merits even though they are well argued and important if doing so would splinter the panel. Because intra-panel disagreements are unpleasant, judges presumably prefer to avoid them, all else being equal. 38See, e.g., Frank B. Cross, Collegial Ideology in the Courts, 103 Nw. U. L. Rev. 1399, 1399–1400 (2009) (“Collegiality and ideology inevitably conflict.”). The panel thus may conclude it is easier to just dismiss the claim as not clearly established, thereby trading law clarification, a public good, for better personal relationships, a private good. While there can be sound reasons not to decide constitutional questions, there also can be bad reasons—i.e., reasons that benefit judges but not the public. Discretion empowers judges to act for both types of reasons.

The prospect of strategic behavior is more pronounced when combined with the power to issue nonprecedential, “unpublished” opinions. 39See, e.g., Lee Epstein & Gary King, The Rules of Inference, 69 U. Chi. L. Rev. 1, 107–08 (2002) (describing unpublished opinions). To date, few scholars or judges have considered the strategic possibilities that arise when a discretionary power to issue nonprecedential opinions is exercised in qualified immunity cases, much less what happens when such discretion is combined with discretion to avoid constitutional questions altogether. Yet flexibility to issue unpublished opinions is, if anything, especially important when it comes to qualified immunity. Because damages can only be awarded if the law was clearly established, the power to decide cases without clearly establishing any law is significant.

These theoretical risks are alarming for a host of reasons. For instance, they suggest not only that “constitutional stagnation” may be real, which itself would be problematic, but also that sometimes it may not happen by accident. Instead, certain judges may be avoiding the constitutional merits on purpose, even in cases in which, per the factors announced in Pearson, the constitutional merits should be decided. 40Nielson & Walker, supra note 33, at 23–24. Moreover, these theoretical risks suggest that how constitutional questions are decided—when they are decided—also may not be by accident. Judges may elect to use their discretion to decide constitutional questions, not based on the neutral characteristics of the case (e.g., whether the question is important and well briefed), but instead based on the ideological homogeneity of the panel. 41See infra Part IV.C.1. Likewise, there is a risk that judges sometimes may use their discretion not to benefit the public but instead to insulate decisions from further review by the en banc court or the Supreme Court. Whether Pearson is sound on its own terms is debatable. But when these potential unintended consequences are considered, Pearson discretion becomes much more disquieting and its ultimate net value becomes much less apparent, at least without further revision by the Justices.

This Article, however, does more than speculate about these theoretical risks. Drawing on data from over 800 published and unpublished circuit court decisions comprising nearly 1500 constitutional claims, this Article suggests that these risks of strategic immunity may be manifesting themselves. For instance, the data suggest that when it comes to constitutional litigation, the ideological composition of a three-judge panel matters. Panels composed unanimously of judges appointed by a Republican President are more likely to exercise this new form of discretion to reach constitutional questions and, when they do, are more likely to find no constitutional violation. 42See id. Conversely, panels entirely appointed by a Democratic President are more likely than any other type of panel to recognize a new constitutional right. And judges appointed by a Democratic President behave differently when they write, rather than join, an opinion. On mixed panels, however, there are no statistically significant differences in how the panels behave along these dimensions—regardless of which party is in the majority on the panel. In other words, when panels are ideologically divided, pronounced differences disappear, which itself could suggest strategic behavior via a collegial concurrence or even a majority compromise to avoid dissent.

The discretionary power not to publish decisions also may reflect strategic behavior. Perhaps most strikingly, one in five decisions recognizing a new constitutional right is not published 43See infra note 276 and accompanying text. —meaning that the panel’s decision may not even create liability in future qualified immunity cases. Moreover, judges appointed by a Republican President seem more likely to deny qualified immunity in unpublished decisions. 44See infra Part IV.C.1. Finally, the unpublished decisions reveal a number of divided opinions where strategic compromise may be taking place. 45See id. After all, a dissent from an opinion is some indication that the majority opinion should have been published. Yet those decisions often are not published, potentially suggesting that compromise or fear of further review—either by the circuit en banc or the Supreme Court—may have led to its nonprecedential status.

These findings counsel in favor of once more revising the procedures for qualified immunity. Elsewhere, we have urged requiring judges to provide reasons when exercising discretion in qualified immunity cases. 46See Nielson & Walker, supra note 33, at 61. The key insight from this Article—that such discretion may be used strategically—further supports a reason-giving requirement. The duty to give reasons, moreover, should extend to the choice not to publish a qualified immunity decision. Finally, in deciding whether to grant certiorari, the Supreme Court should give more weight to panel composition and less weight to whether an opinion is published. Although these solutions will not prevent all strategic behavior, they can help minimize it in a cost-justified way. Indeed, because of its wariness of agency discretion, the Court employs similar devices in the context of administrative law to keep regulatory discretion in check. The Court would do well to apply that same skepticism to this species of judicial discretion. 47Cf. Rachel E. Barkow, Institutional Design and the Policing of Prosecutors: Lessons from Administrative Law, 61 Stan. L. Rev. 869, 871 (2009).

This Article proceeds as follows. Part I sketches how qualified immunity works and explains the broad discretion judges enjoy when it comes to clarifying constitutional law. Part II then focuses on administrative law, explaining why judges and scholars have come to be cautious about agency discretion. Part III, in turn, applies those regulatory insights to the qualified immunity context to show how and why judges might use their discretion for strategic purposes. Part IV then presents the Article’s empirical methodology and findings. Finally, Part V explains how qualified immunity’s procedural rules should be revised to mitigate these unintended consequences.

I. The New Qualified Immunity

The zigzagging story of qualified immunity’s procedural puzzle has been told before. 48See Nielson & Walker, supra note 33, at 15–23. Here, we briefly summarize the puzzle. We first address the substantive test for qualified immunity—whether a federal right was violated and whether that federal right was “clearly established” at the time of violation. Then, we discuss the procedure that judges use to apply that substantive test—whether judges must first decide whether the federal right was violated, or whether they can simply dismiss a claim because it was not clearly established, assuming such a right exists. Both the substantive test for qualified immunity and the procedural steps to apply that substantive test have generated controversy. We conclude by addressing an important but often overlooked aspect of the qualified immunity framework: the role of unpublished opinions.

The key takeaway is that, in Pearson, the Justices—driven by constitutional avoidance on one hand but constitutional stagnation on the other—opted to create a regime of maximalist discretion for judges. The intended result of this discretionary regime is to allow lower courts to clarify constitutional rights when it is appropriate based on certain characteristics of particular cases (e.g., it is an important question with broad applicability that has been well briefed), while giving them flexibility to avoid constitutional questions when such characteristics are absent. 49Pearson v. Callahan, 555 U.S. 223, 236–37 (2009).

A. The Substance of Qualified Immunity

Qualified immunity has rightly been dubbed “the most important doctrine in the law of constitutional torts.” 50John C. Jeffries, Jr., What’s Wrong with Qualified Immunity?, 62 Fla. L. Rev. 851, 851–52 (2010). This immunity exists at the pivot point between two principles. Rights generally have remedies, 51See, e.g., Marbury v. Madison, 5 U.S. (1 Cranch) 137, 163 (1803) (“[This is] a government of laws, and not of men. It will certainly cease to deserve this high appellation, if the laws furnish no remedy for the violation of a vested legal right.”). But see Erwin Chemerinsky, Federal Jurisdiction § 7.1, at 421–25 (Wolters Kluwer Law & Bus. 6th ed. 2012) (noting limits on this principle). but the threat of liability has behavioral consequences that can be problematic. 52See, e.g., United States v. Alvarez, 132 S. Ct. 2537, 2555 (2012) (Breyer, J., concurring in the judgment) (explaining the “chilling” effect liability can have). Qualified immunity attempts to reconcile, or at least accommodate, these principles by declaring that an official can be liable for violating constitutional rights only if those rights were “clearly established” at the time of their violation.

Needless to say, whether qualified immunity ought to exist is controversial. 53See, e.g., Reinhardt, supra note 17, at 1244; Jeffries, supra note 13, at 241 & n.121 (collecting authorities). No one is pleased when constitutional rights are violated and the victim goes uncompensated. The Supreme Court is mindful of that reality but routinely—often without dissent—awards immunity. 54See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1774 n.3 (2015) (collecting authorities). Indeed, despite the fact that certiorari is discretionary, the Court devotes considerable attention to qualified immunity, often to the benefit of officers. 55Id. (explaining that “the Court often corrects lower courts when they wrongly subject individual officers to liability” and collecting authorities). By their own account, the Justices do this because the threat of damages has real-world consequences. 56See Harlow v. Fitzgerald, 457 U.S. 800, 806 (1982) (explaining that the threat of damages can cause “undue interference” in officers’ responsibilities). Officers, fearing personal liability, may hold back rather than act with boldness. 57Id. at 814 (1982). Such diffidence can be dangerous. 58See, e.g., Sheehan, 135 S. Ct. at 1775 (“Reynolds and Holder knew that Sheehan had a weapon and had threatened to use it to kill three people. They also knew that delay could make the situation more dangerous.”); Plumhoff v. Rickard, 134 S. Ct. 2012, 2022 (2014) (“[I]t was Rickard who put Allen in danger by fleeing and refusing to end the chase, and it would be perverse if his disregard for Allen’s safety worked to his benefit.”). Many innocent people depend on the vigorous protection of the state. 59See, e.g., Wilkinson v. Russell, 182 F.3d 89, 99 (2d Cir. 1999) (awarding qualified immunity to officials investigating alleged child sexual abuse).

Nor is vicarious liability a silver bullet. 60This same analysis applies to a regime in which the law imposes strict liability on officers but also provides automatic employer indemnification. Under either regime, the money comes from the same place. See Jeffries, supra note 13, at 232. The idea behind vicarious liability—in particular, strict vicarious liability—is simple: whenever government officials violate constitutional rights, whether clearly established or not, the victim should be able seek compensation from the government itself. This would remedy constitutional violations without subjecting individual officers to liability, in theory curing any concern about “chilled” performance of official duties. But this path is also imperfect, even leaving aside whether the Constitution requires it. Courts may be reluctant to read rights expansively if governments are on the hook for acts that, at the time, seemed reasonable. In close cases, judges may be wary of, for instance, bankrupting a small town. 61See, e.g., Jeffries, supra note 13, at 232, 241 n.122; see also Dave Munday & Glenn Smith, Small Town’s Future Uncertain After ‘Astronomical’ Jury Verdict, Post & Courier (Oct. 16, 2014, 10:00 PM), http://www.postandcourier.com/article/20141016/PC16/141019471/1177 (“A mix of shock, joy and uncertainty buzzed in this tiny, rural community Thursday as residents contemplated how the town might pay a court judgment so large it could fund the local budget for the next 162 years.”); Bruce Schreiner, Kentucky City Files for Bankruptcy to Address Judgment—$11.4M Verdict 4 Times Hillview’s Budget, Lexington Herald-Leader, Aug. 22, 2015, at A5 (“A suburban city near Louisville has filed for bankruptcy, crippled by an $11.4 million jury verdict—nearly four times the city’s budget . . . .”). Judges know that money spent on verdicts is money not spent on schools. 62See Pfander, supra note 25, at 1615 (citing, inter alia, John C. Jeffries, Jr., The Right-Remedy Gap in Constitutional Law, 109 Yale L.J. 87, 98–105 (1999)).

The doctrine of qualified immunity tries to split the difference between chilling official action and leaving citizens without remedies. Today, the substantive test for qualified immunity comes from Harlow v. Fitzgerald, 63457 U.S. 800, 818 (1982). which “shields federal and state officials from money damages unless a plaintiff pleads facts showing (1) that the official violated a statutory or constitutional right, and (2) that the right was ‘clearly established’ at the time of the challenged conduct.” 64Aschroft v. al-Kidd, 563 U.S. 731, 735 (2011) (citing Harlow, 457 U.S. at 818). Needless to say, this standard for establishing qualified immunity varies with the procedural posture of a case; a plaintiff’s burden is different at the motion-to-dismiss stage than at the summary-judgment or trial stage. See, e.g., Saucier v. Katz, 533 U.S. 194, 201 (2001). As to the latter prong, “[a] Government official’s conduct violates clearly established law when, at the time of the challenged conduct, ‘[t]he contours of [a] right [are] sufficiently clear’ that every ‘reasonable official would have understood that what he is doing violates that right.’” 65Al-Kidd, 563 U.S. at 741 (quoting Anderson v. Creighton, 483 U.S. 635, 640 (1987)). While courts “do not require a case directly on point,” they do require that “existing precedent must have placed the statutory or constitutional question beyond debate.” 66Id.

Thus, for instance, even though no case specifically forbade handcuffing a prisoner to a hitching post for hours, such conduct still violated the prisoner’s clearly established rights because the guards had “fair warning” that such conduct was unacceptable. 67Hope v. Pelzer, 536 U.S. 730, 741 (2002). Just how similar another case must be to the one at hand for a right to be “clearly established” is a difficult question 68See, e.g., Michael S. Catlett, Clearly Not Established: Decisional Law and the Qualified Immunity Doctrine, 47 Ariz. L. Rev. 1031, 1034 (2005) (“Although this inquiry sounds somewhat straightforward, in practice it has created a number of recurring issues.”); Diana Hassel, Excessive Reasonableness, 43 Ind. L. Rev. 117 (2009) (discussing qualified immunity in excessive force cases). —though the Court is clear that judges should not rely on principles set at too high a level of abstraction. 69See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1775–76 (2015) (“Qualified immunity is no immunity at all if ‘clearly established’ law can simply be defined as the right to be free from unreasonable searches and seizures.”). Moreover, the “clearly established” analysis is concerned with objective reasonableness, not an officer’s subjective belief, 70Harlow v. Fitzgerald, 457 U.S. 800, 818 (1982). thereby allowing judges to “resolv[e] immunity questions at the earliest possible stage in litigation” and thus often avoiding the burdens of discovery. 71Pearson v. Callahan, 555 U.S. 223, 232 (2009) (quoting Hunter v. Bryant, 502 U.S. 224, 227 (1991)); see also id. at 231 (“[T]he ‘driving force’ behind creation of the qualified immunity doctrine was a desire to ensure that ‘insubstantial claims’ . . . [will] be resolved prior to discovery.”).

B. The Procedure of Qualified Immunity

Qualified immunity also presents a procedural puzzle. Absent extraordinary fact patterns, novel claims are not clearly established. That is why they are novel. So doesn’t that mean that a court, confronted with a novel claim, should simply dismiss it on qualified immunity’s second prong as not clearly established without ever reaching whether the officer violated the Constitution? Courts generally refuse to decide constitutional questions that are irrelevant to a judgment. After all, “the ‘cardinal principle of judicial restraint’ is that ‘if it is not necessary to decide more, it is necessary not to decide more.’” 72Morse v. Frederick, 551 U.S. 393, 431 (2007) (Breyer, J., concurring in the judgment in part and dissenting in part) (quoting PDK Labs., Inc. v. DEA, 362 F.3d 786, 799 (D.C. Cir. 2004) (Roberts, J., concurring in part and concurring in judgment)); see also Ashwander v. Tenn. Valley Auth., 297 U.S. 288, 347 (1936) (Brandeis, J., concurring); Burton v. United States, 196 U.S. 283, 295 (1905); Ex parte Randolph, 20 F. Cas. 242, 254 (C.C.D. Va. 1833) (No. 11,558) (Marshall, J.).

Why not apply the same “cardinal principle” in qualified immunity cases? Resolving these sorts of “constitutional questions [requires considerable] judicial resources.” 73Morse, 551 U.S. at 430 (Breyer, J., concurring in the judgment in part and dissenting in part). Busy judges do not have a lot of time to spend on unnecessary questions, 74Nielson & Walker, supra note 33, at 14 (quoting Nancy Leong, The Saucier Qualified Immunity Experiment: An Empirical Analysis, 36 Pepp. L. Rev. 667, 680–81 (2009)). especially if they are pressed by criminal trials and sentencing, which should be resolved first. 75See, e.g., Fed. R. Crim. P. 50 (“Scheduling preference must be given to criminal proceedings as far as practicable.”); cf. 28 U.S.C. § 2102 (2012) (“Criminal cases . . . shall have priority, on the docket of the Supreme Court, over all cases except cases to which the United States is a party and such other cases as the court may decide to be of public importance.”). Given the large number of civil-rights cases for damages on federal dockets, if courts were forced to decide constitutional questions in all of them, other important work may suffer.

Despite its virtues, however, constitutional avoidance also has downsides, especially when it comes to qualified immunity. Indeed, leapfrogging the constitutional question could “stunt the development of the law and allow government officials to violate constitutional rights with impunity.” 76Beermann, supra note 34, at 149. “Because a great deal of constitutional litigation occurs in cases subject to qualified immunity,” as we have previously explained, “many rights potentially might never be clearly established should a court ‘[simply] skip ahead to the question whether the law clearly established that the officer’s conduct was unlawful in the circumstances of the case.’” 77Nielson & Walker, supra note 33, at 12 (quoting Saucier v. Katz, 533 U.S. 194, 201 (2001)).

Of course, constitutional questions often are resolved in other contexts, such as suits against municipalities or for injunctive relief to stop an ongoing constitutional wrong. 78Id. Likewise, criminal defendants can attempt to suppress evidence via the exclusionary rule, 79Id. and the Constitution can also be used as a substantive defense. 80See, e.g., Steffel v. Thompson, 415 U.S. 452, 482–83 (1974) (Rehnquist, J., concurring) (“[T]he defendant . . . is able to present his case . . . to preserve the defendant’s constitutional rights.”); Wisconsin v. Yoder, 406 U.S. 205, 234 (1972) (recognizing that defendants may raise the First or Fourth Amendments as a defense). Even so, these options are not perfect substitutes for a civil lawsuit for damages. 81Nielson & Walker, supra note 33, at 12. For instance, there is no theory of respondeat superior liability against municipalities. 82See, e.g., Connick v. Thompson, 563 U.S. 51, 54 (2011) (holding that a “failure to train” theory requires more than a single violation of Brady v. Maryland, 373 U.S. 83 (1963), even if the consequence was that an innocent man sat on death row for nearly two decades). Likewise, because of Article III’s standing requirement, suits for injunctive relief often require something akin to an ongoing violation, which does not capture, say, an isolated case of police brutality. 83See, e.g., City of Los Angeles v. Lyons, 461 U.S. 95, 111 (1983). The exclusionary rule, similarly, is a poor fit for many sorts of claims that do not have a tight nexus with criminal investigations. Many First Amendment rights might fall in this category. Moreover, as the Supreme Court cuts back on the exclusionary rule, 84See, e.g., Hudson v. Michigan, 547 U.S. 586, 591 (2006) (limiting the rule’s application to “where its remedial objectives are thought most efficaciously served” (quoting United States v. Calandra, 414 U.S. 338, 348 (1974))). the Court also renders it a less perfect substitute for a damages suit. 85See Pamela S. Karlan, Shoe-Horning, Shell Games, and Enforcing Constitutional Rights in the Twenty-First Century, 78 UMKC L. Rev. 875, 887–88 (2010). Finally, constitutional defenses require prosecutors (with discretion of their own) to bring cases and judges to decide them; neither is inevitable, especially given plea-bargaining. 86See Rachel E. Barkow, Institutional Design and the Policing of Prosecutors: Lessons from Administrative Law, 61 Stan. L. Rev. 869, 871 (2009).

But what can be done, on one hand, to prevent constitutional stagnation without, on the other, overworking judges? There is no great answer. In Saucier v. Katz, the Court declared that “[i]n a suit against an officer for an alleged violation of a constitutional right, the requisites of a qualified immunity defense must be considered in proper sequence.” 87Saucier v. Katz, 533 U.S. 194, 200 (2001). First, a court must decide whether “in the light most favorable to the party asserting the injury, do the facts alleged show the officer’s conduct violated a constitutional right?” 88Id. at 201. And “[i]f no constitutional right would have been violated were the allegations established, there is no necessity for further inquiries concerning qualified immunity,” but “if a violation could be made out on a favorable view of the parties’ submissions, the next, sequential step is to ask whether the right was clearly established.” 89Id. Otherwise, the public “might be deprived of . . . explanation were a court simply to skip ahead to the question whether the law clearly established that the officer’s conduct was unlawful in the circumstances of the case.” 90Id.

Saucier, however, was short lived. From “essentially the moment the Court mandated the two-step sequence, a diverse coalition of Justices, lower-court judges, and scholars began criticizing it.” 91Nielson & Walker, supra note 33, at 17. Within a handful of years, for instance, Justice Stevens called Saucier an “unwise judge-made rule,” 92Bunting v. Mellen, 541 U.S. 1019 (Stevens, J., respecting the denial of certiorari). and Justice Breyer pleaded to “end the failed Saucier experiment now.” 93Morse v. Frederick, 551 U.S. 393, 432 (2007) (Breyer, J., concurring in the judgment in part and dissenting in part).

The result was that just eight years after announcing Saucier, the Justices unanimously overruled it in Pearson. Writing for the Court, Justice Alito explained that the Saucier procedural sequence, although “often beneficial” to “prevent constitutional stagnation,” in some cases could “result[] in a substantial expenditure of scarce judicial resources on difficult questions,” including “cases in which it is plain that a constitutional right is not clearly established but far from obvious whether in fact there is such a right. District courts and courts of appeals with heavy caseloads are often understandably unenthusiastic about what may seem to be an essentially academic exercise.” 94Pearson v. Callahan, 555 U.S. 223, 232, 236–37 (2009). Hence, the Court held that judges “should be permitted to exercise their sound discretion in deciding which of the two prongs of the qualified immunity analysis should be addressed first in light of the circumstances in the particular case at hand.” 95Id. at 236. To help guide that “sound discretion,” moreover, the Court announced factors that counsel in favor 96The factors in favor of deciding a constitutional question include whether: (1) “there would be little if any conservation of judicial resources” in not deciding it; (2) it would “be difficult to decide whether a right is clearly established without deciding precisely what the existing constitutional right happens to be”; (3) the “two-step procedure” would “promote[] the development of constitutional precedent”; and (4) the question is not one that “frequently arise[s] in cases in which a qualified immunity defense is unavailable.” Id. and against 97The factors against deciding a constitutional question include whether: (1) doing so would result in “substantial expenditure of scarce judicial resources”; (2) “it is plain that a constitutional right is not clearly established but far from obvious whether in fact there is such a right”; (3) the court has a “heavy caseload[]”; (4) additional “litigation of constitutional issues [would] waste[] the parties’ resources”; (5) the “question is so factbound that the decision provides little guidance for future cases”; (6) “the question will soon be decided by a higher court”; (7) “an uncertain interpretation of state law” is involved; (8) “‘the parties have provided very few facts to define and limit any holding’ on the constitutional question”; (9) “the briefing of constitutional questions is woefully inadequate”; (10) it would be “hard for affected parties to obtain appellate review of constitutional decisions that may have a serious prospective effect on their operations”; and (11) in a particular case, “constitutional avoidance” is unusually important. Id. at 236–42. deciding a constitutional question.

Since Pearson, the Court has repeatedly exercised this discretion, sometimes opting to decide the constitutional question 98See, e.g., Plumhoff v. Rickard, 134 S. Ct. 2012, 2020 (2014). and sometimes not. 99See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1775 (2015) (declining to exercise Pearson discretion because the constitutional briefing was inadequate). In short, the Court believes that discretion in qualified immunity cases will not lead to undue constitutional stagnation, but will enable a better judicial process. The Court is mindful of constitutional atrophy, but also recognizes that there are categories of cases in which it is imprudent to clarify constitutional law. The intended purpose of this new discretion is thus to strike the best balance between those interests.

C. Unpublished Qualified Immunity Decisions

The story of the Court’s substantive decision in Harlow and its procedural evolution from Saucier to Pearson has been told before. There is another discretionary feature of modern litigation, however, that plays an important role in qualified immunity cases that has not yet received enough attention: unpublished decisions. In deciding cases governed by qualified immunity, judges not only exercise discretion whether to reach the merits of constitutional claims, but they also exercise discretion whether to “publish” their opinions, i.e., give them precedential effect on future panels. 100See Charles Alan Wright et al., Federal Practice and Procedure § 3978.10, at 382 (4th ed. 2008) (discussing unpublished opinions); Deborah Jones Merritt & James J. Brudney, Stalking Secret Law: What Predicts Publication in the United States Courts of Appeal, 54 Vand. L. Rev. 71, 72 (2001) (discussing the prevalence of unpublished opinions). “Unpublished” opinions, like published ones, are publicly available. 101Epstein & King, supra note 39, at 107 n.335 (citing Richard S. Arnold, Unpublished Opinions: A Comment, 1 J. App. Prac. & Process 219, 220 (1999)). But unlike published opinions, they do not carry the same weight in our common law system of precedent. 102See Mata v. Lynch, 135 S. Ct. 2150, 2155 n.3 (2015) (noting that unpublished opinions in some circuits have “no precedential force”); Unpublished Opinion, Black’s Law Dictionary (10th ed. 2014) (explaining that an “unpublished opinion” generally “is considered binding only on the parties to the particular case in which it is issued”).

