Precedent as Permission

Article - - Issue 5


This Article provides an account of precedent that doesn’t call upon it to do the one thing that everyone expects: constrain judicial decision-making. Instead, precedent is tasked to do something else: identify lawful options. So instead of beginning with precedent’s limited ability to constrain, the argument focuses on what precedent enables.

On reflection, precedent has always had two aspects: a permissive aspect that enables certain options and a prohibitory aspect that rules out others. When combined, precedent’s permissive and prohibitory aspects can render certain outcomes mandatory. But that arrangement is contingent, not essential, and precedent’s permissive aspect alone can do a great deal of work. First, precedent’s epistemic value allows it to operate as a shortcut, affording judges an efficient way to arrive at pretty good legal answers. Second, precedent’s rhetorical value allows it to operate as a shield, thereby helping judges resist political and other pressures to deviate from case law. Precedent can thus foster convergence across jurists as well as fidelity to past decisions—even if it imposes no constraint whatsoever.

The upshot is a new “permission model” of precedent, in contrast with the more familiar “binding model.” The permission model challenges longstanding views of stare decisis, particularly horizontal stare decisis in the U.S. Supreme Court. For example, stare decisis is often lambasted for being a malleable doctrine that overlaps with the merits. But the permission model would celebrate that state of affairs. Malleable, merits-sensitive stare decisis helps the Justices manage controversial legal transitions based on their understanding of underlying law. And the binding model could simply be infeasible without enforcement. The permission model also points toward novel reforms, including ways of combining both permissions and mandates. Counterintuitively, the best way to strengthen precedent may be to make it more of a permission.

The permission model may not be the kind of precedent we were looking for, but it usefully informs the mode of precedent we actually have.



The essence of precedent is that it binds judges, requiring certain outcomes and foreclosing others.[1] Or so we are led to believe.[2] But what if precedent does or should instead operate only as a permission? In other words, what if a judge is legally allowed to follow precedent but never required to do so?

The question is timely. As seems to happen whenever there is a consequential change in its composition, the U.S. Supreme Court has become unusually preoccupied with issues of precedent. This attention mainly stems from the possibility that the Court will soon overrule major abortion precedents,[3] but there are also flashes of light around many other topics, ranging from free speech to administrative law.[4] Justice Elena Kagan and other rueful dissenters have come perilously close to announcing—self-defeatingly?—that stare decisis has itself been overruled.[5] Meanwhile, Justice Brett M. Kavanaugh has suggested that the Court’s current approach to stare decisis amounts to a “muddle” that “poses a problem for the rule of law.”[6] Leading observers have sounded similar notes. Professor Frederick Schauer captured the mood in suggesting that it is time, “finally, to face up to the weakness, verging on impotence, of the widely referenced but rarely followed stare decisis norm.”[7] Not to be outdone, the popular Strict Scrutiny podcast is selling shirts, bags, mugs, and sweaters emblazoned with the slogan: “Stare decisis is for suckers.”[8]

Enter Justice Clarence Thomas, whose concurrence in Gamble v. United States[9] appeared to suggest that precedent is, at bottom, a permission.[10] Under certain circumstances, Justice Thomas argued, “the judicial policy of stare decisis permits courts to constitutionally adhere to that interpretation [adopted in case law], even if a later court might have ruled another way as a matter of first impression.”[11] The key word—permits—on its face casts judicial adherence to precedent not as an obligation but rather as an allowable option, and even that only under certain circumstances. When obligation entered into Justice Thomas’s picture, it was to identify when judges have a duty to overrule precedent, not preserve it. For Justice Thomas, this conclusion springs primarily from an originalist inquiry into history and theoretical points about legal interpretation.[12]

Yet the strongest arguments in favor of viewing precedent as a permission are more pragmatic.[13] The core insight here is that behavior sometimes changes when people are more secure in selecting a particular option. If following precedent is guaranteed to be permissible, for instance, then judges might opt in favor of it for the sake of ease or efficiency, yielding convergence across judges even in the absence of any constraint. We can label this reason to favor case law precedent as shortcut. Further, precedent’s permissibility could help a judge resist political and other pressures to rule in more adventurous ways. A judge who bucks the wishes of her political allies, for example, might like to say not “I am right, whereas you are wrong,” but rather “I am following precedent, regardless of who is right.” This basis for using case law can be termed precedent as shield.[14]

Viewing precedent as a shortcut or shield (or both) is significant in part because those considerations in no way require that precedent be mandatory. They rely on the affirmative appeal of following precedent rather than any prohibitory aversion to departing from it. To put the point in crude but familiar terms, precedent might change judicial behavior by acting as a carrot, not a stick. This picture has particular appeal in the U.S. Supreme Court, where sticks are in short supply. Without the benefit of a ready enforcement mechanism, a strongly prohibitory or mandatory form of stare decisis may simply be doomed at the Court.[15] But even if so, the Justices would still see considerable appeal in taking advantage of stare decisis as a helpful permission. Precedent’s attractive force, in other words, can work quite well even in the absence of any binding force. And because both forces point in the same direction, they can work in tandem or be difficult to disentangle. Consider June Medical,[16] where the Chief Justice recently applied a precedent from which he had dissented.[17] Precedent does seem to have had an effect on how the Chief voted. But did it do so by constraining him or by enabling him to moderate the Court at a time of intense political scrutiny?[18]