This discretionary power not to publish a decision, moreover, is especially critical for qualified immunity because the substantive test for immunity turns on whether the right at issue was clearly established at the time of its violation. This means it matters how a right becomes clearly established. Whether a past opinion was designated as precedential can be key to that determination. The debate over published versus unpublished opinions thus is especially relevant in the context of constitutional torts. In this Article, we do not spell out the full history of and controversies surrounding unpublished opinions; others have plowed that ground. 103See, e.g., Wright et al., supra note 100, at § 3978.10, at 382–86 (discussing the history of court rules allowing citation to unpublished opinions); Caleb E. Mason, An Aesthetic Defense of the Nonprecedential Opinion: The Easy Cases Debate in the Wake of the 2007 Amendments to the Federal Rules of Appellate Procedure, 55 UCLA L. Rev. 643, 646 (2008) (collecting authorities relating to “[t]he jurisprudential, political, equitable, ethical, and constitutional issues raised by the use of [unpublished opinions]” (footnotes omitted)). But it is important to briefly explain what these opinions are and why they are important.

Federal circuit courts are common law courts in that their decisions generally have precedential weight on other panels within the circuit. 104See Richard B. Cappalli, The Common Law’s Case Against Non-Precedential Opinions, 76 S. Cal. L. Rev. 755, 759 (2003). As Richard Cappalli has explained, traditionally “the court is obliged to decide under law, which means that the court arrives at a principle of general application that it would be willing to apply in future comparable cases. This decisional rule is called the ‘rule of the case’ or ‘holding of the case.’” 105Id. at 761 (footnotes omitted). Once a holding has been reached, the doctrine of stare decisis applies, meaning that the court’s “decisional rule, whether implicit or express, is to be applied in future cases . . . with comparable facts even if a later . . . court is dissatisfied with the rule.” 106Id. at 762. Indeed, “the most important characteristic of a collegial appellate court is careful attention, respect, and adherence to precedent.” 107UC Health v. NLRB, 803 F.3d 669, 685 (D.C. Cir. 2015) (Silberman, J., dissenting).

In a perfect world, every appellate panel would write a detailed opinion in every case. Such a well-explained opinion could then operate with stare decisis force on all subsequent cases raising the same issue. Unfortunately, in our imperfect world, “[a]ppellate judges continue to labor under the weight of tens of thousands of appeals every year,” many of which do not raise novel questions. 108Boyce F. Martin, Jr., In Defense of Unpublished Opinions, 60 Ohio St. L.J. 177, 177–78 (1999). Indeed, “[a] large proportion of the opinions that have been coming out of American courts add essentially nothing to the corpus of the law. They are of interest and significance to the parties only.” 109Id. at 178 (quoting Paul D. Carrington et al., Justice on Appeal 35 (1976)). It would impose a real burden on the judiciary if circuit courts were required to write full opinions on the tens of thousands of cases filed each year. 110See Philip Nichols, Jr., Selective Publication of Opinions: One Judge’s View, 35 Am. U. L. Rev. 909, 927 (1986) (noting “the absurdity of destroying forests to distribute masses of prolix and repetitious material”).

To deal with this problem, federal appellate courts, primarily in the 1960s and 1970s, began experimenting with “unpublished opinions” that could be used to dispose of appeals quickly. These unpublished opinions often lacked a full articulation of the facts of the case and all the relevant legal principles, and would not receive ordinary precedential effect on future panels. As the Federal Judicial Center explained in 1973, “the judicial time and effort essential for the development of an opinion to be published for posterity and widely distributed is necessarily greater than that sufficient to enable the judge to provide a statement so that the parties can understand the reasons for the decision.” 111Advisory Council on Appellate Justice, FJC Research Series No. 73-2, Standards for Publication of Judicial Opinions 3 (1973). Accordingly, the Federal Rules of Appellate Procedure now recognize that circuit courts may allow panels to designate opinions as “unpublished,” “not for publication,” “non-precedential,” “or the like.” 112See Fed. R. App. P. 32.1.

Since the 1970s, each circuit has developed a scheme for issuing such opinions. 113Martin, supra note 108, at 184–85. Today, moreover, the overwhelming majority of circuit court decisions are unpublished, 114See Joseph L. Gerken, A Librarian’s Guide to Unpublished Judicial Opinions, 96 L. Libr. J. 475, 478 (2004) (noting that about 80% of appellate decisions are unpublished). though there are different percentages across the circuits. 115See Marin K. Levy, The Mechanics of Federal Appeals: Uniformity and Case Management in the Circuit Courts, 61 Duke L.J. 315, 361–64 (2011). Although the circuits have always considered somewhat different factors when deciding whether to publish an opinion, 116See, e.g., William L. Reynolds & William M. Richman, The Non-Precedential Precedent—Limited Publication and No-Citation Rules in the United States Courts of Appeals, 78 Colum. L. Rev. 1167, 1176 (1978). they often boil down to whether publication would be in the “public interest”—a highly discretionary standard. 117See Plumley v. Austin, 135 S. Ct. 828, 831 (2015) (Thomas, J., dissenting from the denial of certiorari) (discussing Fourth Circuit standards for publication, one of which is whether the case “involve[s] a legal issue of continuing public interest”); Epstein & King, supra note 39, at 107 & nn.335–36 (explaining that while judges should publish “opinions that are ‘of general precedential value,’” in practice the standard “is sufficiently vague to permit circuit court judges to publish or not as they see fit”).

Whether unpublished decisions are good policy is a difficult question. On one hand, imagine a lawsuit where the law is clear and it is impossible to imagine how the case’s resolution could be relevant to any other parties in any other cases. 118See, e.g., Mahogany v. La. State Supreme Court, 262 Fed. App’x 636, 637 (5th Cir. 2008) (“Just as one cannot sue . . . the color teal, the number thirteen, or the weather, a complaint filed against a statute fails to state a claim upon which relief can be granted.”). In a case like that, it makes little sense to require busy judges to go through the trouble of preparing a full opinion of the sort that fills law school casebooks. 119See Alex Kozinski & Stephen Reinhardt, Please Don’t Cite This!: Why We Don’t Allow Citation to Unpublished Dispositions, Cal. Law., June 2000, at 43 (“After carefully reviewing the briefs and record, we can succinctly explain who won, who lost, and why. We need not state the facts, as the parties already know them; nor need we announce a rule general enough to apply to future cases.”). Suffice it to say, many cases do not raise novel issues. 120See, e.g., Hurt v. Soc. Sec. Admin., 544 F.3d 308, 308–09 (D.C. Cir. 2008) (“In just the last couple of years, Hurt has sued the Declaration of Independence, Black’s Law Dictionary, the United Nations, agencies of the District of Columbia and the Federal Government, and various courts and their officers. Hurt has claimed the existence of state supreme courts violates the Eighth Amendment, requested the Secret Service and the President’s Cabinet be declared unconstitutional, and demanded the deportation of a Spanish-speaking government employee. Nor are the slights Hurt suffered mere glancing blows; he routinely demands trillions of dollars in damages.” (citations omitted)) (listing forty-four dockets).

On the other hand, unpublished opinions may have downsides: the common law method depends, in large part, on there being precedential cases to cite. 121Cappalli, supra note 104, at 759. Fewer published opinions may lead to less law articulation. Indeed, discomfort with unpublished opinions prompted reform in 2006. 122See Fed. R. App. P. 32.1 & advisory committee’s notes. Building on a recommendation of then-Judge Samuel Alito (who chaired the Federal Advisory Committee on the Federal Rules of Appellate Procedure), 123Patrick J. Schiltz, Much Ado About Little: Explaining the Sturm und Drang over the Citation of Unpublished Opinions, 62 Wash. & Lee L. Rev. 1429, 1444–45 (2005). the Federal Rules now make clear that although courts can continue to issue nonprecedential, unpublished opinions, they “may not prohibit or restrict the citation of federal judicial opinions, orders, judgments, or other written dispositions . . . .” 124Fed. R. App. P. 32.1. This is a compromise; some advocated that unpublished opinions be abolished as contrary to common law principles, whereas others urged that the status quo be retained. 125See Schiltz, supra note 123, at 1470, 1475. The current Federal Rules split the difference while preserving the discretionary power of courts to issue opinions without precedential effect. 126Id. at 1473 (“[T]he Advisory Committee Note to Rule 32.1 went out of its way to be clear that the rule did not forbid any court to issue an unpublished opinion, dictate the circumstances under which a court may choose to designate an opinion as unpublished, or imply anything about what effect a court must give to one of its unpublished opinions.”).

The similarity between the discretion to issue unpublished opinions and the discretion not to resolve constitutional questions in qualified immunity cases is obvious. In both situations, the Court realizes that there are competing values and innumerable fact-specific circumstances in which those values must be reconciled. When it comes to unpublished opinions, for instance, one value is that the public benefits from precedential, well-articulated opinions, while another value is preventing judges from being overloaded by writing full opinions in cases that do not present sufficiently novel legal questions. The same sorts of values apply in qualified immunity cases, where courts must balance the need to prevent constitutional stagnation against the reality that some cases are simply not good vehicles to clearly establish new constitutional rules. Accordingly, federal appellate courts are given discretion to determine—based on the particular characteristics of the case before them—whether it is in the public interest that an opinion be published, just as they are given discretion to leapfrog merits questions in some qualified immunity cases, again depending on the particular circumstances of the case before them.

For the reasons explained above, whether an opinion is published is especially important in the context of qualified immunity. Liability depends on whether a right was clearly established, meaning it matters a great deal whether an unpublished opinion can clearly establish a federal right. Unfortunately, the Supreme Court has never definitively declared what weight, if any, should be given to unpublished opinions in determining whether a right is clearly established. Many circuits hold, however, that unpublished decisions cannot be used to show that an asserted right is clearly established, or, at a minimum, that they are of little value in making that showing. 127See David R. Cleveland, Clear as Mud: How the Uncertain Precedential Status of Unpublished Opinions Muddles Qualified Immunity Determinations, 65 U. Miami L. Rev. 45, 64 (2010) (“[C]ourts have taken radically different interpretations of what sources may clearly establish the law.”). And even in circuits that purport to consider unpublished opinions in determining whether a right is clearly established, 128See id. at 71–74 (describing the Ninth Circuit as the one circuit that “has unequivocally stated a willingness to use unpublished opinions” to show clearly established law, and the Third and Sixth Circuits as “seem[ing] to allow the use of unpublished opinions in clearly establishing the law”). such opinions no doubt carry less weight, at least de facto, since there often is little to no analysis in them. Because whether immunity applies is highly fact-dependent, 129See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1776 (2015) (rejecting the application of legal principles at a high level of abstraction). it follows that unpublished opinions that do not provide the facts necessarily are less able to clearly establish the law. 130Cf. Jackler v. Byrne, 658 F.3d 225, 244 (2d Cir. 2011) (“[S]uch orders, being summary, frequently do not set out the factual background of the case in enough detail to disclose whether its facts are sufficiently similar to those of a subsequent unrelated case to make our summary ruling applicable to the new case.”).

II. Administrative Law and the Danger of Discretion

It is useful to contrast the Supreme Court’s embrace of judicial discretion in qualified immunity cases with how it views administrative discretion in regulatory cases. When it comes to agency discretion, judges and scholars often fear that flexibility may be put to arbitrary ends. 131See, e.g., David S. Rubenstein, “Relative Checks”: Towards Optimal Control of Administrative Power, 51 Wm. & Mary L. Rev. 2169, 2188–89 (2010). Especially relevant here, many worry about “strategic” behavior by agencies, fearing that agencies may use the discretion intended for certain public ends to instead accomplish private ends that benefit the agency. 132See, e.g., Jonathan R. Macey, Administrative Agency Obsolescence and Interest Group Formation: A Case Study of the SEC at Sixty, 15 Cardozo L. Rev. 909, 913 (1994) (“[O]ver time, all bureaucracies will substitute private, bureaucratic objectives for the public objectives that characterized their origination.” (citing Anthony Downs, Inside Bureaucracy (1967))). In fact, from the Administrative Procedure Act (APA) 133See Aaron L. Nielson, In Defense of Formal Rulemaking, 75 Ohio St. L.J. 237, 243–44 (2014) (explaining that in 1946, the Chair of the Senate Judiciary Committee “proclaimed that the APA ‘cut down the “cult of discretion” so far as federal law is concerned’” (quoting George B. Shepherd, Fierce Compromise: The Administrative Procedure Act Emerges from New Deal Politics, 90 Nw. U. L. Rev. 1557, 1666 (1996))). to White House review 134See Cass R. Sunstein, The Office of Information and Regulatory Affairs: Myths and Realities, 126 Harv. L. Rev. 1838, 1845–46 (2013) (discussing OIRA review). to foundational doctrines like the duty of contemporaneous explanation established by SEC v. Chenery Corp., 135318 U.S. 80, 87 (1943); see also Kevin M. Stack, The Constitutional Foundations of Chenery, 116 Yale L.J. 952, 958 (2007) (explaining that Chenery helps “prevent arbitrariness in the agency’s exercise of its discretion”). much of administrative law as we know it is addressed toward “the dangers associated with the exercise of discretion.” 136Barkow, supra note 3, at 1334. Yet when it comes to judicial discretion, at least in this context, such wariness is absent.

A. The Value and Danger of Administrative Discretion

Why is discretion so central to administrative law? Because it is both useful and slippery. In a world in which anything can happen, it is challenging for legislatures to craft rules that make sense in all applications. Even a dutiful legislator often will fail to envision idiosyncratic situations. 137See Charles H. Koch, Jr., Judicial Review of Administrative Discretion, 54 Geo. Wash. L. Rev. 469, 472 n.8 (1986) (“Rules alone, untempered by discretion, cannot cope with the complexities of modern government and of modern justice.” (quoting K. Davis, Discretionary Justice: A Preliminary Inquiry 25 (1969))). It is also difficult to predict all the ways that innovative actors may engage in troublesome behavior, especially because behavior is dynamic, changing in response to enforcement. It thus can be attractive to regulate consequences, rather than actions. 138See, e.g., Bus. Elecs. Corp. v. Sharp Elecs. Corp., 485 U.S. 717, 731 (1988) (“The term ‘restraint of trade’ in the statute, like the term at common law, refers not to a particular list of agreements, but to a particular economic consequence, which may be produced by quite different sorts of agreements in varying times and circumstances.”). So what can legislatures do?

Often, the answer is to bestow discretion on some delegate. 139The delegate, of course, need not be an agency; it can very well be, and often is, a court. See Lemos, supra note 8, at 429; cf. Shaw, supra note 2, at 669 (“[D]iscretion is a necessary component of any legal system, because society’s ability to regulate the future is inherently limited by imperfect information and an imperfect understanding of aims.”). Discretion is valuable because it allows for individualized treatment and can be targeted toward consequences rather than specific actions. Accordingly, the Supreme Court proclaimed in Burlington Truck Lines, Inc. v. United States that “discretion is the lifeblood of the administrative process.” 140371 U.S. 156, 167 (1962); see also Sidney Shapiro, Elizabeth Fisher & Wendy Wagner, The Enlightenment of Administrative Law: Looking Inside the Agency for Legitimacy, 47 Wake Forest L. Rev. 463, 463–64 (2012) (defending this view of administrative law and “the legitimacy of administrative discretion”). This pro-discretion view—associated with New Dealers like James Landis 141Aaron L. Nielson, Visualizing Change in Administrative Law, 49 Ga. L. Rev 757, 768–69 (2015) (citing James M. Landis, The Administrative Process 70 (1966)). —was key to the creation of much of the modern administrative state. 142See Kagan, supra note 1, at 2261 (explaining that Landis thought that expertise could “solv[e] the problem of administrative discretion”).

Unfortunately, discretion comes at a cost, as New Deal excesses came to show. 143See Wong Yang Sung v. McGrath, 339 U.S. 33, 36–37, 41 (1950) (recognizing the “evils” associated with the “[m]ultiplication of federal administrative agencies and expansion of their functions” over the preceding “past half-century”); Kagan, supra note 1, at 2261 (“[Landis’s] idea today seems almost quaint, and even then it provoked strong opposition.”); Shaw, supra note 2, at 668 (“The problem of discretion became more pressing after the New Deal.”). The very flexibility that allows for individualization and adaptation can be used in arbitrary, biased, or self-interested ways. 144See Mark Seidenfeld, A Civic Republican Justification for the Bureaucratic State, 105 Harv. L. Rev. 1511, 1563 (1992) (“[R]egulators will often pursue their own private interests. This pursuit partially explains the failure of the New Deal agency model, which naively assumed that regulators would diligently implement statutory goals.”). For instance, in terms of strategic behavior, an agency may use discretion to “shirk” unpleasant duties 145Id. or to expand its authority to gain increased rents from regulated parties. An example of this is the worry that agency officials may be tempted to increase regulatory complexity so that their bureaucratic knowledge will become more valuable—and lucrative—for later private sector work. 146See id. at 1564 (“In addition to shirking, officials’ self-interest also creates incentives for them to augment their regulatory authority. Greater power allows regulators to increase monopoly rents, which they can then trade to interest groups in return for personal benefits such as future jobs or freedom from criticism.” (footnotes omitted)). Hence, “lifeblood” or not, the Supreme Court in Burlington Truck Lines also warned that “unless we make the requirements for administrative action strict and demanding, expertise, the strength of modern government, can become a monster which rules with no practical limits . . . .” 147Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 167 (1962) (quoting New York v. United States, 342 U.S. 882, 884 (1951) (Douglas, J., dissenting)). After all, “[a]bsolute discretion, like corruption, marks the beginning of the end of liberty.” 148New York, 342 U.S. at 884 (Douglas, J., dissenting).

So here is the central dilemma of administrative law: How to gain the benefits of discretion without surrendering “rule of law” values like legitimacy, consistency, and public spiritedness? 149See Shaw, supra note 2, at 668 (“Rule by arbitrary choice is not rule by law, and thus the lasting legacy of legal realism became a challenge: explain how legal indeterminacy can be reconciled with the rule of law.”). For over a century, that question has dogged the field. In the United States, unchecked discretion arguably reached its zenith during the New Deal, which prompted a recalibration. The resulting “fierce compromise” was the APA, 150Shepherd, supra note 133, at 1557, 1681. which accepted discretion’s utility but also recognized it can be “put to arbitrary and biased use.” 151Wong Yang Sung v. McGrath, 339 U.S. 33, 37 (1950).

B. Administrative Law’s Meta Cost-Benefit Approach

To mitigate the dangers of discretion without snuffing it out altogether, administrative law has developed a “meta” cost-benefit analysis. Agencies are given a great deal of flexibility, but often those same agencies must learn to work within safeguards that “inherently diminish the discretion and flexibility of government officials. Indeed, that is exactly what [they are] designed to do.” 152John H. Jackson, Perspective on the Jurisprudence of International Trade: Costs and Benefits of Legal Procedures in the United States, 82 Mich. L. Rev. 1570, 1578–79 (1984); see also Shapiro et al., supra note 140, at 464 (“[P]rocedure after procedure has been added in a vain effort to eliminate discretion.”). In other words, because discretion is both valuable and dangerous, balances must be struck, depending on whether a particular type of discretion is valuable enough to outweigh the dangers associated with it and, if not, whether cost-justified safeguards can be crafted to mitigate the danger.

All of this is to say that context matters; not all discretion is created equal. For instance, an agency’s choice not to employ notice-and-comment rulemaking when reinterpreting a regulation through an interpretative rule 153See Perez v. Mortg. Bankers Ass’n, 135 S. Ct. 1199, 1203–04 (2015) (upholding discretion without notice-and-comment procedures). is very different from discretion “to regulate the entire economy on the basis of no more precise a standard than stimulating the economy by assuring ‘fair competition.’” 154Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 474 (2001) (discussing unanimous rejection of substantive policy-making discretion as unconstitutional in A.L.A. Schechter Poultry Corp. v. United States, 295 U.S. 495 (1935)). Similarly, because prisons can be hostile, wardens have latitude to set and enforce prison policies. But despite prison dangers, no one thinks a warden’s discretion should be limitless. 155See, e.g., Holt v. Hobbs, 135 S. Ct. 853, 863 (2015) (“We readily agree that the Department has a compelling interest in staunching the flow of contraband into and within its facilities, but the argument that this interest would be seriously compromised by allowing an inmate to grow a ½-inch beard is hard to take seriously.”).

To be sure, the costs of safeguards can exceed their value. This is often true if a generalist judge is the safeguard. For instance, some discretionary acts—like an agency decision not to bring an enforcement action—are effectively not reviewable at all. 156See Heckler v. Chaney, 470 U.S. 821, 831 (1985). Why? Because, on one hand, not enforcing the law is thought to be less a threat to liberty than enforcing the law in an arbitrary way, 157See id. at 832 (“[W]hen an agency refuses to act it generally does not exercise its coercive power over an individual’s liberty or property rights, and thus does not infringe upon areas that courts often are called upon to protect.”). and because, on the other hand, judges are poorly equipped to prioritize the competing resource demands that an agency faces. 158See id. at 831–32 (“The agency is far better equipped than the courts to deal with the many variables involved in the proper ordering of its priorities.”). The dangers of discretion in this context, thus, are not deemed weighty enough to justify the costs of a safeguard to constrain that discretion. Similarly, in informal rulemaking, an agency effectively has absolute discretion to provide procedural protections greater than the baseline requirements of the APA. 159See Vt. Yankee Nuclear Power Corp. v. NRDC, Inc., 435 U.S. 519, 544–45 (1978). This makes sense; the danger that agencies will choose to provide too many procedural protections—and so incur greater burdens for themselves—seems remote.

Other sorts of decisions, however, are reviewed with greater skepticism—particularly where the costs of discretion are substantial. For instance, agencies have a great deal of discretion about the content of legislative rules, but they generally do not have discretion to promulgate such rules with retroactive effect. 160Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208 (1988). The power to regulate retroactively is often too dangerous for agencies to wield because the threat of upsetting expectations is real while the benefits of rulemaking usually can be obtained through prospective rules. 161See id. (“The power to require readjustments for the past is drastic.” (quoting Brimstone R.R. & Canal Co. v. United States, 276 U.S. 104, 122 (1928))).

Similarly, “hard look” review—which, again, generally applies when an agency has exercised discretion in formulating final agency action—requires that “the agency must examine the relevant data and articulate a satisfactory explanation for its action including a ‘rational connection between the facts found and the choice made.’” 162Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962)). This safeguard reflects the realization that policymaking is valuable, but the more power an agency has over policy, the more hazardous it may become. Rather than forbid such discretion—and so lose the good—the Supreme Court crafted a safeguard for it: hard-look review.

C. The Supreme Court’s Concerns About Agency Discretion

The Supreme Court is not blind to the fact that agencies can lose their way. In the real world, the Justices recognize that bureaucrats are not always technocratic; like anyone else, they make mistakes. Thus, despite discretion’s utility, caution is sometimes called for.

Precedent reflects this reality. For instance, the Justices in recent years have reiterated that agency action is presumptively reviewable. A good example is Mach Mining, LLC v. EEOC, decided last year by the unanimous Court. 163135 S. Ct. 1645, 1652 (2015). There, the question was whether the Equal Employment Opportunity Commission’s (EEOC) pre-suit conciliation efforts were subject to judicial review. 164Id. at 1649. The agency opposed such review. 165Id. at 1650. According to the statute, the EEOC can bring suit whenever it “‘has been unable to secure from the respondent a conciliation agreement acceptable to the Commission’ itself.” 166Id. at 1649 (quoting 42 U.S.C. § 2000e–5(f)(1)). Likewise, the statute forbids any evidence regarding conciliation efforts from being introduced in court. 167Id. (citing 42 U.S.C. § 2000e–5(b)). From these textual clues, the EEOC argued that its discretion was not reviewable. But Justice Kagan, writing for the Court, would have none of it, concluding that “the Government takes its observation about discretion too far.” 168Id. at 1652. The Court stressed that there must be a role for judicial review. Otherwise, “the Commission’s compliance with the law would rest in the Commission’s hands alone. We need not doubt the EEOC’s trustworthiness, or its fidelity to law, to shy away from that result.” 169Id. at 1652–53; see also id. (“We need only know—and know that Congress knows—that legal lapses and violations occur, and especially so when they have no consequence.”); EEOC v. Propak Logistics, Inc., 746 F.3d 145, 155–56 (4th Cir. 2014) (Wilkinson, J., concurring) (explaining that to avoid encouraging “sub-optimal agency behavior,” judges must remember the “danger that those inside a public bureaucracy, armed with significant resources, authority, and discretion, may become gradually numb as to how their actions affect those outside” the agency).

Similar discretion-wary analysis drove the Court’s 2012 decision in Sackett v. EPA, which held—also unanimously—that a compliance order under the Clean Water Act was subject to immediate judicial review. 170132 S. Ct. 1367, 1374 (2012). The Sackett Court’s skepticism of discretion in this context was palpable. As Justice Scalia explained, “[t]he APA’s presumption of judicial review is a repudiation of the principle that efficiency of regulation conquers all. And there is no reason to think that the Clean Water Act was uniquely designed to enable the strong-arming of regulated parties into ‘voluntary compliance’ without the opportunity for judicial review . . . .” 171Id. Justice Alito, writing separately, went further: “The position taken in this case by the Federal Government—a position that the Court now squarely rejects—would have put the property rights of ordinary Americans entirely at the mercy of Environmental Protection Agency . . . employees.” 172Id. at 1375 (Alito, J., concurring). The Court adopted a similar analysis in Army Corps of Engineers v. Hawkes Co. in 2016. 173136 S. Ct. 1807, 1816 (2016) (holding that jurisdictional determinations are also subject to immediate review).

Sometimes, moreover, the Court does question an agency’s trustworthiness. Rhetoric surrounding Bowles v. Seminole Rock & Sand Co. deference is a good example. Seminole Rock commands courts to defer to an agency’s interpretation of its own ambiguous regulations. 174325 U.S. 410, 414 (1945). Although this sounds innocent enough, in recent years it has been attacked—repeatedly—on the ground that it creates incentives for agencies to use their discretion in nefarious ways. 175See, e.g., Perez v. Mortg. Bankers Ass’n, 135 S. Ct. 1199, 1210–13, 1217–20 (2015) (separate opinions of Scalia, Thomas, and Alito, J.J.); Decker v. Nw. Envtl. Def. Ctr., 133 S. Ct. 1326, 1338 (2013) (Roberts, C.J., concurring); id. at 1339–42 (Scalia, J., concurring in part and dissenting in part). Indeed, a majority opinion of the Court, while noting the “advantages” of such deference, has stressed the “risk that agencies will promulgate vague and open-ended regulations that they can later interpret as they see fit.” 176Christopher v. SmithKline Beecham Corp., 132 S. Ct. 2156, 2168 (2012). In other words, the Justices do not always trust that agencies, left to their own devices, will unfailingly use discretion for the public good.

III. Strategic Immunity: A Theoretical Examination

The Supreme Court’s increasingly sophisticated view of discretion is nowhere to be found when it comes to qualified immunity. Yet judges, like agency officials, may also use their discretion in suboptimal ways. As Judge Richard Arnold reminded us, judges are “human” too. 177Arnold, supra note 101, at 223. The Court’s presumptive wariness of bureaucratic discretion accordingly should extend to judicial discretion, at least in this context. Indeed, given the Court’s supervisory powers over the lower federal courts, which it does not have over administrative agencies, it makes sense for the Court to be especially conscious of the costs and benefits of discretion when it comes to the procedures used by judges. After all, unlike agency discretion where the Justices must live—more or less—within the confines of the APA, 178But see Gillian E. Metzger, Embracing Administrative Common Law, 80 Geo. Wash. L. Rev. 1293, 1293, 1295 (2012) (“[M]uch of administrative law is really administrative common law, [though] statutes exert some constraining force on judicial creativity.”). when it comes to the federal bench, the Supreme Court has a freer hand. 179See, e.g., Dickerson v. United States, 530 U.S. 428, 437 (2000) (“The law in this area is clear. This Court has supervisory authority over the federal courts, and we may use that authority to prescribe rules of evidence and procedure that are binding in those tribunals.”); see also James E. Pfander, One Supreme Court: Supremacy, Inferiority, and the Judicial Power of the United States 148–49 (2009) (explaining that constitutional structure suggests the Court has a supreme or supervisory role over the lower courts); Tara Leigh Grove, The Structural Case for Vertical Maximalism, 95 Cornell L. Rev. 1, 4 (2009) (arguing that the vertical structure of the judiciary “preserve[s] [the] ‘supreme’ role” of the Court). But see Amy Coney Barrett, The Supervisory Power of the Supreme Court, 106 Colum. L. Rev. 324, 360–66 (2006) (casting some doubt on this power).

In this Part, we set out the theoretical risks posed by qualified immunity’s procedural discretion, especially when combined with discretion not to publish a decision. Our point is not that discretion is a bad thing or that it should be done away with. To the contrary, discretion is valuable and undoubtedly is often used well. But even valuable tools can be abused. To see why, it is helpful to understand just how complex qualified immunity can be, and how that complexity creates opportunities for strategic behavior. To illustrate, this Part begins with a simple example of a qualified immunity case.

A. An Example: How Does a Judge Vote?

Imagine a morning in Ohio. A family is preparing to send a child to school. All of a sudden, a Federal Bureau of Investigation (FBI) task force surrounds the home. Agents suspect a member of the family—the father—has information about a fugitive. But they have no warrant to arrest him or search the home, and the circumstances are not exigent. The agents pound on the door and threaten that unless the family lets them in, the father will go to jail. The agents then enter without permission, but no one orders them to leave. Afterwards, the family sues the agents for violating the Fourth Amendment by breaching the curtilage of their property and entering their home. The agents rejoin that they were allowed to approach as they did and that they had consent to enter because the father began opening the door, and no one told them to get out.

Now imagine you are a judge on the Sixth Circuit confronted with this fact pattern. You have just heard oral argument. After studying the relevant Fourth Amendment cases and principles, you believe that you know the answer to the constitutional questions. How do you vote?

This question might appear to be straightforward, especially under the simple rule from Saucier. If you conclude that the officers violated the Constitution as to both claims—the curtilage claim and the entry claim—you would first hold that the Constitution was violated, and then you would decide whether the rights were clearly established. If, however, you conclude that only one claim is constitutionally cognizable, then you would vote to recognize that claim, state that the other claim is meritless, and then vote on whether the right violated was clearly established. And if you conclude that neither claim actually states a constitutional right, you would vote to dismiss both on the merits. In other words, one of these options would reflect your vote:

nielson-walker-tbl1

Pearson, however, adds a layer of complexity by allowing lower courts to exercise discretion whether to follow the Saucier sequence. 180555 U.S. 223, 236 (2009). Thus under Pearson, you would have to decide whether to resolve the constitutional question, at least in some scenarios. 181See id. at 236–37. This would lead to one of the following votes:

nielson-walker-tbl2

nielson-walker-tbl3

The situation, however, is more complicated still because neither Saucier nor Pearson considered an additional discretionary question: should the opinion be published? That question at least doubles the number of potential outcomes. 182It is possible to partially publish opinions. See, e.g., Adena Reg’l Med. Ctr. v. Leavitt, 527 F.3d 176, 180 n.* (D.C. Cir. 2008) (“The Hospitals’ other arguments are sufficiently lacking in merit as not to warrant consideration in a published opinion”). In theory then, a court could publish its resolution of one claim but not of another.

nielson-walker-tbl4

Finally, on a multi-judge panel, there is at least one more level of complexity. Not only must you decide what your preference is with all else being equal, but you must also decide what your preference is given the preferences of the other voting members. For instance, if both of your colleagues believe that there is no violation as to either claim, while you believe the opposite, perhaps you would opt to vote your preferred outcome, even if it means writing a dissent (note, however, that some judges join opinions they do not completely agree with, especially if the vote will not make a difference 183Richard J. Lazarus, The Opinion Assignment Power, Justice Scalia’s Un-Becoming, and UARG’s Unanticipated Cloud over the Clean Air Act, 39 Harv. Envtl. L. Rev. 37, 45 (2015) (“Justices must frequently join final opinions even though they do not agree with every word or nuance.”). ). But what if the difference between views is not so stark? What if everyone agrees that the plaintiffs should lose on both claims, but for different reasons? One judge is not sure whether the rights exist but is confident neither was clearly established; another is sure that the rights do not exist and so could not have been clearly established; and you agree that neither right was clearly established but believe that both exist and, moreover, that the panel should exercise its discretion under Pearson to clearly establish those rights going forward. Each judge could, in theory, vote her own way. Or the panel might simply say “not clearly established” and move on, perhaps in an unpublished opinion. 184See, e.g., Amy E. Sloan, If You Can’t Beat ‘Em, Join ‘Em: A Pragmatic Approach to Nonprecedential Opinions in the Federal Appellate Courts, 86 Neb. L. Rev. 895, 946 n.308 (2008) (“[J]udges may join an opinion when they agree with the result, even if they do not agree with the reasoning, when they know the opinion will be nonprecedential.” (citing Advisory Comm. on Appellate Rules, Minutes of the Fall 2002 Meeting 36–37 (Nov. 18, 2002), http://www.uscourts.gov/rules/Minutes/appll02.pdf)).

So how do you vote? The answer, of course, depends on how strongly you prefer the “let’s use Pearson discretion to clearly establish these rights” option. For instance, if it takes much effort to write separately, you may push the panel towards an outcome you can live with, even if it is not your first preference.

This example illustrates how resolving these cases is complicated. The Supreme Court’s simple model of qualified immunity is sound enough, as far as it goes. No one should dispute that it makes more sense for a court to use its discretion to decide a constitutional question if it is well briefed, important, and unlikely to arise outside of the qualified immunity context than if it is poorly briefed, idiosyncratic, and subject to another form of litigation. 185See Pearson v. Callahan, 555 U.S. 223, 236–37 (2009). But as this example illustrates, the lists of factors in Pearson do not capture the full range of considerations that go through a judge’s mind. In the real world, things are much more complicated.

This particular illustration, for instance, is a stylized version of a real case: Turk v. Comerford, decided by the Sixth Circuit in 2012. 186488 Fed. App’x 933, 935 (6th Cir. 2012). There, the three-judge panel concluded that the FBI’s entry violated clearly established law but granted qualified immunity as to the curtilage claim without reaching the constitutional question. 187Id. The Sixth Circuit, however, made another decision. Despite hearing oral argument and reversing much of the district court’s judgment, the panel decided that its twenty-five-page opinion should be unpublished. 188Id. at 948. It is possible that that particular outcome was the preferred one of all panel members. But it is also possible that the judges opted to use their discretion to reach a result they all could live with. If so, was this compromise socially optimal or, if the panel’s discretion had been constrained, could a socially better result have emerged? Nothing the Supreme Court has said about qualified immunity even begins to address these questions.

B. Considerations That May Affect How Circuit Court Judges Resolve Qualified Immunity Cases

No one has perfectly mapped the utility function for circuit court judges. 189See, e.g., Solum, supra note 9, at 2484–85 (discussing the nuanced factors that must go into a correct judicial utility function). Indeed, there surely is no one such utility function. All judges are different. 190See, e.g., id. at 2489 (“Of course, the relative strength of legal and policy preferences could vary from judge to judge.”). (The same, of course, could be said for agency officials.) Nonetheless, it is still useful to think through the sorts of considerations that are likely to go through a typical judge’s mind.

Our intuition is that judges try to “get the law right,” in that they apply traditional legal principles to reach predictable results. 191Id.; see also Pauline T. Kim, Lower Court Discretion, 82 N.Y.U. L. Rev. 383, 396 (2007) (“The theory that judges single-mindedly pursue their policy preferences is incompatible with the numerous studies concluding that judges shift the nature of their reasoning and the general trend of their decisions in response to Supreme Court precedent.”); Alex Kozinski & Fred Bernstein, Clerkship Politics, 2 Green Bag 57, 59–60 (1998) (noting that ideology’s impact on decisionmaking is overstated). But what other sorts of considerations pass through the minds of circuit court judges deciding qualified immunity cases? This Article does not offer a comprehensive list; the literature on judicial decisionmaking is expansive and conflicting. 192See, e.g., Solum, supra note 9, at 2467 (discussing literature and critiquing Lee Epstein, William M. Landes & Richard A. Posner, The Behavior of Federal Judges: A Theoretical and Empirical Study of Rational Choice (2013)). We do not wade too deeply into this complex and growing field of scholarship. 193See, e.g., Michael A. Bailey & Forrest Maltzman, The Constrained Court: Law, Politics, and the Decisions Justices Make 2 (2011); Frank B. Cross, Decision Making in the U.S. Courts of Appeals 18 (2007); Pauline T. Kim, Deliberation and Strategy on the United States Courts of Appeals: An Empirical Exploration of Panel Effects, 157 U. Pa. L. Rev. 1319 (2009); McNollgast, Politics and the Courts: A Positive Theory of Judicial Doctrine and the Rule of Law, 68 S. Cal. L. Rev. 1631 (1995). Instead, our analysis is meant to be commonsensical. As a thought experiment, what seems plausible? In thinking about how judges exercise their discretion in qualified immunity cases, the following sorts of considerations seem like realistic candidates.

1. Substantive Legal Outcomes. Few issues are more hotly contested than the extent to which ideological preferences determine judicial behavior. In our experience, models woefully miss the mark if they do not account for the fact that judges in good faith try to follow the law. Indeed, as Larry Solum has noted, a “purely ideological model does not fare well” in the real world. 194Solum, supra note 9, at 2487. That said, judges are not indifferent to the ideological outcomes of their decisions, even though they are often willing to rule contrary to them.

2. Future Effects of Precedent. Judges also care about more than just the case before them. In a common law system, they know that precedent matters. They also know other judges will apply their precedent. Accordingly, even if a judge believes she can properly apply a complex rule in the next case, she also knows there is a real chance that whatever rule she crafts will be misapplied (from her perspective) by others. After all, it is not uncommon for judges to apply a precedent in a way that the precedent’s author did not intend. 195See, e.g., Walker v. Tex. Div., Sons of Confederate Veterans, Inc., 135 S. Ct. 2239, 2248–51 (2015) (reading Pleasant Grove City v. Summum, 555 U.S. 460 (2009), more broadly than its author intended); Ashcroft v. Iqbal, 556 U.S. 662, 684 (2009) (reading Bell Atlantic Corp. v. Twombly, 550 U.S. 544 (2007), more broadly than its author intended); SEC v. Chenery Corp., 332 U.S. 194, 207 (1947) (reading SEC v. Chenery Corp., 318 U.S. 80 (1943), more narrowly than its author intended). Hence, if a judge is worried that others will misapply precedent, this may influence how an opinion is written.

3. Prospects of Further Review. Aside from getting the law right, judges worry about being reversed. While this concern is not the only motivation driving behavior (it may not even be a strong one), 196David E. Klein & Robert J. Hume, Fear of Reversal as an Explanation of Lower Court Compliance, 37 Law & Soc’y Rev. 579, 597 (2003) (explaining with empirical support that fear of reversal cannot explain all aspects of law compliance). it is hardly surprising that judges, like anyone else, dislike having their reasoning rejected, especially if reversal imposes reputational harm. 197Mark Herrmann, The Curmudgeon’s Guide to Practicing Law 4 (2006) (“Judges do not like to be reversed. Accordingly, if a precedent contains the implicit threat of reversal, I will use that threat (gently, of course) when I discuss that case.”); Nou, supra note 9, at 1771–72 (explaining, in the context of district courts, that reversals impose “reputational costs”) (collecting authorities). Indeed, just as administrative agencies may try to avoid having their handiwork second-guessed by courts, judges also do not like being reversed. 198Nou, supra note 9, at 1756 (analogizing agencies and courts).

4. Collegiality. It is widely acknowledged that judges care about collegiality on the bench. 199See Virginia A. Hettinger, Stefanie A. Lindquist & Wendy L. Martinek, Judging on a Collegial Court: Influences on Federal Appellate Decision Making 112 (2006). Again, judges are people. For many, it is unpleasant to disagree with colleagues, especially if that disagreement requires more work, such as a dissent. 200See Frank B. Cross, Collegial Ideology in the Courts, 103 Nw. U. L. Rev. 1399, 1399–1400 (2009) (“Collegiality and ideology inevitably conflict. Given the range of ideological preferences and political ‘hot button’ issues facing courts today, one would expect judges to disagree about the resolution of certain legal issues. Because such disagreement puts collegiality and individuality in conflict, any model of judicial decisionmaking must account for the interaction of these potentially opposing factors.” (footnote omitted)); Frank B. Cross & Emerson H. Tiller, Judicial Partisanship and Obedience to Legal Doctrine: Whistleblowing on the Federal Courts of Appeals, 107 Yale L.J. 2155 (1998) (discussing the effect of a “whistleblower” on the court); Solum, supra note 9, at 2471 (“[D]issents will be more likely when the ideological stakes are high, because writing dissents may entail collegiality costs and always involves an effort cost . . . .”). Accordingly, if it is possible for judges to reach a resolution that does not spur animosity across the panel, that path may be attractive.

5. Public Reaction. Finally, there is reason to think that judges, as human beings, enjoy praise and dislike criticism. 201See Aaron L. Nielson, Reflections on the End of the Federal Law Clerk Hiring Plan, 112 Mich. L. Rev. First Impressions 22, 27 (2013) (explaining the “human nature” of judges). Usually, this instinct is irrelevant to adjudication because there is no public attention; the vast majority of cases decided in circuit courts are relevant only to the parties. Judges do not need to think twice about what a reporter is likely to think about most contract cases because no reporter will ever think of it at all. But what if a decision will receive public scrutiny? Allegations of police brutality, for instance, catch people’s attention, especially today.

C. Theoretical Risks of Judicial Discretion

When all of these factors are considered, there is a risk that judges may use their discretion in qualified immunity cases in ways that are, from the public’s perspective, suboptimal. Indeed, they may use discretion in self-serving ways—for instance, by maximizing private advantage.

To be clear, by using the word “strategic,” we do not cast aspersions on the good faith of judges. 202Cf. Mach Mining, LLC v. EEOC, 135 S. Ct. 1645, 1652 (2015) (explaining that the Court “need not doubt the EEOC’s trustworthiness” to worry about discretion). No doubt almost all judges always try to do their best. The same is true for agency officials. Even so, incentives matter, and, in any event, “the standard set for men of good will is even more useful to the venal.” 203New York v. United States, 342 U.S. 882, 884 (1951) (Douglas, J., dissenting). Humans are tempted to act in self-interested ways and can, in complete good faith, rationalize away self-interested behavior. 204See, e.g., Arnold, supra note 101, at 223 (noting even good-faith self-deception). Just as agency officials may come to believe that self-interested acts are, in fact, public-spirited, judges may come to believe that what is good for the bench is good for the bar. That reality may lead to the following types of suboptimal outcomes, which can be grouped into two broad categories.

1. Strategic (Non)Use of Pearson Discretion

An obvious danger is that judges may use their discretion to decline to clarify law when they should clarify it. Imagine a case that would be an ideal candidate for clarifying constitutional law per the Pearson factors: a case in which the constitutional question, although not clearly established, is well briefed, important, and unlikely to arise outside of the qualified immunity context. 205See Plumhoff v. Rickard, 134 S. Ct. 2012, 2020 (2014) (exercising Pearson discretion because those qualities existed). Although the plaintiff’s claim must be dismissed, the panel should nevertheless resolve the constitutional question to provide clarity to the public. After all, applying Pearson’s weighing of the costs and benefits, clarification in a case like this would be the socially-optimal outcome.

But imagine further that the case also involves a controversial question, and the panel is not ideologically homogeneous. If the panel was forced to reach the merits, the law would be clarified (in one direction or the other), but not without prompting a fierce dissent. In such a scenario, it is not unthinkable that the panel might opt simply to dismiss the claim as not clearly established and say nothing more, in effect trading the public good of constitutional clarity for the private benefit of better collegiality. Likewise, if the issue is controversial, the panel may fear reversal. One way to minimize that chance would be not to reach the constitutional merits. By deciding an issue relevant only in a particular case (e.g., the claim was not clearly established), as opposed to announcing a constitutional principle of broad application, the decision would be less certworthy. 206See Sup. Ct. R. 10 (“A petition for a writ of certiorari is rarely granted when the asserted error consists of . . . misapplication of a properly stated rule of law.”).

The inverse is also true. There may be cases in which doctrine should not be clarified, even if discretion exists to do so. For instance, imagine a case in which the constitutional question is not well briefed, 207Cf. City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1774 (2015) (noting poor briefing). and is likely to arise outside of the qualified immunity context, yet is important enough for the panel to care about. Applying Pearson, the panel should not resolve the constitutional question. 208Id. (explaining that although the Court has discretion to decide unbriefed constitutional questions, it is not prudent to do so). Nonetheless, if a judge were to think, “well, you know, we have a pretty good panel,” he may press to resolve the constitutional question once and for all. Such a judge would be trading a sound judicial process, which benefits the public, for an ideologically pleasing result from the perspective of the panel. Such a result seems contrary to how the public—and the Supreme Court—would wish cases to be adjudicated.

2. Strategic (Non)Publication of Immunity Decisions

Judges also might decide not to publish a decision that, from the public’s perspective, should be published. Again, imagine a case that should result in an opinion, per Pearson, that clarifies the law going forward. Yet also imagine that the panel is split on the merits and deciding them in a published opinion that sets the law for the circuit would prompt animosity. One can imagine judges opting to go along with an opinion going one way or the other so long as it is not published.

The idea that judges may use their publication discretion in socially suboptimal ways is not new. Judge Patricia Wald, formerly of the D.C. Circuit, has observed that this discretion “allows for deviousness and abuse. I have seen judges purposely compromise on an unpublished decision incorporating an agreed-upon result in order to avoid a time-consuming public debate about what law controls.” 209Patricia M. Wald, The Rhetoric of Results and the Results of Rhetoric: Judicial Writings, 62 U. Chi. L. Rev. 1371, 1374 (1995). Moreover, Judge Wald noted that she had “even seen wily would-be dissenters go along with a result they do not like so long as it is not elevated to a precedent.” 210Id. Judge Arnold, formerly of the Eighth Circuit, made a similar observation. 211Arnold, supra note 101, at 223 (noting the risk that judges who cannot distinguish a case may opt for “an abbreviated, unpublished opinion, and no one will ever be the wiser”). And Justice Clarence Thomas—no novice when it comes to reading circuit court decisions—has suggested the same. Dissenting from the Court’s denial of certiorari, he included this forceful attack:

True enough, the decision below is unpublished and therefore lacks precedential force in the Fourth Circuit. But that in itself is yet another disturbing aspect of the Fourth Circuit’s decision, and yet another reason to grant review. The Court of Appeals had full briefing and argument on Austin’s claim of judicial vindictiveness. It analyzed the claim in a 39-page opinion written over a dissent. By any standard—and certainly by the Fourth Circuit’s own—this decision should have been published. . . . It is hard to imagine a reason that the Court of Appeals would not have published this opinion except to avoid creating binding law for the Circuit. 212Plumley v. Austin, 135 S. Ct. 828, 831 (2015) (Thomas, J., dissenting from the denial of certiorari) (citations omitted).

Scholars too have observed that judges may “make strategic use of their discretion—publishing opinions that follow Supreme Court precedent . . . and failing to publish those opinions that do not, with the goal of avoiding reversal.” 213Epstein & King, supra note 39, at 108 (footnote omitted). The worry is that unpublished opinions “allow courts to engage in ad hoc decision-making and avoid accountability for so doing,” for instance, by further insulating their decision from review or backtracking from precedent. 214David C. Vladeck & Mitu Gulati, Judicial Triage: Reflections on the Debate over Unpublished Opinions, 62 Wash. & Lee L. Rev. 1667, 1680 (2005). Of course, even if unpublished opinions are not used strategically, they may still be problematic. See, e.g., Scott E. Gant, Missing the Forest for a Tree: Unpublished Opinions and New Federal Rule of Appellate Procedure 32.1, 47 B.C. L. Rev. 705, 735 (2006) (arguing that judges do not know whether an opinion will be important); Penelope Pether, Inequitable Injunctions: The Scandal of Private Judging in the U.S. Courts, 56 Stan. L. Rev. 1435, 1484 (2004) (“Even if there were no credible evidence of misconduct or structural inequality in the operation of unpublication, the lack of transparency it produces would damage the legitimacy of the judicial system . . . .”). But there are benefits too. The issue is whether the costs outweigh the benefits. Indeed, one study showed that “judges made a higher percentage of decisions that agreed with their ideological preferences in published opinions, and a higher percentage of decisions that disagreed with their ideological preferences in unpublished opinions,” though finding no evidence that “judges exploit nonpublication to hide ideologically-driven decisions.” 215Denise M. Keele et al., An Analysis of Ideological Effects in Published Versus Unpublished Judicial Opinions, 6 J. Empirical Legal Stud. 213, 230, 234 (2009) (emphasis added); see also David S. Law, Strategic Judicial Lawmaking: Ideology, Publication, and Asylum Law in the Ninth Circuit, 73 U. Cin. L. Rev. 817, 820 (2005) (explaining that judges may try to “steer the evolution of the law in a particular ideological direction by publishing opinions they like” and offering some empirical data). While some question whether judges hide decisions, 216Alex Kozinski, In Opposition to Proposed Federal Rule of Appellate Procedure 32.1, 51 Fed. Law., June 2004, at 36, 39. many believe that at least some strategic use of unpublished opinions occurs. 217See, e.g., Marla Brooke Tusk, Note, No-Citation Rules as a Prior Restraint on Attorney Speech, 103 Colum. L. Rev. 1202, 1216 (2003) (“There is even some speculation that, because the Supreme Court is less likely to grant certiorari to an appeal from an unpublished opinion, appellate judges may decide controversial cases via unpublished opinions simply to insulate those decisions from Supreme Court review.”).

All of these strategic dangers are especially potent when it comes to qualified immunity because damages are only possible for “clearly established” law. To be clearly established, an earlier case does not need to be factually identical to the subsequent case—indeed, that is impossible. 218See, e.g., Hope v. Pelzer, 536 U.S. 730, 743 (2002) (not requiring “identical” facts). There thus is a sort of penumbral effect around cases that find violations of clearly established rights; those cases clearly establish the law for closely related contexts too. It follows that if a judge wants to clearly establish more law going forward, she could write a broad, published opinion. By contrast, if a judge worries about clearly establishing law, she could write a narrow, unpublished opinion.

Finally, some decisions arguably should be unpublished. For instance, if the law was already clearly established, a decision applying that clearly established law may serve no function going forward for the public. If the panel spends too much time on such a decision—for instance, to “make a point”—it is possible that other work may be shortchanged. Likewise, because precedent has penumbral effects, especially in this context, publishing an opinion may chill behavior because regulated parties are unsure whether a related but arguably distinct practice is covered by the published decision.

* * *

For all these reasons and others, there is a risk that discretion in qualified immunity cases sometimes may be put to strategic ends. We do not argue that this happens all or even most of the time. Nor do we suggest that judges act in bad faith. What we do argue, however, is that, just as in administrative law, discretion in this context may have both intended and unintended consequences. In evaluating whether discretion makes sense, it is necessary to weigh all the costs and benefits and to consider whether safeguards are necessary.

IV. Strategic Immunity: An Empirical Examination

For all the reasons explained in Part III, there is a danger that Pearson discretion, especially when combined with discretion not to publish a decision, may be put to strategic ends. The danger, however, might be more than just theoretical. Using a data set of over 800 cases and nearly 1500 distinct constitutional claims, this Part reviews how civil-rights claims for damages are decided in the real world. Although we cannot say for certain that discretion is used strategically, the data suggest that such use may be occurring.

A. Study Methodology

The purpose of this study is to assess how federal appellate courts are deciding civil-rights cases in which qualified immunity is a defense. To do this, we reviewed every federal appellate decision in Westlaw—including unpublished decisions—issued from 2009 through the end of 2012 that cited Pearson. 219As detailed in The New Qualified Immunity, we focused on 2009 to 2012 because those cases created a data set large enough for analysis; cases nearer in time to Pearson are more likely to cite Pearson instead of circuit-specific precedent; and we wanted to avoid cases that may have come before the Supreme Court while one of us, Nielson, was clerking there. See Nielson & Walker, supra note 33, at 30 n.185. For the purposes of The New Qualified Immunity, the focus on cases citing Pearson made sense as we were trying to assess the effect of Pearson in the lower courts. As discussed below, the exclusive focus on Pearson in this study imposes methodological limitations to assess strategic judicial behavior. After all, an important strategic behavior not captured by the cases reviewed is a court’s decision not to cite Pearson at all when deciding whether to reach constitutional questions in the qualified immunity context. Moreover, we focused on circuit courts because—outside of the Supreme Court—only those courts can clearly establish law. To be sure, there is some uncertainty on this score. See, e.g., Carroll v. Carman, 135 S. Ct. 348, 350 (2014) (declining to decide whether circuit precedent can clearly establish rights). Even so, it is conventionally understood that they can. See, e.g., Ashcroft v. al-Kidd, 563 U.S. 731, 741–42 (2011) (distinguishing district and circuit courts). We then excluded cases that did not involve qualified immunity but that nonetheless mentioned Pearson. We also excluded en banc decisions. In short, we only coded circuit court cases in which:

(1) a plaintiff brought at least one constitutional or federal statutory claim seeking money damages against an individual government-official defendant;

(2) the defendant raised a qualified immunity defense against that claim; and

(3) the court decided the merits of either the constitutional or statutory claim, the qualified immunity claim, or both. 220Leong, supra note 74, at 685–86.

This generated 844 cases. We then further broke down cases to reflect the reality that often a single case presents more than one claim subject to a qualified immunity defense: “Many cases, for instance, involve multiple defendants, multiple claims (by which we mean an independent potential constitutional violation considered by the appellate court, regardless of whether the potential violation was labeled as a separate count in a party’s complaint), or both.” 221Nielson & Walker, supra note 33, at 31. For cases that “involved the same claim brought against multiple defendants,” we listed the claims “separately only where the court reached different results for different defendants.” 222Leong, supra note 74, at 686. The result was 1460 claims. 223The study methodology and its limitations are described in greater detail—and compared to prior empirical studies of qualified immunity—in The New Qualified Immunity. See Nielson & Walker, supra note 33, at 27–33.

B. The New Qualified Immunity Findings

In The New Qualified Immunity, we explored four findings from the cases reviewed: (1) overall findings on the exercise of Pearson discretion and constitutional stagnation; (2) circuit disparities in exercising Pearson discretion; (3) preliminary findings on the potential political disparities in exercising Pearson discretion; and (4) the rate of providing reasons for exercising, or not exercising, Pearson discretion. Each will be briefly summarized in turn.

1. Pearson Discretion and Constitutional Stagnation. We looked at the overall results to understand how circuit courts exercised their Pearson discretion and whether constitutional stagnation concerns raised in the literature are well-founded. 224See id. at 33–38. Roughly half of the time (45.5%) circuit courts exercised their Pearson discretion to reach constitutional questions, whereas a quarter of the time (26.7%) the courts opted just to declare that the right is not clearly established. 225Id. at 33–34 & fig.1. As noted in The New Qualified Immunity, “[w]e label this option ‘not exercising Pearson discretion’ for ease of reference, recognizing that the decision not to reach a constitutional question is still, in a sense, an exercise of discretion provided by Pearson.” Id. at 34. With respect to the remaining claims (27.7%), the court determined it lacked any Pearson discretion because “the constitutional right was clearly established at the time of the violation.” 226Id. at 34 & fig.1. In other words, in those cases, the court denied qualified immunity. When circuits exercised their Pearson discretion to decide constitutional questions, in less than one in ten instances (8.0%) did the court find a constitutional violation (that was not clearly established)—what we call a “pure Saucier” exercise of discretion to recognize new constitutional rights. The rest of the time (92.0%) the court found no constitutional violation. 227Id. at 35 & fig.2.

In comparing these numbers to pre-Pearson empirical studies of qualified immunity, 228See Paul W. Hughes, Not a Failed Experiment: Wilson-Saucier Sequencing and the Articulation of Constitutional Rights, 80 U. Colo. L. Rev. 401, 418–19, 421, 423 tbl.1 (2009); Leong, supra note 74, at 684–91; Colin Rolfs, Comment, Qualified Immunity After Pearson v. Callahan, 59 UCLA L. Rev. 468, 490–97 (2011); Ted Sampsell-Jones & Jenna Yauch, Measuring Pearson in the Circuits, 80 Fordham L. Rev. 623, 638–39 (2011); Greg Sobolski & Matt Steinberg, Note, An Empirical Analysis of Section 1983 Qualified Immunity Actions and Implications of Pearson v. Callahan, 62 Stan. L. Rev. 523, 525 (2010). we concluded “that courts decline to decide constitutional questions at a rate similar to the pre-Saucier period—in around one in four cases, as opposed to less than six percent during the Saucier regime.” 229Nielson & Walker, supra note 33, at 37. The overall rate of reaching constitutional questions accordingly has decreased after Pearson. Moreover, as to constitutional stagnation in the pure Saucier sense, the concern about post-Pearson stagnation appears well-founded. “[A]ll of the post-Pearson studies . . . found that circuit courts found constitutional violations of rights that were not clearly established in 3.6%, 7.9%, and 2.5%, respectively, of the total claims reviewed, whereas the three pre-Pearson studies found rates ranging from 6.5% to 13.9% during the Saucier mandatory sequencing regime. Our findings suggest something has changed.” 230Nielson & Walker, supra note 33, at 37–38 (footnote omitted); see also id. at 37 tbl.1 (comparing all six studies cited supra note 228).

2. Circuit Disparities. The cases reviewed also reveal great disparities among circuits in how courts exercise Pearson discretion. 231See id. at 39–42. We focused on the Fifth, Sixth, and Ninth Circuits—three of the circuits with the largest number of claims in the cases reviewed. For instance, the rates of exercising Pearson discretion in both the Fifth Circuit (57.6%) and the Ninth Circuit (36.0%) are statistically significant (p < 0.05) in comparison with the national circuit average (46.4%)—just in opposite directions. 232Id. at 39–40 & fig.3. The rate in the Sixth Circuit is 47.7%. Id. The same is true of the rates for not exercising Pearson discretion and just finding that any right is not clearly established. When zeroing in on how the circuits exercise Pearson discretion, there are no statistically significant differences between these three circuits and the national average (90.5%) with respect to finding no constitutional violation. But with respect to recognizing new constitutional rights (that are not clearly established), the Fifth Circuit (1.3%), the Sixth Circuit (0.8%), and the Ninth Circuit (16.4%) “all differ dramatically from the national average (9.5%), by a statistically significant margin (p < 0.05).” 233Id. at 40–41 & fig.4. The national average here of 9.5% differs from the overall 8.0% figure provided above because, for purposes of comparing circuits, “the national average is the average of the percentage for each circuit, excluding the D.C. Circuit because it handled far fewer claims (only a dozen) in the sample and seems to be an outlier.” Id. at 39 n.209.

Based on these findings, we concluded that “to the extent the numbers are generalizable, these circuit-by-circuit disparities may reveal a geographic distortion in the development of constitutional law,” suggesting that “constitutional law may develop quite differently in the various circuits, such that the Constitution means something different—and government actors are constrained by different clearly established rights—among the fifty states.” 234Id. at 41–42. Importantly, Pearson’s procedural grant of discretion could have systemic effects in that “those [substantive] differences can be tied at least in part to whether and how the circuits exercise their Pearson discretion differently to decide constitutional questions.” 235Id. at 42.

3. Politics and Pearson. In The New Qualified Immunity, we also conducted a preliminary inquiry into whether the political ideology of a judge affects how the judge exercises Pearson discretion. 236See id. at 43–49. To explore this issue, we used the political party affiliation of the President who appointed the judge and focused on the judge who authored the opinion. With respect to the rate of exercising Pearson discretion to reach constitutional questions, there are no statistically significant differences (p < 0.05) based on party affiliation: “authors nominated by a Democratic president (‘D judges’) and those nominated by a Republican president (‘R judges’) both exercised Pearson discretion 43.3% and 45.0% of the time, respectively.” 237Id. at 45–46 & fig.5. Yet as to not exercising Pearson discretion (just finding a right not clearly established) or finding no discretion to exercise (thus denying qualified immunity), the differences are statistically significant: “R judges were much more likely (27.4% to 18.0%) to decide not to exercise their Pearson discretion,” whereas “D judges were more likely (38.8% to 27.1%) to deny qualified immunity . . . .” 238Id.

Moreover, when exercising Pearson discretion, there are no statistically significant differences (p < 0.05) between D and R judges with respect to finding no constitutional violation, but the same cannot be said of the rate of finding constitutional violations that are not clearly established: “D judges found constitutional violations nearly twice as often as R judges (13.1% and 6.7%, respectively) . . . .” 239Id. at 47–48 & fig.6. Based on these findings, we posited that after Pearson “the development of constitutional law might be shifting away from those with certain substantive constitutional views and toward those with others”—a shift that may “not be due to a change in the substantive views of the federal judiciary at large, but instead due to philosophical inclinations not to reach constitutional questions unnecessarily.” 240Id. at 48. Part IV.C.1 of this Article explores the potential role of politics in the exercise of Pearson discretion in more nuanced ways—looking at each panel member’s vote as well as at potential panel effects on the exercise of discretion.

4. Rate of Reason-Giving. Finally, we documented the dearth of reason-giving by the circuit courts when exercising, or not exercising, their Pearson discretion to reach constitutional questions. 241See id. at 49–51. When courts determined they had discretion to reach the constitutional question, “[i]n less than one in ten of those instances (8.1% or 85 claims) did the court provide any reason for why it had decided to exercise (or not) its Pearson discretion.” 242Id. at 49 (emphasis added). When giving a reason, “courts more often than not (69.4% or 59 claims) included at least one reason expressly provided in the Pearson opinion.” 243Id. at 49–51 & figs.8–9. Among the reasons given that were not explicitly mentioned in the Pearson opinion, “a recurring theme was that the circuit court was just following the district court’s sequencing.” 244Id. at 51.

Moreover, courts were four times more likely to provide a reason when deciding to avoid the constitutional question (15.4%) than when exercising their discretion to decide the constitutional question (3.8%). 245Id. at 49. Perhaps this difference is not a coincidence, as it is consistent with the lopsided provision of reasons against (eleven reasons) and for (four reasons) reaching a constitutional question provided in the Pearson opinion itself. 246See Pearson v. Callahan, 555 U.S. 223, 236–42 (2009); see also supra notes 96–97. Based on these findings we recommended that the Supreme Court “require lower courts—both trial and appellate courts—to give reasons for exercising (or not) their Pearson discretion to reach constitutional questions.” 247Nielson & Walker, supra note 33, at 52; accord Beermann, supra note 34, at 175 (“At a minimum, in light of the strong reasons for reaching the constitutional merits, courts should be required to give reasons for not doing so.”).

C. Strategic Immunity Findings

The New Qualified Immunity examines some of the dangers of Pearson discretion, including whether the lower courts are striking the right balance between constitutional avoidance and stagnation. We concluded that because there is a risk of stagnation and disuniform development of constitutional law, the Court should revise Pearson to require judges to give reasons for the exercise of their discretion. 248Nielson & Walker, supra note 33, at 65. In this way, discretion can better accomplish its intended purpose: preventing constitutional stagnation without imposing shackles that are imprudent in particular cases.

This Article, by contrast, examines other dangers that the Justices did not consider, much less intend: the opportunities for strategic judicial behavior that may be caused—or at least exacerbated by—Pearson’s grant of maximalist discretion. This theoretical risk is particularly troubling because it suggests, for instance, that not only is constitutional stagnation real, but that sometimes it may not be by accident. Instead, judges may be using their discretion in strategic ways, including by deliberately avoiding the recognition of new constitutional rights, even in situations where an application of the Pearson factors would suggest that the constitutional merits should be reached. This is a novel fear that has not been explored in the literature. As discussed in Part III.C, suboptimal behavior in the qualified immunity context could include at least two strategies: the improper exercise (or not) of Pearson discretion to reach constitutional questions, perhaps due to panel compromises, collegiality concerns, or a desire to avoid further judicial review (Part IV.C.1); and the improper publication (or not) of the panel’s decision, perhaps for similar reasons noted above (Part IV.C.2). This Part empirically assesses, at least as a preliminary matter, 249In this preliminary inquiry we rely solely on basic tests of statistical significance. As further research is undertaken, however, one should incorporate regression analysis, and it may be useful to employ semiparametric matching rather than standard regression analysis. Cf. Christina L. Boyd, Lee Epstein & Andrew D. Martin, Untangling the Causal Effects of Sex on Judging, 54 Am. J. Pol. Sci. 389, 389 (2010). In other words, our study’s tentative analysis should not be the final word on this issue. those theoretical risks in the qualified immunity context.

Before turning to the empirical findings with respect to these two types of strategic behavior, however, it is important to underscore one important methodological limitation: the study focuses exclusively on circuit decisions that cite Pearson. 250These are, of course, not the only methodological limitations. For instance, any attempt to draw inferences about the legal system from a selection of trial or appellate opinions has problems. See, e.g., George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13 J. Legal Stud. 1, 1 (1984); Theodore Eisenberg & Michael Heise, Plaintiphobia in State Courts Redux? An Empirical Study of State Court Trials on Appeal, 12 J. Empirical Legal Stud. 100, 100 (2015); Theodore Eisenberg & Stewart J. Schwab, What Shapes Perceptions of the Federal Court System?, 56 U. Chi. L. Rev. 501, 503–04 (1989). The purpose of this study, however, is not to draw inferences about the larger legal system but to better understand the strategic considerations that may affect how circuit courts exercise their discretion to develop constitutional law based on the cases that reach their dockets—recognizing that there are selection effects embedded in the sample of legal disputes that reach the appellate stage. As such, the data set does not include the entire universe of qualified immunity decisions issued during the relevant time period, and it is possible—though based on our experience not likely—that the sample of cases citing Pearson is not representative of qualified immunity decisions more generally. Likewise, this study does not capture a potentially important strategic behavior: a court’s decision not to cite Pearson at all when deciding whether to reach constitutional questions in the qualified immunity context.

1. Panel Effects and Strategic (Non)Use of Discretion

As discussed in Part IV.B, to assess the effect of the political ideology of a judge on the exercise of Pearson discretion, we limited our preliminary inquiry in The New Qualified Immunity to the political affiliation of the President who appointed the judge who authored the opinion. Aside from the fact that the political affiliation of the nominating President may be a poor proxy for the political ideology of the authoring judge, 251See Christopher A. Whytock, The Evolving Forum Shopping System, 96 Cornell L. Rev. 481, 521 n.181 (2011) (“[T]ests using the party of the nominating president as a proxy for a judge’s ideological attitudes may underestimate the impact of those attitudes.” (citing Joshua B. Fischman & David S. Law, What Is Judicial Ideology, and How Should We Measure It?, 29 Wash. U. J.L. & Pol’y 133, 170–71 (2009))). But see Jeremy D. Horowitz, Not Taking “No” for an Answer: An Empirical Assessment of Dissents from Denial of Rehearing En Banc, 102 Geo. L.J. 59, 73 (2013) (noting that the party of a judge’s nominating President can shed some insight “when looking at a large number of judges and decisions” (citing Cass R. Sunstein et al., Ideological Voting on Federal Courts of Appeals: A Preliminary Investigation, 90 Va. L. Rev. 301, 303 (2004))); Daniel R. Pinello, Linking Party to Judicial Ideology in American Courts: A Meta-Analysis, 20 Just. Sys. J. 219, 243 (1999) (observing that party can be a useful indicator of ideology after conducting a meta-analysis of studies). an exclusive focus on the authoring judge in appellate panels that typically consist of three judges does not fully capture the effects that ideological preferences may play in the exercise of Pearson discretion.

Here, accordingly, we offer a more nuanced evaluation of judicial behavior by focusing on the conduct of all judges on the panel. In particular, Figure 1 depicts how circuit courts ruled on qualified immunity grounds, broken down by the party affiliation of the President who appointed the opinion author as well as by the party affiliation of the President who appointed all of the judges in each case. In other words, the latter category includes as separate data points the vote of each judge on the panel. 252The numbers of all judges reported in Figures 1 and 2 include all judges on unsigned or per curiam opinions.

nielson-walker-fig1

Figure 1. Post-Pearson Qualified Immunity and Party Affiliation of Judges’ Nominating President

As Figure 1 details, the rate of exercising Pearson discretion is not different based on the party affiliation of the President who nominated the authoring judge: authors nominated by a Democratic President (D authoring judges) and those nominated by a Republican President (R authoring judges) exercised Pearson discretion 43.3% and 45.0% of the time, respectively. Nor is there a difference between D authoring judges (43.3%) and all D judges (43.1%), or R authoring judges (45.0%) and all R judges (43.1%).

As highlighted in The New Qualified Immunity, however, the differences in the other two categories are statistically significant (p < 0.05) as to the authoring judges. 253Nielson & Walker, supra note 33, at 45–46. To assess statistical significance for political affiliation (p < 0.05), we assume that if there is no bias in the distribution, the results would be 50/50. That is, if observing opinions through the lens of political affiliation and assuming no bias, a judge nominated by a Democratic President would have the same odds of deciding a given question as a judge nominated by a Republican President. The results will split down the middle. If, however, the political affiliation of the nominating President affects the way judges decide opinions, the odds will be skewed according to that variable. For determining statistical relevance, we compare the percentage of events occurring given a specific categorization to the expected value of an unbiased—50/50—result. Some normalization was required (i.e., D to R), since there is a disparity between the total number of claims decided by each party. Otherwise, a 50/50 split is already skewed. This has some caveats, so normalization was always done in the most conservative manner possible. We used Prism software to run parts-of-a-whole comparisons (actually a two-tailed binomial distribution given the input value versus the expected output value, which is why we arrive at a P-value) on the positive values observed from the dominant population in the data against the normalized values of the minority group in the data. Moreover, chi-squared tests for independence were performed to assess whether there was statistical evidence to suggest that nominating party mattered as to the three types of decisions depicted in Figure 1 as well as the two types of decisions depicted in Figure 2. The results were consistent with those of the binomial test reported herein. In contrast to D authoring judges, R authoring judges were more likely (27.4% to 18.0%) to decide not to exercise their Pearson discretion and just hold that the right was not clearly established. Conversely, D authoring judges were more likely (38.8% to 27.1%) than R authoring judges to deny qualified immunity by finding a violation of a clearly established right. 254See supra note 238 and accompanying text.

Yet when one examines the votes of all judges—and not just the authoring judges—there is a surprising shift. To be sure, there is no noteworthy shift with respect to R authoring judges compared to all R judges as to not exercising Pearson discretion (27.4% to 27.2%) or to finding no discretion to exercise (27.1% to 29.7%). With respect to D judges, however, there is a change. In particular, when the universe of D judges is examined, it appears that D judges as a whole are much more similar to R judges as a whole than they are to D authoring judges as to (1) not exercising Pearson discretion (27.2% to 18.0%), and as to (2) finding no discretion to exercise (29.7% to 38.8%). Put another way, any significant difference between R and D judges disappears when one examines all judges as opposed to limiting the analysis to the authoring judges.

What’s going on? It appears that when a D judge is assigned an opinion, she behaves differently than when asked simply to join an opinion. It is unclear why this is so, but one potential reason, at least in some cases, might be strategy. A D judge may have certain preferences, all else being equal, and when that judge can author the opinion, she may try to capture some of those preferences (realizing that other judges will have to incur costs if they want to change the opinion). But when another judge has authored the opinion, the costs of disagreeing are too significant to rock the boat. Or, conversely, when a D judge has indicated a willingness to dissent, the majority compromises in its opinion to avoid a divided opinion. It also, of course, may suggest that panels are more willing to allow D nominated judges to author opinions that find liability, perhaps because D nominated judges prefer such assignments.

By contrast, with respect to how D and R judges exercise Pearson discretion—in particular, whether they recognize a new constitutional right in the pure Saucier sense or find no constitutional violation—the political findings stay roughly the same whether one looks at authoring judge or all judges. Figure 2 breaks down the exercise of Pearson discretion, again based both on the party affiliation of the President who appointed the opinion author and the party affiliation of the President who appointed all of the judges in each case.

nielson-walker-fig2

Figure 2. Exercise of Pearson Discretion and Party Affiliation of Judges’ Nominating President

As Figure 2 depicts, the differences between D and R authoring judges with respect to finding no violation are not statistically significant. When exercising Pearson discretion to reach constitutional questions, R authoring judges found no violation in 93.3% of the cases, as opposed to 86.0% for D authoring judges. The same is not true, however, regarding the recognition of new constitutional rights. When exercising Pearson discretion to reach constitutional questions, D authoring judges found constitutional violations nearly twice as often as R authoring judges (14.1% and 6.7%, respectively)—a statistically significant difference (p < 0.05). Moreover, comparing the authoring judge to all judges does not make much of a difference. R authoring judges and all R judges found no constitutional violation (93.3% to 93.2%) and recognized new constitutional rights (6.7% to 6.8%) at roughly the same rates. And there is only a slight difference in D authoring judges and all D judges as to finding no constitutional violation (86.0% to 89.9%) as well as to recognizing new constitutional rights (14.1% to 10.1%). This may suggest that because it is especially significant to recognize a new constitutional right, judges may have stronger preferences regarding them.

In light of these differences in the data based on the authoring judge or all judges, one may be tempted to conclude that an assessment of the votes of all judges more accurately captures the effects of ideology on the exercise of Pearson discretion. That too is overly simplistic. As scholars have documented in empirical studies outside of the qualified immunity context, there may be “panel effects” in federal circuit court decisionmaking. 255See, e.g., Cass R. Sunstein et al., Are Judges Political?: An Empirical Analysis of the Federal Judiciary vii (2006); Boyd, Epstein & Martin, supra note 249; David S. Law, Strategic Judicial Lawmaking: Ideology, Publication, and Asylum Law in the Ninth Circuit, 73 U. Cin. L. Rev. 817, 819 (2005); Thomas J. Miles & Cass R. Sunstein, Do Judges Make Regulatory Policy? An Empirical Investigation of Chevron, 73 U. Chi. L. Rev. 823, 851 (2006); Jennifer L. Peresie, Note, Female Judges Matter: Gender and Collegial Decisionmaking in the Federal Appellate Courts, 114 Yale L.J. 1759, 1762 (2005); Richard L. Revesz, Litigation and Settlement in the Federal Appellate Courts: Impact of Panel Selection Procedures on Ideologically Divided Courts, 29 J. Legal Stud. 685, 685 (2000); Gregory C. Sisk, Michael Heise & Andrew P. Morriss, Searching for the Soul of Judicial Decisionmaking: An Empirical Study of Religious Freedom Decisions, 65 Ohio St. L.J. 491, 491 (2004). Matthew Stephenson has explained that these panel effects come in two distinct flavors that “are sometimes conflated even though they are quite different”: (1) “judges in the ideological minority on a three-judge panel tend to vote with their colleagues instead of dissenting”; and (2) “courts behave differently when there is at least one judge on the panel who has political views that differ from the panel majority,” such that “[m]ixed panels tend to be more moderate than homogenous panels . . . .” 256Matthew C. Stephenson, Statutory Interpretation by Agencies, in Research Handbook on Public Choice and Public Law 285, 307 (Daniel A. Farber & Anne Joseph O’Connell eds., 2010).

For example, in a study of over 6000 published three-judge panel decisions (with nearly 20,000 total votes of individual judges), Cass Sunstein, David Schkade, Lisa Ellman, and Andrew Sawicki uncovered the following main findings as to their entire sample:

[F]or purposes of analysis, we are taking, as the baseline, cases in which a judge is sitting with one Democrat and one Republican, and we are examining how voting patterns shift when a judge is sitting instead with two Democratic appointees or two Republican appointees. We can readily see that a Democrat, in the baseline condition, casts a liberal vote 52 percent of the time, whereas a Republican does so 37 percent of the time. Sitting with two Democratic appointees, Democratic appointees cast liberal votes 64 percent of the time (an increase of 12 percent over baseline), whereas Republican appointees do so 46 percent of the time (an increase of 9 percent over baseline). Sitting with two Republican appointees, Democratic appointees cast liberal votes 44 percent of the time (a decrease of 8 percent), whereas Republican appointees do so only 31 percent of the time (a decrease of 6 percent). Thus, Republican appointees sitting with two Democratic appointees show the same basic pattern of votes (46 percent liberal votes) as do Democratic appointees sitting with two Republican appointees (44 percent liberal votes). 257Sunstein et al., supra note 255, at 23–24.

As Professor Stephenson has explained, “judges on court of appeals panels sometimes make concessions to their colleagues,” including that “minority judges will sometimes go along with the majority” (majority effect) and that “majority judges will sometimes make concessions to the minority judge on the panel” perhaps to avoid a separate dissent (diversity effect). 258Stephenson, supra note 256, at 308.

In the post-Pearson qualified immunity context, these panel effects may be even more pronounced in circumstances where the court does not conclude that any constitutional right is clearly established. As discussed in Part III.C.1, because such constitutional ruling has no effect on the parties before the court, a judge in those circumstances may be more willing to compromise as to whether to exercise Pearson discretion to decide the constitutional question. To explore this idea, Figure 3 depicts how the three-judge panels ruled on qualified immunity grounds, broken down by the four possible political configurations of the panel. 259A number of cases in the sample were decided by two-judge panels and thus are not included in this analysis. As noted in Part IV.A, en banc decisions have also been excluded.

nielson-walker-fig3

Figure 3. Post-Pearson Qualified Immunity with Panel Effects

If we analyze these panel effects on the gross assumption that each group of judges has an equal number of D or R judges—basically the court is like a bag of judges and there is an equal probability (1:2) of pulling a D or R judge from the bag—there is statistically significant evidence (p < 0.0001) to suggest that each type of decision is not independent of the panel composition. 260This is based on a chi-squared test to compare all four potential panel compositions with respect to the probabilities that they exercised each of the three types of decisions depicted in Figure 3 as well as the two types of decisions depicted in Figure 4. The chi-squared tests do not allow for examination of each type of panel composition individually. Similarly, when one panel composition is compared to the rest of the panel compositions, there are statistically significant differences (p < 0.0001) as to all three categories for the panels of all D judges (3D) and of all R judges (3R). As for mixed panels, both the 2D-1R and 2R-1D panels have statistically significant differences (p < 0.05) as to not exercising Pearson discretion. With 2D-1R, there was also a statistically significant difference (p < 0.0001) as to exercising Pearson discretion to decide a constitutional question.

If ideological composition did not matter at all, one would expect no differences in these decisions based on the composition of the panel. As the prior studies on panel effects demonstrate, however, we live in a world where ideology matters. When the composition of the panel changes, the data suggest that the behavior of the panel changes.

Moreover, this 50/50 approach is overly simplistic for the additional reason that it is not representative of the real-world divide between the number of D and R circuit judges (at least as observed in the cases reviewed). In particular, in our data set there are votes from a total of 4299 judges—of which 1696 (39.5%) are D judges and 2603 (60.5%) are R judges. Accordingly, by normalizing the expected panel compositions based on the 40/60 occurrence of D and R judges in the sample, we get a more accurate picture of the panel effects in the cases reviewed. Under this normalized approach, there remains statistically significant evidence (p < 0.01) to suggest that a panel’s decision to exercise its Pearson discretion to reach a constitutional decision as well as a panel’s finding of no discretion to exercise (a denial of qualified immunity) are not independent of the panel composition. But the same is no longer true as to a panel’s decision not to exercise Pearson discretion and instead hold that a right is not clearly established. This suggests that judges, when joined by ideologically similar colleagues, may place special value on cases involving constitutional violations.

Similarly, when one panel composition is compared to the rest of the panel compositions under this normalized model, other statistically significant results emerge. As for unanimous 3D or 3R panels, the only statistically significant difference (p < 0.001) is that 3R panels are more likely to exercise their Pearson discretion to reach a constitutional question. As for mixed panels, the only statistically significant results concern finding no discretion to exercise and thus denying immunity. These results are significant, however, in opposite directions: 2D-1R panels (p < 0.001) are statistically more likely to deny qualified immunity as compared to the rest of the panel compositions, whereas 2R-1D panels (p < 0.05) panels are less likely to deny immunity.

These latter findings are interesting, though they may not tell us too much about strategic approaches to exercising Pearson discretion. After all, a finding of no qualified immunity, by definition, means the panel determined there was no discretion to exercise. But it is interesting that 3R panels are more likely to exercise Pearson discretion to reach constitutional questions and thus develop constitutional law. As we posited in The New Qualified Immunity, this finding is in tension with the notion of “R judges’ minimalist inclination to avoid constitutional rulings by just holding the right is not clearly established.” 261Nielson & Walker, supra note 33, at 46. See generally Cass R. Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court 3–4 (1999) (describing judicial minimalism). On the other hand, perhaps it is not surprising that unanimous 3R panels would be more aggressive in exercising their Pearson discretion if they did so to find no constitutional violation. In other words, “[i]n grossly oversimplified terms, some ‘conservative’ judges may want to rule that the Constitution does not prohibit the government action at issue to allow such action to continue . . . .” 262Nielson & Walker, supra note 33, at 47.

For these reasons, perhaps the more important inquiry is how D and R judges rule when they decide to exercise their Pearson discretion. Figure 4 breaks down the exercise of Pearson discretion by the four possible political configurations of the panel.

nielson-walker-fig4

Figure 4. Exercise of Pearson Discretion with Panel Effects

After normalizing these results based on the 40/60 D-R composition of the sample, a number of interesting panel effects emerge. 263Under the overly simplistic 50/50 model, there is statistically significant evidence (p < 0.0001) to suggest that a panel’s decision to find no constitutional violation is not independent of the panel composition, whereas the differences are not statistically significant (p < 0.05) as to a panel’s decision to recognize a new constitutional right. Moreover, when one panel composition is compared to the rest of the panel compositions, there are statistically significant differences (p < 0.0001) as to 3D, 3R, and 2D-1R panels in finding no constitutional violations (but not as to 2R-1D panels), and there are no statistically significant differences (p < 0.05) as to any panel composition’s decision to recognize a new constitutional right. First, there is statistically significant evidence (p < 0.0001) to suggest that a panel’s decision to find no constitutional violation is not independent of the panel composition, whereas the differences are not statistically significant (p < 0.05) as to a panel’s decision to recognize a new constitutional right. Comparing each panel composition to the rest of the panel compositions yields a number of statistically significant results that are consistent with prior empirical studies on panel effects. As for unanimous panels, 3D panels are more likely (p < 0.01) to exercise their Pearson discretion to recognize new constitutional rights than the other three panel compositions. Conversely, 3R panels are more likely (p < 0.001) to exercise their Pearson discretion to find no constitutional violation. In other words, “on unified panels, ideological tendencies are amplified,” with “panels consisting of three Republican appointees show[ing] systemically different outcomes than panels consisting of three Democratic appointees.” 264Sunstein et al., supra note 255, at 129.

Moreover, perhaps consistent with the “diversity effect” discussed above, on mixed panels—either 2D-1R or 2R-1D—there are no statistically significant differences as to either creating new constitutional rights or finding no constitutional violation. In other words, it is quite possible that the panel effects of a mixed panel in this context results in “the majority judges generally moving toward the center just enough that the minority judge is willing to go along without dissent.” 265Stephenson, supra note 256, at 308. Or, perhaps similar to what Tom Miles and Cass Sunstein found in the administrative-law context, “[t]he most important lesson is that both Democratic and Republican appointees show far more political voting patterns when they are sitting on unified panels. When the panels are divided, the role of politics is greatly dampened.” 266Miles & Sunstein, supra note 255, at 852.

In sum, although this is just a preliminary inquiry into the strategic judicial behavior that may be at play, the panel effects suggest that suboptimal exercise of Pearson discretion may occur based on the composition of the panel. In particular, D judges behave differently when they are the author of an opinion rather than a judge joining an opinion. Panels composed of all D or all R judges behave differently from mixed panels. Unanimous R-judge panels are more likely to decide constitutional questions than unanimous D-judge panels (though 3R panels are also more likely to exercise such discretion to find no constitutional violation). And all D-judge panels are most likely to recognize new constitutional rights. None of this proves strategic behavior on the part of judges, but the behavior is consistent with it.

2. Strategic (Non)Publication of Immunity Decisions

There is another potential avenue for strategic behavior: the ability not to publish an immunity decision. As discussed in Part III.C.2, the ability not to publish a decision allows for further compromise on a three-judge panel. In other words, a judge can compromise by joining a panel decision that is not optimal or by compromising on a narrower unanimous panel decision, but part of the compromise could also be that the panel agrees to issue an unpublished, nonprecedential decision. Likewise, a panel that fears further review (from the circuit en banc or the Supreme Court) may elect not to publish its decision, thus making the decision less important going forward. As discussed in Part I.C, unpublished decisions are particularly important in the qualified immunity context because of their uneasy relationship to the creation of clearly established law. In other words, strategic immunity does not just concern how panels exercise their Pearson discretion to reach constitutional questions, but also whether they do so in published, precedential opinions. Figure 5 depicts how circuit courts ruled on qualified immunity grounds, broken down by published and unpublished decisions.

nielson-walker-fig5

Figure 5. Post-Pearson Qualified Immunity and Publication Decisions

As Figure 5 details, there is nearly a 50/50 split between published decisions (773 claims or 52.9%) and unpublished decisions (689 claims or 47.1%). 267For determining statistical relevance, we utilized the same method as for political affiliation, and compared the percentage of events occurring given a specific categorization to the expected value of 50/50. Some normalization was required, since there is a disparity between the total number of claims decided in published and unpublished decisions. Otherwise, a 50/50 split is already skewed. Although there is no real difference (p < 0.05) in the rate of deciding not to exercise Pearson discretion (just holding any right is not clearly established) in published decisions (25.1%) versus unpublished decisions (28.6%), there are statistically significant differences as to the other two categories. Courts were more likely (p < 0.0001) to find no discretion to exercise, and thus deny immunity, in published decisions (33.4% to 21.3%) (i.e., findings against immunity for government actors are often published), whereas they were more likely (p < 0.05) to exercise Pearson discretion to reach constitutional questions in unpublished decisions (50.1% to 41.5%) (i.e., constitutional decisions are less likely to be published).

At first blush, both of these findings may seem surprising. After all, to deny qualified immunity, the court must find that the government actor violated clearly established constitutional law, and that law must be established by prior precedent. In other words, by definition, all the court is doing is applying established precedent without creating new law—something that should be quite suitable for an unpublished decision. Yet nearly two-thirds (63.7%) of all decisions denying qualified immunity are published. On the other hand, stripping a government actor of immunity for violating clearly established law is a significant judicial action. The court may well want to send a strong and clear measure to other government actors via a published decision.

These results are somewhat surprising when broken down by R and D authoring judges. When utilizing unpublished decisions, R authoring judges seemed more likely than D authoring judges to deny qualified immunity. 268This finding is statistically significant only at a lower level (p < 0.1) than we set as the baseline for the rest of the findings (p < 0.05). Put differently, it appears that R authoring judges were more likely to rule against the government official in a nonprecedential opinion. Indeed, the cases reviewed reveal a striking number of 3R unpublished decisions—oftentimes quite lengthy—where the panel denied qualified immunity. 269See, e.g., Anderson v. McCaleb, 480 F. App’x 768, 773 (5th Cir. 2012) (per curiam) (reversing the district court’s grant of qualified immunity in the excessive force context by 3R panel in a somewhat lengthy, unpublished, unsigned decision); Quint v. Vill. of Deerfield, 365 F. App’x 697, 701 (7th Cir. 2010) (reversing the grant of qualified immunity by 3R panel in the unlawful search and seizure context); Evans v. City of Etowah, 312 F. App’x 767, 768 (6th Cir. 2009) (affirming the denial of qualified immunity in the unlawful arrest context by 3R panel); Haley v. Elsmere Police Dep’t, 452 F. App’x 623, 629 (6th Cir. 2011) (affirming the denial of qualified immunity in the unlawful arrest context by 3R panel though over a dissent); see also Comeaux v. Sutton, 496 F. App’x 368, 375 (5th Cir. 2012) (per curiam) (reversing the district court’s grant of immunity in the excessive force context by 2D-1R in a somewhat lengthy, unpublished, unsigned opinion). These cases suggest some strategic use of unpublished decisions. For instance, these judges may fear the penumbral effect of a published opinion—i.e., the fact that a finding of no immunity in one case can be used in subsequent cases to say the law was clearly established, even if those subsequent cases have slightly different facts.

Similarly, it may seem odd that courts more often exercise Pearson discretion to answer constitutional questions in unpublished decisions. But that finding is only noteworthy when placed in the context of how the court exercised Pearson discretion. It would not be unusual for a court to hold in an unpublished opinion that there is no constitutional violation because, for instance, existing precedent has already established that the government action is constitutional. Conversely, it would seem quite odd for a court to recognize a new constitutional right in an unpublished decision, as such rights recognition arguably would not create clearly established law. Figure 6 depicts how courts exercise Pearson discretion in published and unpublished opinions.

nielson-walker-fig6

Figure 6. Exercise of Pearson Discretion and Publication Decisions

As Figure 6 details, whereas there is no statistically significant difference (p < 0.05) as to courts finding no constitutional violation in published (86.9%) or unpublished (96.8%) decisions, the results show statistically significant differences as to pure Saucier decisions: courts were more likely to exercise their Pearson discretion to recognize new constitutional rights in published opinions (13.1%) as opposed to unpublished opinions (3.2%). Put differently, four in five pure Saucier decisions (42 claims or 79.3%) were published.

What do these publication rates tell us about strategic immunity? Panel decisions not to exercise Pearson discretion, and instead just to hold that a right is not clearly established, are pretty evenly distributed between published (49.6%) and unpublished (50.4%) decisions. The act of not publishing one of these decisions could implicate strategic immunity as it would allow another panel—in a case with perhaps better facts or better lawyering—to declare that the right is indeed clearly established. From these data, it is difficult to tell whether opinions are being unpublished for strategic reasons.

Similarly, panel decisions to exercise discretion to answer the constitutional questions are also pretty evenly distributed between published (48.2%) and unpublished (51.8%) decisions. And, moreover, when panels exercise Pearson discretion to find no constitutional violation, there is no statistically significant difference (p < 0.05) between their use of published (45.5%) or unpublished (55.5%) decisions. Again, the act of not publishing a decision finding no constitutional violation could involve strategic compromise, as a subsequent panel would not be bound by the panel’s nonprecedential constitutional ruling. Once more, however, it is difficult to tell from the data whether that is occurring, but there certainly is some evidence for it.

A closer look at the unpublished decisions, moreover, provides strong anecdotal support for strategic use of unpublished decisions. For instance, there are a surprising number of divided opinions—i.e., opinions with a dissent—that are not published. Many judges would have denied qualified immunity entirely because there was a violation of a clearly established constitutional right—at least at the pleading or summary judgment stage. 270See, e.g., Anderson v. City of Bainbridge Island, 472 F. App’x 538, 541 & n.2 (9th Cir. 2012) (Gould, J., dissenting) (dissenting in excessive force context as to the majority’s finding of no constitutional violation); Jordan v. Brown, 489 F. App’x 190, 191 (9th Cir. 2012) (Thomas, J., dissenting) (“I would affirm the district court’s denial of qualified immunity at this stage of the litigation.”); Smith v. Cty. of Lenawee, 505 F. App’x 526, 540–41 (6th Cir. 2012) (Kethledge, J., concurring in part and dissenting in part) (“I would affirm the district court’s denial of qualified immunity to Westgate. Unlike Dye—who I think presents a close case, but who at least showed concern by putting Smith in a padded cell and calling Dr. Stickney—Westgate basically took no action at all during his four-hour shift on the night before Smith died.”); Williams v. Port Huron Sch. Dist., 455 F. App’x 612, 622 (6th Cir. 2012) (Moore, J., dissenting) (“I agree with the district court that, given the facts as pleaded by the Plaintiffs, a reasonable jury could conclude that Dalke and Jones were deliberately indifferent to the harassment at Port Huron Northern.”); Estate of Henson v. Callahan, 440 F. App’x 352, 359 (5th Cir. 2011) (Dennis, J., dissenting) (“I respectfully dissent. In my view, there is evidence from which a reasonable jury could conclude that each of the three elements of supervisory liability is satisfied, and thus that the district court’s denial of qualified immunity to Sheriff Callahan should be affirmed.”); Neal v. Melton, 453 F. App’x 572, 581 (6th Cir. 2011) (Moore, J., concurring in part and dissenting in part) (“Taking the facts in the light most favorable to the Plaintiffs, I believe that the officers lacked reasonable suspicion to detain the Plaintiffs after completion of the vehicle-registration check in order to conduct a canine sweep.”); Rodriguez v. City of Cleveland, 439 F. App’x 433, 459 (6th Cir. 2011) (White, J., concurring in part and dissenting in part) (“I respectfully dissent from the majority’s determination that the individual defendants are entitled to qualified immunity as to Rodriguez’s claims of retaliation and harassment under the First Amendment.”); Poulakis v. Rogers, 341 F. App’x 523, 540 (11th Cir. 2009) (Quist, J., dissenting) (“My larger concern is that this Court’s decision sends a signal to police officers that they are free to ignore the law of the intermediate state appellate courts by which they are otherwise bound, and an unambiguous statute, without concern for violating an individual’s federal constitutional rights. Therefore, I would reverse on qualified immunity and remand for consideration of the other defenses Defendants raised in their summary judgment motion.”). Moreover, some agreed to the grant of qualified immunity but disagreed with the majority that there was an underlying constitutional violation. 271See, e.g., Koch v. Lockyer, 340 F. App’x 372, 375 (9th Cir. 2009) (Clifton, J., concurring in part and dissenting in part) (“I agree with my colleagues that the individual defendants are entitled to qualified immunity . . . , but I strongly disagree with the conclusions that the collection of Koch’s DNA violated the Fourth Amendment . . . .”). Finally, other dissenters would have held that there was no constitutional violation at all. 272See, e.g., Estate of Henson v. Callahan, 440 F. App’x 352, 358 (5th Cir. 2011) (Owen, J., concurring) (“I join Judge Southwick’s holding that Sheriff Thomas Callahan is entitled to qualified immunity. With respect, however, I do not join his opinion.”); Haley v. Elsmere Police Dep’t, 452 F. App’x 623, 629 (6th Cir. 2011) (Rogers, J., dissenting) (“Because Markesbery had probable cause to arrest Haley, I would reverse the district court’s denial of summary judgment on the unlawful arrest claim.”).

Any dissent from an unpublished decision should raise some concerns that perhaps the opinion should have been published, and thus a compromise not to publish had been reached. 273See, e.g., Plumley v. Austin, 135 S. Ct. 828, 831 (2015) (Thomas, J., dissenting from the denial of certiorari) (focusing on the presence of a dissent from a lengthy majority opinion as evidence that the decision should have been published). After all, unpublished opinions are generally understood to apply settled law, 274See supra Part I.C. and a dissent suggests that may not be the case—though, to be sure, some dissents may agree as to the applicable law but disagree as to its application with regard to particular facts. In the qualified immunity context, such divided, unpublished decisions may well be even stronger evidence of strategic judicial behavior where, for instance, the majority has conceded that its denial of qualified immunity—and thus holding that a clearly established constitutional right had been violated—would not be issued in a published decision. The fight still takes place, but it takes place in the nonprecedential shadows.

What is perhaps most interesting, and which may provide compelling evidence of strategic behavior, are the eleven unpublished pure Saucier decisions. 275Akrawi v. Remillet, 504 F. App’x 450, 453 (6th Cir. 2012) (rejecting defendants’ challenge to the award of attorney’s fees where the grant of qualified immunity was affirmed because plaintiff “prevailed on his central claim—that the defendants violated his right to due process”); Escobar v. Mora, 496 F. App’x 806, 813 (10th Cir. 2012) (“[W]hile we have extrapolated from general principles here to hold that Mr. Escobar stated an Eighth Amendment claim, we cannot say that the analytical bridge from the former principles to the latter holding impressed itself upon us with the clarity necessary to conclude that defendants were on fair notice that their conduct rose to the level of a constitutional violation.”); Saavedra v. Scribner, 482 F. App’x 268 (9th Cir. 2012) (per curiam) (finding two separate due process violations in the administrative-segregation context but granting qualified immunity); Kozel v. Duncan, 421 F. App’x 843, 850 (10th Cir. 2011) (“Although we have determined that there was a constitutional violation in the prolonged sobriety detention, we must still consider whether a reasonable officer in Sheriff Duncan’s position would have been on notice that his conduct was unlawful. Mr. Kozel has not met his burden in this regard.”); Rivers v. Fischer, 390 F. App’x 22, 24 (2nd Cir. 2010) (per curiam) (concluding that “there is no doubt that Rivers’s constitutional rights were violated when the Department of Corrections administratively imposed a 5-year term of supervised release that was not orally pronounced by the sentencing judge” but concluding that right was not clearly established at time of the incident); Rock for Life–UMBC v. Hrabowski, 411 F. App’x 541, 555 (4th Cir. 2010) (“Although the district court erred by holding that the plaintiffs failed to demonstrate a triable issue of fact whether the defendants regulated their speech based on its content, the defendants are nevertheless entitled to qualified immunity from 42 U.S.C. § 1983 claims brought against them in their individual capacities.”); id. at 558 (King, J., concurring in part, dissenting in part, and concurring in the judgment) (“I agree with the majority that we should award qualified immunity to the defendants on the as-applied First Amendment challenge to UMBC’s policy on facilities use, but I would get there by a different route—namely, by concluding that a First Amendment violation has not been shown.”); Schwenk v. Cty. of Alameda, 364 F. App’x 336, 337 (9th Cir. 2010) (per curiam) (“We affirm the dismissal as to Lacativo on the basis of qualified immunity. At the time of the alleged conduct, we had not yet decided Burke, which expressly held that ‘parents with legal custody, regardless of whether they also possess physical custody of their children’ have a liberty interest in the care, custody, and management of their children.” (citing Burke v. Cty. of Alameda, 586 F.3d 725, 731, 733–34 (9th Cir. 2009))); Solis v. Oules, 378 F. App’x 642, 644 (9th Cir. 2010) (per curiam) (granting immunity yet finding that the officer “violated Solis’s Fourth Amendment right to be free from unreasonable searches and seizures when he stopped her vehicle and apparently removed her from it under a law that did not criminalize her behavior”); Koch, 340 F. App’x at 374 (per curiam) (“While the compulsory DNA collection violated Koch’s Fourth Amendment rights, Appellees are entitled to qualified immunity against damages.”); Thompson v. Williams, 320 F. App’x 678, 679 (9th Cir. 2009) (per curiam) (granting qualified immunity but affirming “that Thompson’s religious beliefs are sincerely held and that there are triable issues as to whether the prison’s policy violates the First Amendment or RLUIPA”). In the cases reviewed, nearly one in five pure Saucier decisions is not published. The fact that there are any such unpublished decisions should be surprising. 276To be sure, two of these decisions noted that the right had already been recognized in a prior published decision but that the incident at issue took place prior to that decision’s publication and thus was not clearly established at the time of the incident. Rivers, 390 F. App’x at 24; Schwenk, 364 F. App’x at 337. After all, pure Saucier decisions are intended to recognize a new constitutional right and thus develop constitutional law, yet announcing such new law in an unpublished opinion generally does not create clearly established constitutional law for future litigants, or at least, at a minimum, does not do so as emphatically. Why would a panel not publish a pure Saucier opinion since their whole point is to clearly establish law?

A closer look at the panel composition with respect to these eleven claims provides further support for strategic immunity. Four of the eleven claims are included in unpublished opinions authored by R judges, whereas the remaining seven claims are included in unsigned opinions (also known as per curiam opinions). In other words, no D judge authored any of these opinions. Yet, of the thirty-three judges ruling on these claims, fourteen were D judges. It should be noted that one of those D judges actually dissented in part from the unpublished opinion because he would “conclude[] that a First Amendment violation has not been shown” (interestingly, with regards to a “conservative” speech issue, namely, anti-abortion speech). 277Rock for Life–UMBC, 411 F. App’x at 558 (King, J., concurring in part, dissenting in part, and concurring in the judgment). It may also be worth noting that six of these eleven claims were decided in unpublished decisions issued by the Ninth Circuit. 278See supra note 275.

In sum, to fully appreciate the suboptimal judicial behavior for which the post-Pearson new qualified immunity standard provides, it is important to look not only at the panel effects involved in exercising (or not exercising) Pearson discretion, but also the panel’s strategic decision to publish (or not to publish) its immunity decision. As discussed in Part IV.C.1, the panel effects are pronounced and statistically significant—suggesting that the unified panels are more aggressive in finding no constitutional violations (for 3R panels) and in recognizing new constitutional rights (for 3D panels), and that mixed panels lead to more normal patterns in the use of Pearson discretion. It is more difficult to ascertain from this data set the effect of unpublished decisions. That said, R authoring judges seem more likely to put immunity denials in unpublished decisions, and the set of unpublished decisions reveal a number of divided opinions where strategic compromise may have taken place. And, of course, there is the surprising finding that one in five pure Saucier decisions is not published.

V. Counteracting Strategic Behavior

For the foregoing reasons, there are both theoretical and empirical grounds to fear that courts may use their Pearson discretion in qualified immunity cases in suboptimal ways. This Article’s basic point is that in crafting the procedures for qualified immunity, the Supreme Court did not pay enough attention to an important insight from administrative law: When it comes to discretion, unintended consequences can be just as important as intended consequences, and one should never lose sight of incentives.

The question becomes: What, if anything, should be done to combat such strategic behavior? On one hand, sometimes the best response to the risk of discretion’s abuse is to eliminate discretion altogether. Just as agencies—as a rule—lack discretion to promulgate regulations with retroactive effect, 279See Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208 (1988). perhaps judges should never clarify law in cases where the right was not clearly established. 280See Camreta v. Greene, 563 U.S. 692, 714 (2011) (Scalia, J., concurring); Nielson & Walker, supra note 33, at 52 (citing Camreta, 563 U.S. at 727 (Kennedy, J., dissenting)). Or, in the same spirit, perhaps the Court should return to Saucier, which would also deprive judges of discretion, albeit in the other direction. 281See Reinhardt, supra note 17, at 1250.

On the other hand, sometimes the danger of discretion does outweigh its utility and no safeguard is cost justified. Just as agencies generally have total discretion when it comes to whether to bring enforcement actions, 282See Heckler v. Chaney, 470 U.S. 821, 831 (1985). perhaps the Supreme Court should conclude that notwithstanding the risk of strategic behavior, there is no need for safeguards in the qualified immunity context.

We recommend a middle position. In administrative law, the optimal solution is often to mitigate the danger of discretion through safeguards. Safeguards are not perfect; some misuses of discretion slip through the cracks, and sometimes agency action is wrongly condemned. But in terms of a meta cost-benefit analysis, this middle-ground approach is often a better solution than the alternatives: unchecked discretion or no discretion at all. In our view, the same sort of middle-ground approach makes sense in the qualified immunity context. In Pearson, the Justices’ unanimous opinion stated that “constitutional stagnation” is a legitimate concern. 283See Pearson v. Callahan, 555 U.S. 223, 232 (2009). The opinion also recognized, however, that Saucier was too costly. 284Id. at 236–37. Hence the Court seized on discretion as the best tool available to sort out cases in which it makes sense to decide constitutional questions from those in which it does not. 285See id. at 236. The Court’s instinct was wise; discretion can be useful. The Court erred, however, by not considering how this new discretion might be abused and by not thinking about ways to minimize that risk.

We recommend at least three devices to safeguard against these abuses: a reason-giving requirement, revised certiorari criteria, and greater informal vigilance. While these safeguards will not prevent all risks of strategic behavior, they should reduce it in a way that, on the whole, is cost justified.

A. Courts Should Be Required to Give Reasons

In The New Qualified Immunity, we explained why requiring reason-giving frequently makes sense, including in the qualified immunity context. 286Nielson & Walker, supra note 33, at 52–65. Because this recommendation was explored at length in The New Qualified Immunity, we will not repeat that discussion here. Reason-giving is often a sensible, low-cost way to mitigate concerns that discretion will be abused. Reason-giving may be more useful when the concern is that certain important values may be overlooked (which we discussed in The New Qualified Immunity) rather than when the concern is that the delegate has incentives to act strategically (the focus here). But giving reasons has value in both contexts. Again, consider administrative law. Agencies also sometimes have self-interested incentives. Nonetheless, reason-giving is still a useful curative. Some explanations, for instance, do not have the ring of truth to them. 287See, e.g., EEOC v. Asplundh Tree Expert Co., 340 F.3d 1256, 1261 (11th Cir. 2003) (rejecting agency excuses for failure to consult per the statute’s requirement).

In light of the findings in this Article, moreover, the Supreme Court should also require lower courts to give reasons for not publishing opinions, at least for opinions in which the court exercises its Pearson discretion to decide a constitutional question. Unpublished decisions, de facto and often de jure, do not carry the same weight in creating clearly established law as published opinions. Accordingly, before a court issues an unpublished qualified immunity opinion, it should explain itself. Again, this Article does not claim that this requirement will be a silver bullet. But it may help mitigate this discretionary danger.

B. For Certiorari Review, Panel Composition Should Matter More, Publication Status Less

To reform qualified immunity, moreover, we also urge the Court to be mindful of this Article’s findings when it considers whether to grant certiorari. In particular, the Justices should pay more attention to panel composition than they do today and less attention to whether an opinion is published—at least in the qualified immunity context.

For reasons explained above, homogeneous panels pose risks, at least in some respects; they are more likely to decide a constitutional question—3R panels in finding no constitutional violation, and 3D panels in recognizing new constitutional rights—including perhaps in situations where, applying the Pearson factors, a constitutional decision should not be made. At the same time, heterogeneous panels may also be dangerous in some respects, in that the panel majority may have compromised its position to avoid a dissent. Or, perhaps, the minority judge may have joined the majority as a collegial concurrence. Homogenous panels may be more or less likely than a heterogeneous panel to publish a decision finding a constitutional violation; this matters because some types of decisions should be published but are not potentially because of strategic compromise.

Similarly, the fact that a decision is unpublished should carry little weight in certiorari review. To be sure, as a formal matter, the Supreme Court can review unpublished decisions and, indeed, has already stated that “the fact that the Court of Appeals’ order under challenge here is unpublished carries no weight in our decision to review the case.” 288Comm’r v. McCoy, 484 U.S. 3, 7 (1987). Yet in practice, whether a decision is published surely does play a role in whether the Court decides to review a decision. For instance, the Office of the U.S. Solicitor General—the preeminent expert on this subject—urges the fact that a decision is unpublished as a reason to reject certiorari. 289See, e.g., Brief for the United States in Opposition at 12, Johnson v. United States, 136 S. Ct. 318 (2015) (No. 15-16) (“The unpublished decision below . . . sets no precedent.”); Brief for the Respondent in Opposition at 14, Rangolan v. Mukasey, 554 U.S. 903 (2008) (No. 07-1169) (“[T]he court of appeals’ brief, unpublished decision would not be an appropriate vehicle for resolving the question presented, because the court did not provide an explanation for why it dismissed the petition for review.”). And, indeed, the Court seldom grants certiorari for unpublished opinions. 290See Gerken, supra note 114, at 486 (“[A] review of the cases considered by the Court in any recent term demonstrates that virtually all of those cases involve a review of published courts of appeals decisions.”).

Unpublished opinions, however, may be where strategic decisionmaking often occurs, at least in the qualified immunity context. Even if judges are not hiding opinions that flout Supreme Court precedent, they may use the nonpublishing power to make them less attractive for Supreme Court review. For instance, the data show that one in five pure Saucier decisions is unpublished. This means panels are deciding novel constitutional questions, finding violations, but yet not going all the way and publishing the opinion. Likewise, certain panels with R authoring judges appear less likely to publish decisions stripping government actors of qualified immunity than other panels.

To be clear, this Article is not proposing radical change. Even in the context of qualified immunity, most lower-court decisions are sound. But at the margins, a shift in certiorari focus may reduce the incentives that judges have to use their discretion in suboptimal ways.

C. The Supreme Court Should Police Discretion Informally

Finally, there is much to be said for informal sanctions. If the Supreme Court is aware of the risk that discretion presents, it will pay more attention to it generally. If the Justices were to speak critically of using discretion in strategic ways, that itself may change practices. Judges value their reputations.

Again, administrative law provides a good comparison. For agencies, judicial review reduces “the likelihood that the agency will exercise its discretion in a way that . . . reflect[s] the agency’s self-interest.” 291Cass R. Sunstein, Factions, Self-Interest, and the APA: Four Lessons Since 1946, 72 Va. L. Rev. 271, 289 (1986). Indeed, “[i]t is a great tonic to a program to discover that even if a regulation can be slipped or wrestled through various layers of internal or external review without significant change, the final and most prestigious reviewing forum of all” will cast a skeptical eye on the agency’s action. 292William F. Pedersen, Jr., Formal Records and Informal Rulemaking, 85 Yale L.J. 38, 60 (1975) (discussing circuit courts). Review therefore “give[s] those who care about well-documented and well-reasoned decisionmaking a lever with which to move those who do not.” 293Id.

Even though the Supreme Court should not reverse a judgment because of strategic use of discretion in the qualified immunity context (after all, the same party presumably should win regardless of whether the circuit court reached the constitutional merits or whether the decision was published), it can increase the costs of strategic behavior by imposing a stigma on it. The Court, for instance, sometimes calls out a specific court by name. 294See, e.g., White v. Wheeler, 136 S. Ct. 456, 462 (2015) (per curiam) (“As a final matter, this Court again advises the Court of Appeals [for the Sixth Circuit] that the provisions of AEDPA apply with full force even when reviewing a conviction and sentence imposing the death penalty.”); Lopez v. Smith, 135 S. Ct. 1, 4 (2014) (“We have before cautioned the lower courts—and the Ninth Circuit in particular—against ‘framing our precedents at such a high level of generality.’” (citation omitted)); Ashcroft v. al-Kidd, 563 U.S. 731, 742 (2011) (“We have repeatedly told courts—and the Ninth Circuit in particular . . .—not to define clearly established law at a high level of generality.” (citation omitted)). In this context too, such public shaming strikes us as a useful weapon in the Court’s arsenal.

Conclusion

If we were all angels, there would be no need for judicial review. When it comes to discretion, however, the experience of the modern administrative state confirms that skepticism is sometimes warranted. This is why the Supreme Court is often distrustful of discretion. That distrust does not mean that discretion is not valuable or that its benefits never outweigh its costs. To the contrary, discretion is very valuable and regularly exceeds its costs. It does mean, however, that no one should blink away the reality that even dutiful agents trying to do their best may end up using discretion in suboptimal ways, even sometimes to benefit themselves.

Why are judges different? Judges deciding qualified immunity cases have been entrusted with a great deal of discretion, and they too face complex incentives. Nonetheless, when it comes to clarifying constitutional rights, the Supreme Court’s current procedural doctrine trusts judges to make decisions based solely on what is good for the law. Judges, no doubt, also generally use their discretion to serve the public interest, but there is a danger that they may not always do so. Because judges, like agency officials, are people, they face temptations to use their discretion poorly. Constitutional litigation is arguably the most important part of a judge’s job. It is essential that the Supreme Court strike the right balance when it devises procedural rules for the exercise of that awesome power. The Court has not yet done so. The empirical findings explored in this Article reinforce the first rule of administrative law: Although discretion is valuable, a healthy skepticism of it is valuable too.

Footnotes

Aaron L. Nielson is Associate Professor of Law, J. Reuben Clark Law School, Brigham Young University.Christopher J. Walker is Associate Professor of Law, Michael E. Moritz College of Law, The Ohio State University. For helpful feedback on project design and framing, thanks are due to Kent Barnett, Tara Leigh Grove, John Harrison, Michael Heise, Daniel Hemel, Randy Kozel, Nancy Leong, James Lindgren, Janet Moore, Jeff Pojanowski, Richard Re, Dan Tokaji, and Amanda Tyler. For helpful comments on earlier drafts, thanks are due to Ian Farrell, Andy Hessick, Christine Hurt, RonNell Andersen Jones, Gillian Metzger, David Moore, Paul Stancil, Peter Strauss, Mark Seidenfeld, Lisa Grow Sun, and participants at the AALS New Voices in Administrative Law Workshop, BYU Law Work-in-Progress Series, J. Reuben Clark Law Society Faculty Conference, and Rocky Mountain Junior Scholars Forum. Thanks are also due to J. Paul Norton for assistance with statistical analysis and to our research assistants Andrew Bean, Jade Bollinger, Trevor Crowley, Roman Harper, James Heilpern, Bryant Hendriksen, Ellysa Hendriksen, Andrew Henley, Steven Hilton, Neal Hoopes, Jarom Jones, Austin Martineau, Perris Nelson, Robert Patterson, Rebecca Perez, Richard Storms, Mark Tyler, Morgen Weeks, and Lani Wilson for help with case analysis. This study was funded in part by The Ohio State University Center for Interdisciplinary Law and Policy Studies and an internal research grant from Brigham Young University.

1Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2254 (2001). See also Elizabeth Magill, Foreword: Agency Self-Regulation, 77 Geo. Wash. L. Rev. 859, 859 (2009) (“Discretion is at the center of most accounts of bureaucracy.”).

2See, e.g., Geoffrey C. Shaw, H. L. A. Hart’s Lost Essay: Discretion and the Legal Process School, 127 Harv. L. Rev. 666, 668 (2013).

3See, e.g., Rachel E. Barkow, The Ascent of the Administrative State and the Demise of Mercy, 121 Harv. L. Rev. 1332, 1334 (2008) (“The expansion of the administrative state has showcased the dangers associated with the exercise of discretion.”).

4See, e.g., NetworkIP, LLC v. FCC, 548 F.3d 116, 127 (D.C. Cir. 2008) (“Complainants the agency ‘likes’ can be excused, while ‘difficult’ defendants can find themselves drawing the short straw.”); Lisa Schultz Bressman, Beyond Accountability: Arbitrariness and Legitimacy in the Administrative State, 78 N.Y.U. L. Rev. 461, 472–73 (2003) (“[A]gency discretion [may be used] to satisfy private or selfish interests at public expense.”).

5Mach Mining, LLC v. EEOC, 135 S. Ct. 1645, 1652 (2015). But see Nicholas Bagley, The Puzzling Presumption of Reviewability, 127 Harv. L. Rev. 1285, 1287 (2014) (questioning the presumption).

6See, e.g., Felix Frankfurter, The Task of Administrative Law, 75 U. Pa. L. Rev. 614, 619 (1927).

7See, e.g., Thomas W. Merrill, The Mead Doctrine: Rules and Standards, Meta-Rules and Meta-Standards, 54 Admin. L. Rev. 807, 821 (2002) (observing the “danger of lower court discretion”); cf. Kathryn A. Watts, Constraining Certiorari Using Administrative Law Principles, 160 U. Pa. L. Rev. 1, 4 (2011) (noting the dangers of Supreme Court discretion).

8See Margaret H. Lemos, The Other Delegate: Judicially Administered Statutes and the Nondelegation Doctrine, 81 S. Cal. L. Rev. 405, 407–08 (2008).

9See, e.g., Jennifer Nou, Agency Self-Insulation Under Presidential Review, 126 Harv. L. Rev. 1755, 1759–60 (2013) (discussing agency incentives); Lawrence B. Solum, The Positive Foundations of Formalism: False Necessity and American Legal Realism, 127 Harv. L. Rev. 2464, 2494–95 (2014) (book review) (discussing judicial incentives).

10Whitman v. Am. Trucking Ass’ns, Inc., 531 U.S. 457, 475 (2001) (emphasis added).

11Pearson v. Callahan, 555 U.S. 223, 236–37 (2009).

12See id. at 239 (noting that “woefully inadequate” briefs foster “bad decisionmaking”).

13John C. Jeffries, Jr., The Liability Rule for Constitutional Torts, 99 Va. L. Rev. 207, 244 (2013). But see Joanna C. Schwartz, Police Indemnification, 89 N.Y.U. L. Rev. 885, 952–60 (2014) (questioning this “chill” argument regarding whether the risk of civil liability makes government officials too cautious).

14Harlow v. Fitzgerald, 457 U.S. 800, 814 (1982).

15Messerschmidt v. Millender, 132 S. Ct. 1235, 1244 (2012).

16City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1774 n.3 (2015) (quoting Harlow, 457 U.S. at 814).

17See, e.g., John Paul Stevens, Six Amendments: How and Why We Should Change the Constitution 106 (2014) (urging that qualified immunity be banned by constitutional amendment); Stephen R. Reinhardt, The Demise of Habeas Corpus and the Rise of Qualified Immunity: The Court’s Ever Increasing Limitations on the Development and Enforcement of Constitutional Rights and Some Particularly Unfortunate Consequences, 113 Mich. L. Rev. 1219, 1244 (2015) (denouncing qualified immunity for “limiting the ability of all persons to receive the protections of the Constitution”).

18557 U.S. 364, 379 (2009).

19Sheehan, 135 S. Ct. at 1775.

20See id. at 1774 n.3 (citing five separate cases from the preceding three years).

21William Baude, Foreword: The Supreme Court’s Shadow Docket, 9 N.Y.U. J.L. & Liberty 1, 1 (2015); see also Scott Michelman, Taylor v. Barkes: Summary Reversal Is Part of a Qualified Immunity Trend, SCOTUSblog.com (June 2, 2015, 11:17 AM), http://www.scotusblog.com/2015/06/taylor-v-barkes-summary-reversal-is-part-of-a-qualified-immunity-trend/.

22Taylor v. Barkes, 135 S. Ct. 2042, 2044 (2015) (per curiam).

23See Mullenix v. Luna, 136 S. Ct. 305, 312 (2015) (per curiam). At the same time, the Court summarily vacated and remanded another case in light of Mullenix. Pickens v. Aldaba, 136 S. Ct. 479 (2015) (mem.).

24Mullenix, 136 S. Ct. at 316 (Sotomayor, J., dissenting).

25See James E. Pfander, Resolving the Qualified Immunity Dilemma: Constitutional Tort Claims for Nominal Damages, 111 Colum. L. Rev. 1601, 1604 (2011).

26See Pearson v. Callahan, 555 U.S. 223, 241 (2009) (discussing “the ‘older, wiser . . . counsel “not to pass . . . constitutionality . . . unless such adjudication is unavoidable”’” (quoting Scott v. Harris, 550 U.S. 372, 388 (2007) (Breyer, J., concurring))).

27Pearson, 555 U.S. at 232.

28Id. at 236, 239.

29See, e.g., Wesby v. District of Columbia, 816 F.3d 96, 102 (D.C. Cir. 2016) (Kavanaugh, J., dissenting from the denial of rehearing en banc) (“Indeed, in just the past five years, the Supreme Court has issued 11 decisions reversing federal courts of appeals in qualified immunity cases, including five strongly worded summary reversals.”) (collecting authorities).

30See, e.g., Nancy Leong, Improving Rights, 100 Va. L. Rev. 377, 396–409 (2014); Nancy Leong, Rethinking the Order of Battle in Constitutional Torts, 105 Nw. U. L. Rev. 969, 973–77 (2011) [hereinafter Leong, Rethinking the Order of Battle]; Reinhardt, supra note 17, at 1250.

31Reinhardt, supra note 17, at 1249–50 (criticizing Pearson).

32Camreta v. Greene, 563 U.S. 692, 726 (2011) (Kennedy, J., dissenting); id. at 714 (Scalia, J., concurring).

33Aaron L. Nielson & Christopher J. Walker, The New Qualified Immunity, 89 S. Cal. L. Rev. 1, 6 (2015). Part IV.B provides an overview of the findings from The New Qualified Immunity, which examines post-Pearson constitutional stagnation concerns, disparities across circuits, and a preliminary inquiry into the role of ideology. See also Nancy Leong, A Fresh Look at Qualified Immunity, Jotwell (Dec. 3, 2015), http://courtslaw.jotwell.com/?s=a+fresh+look+at+qualified+immunity (reviewing Nielson & Walker, supra). This Article leverages that data to look beyond how Pearson discretion has been used in the circuit courts to explore theoretically and empirically the more troubling unintended consequence of this discretion: how it encourages strategic judicial behavior.

34See, e.g., Jack M. Beermann, Qualified Immunity and Constitutional Avoidance, 2009 Sup. Ct. Rev. 139, 143 (noting the risk of strategic behavior).

35See id. at 142.

36Pearson v. Callahan, 555 U.S. 223, 236 (2009).

37Id. at 237.

38See, e.g., Frank B. Cross, Collegial Ideology in the Courts, 103 Nw. U. L. Rev. 1399, 1399–1400 (2009) (“Collegiality and ideology inevitably conflict.”).

39See, e.g., Lee Epstein & Gary King, The Rules of Inference, 69 U. Chi. L. Rev. 1, 107–08 (2002) (describing unpublished opinions).

40Nielson & Walker, supra note 33, at 23–24.

41See infra Part IV.C.1.

42See id.

43See infra note 276 and accompanying text.

44See infra Part IV.C.1.

45See id.

46See Nielson & Walker, supra note 33, at 61.

47Cf. Rachel E. Barkow, Institutional Design and the Policing of Prosecutors: Lessons from Administrative Law, 61 Stan. L. Rev. 869, 871 (2009).

48See Nielson & Walker, supra note 33, at 15–23.

49Pearson v. Callahan, 555 U.S. 223, 236–37 (2009).

50John C. Jeffries, Jr., What’s Wrong with Qualified Immunity?, 62 Fla. L. Rev. 851, 851–52 (2010).

51See, e.g., Marbury v. Madison, 5 U.S. (1 Cranch) 137, 163 (1803) (“[This is] a government of laws, and not of men. It will certainly cease to deserve this high appellation, if the laws furnish no remedy for the violation of a vested legal right.”). But see Erwin Chemerinsky, Federal Jurisdiction § 7.1, at 421–25 (Wolters Kluwer Law & Bus. 6th ed. 2012) (noting limits on this principle).

52See, e.g., United States v. Alvarez, 132 S. Ct. 2537, 2555 (2012) (Breyer, J., concurring in the judgment) (explaining the “chilling” effect liability can have).

53See, e.g., Reinhardt, supra note 17, at 1244; Jeffries, supra note 13, at 241 & n.121 (collecting authorities).

54See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1774 n.3 (2015) (collecting authorities).

55Id. (explaining that “the Court often corrects lower courts when they wrongly subject individual officers to liability” and collecting authorities).

56See Harlow v. Fitzgerald, 457 U.S. 800, 806 (1982) (explaining that the threat of damages can cause “undue interference” in officers’ responsibilities).

57Id. at 814 (1982).

58See, e.g., Sheehan, 135 S. Ct. at 1775 (“Reynolds and Holder knew that Sheehan had a weapon and had threatened to use it to kill three people. They also knew that delay could make the situation more dangerous.”); Plumhoff v. Rickard, 134 S. Ct. 2012, 2022 (2014) (“[I]t was Rickard who put Allen in danger by fleeing and refusing to end the chase, and it would be perverse if his disregard for Allen’s safety worked to his benefit.”).

59See, e.g., Wilkinson v. Russell, 182 F.3d 89, 99 (2d Cir. 1999) (awarding qualified immunity to officials investigating alleged child sexual abuse).

60This same analysis applies to a regime in which the law imposes strict liability on officers but also provides automatic employer indemnification. Under either regime, the money comes from the same place. See Jeffries, supra note 13, at 232.

61See, e.g., Jeffries, supra note 13, at 232, 241 n.122; see also Dave Munday & Glenn Smith, Small Town’s Future Uncertain After ‘Astronomical’ Jury Verdict, Post & Courier (Oct. 16, 2014, 10:00 PM), http://www.postandcourier.com/article/20141016/PC16/141019471/1177 (“A mix of shock, joy and uncertainty buzzed in this tiny, rural community Thursday as residents contemplated how the town might pay a court judgment so large it could fund the local budget for the next 162 years.”); Bruce Schreiner, Kentucky City Files for Bankruptcy to Address Judgment—$11.4M Verdict 4 Times Hillview’s Budget, Lexington Herald-Leader, Aug. 22, 2015, at A5 (“A suburban city near Louisville has filed for bankruptcy, crippled by an $11.4 million jury verdict—nearly four times the city’s budget . . . .”).

62See Pfander, supra note 25, at 1615 (citing, inter alia, John C. Jeffries, Jr., The Right-Remedy Gap in Constitutional Law, 109 Yale L.J. 87, 98–105 (1999)).

63457 U.S. 800, 818 (1982).

64Aschroft v. al-Kidd, 563 U.S. 731, 735 (2011) (citing Harlow, 457 U.S. at 818). Needless to say, this standard for establishing qualified immunity varies with the procedural posture of a case; a plaintiff’s burden is different at the motion-to-dismiss stage than at the summary-judgment or trial stage. See, e.g., Saucier v. Katz, 533 U.S. 194, 201 (2001).

65Al-Kidd, 563 U.S. at 741 (quoting Anderson v. Creighton, 483 U.S. 635, 640 (1987)).

66Id.

67Hope v. Pelzer, 536 U.S. 730, 741 (2002).

68See, e.g., Michael S. Catlett, Clearly Not Established: Decisional Law and the Qualified Immunity Doctrine, 47 Ariz. L. Rev. 1031, 1034 (2005) (“Although this inquiry sounds somewhat straightforward, in practice it has created a number of recurring issues.”); Diana Hassel, Excessive Reasonableness, 43 Ind. L. Rev. 117 (2009) (discussing qualified immunity in excessive force cases).

69See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1775–76 (2015) (“Qualified immunity is no immunity at all if ‘clearly established’ law can simply be defined as the right to be free from unreasonable searches and seizures.”).

70Harlow v. Fitzgerald, 457 U.S. 800, 818 (1982).

71Pearson v. Callahan, 555 U.S. 223, 232 (2009) (quoting Hunter v. Bryant, 502 U.S. 224, 227 (1991)); see also id. at 231 (“[T]he ‘driving force’ behind creation of the qualified immunity doctrine was a desire to ensure that ‘insubstantial claims’ . . . [will] be resolved prior to discovery.”).

72Morse v. Frederick, 551 U.S. 393, 431 (2007) (Breyer, J., concurring in the judgment in part and dissenting in part) (quoting PDK Labs., Inc. v. DEA, 362 F.3d 786, 799 (D.C. Cir. 2004) (Roberts, J., concurring in part and concurring in judgment)); see also Ashwander v. Tenn. Valley Auth., 297 U.S. 288, 347 (1936) (Brandeis, J., concurring); Burton v. United States, 196 U.S. 283, 295 (1905); Ex parte Randolph, 20 F. Cas. 242, 254 (C.C.D. Va. 1833) (No. 11,558) (Marshall, J.).

73Morse, 551 U.S. at 430 (Breyer, J., concurring in the judgment in part and dissenting in part).

74Nielson & Walker, supra note 33, at 14 (quoting Nancy Leong, The Saucier Qualified Immunity Experiment: An Empirical Analysis, 36 Pepp. L. Rev. 667, 680–81 (2009)).

75See, e.g., Fed. R. Crim. P. 50 (“Scheduling preference must be given to criminal proceedings as far as practicable.”); cf. 28 U.S.C. § 2102 (2012) (“Criminal cases . . . shall have priority, on the docket of the Supreme Court, over all cases except cases to which the United States is a party and such other cases as the court may decide to be of public importance.”).

76Beermann, supra note 34, at 149.

77Nielson & Walker, supra note 33, at 12 (quoting Saucier v. Katz, 533 U.S. 194, 201 (2001)).

78Id.

79Id.

80See, e.g., Steffel v. Thompson, 415 U.S. 452, 482–83 (1974) (Rehnquist, J., concurring) (“[T]he defendant . . . is able to present his case . . . to preserve the defendant’s constitutional rights.”); Wisconsin v. Yoder, 406 U.S. 205, 234 (1972) (recognizing that defendants may raise the First or Fourth Amendments as a defense).

81Nielson & Walker, supra note 33, at 12.

82See, e.g., Connick v. Thompson, 563 U.S. 51, 54 (2011) (holding that a “failure to train” theory requires more than a single violation of Brady v. Maryland, 373 U.S. 83 (1963), even if the consequence was that an innocent man sat on death row for nearly two decades).

83See, e.g., City of Los Angeles v. Lyons, 461 U.S. 95, 111 (1983).

84See, e.g., Hudson v. Michigan, 547 U.S. 586, 591 (2006) (limiting the rule’s application to “where its remedial objectives are thought most efficaciously served” (quoting United States v. Calandra, 414 U.S. 338, 348 (1974))).

85See Pamela S. Karlan, Shoe-Horning, Shell Games, and Enforcing Constitutional Rights in the Twenty-First Century, 78 UMKC L. Rev. 875, 887–88 (2010).

86See Rachel E. Barkow, Institutional Design and the Policing of Prosecutors: Lessons from Administrative Law, 61 Stan. L. Rev. 869, 871 (2009).

87Saucier v. Katz, 533 U.S. 194, 200 (2001).

88Id. at 201.

89Id.

90Id.

91Nielson & Walker, supra note 33, at 17.

92Bunting v. Mellen, 541 U.S. 1019 (Stevens, J., respecting the denial of certiorari).

93Morse v. Frederick, 551 U.S. 393, 432 (2007) (Breyer, J., concurring in the judgment in part and dissenting in part).

94Pearson v. Callahan, 555 U.S. 223, 232, 236–37 (2009).

95Id. at 236.

96The factors in favor of deciding a constitutional question include whether: (1) “there would be little if any conservation of judicial resources” in not deciding it; (2) it would “be difficult to decide whether a right is clearly established without deciding precisely what the existing constitutional right happens to be”; (3) the “two-step procedure” would “promote[] the development of constitutional precedent”; and (4) the question is not one that “frequently arise[s] in cases in which a qualified immunity defense is unavailable.” Id.

97The factors against deciding a constitutional question include whether: (1) doing so would result in “substantial expenditure of scarce judicial resources”; (2) “it is plain that a constitutional right is not clearly established but far from obvious whether in fact there is such a right”; (3) the court has a “heavy caseload[]”; (4) additional “litigation of constitutional issues [would] waste[] the parties’ resources”; (5) the “question is so factbound that the decision provides little guidance for future cases”; (6) “the question will soon be decided by a higher court”; (7) “an uncertain interpretation of state law” is involved; (8) “‘the parties have provided very few facts to define and limit any holding’ on the constitutional question”; (9) “the briefing of constitutional questions is woefully inadequate”; (10) it would be “hard for affected parties to obtain appellate review of constitutional decisions that may have a serious prospective effect on their operations”; and (11) in a particular case, “constitutional avoidance” is unusually important. Id. at 236–42.

98See, e.g., Plumhoff v. Rickard, 134 S. Ct. 2012, 2020 (2014).

99See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1775 (2015) (declining to exercise Pearson discretion because the constitutional briefing was inadequate).

100See Charles Alan Wright et al., Federal Practice and Procedure § 3978.10, at 382 (4th ed. 2008) (discussing unpublished opinions); Deborah Jones Merritt & James J. Brudney, Stalking Secret Law: What Predicts Publication in the United States Courts of Appeal, 54 Vand. L. Rev. 71, 72 (2001) (discussing the prevalence of unpublished opinions).

101Epstein & King, supra note 39, at 107 n.335 (citing Richard S. Arnold, Unpublished Opinions: A Comment, 1 J. App. Prac. & Process 219, 220 (1999)).

102See Mata v. Lynch, 135 S. Ct. 2150, 2155 n.3 (2015) (noting that unpublished opinions in some circuits have “no precedential force”); Unpublished Opinion, Black’s Law Dictionary (10th ed. 2014) (explaining that an “unpublished opinion” generally “is considered binding only on the parties to the particular case in which it is issued”).

103See, e.g., Wright et al., supra note 100, at § 3978.10, at 382–86 (discussing the history of court rules allowing citation to unpublished opinions); Caleb E. Mason, An Aesthetic Defense of the Nonprecedential Opinion: The Easy Cases Debate in the Wake of the 2007 Amendments to the Federal Rules of Appellate Procedure, 55 UCLA L. Rev. 643, 646 (2008) (collecting authorities relating to “[t]he jurisprudential, political, equitable, ethical, and constitutional issues raised by the use of [unpublished opinions]” (footnotes omitted)).

104See Richard B. Cappalli, The Common Law’s Case Against Non-Precedential Opinions, 76 S. Cal. L. Rev. 755, 759 (2003).

105Id. at 761 (footnotes omitted).

106Id. at 762.

107UC Health v. NLRB, 803 F.3d 669, 685 (D.C. Cir. 2015) (Silberman, J., dissenting).

108Boyce F. Martin, Jr., In Defense of Unpublished Opinions, 60 Ohio St. L.J. 177, 177–78 (1999).

109Id. at 178 (quoting Paul D. Carrington et al., Justice on Appeal 35 (1976)).

110See Philip Nichols, Jr., Selective Publication of Opinions: One Judge’s View, 35 Am. U. L. Rev. 909, 927 (1986) (noting “the absurdity of destroying forests to distribute masses of prolix and repetitious material”).

111Advisory Council on Appellate Justice, FJC Research Series No. 73-2, Standards for Publication of Judicial Opinions 3 (1973).

112See Fed. R. App. P. 32.1.

113Martin, supra note 108, at 184–85.

114See Joseph L. Gerken, A Librarian’s Guide to Unpublished Judicial Opinions, 96 L. Libr. J. 475, 478 (2004) (noting that about 80% of appellate decisions are unpublished).

115See Marin K. Levy, The Mechanics of Federal Appeals: Uniformity and Case Management in the Circuit Courts, 61 Duke L.J. 315, 361–64 (2011).

116See, e.g., William L. Reynolds & William M. Richman, The Non-Precedential Precedent—Limited Publication and No-Citation Rules in the United States Courts of Appeals, 78 Colum. L. Rev. 1167, 1176 (1978).

117See Plumley v. Austin, 135 S. Ct. 828, 831 (2015) (Thomas, J., dissenting from the denial of certiorari) (discussing Fourth Circuit standards for publication, one of which is whether the case “involve[s] a legal issue of continuing public interest”); Epstein & King, supra note 39, at 107 & nn.335–36 (explaining that while judges should publish “opinions that are ‘of general precedential value,’” in practice the standard “is sufficiently vague to permit circuit court judges to publish or not as they see fit”).

118See, e.g., Mahogany v. La. State Supreme Court, 262 Fed. App’x 636, 637 (5th Cir. 2008) (“Just as one cannot sue . . . the color teal, the number thirteen, or the weather, a complaint filed against a statute fails to state a claim upon which relief can be granted.”).

119See Alex Kozinski & Stephen Reinhardt, Please Don’t Cite This!: Why We Don’t Allow Citation to Unpublished Dispositions, Cal. Law., June 2000, at 43 (“After carefully reviewing the briefs and record, we can succinctly explain who won, who lost, and why. We need not state the facts, as the parties already know them; nor need we announce a rule general enough to apply to future cases.”).

120See, e.g., Hurt v. Soc. Sec. Admin., 544 F.3d 308, 308–09 (D.C. Cir. 2008) (“In just the last couple of years, Hurt has sued the Declaration of Independence, Black’s Law Dictionary, the United Nations, agencies of the District of Columbia and the Federal Government, and various courts and their officers. Hurt has claimed the existence of state supreme courts violates the Eighth Amendment, requested the Secret Service and the President’s Cabinet be declared unconstitutional, and demanded the deportation of a Spanish-speaking government employee. Nor are the slights Hurt suffered mere glancing blows; he routinely demands trillions of dollars in damages.” (citations omitted)) (listing forty-four dockets).

121Cappalli, supra note 104, at 759.

122See Fed. R. App. P. 32.1 & advisory committee’s notes.

123Patrick J. Schiltz, Much Ado About Little: Explaining the Sturm und Drang over the Citation of Unpublished Opinions, 62 Wash. & Lee L. Rev. 1429, 1444–45 (2005).

124Fed. R. App. P. 32.1.

125See Schiltz, supra note 123, at 1470, 1475.

126Id. at 1473 (“[T]he Advisory Committee Note to Rule 32.1 went out of its way to be clear that the rule did not forbid any court to issue an unpublished opinion, dictate the circumstances under which a court may choose to designate an opinion as unpublished, or imply anything about what effect a court must give to one of its unpublished opinions.”).

127See David R. Cleveland, Clear as Mud: How the Uncertain Precedential Status of Unpublished Opinions Muddles Qualified Immunity Determinations, 65 U. Miami L. Rev. 45, 64 (2010) (“[C]ourts have taken radically different interpretations of what sources may clearly establish the law.”).

128See id. at 71–74 (describing the Ninth Circuit as the one circuit that “has unequivocally stated a willingness to use unpublished opinions” to show clearly established law, and the Third and Sixth Circuits as “seem[ing] to allow the use of unpublished opinions in clearly establishing the law”).

129See, e.g., City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1776 (2015) (rejecting the application of legal principles at a high level of abstraction).

130Cf. Jackler v. Byrne, 658 F.3d 225, 244 (2d Cir. 2011) (“[S]uch orders, being summary, frequently do not set out the factual background of the case in enough detail to disclose whether its facts are sufficiently similar to those of a subsequent unrelated case to make our summary ruling applicable to the new case.”).

131See, e.g., David S. Rubenstein, “Relative Checks”: Towards Optimal Control of Administrative Power, 51 Wm. & Mary L. Rev. 2169, 2188–89 (2010).

132See, e.g., Jonathan R. Macey, Administrative Agency Obsolescence and Interest Group Formation: A Case Study of the SEC at Sixty, 15 Cardozo L. Rev. 909, 913 (1994) (“[O]ver time, all bureaucracies will substitute private, bureaucratic objectives for the public objectives that characterized their origination.” (citing Anthony Downs, Inside Bureaucracy (1967))).

133See Aaron L. Nielson, In Defense of Formal Rulemaking, 75 Ohio St. L.J. 237, 243–44 (2014) (explaining that in 1946, the Chair of the Senate Judiciary Committee “proclaimed that the APA ‘cut down the “cult of discretion” so far as federal law is concerned’” (quoting George B. Shepherd, Fierce Compromise: The Administrative Procedure Act Emerges from New Deal Politics, 90 Nw. U. L. Rev. 1557, 1666 (1996))).

134See Cass R. Sunstein, The Office of Information and Regulatory Affairs: Myths and Realities, 126 Harv. L. Rev. 1838, 1845–46 (2013) (discussing OIRA review).

135318 U.S. 80, 87 (1943); see also Kevin M. Stack, The Constitutional Foundations of Chenery, 116 Yale L.J. 952, 958 (2007) (explaining that Chenery helps “prevent arbitrariness in the agency’s exercise of its discretion”).

136Barkow, supra note 3, at 1334.

137See Charles H. Koch, Jr., Judicial Review of Administrative Discretion, 54 Geo. Wash. L. Rev. 469, 472 n.8 (1986) (“Rules alone, untempered by discretion, cannot cope with the complexities of modern government and of modern justice.” (quoting K. Davis, Discretionary Justice: A Preliminary Inquiry 25 (1969))).

138See, e.g., Bus. Elecs. Corp. v. Sharp Elecs. Corp., 485 U.S. 717, 731 (1988) (“The term ‘restraint of trade’ in the statute, like the term at common law, refers not to a particular list of agreements, but to a particular economic consequence, which may be produced by quite different sorts of agreements in varying times and circumstances.”).

139The delegate, of course, need not be an agency; it can very well be, and often is, a court. See Lemos, supra note 8, at 429; cf. Shaw, supra note 2, at 669 (“[D]iscretion is a necessary component of any legal system, because society’s ability to regulate the future is inherently limited by imperfect information and an imperfect understanding of aims.”).

140371 U.S. 156, 167 (1962); see also Sidney Shapiro, Elizabeth Fisher & Wendy Wagner, The Enlightenment of Administrative Law: Looking Inside the Agency for Legitimacy, 47 Wake Forest L. Rev. 463, 463–64 (2012) (defending this view of administrative law and “the legitimacy of administrative discretion”).

141Aaron L. Nielson, Visualizing Change in Administrative Law, 49 Ga. L. Rev 757, 768–69 (2015) (citing James M. Landis, The Administrative Process 70 (1966)).

142See Kagan, supra note 1, at 2261 (explaining that Landis thought that expertise could “solv[e] the problem of administrative discretion”).

143See Wong Yang Sung v. McGrath, 339 U.S. 33, 36–37, 41 (1950) (recognizing the “evils” associated with the “[m]ultiplication of federal administrative agencies and expansion of their functions” over the preceding “past half-century”); Kagan, supra note 1, at 2261 (“[Landis’s] idea today seems almost quaint, and even then it provoked strong opposition.”); Shaw, supra note 2, at 668 (“The problem of discretion became more pressing after the New Deal.”).

144See Mark Seidenfeld, A Civic Republican Justification for the Bureaucratic State, 105 Harv. L. Rev. 1511, 1563 (1992) (“[R]egulators will often pursue their own private interests. This pursuit partially explains the failure of the New Deal agency model, which naively assumed that regulators would diligently implement statutory goals.”).

145Id.

146See id. at 1564 (“In addition to shirking, officials’ self-interest also creates incentives for them to augment their regulatory authority. Greater power allows regulators to increase monopoly rents, which they can then trade to interest groups in return for personal benefits such as future jobs or freedom from criticism.” (footnotes omitted)).

147Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 167 (1962) (quoting New York v. United States, 342 U.S. 882, 884 (1951) (Douglas, J., dissenting)).

148New York, 342 U.S. at 884 (Douglas, J., dissenting).

149See Shaw, supra note 2, at 668 (“Rule by arbitrary choice is not rule by law, and thus the lasting legacy of legal realism became a challenge: explain how legal indeterminacy can be reconciled with the rule of law.”).

150Shepherd, supra note 133, at 1557, 1681.

151Wong Yang Sung v. McGrath, 339 U.S. 33, 37 (1950).

152John H. Jackson, Perspective on the Jurisprudence of International Trade: Costs and Benefits of Legal Procedures in the United States, 82 Mich. L. Rev. 1570, 1578–79 (1984); see also Shapiro et al., supra note 140, at 464 (“[P]rocedure after procedure has been added in a vain effort to eliminate discretion.”).

153See Perez v. Mortg. Bankers Ass’n, 135 S. Ct. 1199, 1203–04 (2015) (upholding discretion without notice-and-comment procedures).

154Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 474 (2001) (discussing unanimous rejection of substantive policy-making discretion as unconstitutional in A.L.A. Schechter Poultry Corp. v. United States, 295 U.S. 495 (1935)).

155See, e.g., Holt v. Hobbs, 135 S. Ct. 853, 863 (2015) (“We readily agree that the Department has a compelling interest in staunching the flow of contraband into and within its facilities, but the argument that this interest would be seriously compromised by allowing an inmate to grow a ½-inch beard is hard to take seriously.”).

156See Heckler v. Chaney, 470 U.S. 821, 831 (1985).

157See id. at 832 (“[W]hen an agency refuses to act it generally does not exercise its coercive power over an individual’s liberty or property rights, and thus does not infringe upon areas that courts often are called upon to protect.”).

158See id. at 831–32 (“The agency is far better equipped than the courts to deal with the many variables involved in the proper ordering of its priorities.”).

159See Vt. Yankee Nuclear Power Corp. v. NRDC, Inc., 435 U.S. 519, 544–45 (1978).

160Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208 (1988).

161See id. (“The power to require readjustments for the past is drastic.” (quoting Brimstone R.R. & Canal Co. v. United States, 276 U.S. 104, 122 (1928))).

162Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962)).

163135 S. Ct. 1645, 1652 (2015).

164Id. at 1649.

165Id. at 1650.

166Id. at 1649 (quoting 42 U.S.C. § 2000e–5(f)(1)).

167Id. (citing 42 U.S.C. § 2000e–5(b)).

168Id. at 1652.

169Id. at 1652–53; see also id. (“We need only know—and know that Congress knows—that legal lapses and violations occur, and especially so when they have no consequence.”); EEOC v. Propak Logistics, Inc., 746 F.3d 145, 155–56 (4th Cir. 2014) (Wilkinson, J., concurring) (explaining that to avoid encouraging “sub-optimal agency behavior,” judges must remember the “danger that those inside a public bureaucracy, armed with significant resources, authority, and discretion, may become gradually numb as to how their actions affect those outside” the agency).

170132 S. Ct. 1367, 1374 (2012).

171Id.

172Id. at 1375 (Alito, J., concurring).

173136 S. Ct. 1807, 1816 (2016) (holding that jurisdictional determinations are also subject to immediate review).

174325 U.S. 410, 414 (1945).

175See, e.g., Perez v. Mortg. Bankers Ass’n, 135 S. Ct. 1199, 1210–13, 1217–20 (2015) (separate opinions of Scalia, Thomas, and Alito, J.J.); Decker v. Nw. Envtl. Def. Ctr., 133 S. Ct. 1326, 1338 (2013) (Roberts, C.J., concurring); id. at 1339–42 (Scalia, J., concurring in part and dissenting in part).

176Christopher v. SmithKline Beecham Corp., 132 S. Ct. 2156, 2168 (2012).

177Arnold, supra note 101, at 223.

178But see Gillian E. Metzger, Embracing Administrative Common Law, 80 Geo. Wash. L. Rev. 1293, 1293, 1295 (2012) (“[M]uch of administrative law is really administrative common law, [though] statutes exert some constraining force on judicial creativity.”).

179See, e.g., Dickerson v. United States, 530 U.S. 428, 437 (2000) (“The law in this area is clear. This Court has supervisory authority over the federal courts, and we may use that authority to prescribe rules of evidence and procedure that are binding in those tribunals.”); see also James E. Pfander, One Supreme Court: Supremacy, Inferiority, and the Judicial Power of the United States 148–49 (2009) (explaining that constitutional structure suggests the Court has a supreme or supervisory role over the lower courts); Tara Leigh Grove, The Structural Case for Vertical Maximalism, 95 Cornell L. Rev. 1, 4 (2009) (arguing that the vertical structure of the judiciary “preserve[s] [the] ‘supreme’ role” of the Court). But see Amy Coney Barrett, The Supervisory Power of the Supreme Court, 106 Colum. L. Rev. 324, 360–66 (2006) (casting some doubt on this power).

180555 U.S. 223, 236 (2009).

181See id. at 236–37.

182It is possible to partially publish opinions. See, e.g., Adena Reg’l Med. Ctr. v. Leavitt, 527 F.3d 176, 180 n.* (D.C. Cir. 2008) (“The Hospitals’ other arguments are sufficiently lacking in merit as not to warrant consideration in a published opinion”). In theory then, a court could publish its resolution of one claim but not of another.

183Richard J. Lazarus, The Opinion Assignment Power, Justice Scalia’s Un-Becoming, and UARG’s Unanticipated Cloud over the Clean Air Act, 39 Harv. Envtl. L. Rev. 37, 45 (2015) (“Justices must frequently join final opinions even though they do not agree with every word or nuance.”).

184See, e.g., Amy E. Sloan, If You Can’t Beat ‘Em, Join ‘Em: A Pragmatic Approach to Nonprecedential Opinions in the Federal Appellate Courts, 86 Neb. L. Rev. 895, 946 n.308 (2008) (“[J]udges may join an opinion when they agree with the result, even if they do not agree with the reasoning, when they know the opinion will be nonprecedential.” (citing Advisory Comm. on Appellate Rules, Minutes of the Fall 2002 Meeting 36–37 (Nov. 18, 2002), http://www.uscourts.gov/rules/Minutes/appll02.pdf)).

185See Pearson v. Callahan, 555 U.S. 223, 236–37 (2009).

186488 Fed. App’x 933, 935 (6th Cir. 2012).

187Id.

188Id. at 948.

189See, e.g., Solum, supra note 9, at 2484–85 (discussing the nuanced factors that must go into a correct judicial utility function).

190See, e.g., id. at 2489 (“Of course, the relative strength of legal and policy preferences could vary from judge to judge.”).

191Id.; see also Pauline T. Kim, Lower Court Discretion, 82 N.Y.U. L. Rev. 383, 396 (2007) (“The theory that judges single-mindedly pursue their policy preferences is incompatible with the numerous studies concluding that judges shift the nature of their reasoning and the general trend of their decisions in response to Supreme Court precedent.”); Alex Kozinski & Fred Bernstein, Clerkship Politics, 2 Green Bag 57, 59–60 (1998) (noting that ideology’s impact on decisionmaking is overstated).

192See, e.g., Solum, supra note 9, at 2467 (discussing literature and critiquing Lee Epstein, William M. Landes & Richard A. Posner, The Behavior of Federal Judges: A Theoretical and Empirical Study of Rational Choice (2013)).

193See, e.g., Michael A. Bailey & Forrest Maltzman, The Constrained Court: Law, Politics, and the Decisions Justices Make 2 (2011); Frank B. Cross, Decision Making in the U.S. Courts of Appeals 18 (2007); Pauline T. Kim, Deliberation and Strategy on the United States Courts of Appeals: An Empirical Exploration of Panel Effects, 157 U. Pa. L. Rev. 1319 (2009); McNollgast, Politics and the Courts: A Positive Theory of Judicial Doctrine and the Rule of Law, 68 S. Cal. L. Rev. 1631 (1995).

194Solum, supra note 9, at 2487.

195See, e.g., Walker v. Tex. Div., Sons of Confederate Veterans, Inc., 135 S. Ct. 2239, 2248–51 (2015) (reading Pleasant Grove City v. Summum, 555 U.S. 460 (2009), more broadly than its author intended); Ashcroft v. Iqbal, 556 U.S. 662, 684 (2009) (reading Bell Atlantic Corp. v. Twombly, 550 U.S. 544 (2007), more broadly than its author intended); SEC v. Chenery Corp., 332 U.S. 194, 207 (1947) (reading SEC v. Chenery Corp., 318 U.S. 80 (1943), more narrowly than its author intended).

196David E. Klein & Robert J. Hume, Fear of Reversal as an Explanation of Lower Court Compliance, 37 Law & Soc’y Rev. 579, 597 (2003) (explaining with empirical support that fear of reversal cannot explain all aspects of law compliance).

197Mark Herrmann, The Curmudgeon’s Guide to Practicing Law 4 (2006) (“Judges do not like to be reversed. Accordingly, if a precedent contains the implicit threat of reversal, I will use that threat (gently, of course) when I discuss that case.”); Nou, supra note 9, at 1771–72 (explaining, in the context of district courts, that reversals impose “reputational costs”) (collecting authorities).

198Nou, supra note 9, at 1756 (analogizing agencies and courts).

199See Virginia A. Hettinger, Stefanie A. Lindquist & Wendy L. Martinek, Judging on a Collegial Court: Influences on Federal Appellate Decision Making 112 (2006).

200See Frank B. Cross, Collegial Ideology in the Courts, 103 Nw. U. L. Rev. 1399, 1399–1400 (2009) (“Collegiality and ideology inevitably conflict. Given the range of ideological preferences and political ‘hot button’ issues facing courts today, one would expect judges to disagree about the resolution of certain legal issues. Because such disagreement puts collegiality and individuality in conflict, any model of judicial decisionmaking must account for the interaction of these potentially opposing factors.” (footnote omitted)); Frank B. Cross & Emerson H. Tiller, Judicial Partisanship and Obedience to Legal Doctrine: Whistleblowing on the Federal Courts of Appeals, 107 Yale L.J. 2155 (1998) (discussing the effect of a “whistleblower” on the court); Solum, supra note 9, at 2471 (“[D]issents will be more likely when the ideological stakes are high, because writing dissents may entail collegiality costs and always involves an effort cost . . . .”).

201See Aaron L. Nielson, Reflections on the End of the Federal Law Clerk Hiring Plan, 112 Mich. L. Rev. First Impressions 22, 27 (2013) (explaining the “human nature” of judges).

202Cf. Mach Mining, LLC v. EEOC, 135 S. Ct. 1645, 1652 (2015) (explaining that the Court “need not doubt the EEOC’s trustworthiness” to worry about discretion).

203New York v. United States, 342 U.S. 882, 884 (1951) (Douglas, J., dissenting).

204See, e.g., Arnold, supra note 101, at 223 (noting even good-faith self-deception).

205See Plumhoff v. Rickard, 134 S. Ct. 2012, 2020 (2014) (exercising Pearson discretion because those qualities existed).

206See Sup. Ct. R. 10 (“A petition for a writ of certiorari is rarely granted when the asserted error consists of . . . misapplication of a properly stated rule of law.”).

207Cf. City of San Francisco v. Sheehan, 135 S. Ct. 1765, 1774 (2015) (noting poor briefing).

208Id. (explaining that although the Court has discretion to decide unbriefed constitutional questions, it is not prudent to do so).

209Patricia M. Wald, The Rhetoric of Results and the Results of Rhetoric: Judicial Writings, 62 U. Chi. L. Rev. 1371, 1374 (1995).

210Id.

211Arnold, supra note 101, at 223 (noting the risk that judges who cannot distinguish a case may opt for “an abbreviated, unpublished opinion, and no one will ever be the wiser”).

212Plumley v. Austin, 135 S. Ct. 828, 831 (2015) (Thomas, J., dissenting from the denial of certiorari) (citations omitted).

213Epstein & King, supra note 39, at 108 (footnote omitted).

214David C. Vladeck & Mitu Gulati, Judicial Triage: Reflections on the Debate over Unpublished Opinions, 62 Wash. & Lee L. Rev. 1667, 1680 (2005). Of course, even if unpublished opinions are not used strategically, they may still be problematic. See, e.g., Scott E. Gant, Missing the Forest for a Tree: Unpublished Opinions and New Federal Rule of Appellate Procedure 32.1, 47 B.C. L. Rev. 705, 735 (2006) (arguing that judges do not know whether an opinion will be important); Penelope Pether, Inequitable Injunctions: The Scandal of Private Judging in the U.S. Courts, 56 Stan. L. Rev. 1435, 1484 (2004) (“Even if there were no credible evidence of misconduct or structural inequality in the operation of unpublication, the lack of transparency it produces would damage the legitimacy of the judicial system . . . .”). But there are benefits too. The issue is whether the costs outweigh the benefits.

215Denise M. Keele et al., An Analysis of Ideological Effects in Published Versus Unpublished Judicial Opinions, 6 J. Empirical Legal Stud. 213, 230, 234 (2009) (emphasis added); see also David S. Law, Strategic Judicial Lawmaking: Ideology, Publication, and Asylum Law in the Ninth Circuit, 73 U. Cin. L. Rev. 817, 820 (2005) (explaining that judges may try to “steer the evolution of the law in a particular ideological direction by publishing opinions they like” and offering some empirical data).

216Alex Kozinski, In Opposition to Proposed Federal Rule of Appellate Procedure 32.1, 51 Fed. Law., June 2004, at 36, 39.

217See, e.g., Marla Brooke Tusk, Note, No-Citation Rules as a Prior Restraint on Attorney Speech, 103 Colum. L. Rev. 1202, 1216 (2003) (“There is even some speculation that, because the Supreme Court is less likely to grant certiorari to an appeal from an unpublished opinion, appellate judges may decide controversial cases via unpublished opinions simply to insulate those decisions from Supreme Court review.”).

218See, e.g., Hope v. Pelzer, 536 U.S. 730, 743 (2002) (not requiring “identical” facts).

219As detailed in The New Qualified Immunity, we focused on 2009 to 2012 because those cases created a data set large enough for analysis; cases nearer in time to Pearson are more likely to cite Pearson instead of circuit-specific precedent; and we wanted to avoid cases that may have come before the Supreme Court while one of us, Nielson, was clerking there. See Nielson & Walker, supra note 33, at 30 n.185. For the purposes of The New Qualified Immunity, the focus on cases citing Pearson made sense as we were trying to assess the effect of Pearson in the lower courts. As discussed below, the exclusive focus on Pearson in this study imposes methodological limitations to assess strategic judicial behavior. After all, an important strategic behavior not captured by the cases reviewed is a court’s decision not to cite Pearson at all when deciding whether to reach constitutional questions in the qualified immunity context. Moreover, we focused on circuit courts because—outside of the Supreme Court—only those courts can clearly establish law. To be sure, there is some uncertainty on this score. See, e.g., Carroll v. Carman, 135 S. Ct. 348, 350 (2014) (declining to decide whether circuit precedent can clearly establish rights). Even so, it is conventionally understood that they can. See, e.g., Ashcroft v. al-Kidd, 563 U.S. 731, 741–42 (2011) (distinguishing district and circuit courts).

220Leong, supra note 74, at 685–86.

221Nielson & Walker, supra note 33, at 31.

222Leong, supra note 74, at 686.

223The study methodology and its limitations are described in greater detail—and compared to prior empirical studies of qualified immunity—in The New Qualified Immunity. See Nielson & Walker, supra note 33, at 27–33.

224See id. at 33–38.

225Id. at 33–34 & fig.1. As noted in The New Qualified Immunity, “[w]e label this option ‘not exercising Pearson discretion’ for ease of reference, recognizing that the decision not to reach a constitutional question is still, in a sense, an exercise of discretion provided by Pearson.” Id. at 34.

226Id. at 34 & fig.1.

227Id. at 35 & fig.2.

228See Paul W. Hughes, Not a Failed Experiment: Wilson-Saucier Sequencing and the Articulation of Constitutional Rights, 80 U. Colo. L. Rev. 401, 418–19, 421, 423 tbl.1 (2009); Leong, supra note 74, at 684–91; Colin Rolfs, Comment, Qualified Immunity After Pearson v. Callahan, 59 UCLA L. Rev. 468, 490–97 (2011); Ted Sampsell-Jones & Jenna Yauch, Measuring Pearson in the Circuits, 80 Fordham L. Rev. 623, 638–39 (2011); Greg Sobolski & Matt Steinberg, Note, An Empirical Analysis of Section 1983 Qualified Immunity Actions and Implications of Pearson v. Callahan, 62 Stan. L. Rev. 523, 525 (2010).

229Nielson & Walker, supra note 33, at 37.

230Nielson & Walker, supra note 33, at 37–38 (footnote omitted); see also id. at 37 tbl.1 (comparing all six studies cited supra note 228).

231See id. at 39–42.

232Id. at 39–40 & fig.3. The rate in the Sixth Circuit is 47.7%. Id.

233Id. at 40–41 & fig.4. The national average here of 9.5% differs from the overall 8.0% figure provided above because, for purposes of comparing circuits, “the national average is the average of the percentage for each circuit, excluding the D.C. Circuit because it handled far fewer claims (only a dozen) in the sample and seems to be an outlier.” Id. at 39 n.209.

234Id. at 41–42.

235Id. at 42.

236See id. at 43–49.

237Id. at 45–46 & fig.5.

238Id.

239Id. at 47–48 & fig.6.

240Id. at 48.

241See id. at 49–51.

242Id. at 49 (emphasis added).

243Id. at 49–51 & figs.8–9.

244Id. at 51.

245Id. at 49.

246See Pearson v. Callahan, 555 U.S. 223, 236–42 (2009); see also supra notes 96–97.

247Nielson & Walker, supra note 33, at 52; accord Beermann, supra note 34, at 175 (“At a minimum, in light of the strong reasons for reaching the constitutional merits, courts should be required to give reasons for not doing so.”).

248Nielson & Walker, supra note 33, at 65.

249In this preliminary inquiry we rely solely on basic tests of statistical significance. As further research is undertaken, however, one should incorporate regression analysis, and it may be useful to employ semiparametric matching rather than standard regression analysis. Cf. Christina L. Boyd, Lee Epstein & Andrew D. Martin, Untangling the Causal Effects of Sex on Judging, 54 Am. J. Pol. Sci. 389, 389 (2010). In other words, our study’s tentative analysis should not be the final word on this issue.

250These are, of course, not the only methodological limitations. For instance, any attempt to draw inferences about the legal system from a selection of trial or appellate opinions has problems. See, e.g., George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13 J. Legal Stud. 1, 1 (1984); Theodore Eisenberg & Michael Heise, Plaintiphobia in State Courts Redux? An Empirical Study of State Court Trials on Appeal, 12 J. Empirical Legal Stud. 100, 100 (2015); Theodore Eisenberg & Stewart J. Schwab, What Shapes Perceptions of the Federal Court System?, 56 U. Chi. L. Rev. 501, 503–04 (1989). The purpose of this study, however, is not to draw inferences about the larger legal system but to better understand the strategic considerations that may affect how circuit courts exercise their discretion to develop constitutional law based on the cases that reach their dockets—recognizing that there are selection effects embedded in the sample of legal disputes that reach the appellate stage.

251See Christopher A. Whytock, The Evolving Forum Shopping System, 96 Cornell L. Rev. 481, 521 n.181 (2011) (“[T]ests using the party of the nominating president as a proxy for a judge’s ideological attitudes may underestimate the impact of those attitudes.” (citing Joshua B. Fischman & David S. Law, What Is Judicial Ideology, and How Should We Measure It?, 29 Wash. U. J.L. & Pol’y 133, 170–71 (2009))). But see Jeremy D. Horowitz, Not Taking “No” for an Answer: An Empirical Assessment of Dissents from Denial of Rehearing En Banc, 102 Geo. L.J. 59, 73 (2013) (noting that the party of a judge’s nominating President can shed some insight “when looking at a large number of judges and decisions” (citing Cass R. Sunstein et al., Ideological Voting on Federal Courts of Appeals: A Preliminary Investigation, 90 Va. L. Rev. 301, 303 (2004))); Daniel R. Pinello, Linking Party to Judicial Ideology in American Courts: A Meta-Analysis, 20 Just. Sys. J. 219, 243 (1999) (observing that party can be a useful indicator of ideology after conducting a meta-analysis of studies).

252The numbers of all judges reported in Figures 1 and 2 include all judges on unsigned or per curiam opinions.

253Nielson & Walker, supra note 33, at 45–46. To assess statistical significance for political affiliation (p < 0.05), we assume that if there is no bias in the distribution, the results would be 50/50. That is, if observing opinions through the lens of political affiliation and assuming no bias, a judge nominated by a Democratic President would have the same odds of deciding a given question as a judge nominated by a Republican President. The results will split down the middle. If, however, the political affiliation of the nominating President affects the way judges decide opinions, the odds will be skewed according to that variable. For determining statistical relevance, we compare the percentage of events occurring given a specific categorization to the expected value of an unbiased—50/50—result. Some normalization was required (i.e., D to R), since there is a disparity between the total number of claims decided by each party. Otherwise, a 50/50 split is already skewed. This has some caveats, so normalization was always done in the most conservative manner possible. We used Prism software to run parts-of-a-whole comparisons (actually a two-tailed binomial distribution given the input value versus the expected output value, which is why we arrive at a P-value) on the positive values observed from the dominant population in the data against the normalized values of the minority group in the data. Moreover, chi-squared tests for independence were performed to assess whether there was statistical evidence to suggest that nominating party mattered as to the three types of decisions depicted in Figure 1 as well as the two types of decisions depicted in Figure 2. The results were consistent with those of the binomial test reported herein.

254See supra note 238 and accompanying text.

255See, e.g., Cass R. Sunstein et al., Are Judges Political?: An Empirical Analysis of the Federal Judiciary vii (2006); Boyd, Epstein & Martin, supra note 249; David S. Law, Strategic Judicial Lawmaking: Ideology, Publication, and Asylum Law in the Ninth Circuit, 73 U. Cin. L. Rev. 817, 819 (2005); Thomas J. Miles & Cass R. Sunstein, Do Judges Make Regulatory Policy? An Empirical Investigation of Chevron, 73 U. Chi. L. Rev. 823, 851 (2006); Jennifer L. Peresie, Note, Female Judges Matter: Gender and Collegial Decisionmaking in the Federal Appellate Courts, 114 Yale L.J. 1759, 1762 (2005); Richard L. Revesz, Litigation and Settlement in the Federal Appellate Courts: Impact of Panel Selection Procedures on Ideologically Divided Courts, 29 J. Legal Stud. 685, 685 (2000); Gregory C. Sisk, Michael Heise & Andrew P. Morriss, Searching for the Soul of Judicial Decisionmaking: An Empirical Study of Religious Freedom Decisions, 65 Ohio St. L.J. 491, 491 (2004).

256Matthew C. Stephenson, Statutory Interpretation by Agencies, in Research Handbook on Public Choice and Public Law 285, 307 (Daniel A. Farber & Anne Joseph O’Connell eds., 2010).

257Sunstein et al., supra note 255, at 23–24.

258Stephenson, supra note 256, at 308.

259A number of cases in the sample were decided by two-judge panels and thus are not included in this analysis. As noted in Part IV.A, en banc decisions have also been excluded.

260This is based on a chi-squared test to compare all four potential panel compositions with respect to the probabilities that they exercised each of the three types of decisions depicted in Figure 3 as well as the two types of decisions depicted in Figure 4. The chi-squared tests do not allow for examination of each type of panel composition individually.

261Nielson & Walker, supra note 33, at 46. See generally Cass R. Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court 3–4 (1999) (describing judicial minimalism).

262Nielson & Walker, supra note 33, at 47.

263Under the overly simplistic 50/50 model, there is statistically significant evidence (p < 0.0001) to suggest that a panel’s decision to find no constitutional violation is not independent of the panel composition, whereas the differences are not statistically significant (p < 0.05) as to a panel’s decision to recognize a new constitutional right. Moreover, when one panel composition is compared to the rest of the panel compositions, there are statistically significant differences (p < 0.0001) as to 3D, 3R, and 2D-1R panels in finding no constitutional violations (but not as to 2R-1D panels), and there are no statistically significant differences (p < 0.05) as to any panel composition’s decision to recognize a new constitutional right.

264Sunstein et al., supra note 255, at 129.

265Stephenson, supra note 256, at 308.

266Miles & Sunstein, supra note 255, at 852.

267For determining statistical relevance, we utilized the same method as for political affiliation, and compared the percentage of events occurring given a specific categorization to the expected value of 50/50. Some normalization was required, since there is a disparity between the total number of claims decided in published and unpublished decisions. Otherwise, a 50/50 split is already skewed.

268This finding is statistically significant only at a lower level (p < 0.1) than we set as the baseline for the rest of the findings (p < 0.05).

269See, e.g., Anderson v. McCaleb, 480 F. App’x 768, 773 (5th Cir. 2012) (per curiam) (reversing the district court’s grant of qualified immunity in the excessive force context by 3R panel in a somewhat lengthy, unpublished, unsigned decision); Quint v. Vill. of Deerfield, 365 F. App’x 697, 701 (7th Cir. 2010) (reversing the grant of qualified immunity by 3R panel in the unlawful search and seizure context); Evans v. City of Etowah, 312 F. App’x 767, 768 (6th Cir. 2009) (affirming the denial of qualified immunity in the unlawful arrest context by 3R panel); Haley v. Elsmere Police Dep’t, 452 F. App’x 623, 629 (6th Cir. 2011) (affirming the denial of qualified immunity in the unlawful arrest context by 3R panel though over a dissent); see also Comeaux v. Sutton, 496 F. App’x 368, 375 (5th Cir. 2012) (per curiam) (reversing the district court’s grant of immunity in the excessive force context by 2D-1R in a somewhat lengthy, unpublished, unsigned opinion).

270See, e.g., Anderson v. City of Bainbridge Island, 472 F. App’x 538, 541 & n.2 (9th Cir. 2012) (Gould, J., dissenting) (dissenting in excessive force context as to the majority’s finding of no constitutional violation); Jordan v. Brown, 489 F. App’x 190, 191 (9th Cir. 2012) (Thomas, J., dissenting) (“I would affirm the district court’s denial of qualified immunity at this stage of the litigation.”); Smith v. Cty. of Lenawee, 505 F. App’x 526, 540–41 (6th Cir. 2012) (Kethledge, J., concurring in part and dissenting in part) (“I would affirm the district court’s denial of qualified immunity to Westgate. Unlike Dye—who I think presents a close case, but who at least showed concern by putting Smith in a padded cell and calling Dr. Stickney—Westgate basically took no action at all during his four-hour shift on the night before Smith died.”); Williams v. Port Huron Sch. Dist., 455 F. App’x 612, 622 (6th Cir. 2012) (Moore, J., dissenting) (“I agree with the district court that, given the facts as pleaded by the Plaintiffs, a reasonable jury could conclude that Dalke and Jones were deliberately indifferent to the harassment at Port Huron Northern.”); Estate of Henson v. Callahan, 440 F. App’x 352, 359 (5th Cir. 2011) (Dennis, J., dissenting) (“I respectfully dissent. In my view, there is evidence from which a reasonable jury could conclude that each of the three elements of supervisory liability is satisfied, and thus that the district court’s denial of qualified immunity to Sheriff Callahan should be affirmed.”); Neal v. Melton, 453 F. App’x 572, 581 (6th Cir. 2011) (Moore, J., concurring in part and dissenting in part) (“Taking the facts in the light most favorable to the Plaintiffs, I believe that the officers lacked reasonable suspicion to detain the Plaintiffs after completion of the vehicle-registration check in order to conduct a canine sweep.”); Rodriguez v. City of Cleveland, 439 F. App’x 433, 459 (6th Cir. 2011) (White, J., concurring in part and dissenting in part) (“I respectfully dissent from the majority’s determination that the individual defendants are entitled to qualified immunity as to Rodriguez’s claims of retaliation and harassment under the First Amendment.”); Poulakis v. Rogers, 341 F. App’x 523, 540 (11th Cir. 2009) (Quist, J., dissenting) (“My larger concern is that this Court’s decision sends a signal to police officers that they are free to ignore the law of the intermediate state appellate courts by which they are otherwise bound, and an unambiguous statute, without concern for violating an individual’s federal constitutional rights. Therefore, I would reverse on qualified immunity and remand for consideration of the other defenses Defendants raised in their summary judgment motion.”).

271See, e.g., Koch v. Lockyer, 340 F. App’x 372, 375 (9th Cir. 2009) (Clifton, J., concurring in part and dissenting in part) (“I agree with my colleagues that the individual defendants are entitled to qualified immunity . . . , but I strongly disagree with the conclusions that the collection of Koch’s DNA violated the Fourth Amendment . . . .”).

272See, e.g., Estate of Henson v. Callahan, 440 F. App’x 352, 358 (5th Cir. 2011) (Owen, J., concurring) (“I join Judge Southwick’s holding that Sheriff Thomas Callahan is entitled to qualified immunity. With respect, however, I do not join his opinion.”); Haley v. Elsmere Police Dep’t, 452 F. App’x 623, 629 (6th Cir. 2011) (Rogers, J., dissenting) (“Because Markesbery had probable cause to arrest Haley, I would reverse the district court’s denial of summary judgment on the unlawful arrest claim.”).

273See, e.g., Plumley v. Austin, 135 S. Ct. 828, 831 (2015) (Thomas, J., dissenting from the denial of certiorari) (focusing on the presence of a dissent from a lengthy majority opinion as evidence that the decision should have been published).

274See supra Part I.C.

275Akrawi v. Remillet, 504 F. App’x 450, 453 (6th Cir. 2012) (rejecting defendants’ challenge to the award of attorney’s fees where the grant of qualified immunity was affirmed because plaintiff “prevailed on his central claim—that the defendants violated his right to due process”); Escobar v. Mora, 496 F. App’x 806, 813 (10th Cir. 2012) (“[W]hile we have extrapolated from general principles here to hold that Mr. Escobar stated an Eighth Amendment claim, we cannot say that the analytical bridge from the former principles to the latter holding impressed itself upon us with the clarity necessary to conclude that defendants were on fair notice that their conduct rose to the level of a constitutional violation.”); Saavedra v. Scribner, 482 F. App’x 268 (9th Cir. 2012) (per curiam) (finding two separate due process violations in the administrative-segregation context but granting qualified immunity); Kozel v. Duncan, 421 F. App’x 843, 850 (10th Cir. 2011) (“Although we have determined that there was a constitutional violation in the prolonged sobriety detention, we must still consider whether a reasonable officer in Sheriff Duncan’s position would have been on notice that his conduct was unlawful. Mr. Kozel has not met his burden in this regard.”); Rivers v. Fischer, 390 F. App’x 22, 24 (2nd Cir. 2010) (per curiam) (concluding that “there is no doubt that Rivers’s constitutional rights were violated when the Department of Corrections administratively imposed a 5-year term of supervised release that was not orally pronounced by the sentencing judge” but concluding that right was not clearly established at time of the incident); Rock for Life–UMBC v. Hrabowski, 411 F. App’x 541, 555 (4th Cir. 2010) (“Although the district court erred by holding that the plaintiffs failed to demonstrate a triable issue of fact whether the defendants regulated their speech based on its content, the defendants are nevertheless entitled to qualified immunity from 42 U.S.C. § 1983 claims brought against them in their individual capacities.”); id. at 558 (King, J., concurring in part, dissenting in part, and concurring in the judgment) (“I agree with the majority that we should award qualified immunity to the defendants on the as-applied First Amendment challenge to UMBC’s policy on facilities use, but I would get there by a different route—namely, by concluding that a First Amendment violation has not been shown.”); Schwenk v. Cty. of Alameda, 364 F. App’x 336, 337 (9th Cir. 2010) (per curiam) (“We affirm the dismissal as to Lacativo on the basis of qualified immunity. At the time of the alleged conduct, we had not yet decided Burke, which expressly held that ‘parents with legal custody, regardless of whether they also possess physical custody of their children’ have a liberty interest in the care, custody, and management of their children.” (citing Burke v. Cty. of Alameda, 586 F.3d 725, 731, 733–34 (9th Cir. 2009))); Solis v. Oules, 378 F. App’x 642, 644 (9th Cir. 2010) (per curiam) (granting immunity yet finding that the officer “violated Solis’s Fourth Amendment right to be free from unreasonable searches and seizures when he stopped her vehicle and apparently removed her from it under a law that did not criminalize her behavior”); Koch, 340 F. App’x at 374 (per curiam) (“While the compulsory DNA collection violated Koch’s Fourth Amendment rights, Appellees are entitled to qualified immunity against damages.”); Thompson v. Williams, 320 F. App’x 678, 679 (9th Cir. 2009) (per curiam) (granting qualified immunity but affirming “that Thompson’s religious beliefs are sincerely held and that there are triable issues as to whether the prison’s policy violates the First Amendment or RLUIPA”).

276To be sure, two of these decisions noted that the right had already been recognized in a prior published decision but that the incident at issue took place prior to that decision’s publication and thus was not clearly established at the time of the incident. Rivers, 390 F. App’x at 24; Schwenk, 364 F. App’x at 337.

277Rock for Life–UMBC, 411 F. App’x at 558 (King, J., concurring in part, dissenting in part, and concurring in the judgment).

278See supra note 275.

279See Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208 (1988).

280See Camreta v. Greene, 563 U.S. 692, 714 (2011) (Scalia, J., concurring); Nielson & Walker, supra note 33, at 52 (citing Camreta, 563 U.S. at 727 (Kennedy, J., dissenting)).

281See Reinhardt, supra note 17, at 1250.

282See Heckler v. Chaney, 470 U.S. 821, 831 (1985).

283See Pearson v. Callahan, 555 U.S. 223, 232 (2009).

284Id. at 236–37.

285See id. at 236.

286Nielson & Walker, supra note 33, at 52–65. Because this recommendation was explored at length in The New Qualified Immunity, we will not repeat that discussion here.

287See, e.g., EEOC v. Asplundh Tree Expert Co., 340 F.3d 1256, 1261 (11th Cir. 2003) (rejecting agency excuses for failure to consult per the statute’s requirement).

288Comm’r v. McCoy, 484 U.S. 3, 7 (1987).

289See, e.g., Brief for the United States in Opposition at 12, Johnson v. United States, 136 S. Ct. 318 (2015) (No. 15-16) (“The unpublished decision below . . . sets no precedent.”); Brief for the Respondent in Opposition at 14, Rangolan v. Mukasey, 554 U.S. 903 (2008) (No. 07-1169) (“[T]he court of appeals’ brief, unpublished decision would not be an appropriate vehicle for resolving the question presented, because the court did not provide an explanation for why it dismissed the petition for review.”).

290See Gerken, supra note 114, at 486 (“[A] review of the cases considered by the Court in any recent term demonstrates that virtually all of those cases involve a review of published courts of appeals decisions.”).

291Cass R. Sunstein, Factions, Self-Interest, and the APA: Four Lessons Since 1946, 72 Va. L. Rev. 271, 289 (1986).

292William F. Pedersen, Jr., Formal Records and Informal Rulemaking, 85 Yale L.J. 38, 60 (1975) (discussing circuit courts).

293Id.

294See, e.g., White v. Wheeler, 136 S. Ct. 456, 462 (2015) (per curiam) (“As a final matter, this Court again advises the Court of Appeals [for the Sixth Circuit] that the provisions of AEDPA apply with full force even when reviewing a conviction and sentence imposing the death penalty.”); Lopez v. Smith, 135 S. Ct. 1, 4 (2014) (“We have before cautioned the lower courts—and the Ninth Circuit in particular—against ‘framing our precedents at such a high level of generality.’” (citation omitted)); Ashcroft v. al-Kidd, 563 U.S. 731, 742 (2011) (“We have repeatedly told courts—and the Ninth Circuit in particular . . .—not to define clearly established law at a high level of generality.” (citation omitted)).