Even more importantly, permissions can help strengthen stare decisis. Consider: when the Court concludes that there is a “special justification” for overruling precedent, does stare decisis simply evaporate, such that any erroneous precedent must be overruled?[19] An affirmative answer would leave judges hard-pressed to waive off strong merits-based challenges to precedent. That is because, as we will see, a suitably adjusted argument on the merits can often double as a special justification; indeed, merits-sensitive stare decisis may be unavoidable.[20] But that problem would not arise if the presence of special justifications simply caused adherence to precedent to shift from mandatory to permissible: a jurist who knows that precedent always affords a lawful option can confidently sidestep even the most forceful merits claims. And the jurist would often want to do just that, even if precedent is formally or effectively nonbinding. In this and other ways, the law of stare decisis can usefully blend permissions with more familiar mandatory rules.

This Article argues that we should alter our most fundamental expectations about what precedent both is and does. Even if it has no binding force—that is, even if it is in practice purely permissive—precedent can still do a lot of work. In that sense, it is possible to care about stare decisis at the Court without being a “sucker.” And permission-oriented reforms can make precedent matter still more.

The argument proceeds in three stages. First, precedent has both permissive and prohibitory aspects in that it supports certain outcomes while cutting against others. Further, these dual aspects, while usually linked, are separable. This analytical point illuminates how precedent normally works while also pointing toward a novel form of stare decisis. In particular, precedent could be purely permissive and so not mandatory at all.

Second, the initially shocking idea that precedent is purely permissive turns out to have descriptive and normative appeal, at least with respect to the Court. Rather than being a constraint, precedent might facilitate “good enough” dispositions in normal cases while offering rhetorical insulation from political and psychological pressures in high-salience cases. This is the permission model of precedent.[21]

Finally, the permission model has important implications for the content of stare decisis. Some of stare decisis’s most lamented shortcomings, namely its malleability and merits-sensitivity,[22] may actually be underappreciated strengths. And the permission model can be blended with binding approaches to precedent, thereby yielding new ways of structuring and reforming the stare decisis inquiry.

Ultimately, the point of exploring permissive precedent is not to abandon binding stare decisis so much as to understand and use it better. If precedent mainly works not by constraining but—counterintuitively—by enabling, then reformers’ priorities must change. Rather than aspiring to create a more merits-neutral test for overruling, reformers should look for ways to strengthen precedent’s permissive aspect. So while it may not be the kind of precedent we were looking for, the permission model usefully informs the mode of precedent we actually have.

I. Permission as an Ingredient of Precedent

This Part introduces the idea that precedents, like many legal authorities, have both permissive and prohibitory aspects. To that end, subpart I(A) introduces a framework for decomposing precedential rules, and subpart I(B) critically discusses Justice Thomas’s recent Gamble opinion.

A. The Permissive Aspect of Precedent

The conventional, if not universal, picture of precedent as binding combines what could be called the permissive and prohibitory aspects of precedential authority.[23] In its permissive aspect, precedent supports the lawfulness of a particular course of action—namely, adhering to past decisions. And in its prohibitory aspect, precedent cuts against all courses of action other than following the precedent.

This push-pull dynamic is characteristic of many rules and open to innumerable variations.[24] For instance, if precedent were viewed as a reason or factor counseling in favor of adhering to previous decisions, it would sometimes not only allow but also require a certain outcome—and so would be both more and less than a permission.[25]

Precedent’s two aspects are not only analytically distinct but also practically separable. That is, it is possible to imagine a precedential regime that is exclusively permissive or exclusively prohibitory. Further, each aspect can have its own degree of strength. That point is easily lost because a strongly prohibitory form of precedent will frequently guarantee or entail a strong permissive aspect: if all actions other than following precedent are prohibited, as when compliance with precedent is viewed as mandatory, then a workable regime must treat that action as permitted. Because prohibitions can engender mandatoriness, precedent’s “prohibitory” aspect can alternatively be called its “mandatory” aspect.

A schematic can help illuminate the range of options. Assume that precedent’s permissive aspect is “strong” when following precedent is categorically lawful and “weak” when precedent is merely a slight factor in favor of a result’s lawfulness. And further assume that precedent’s prohibitory aspect is “strong” when deviating from precedent is lawful only if stringent stare decisis requirements are met and “weak” when stare decisis hardly counts against any action at all. Those assumptions yield the four options set out below.


Figure 1: The Two Aspects of Precedential Authority

Permissive Aspect
Weak Strong
Mandatory Aspect Weak Option 1. Precedent neither limits nor allows, and so operates, at most, as one factor among others. Option 2. Courts are always allowed to follow precedent or, instead, to apply their view of under-lying law.
Strong Option 3. In general, courts must follow precedent. But if stringent requirements of stare decisis are met, precedent ceases to matter—leaving courts to apply their view of underlying law. Option 4. In general, courts must follow precedent. But if stringent requirements of stare decisis are met, courts have permission to follow precedent; alternatively, courts may apply their view of underlying law.


In the above table, the top row is less familiar because it focuses on precedent where its mandatory aspect is weak. In Option 1, precedent is weak in both its aspects and so hardly matters at all. By contrast, Option 2 features a strong permissive aspect and a weak mandatory one, such that courts are always allowed—but never required—to follow precedent. The bottom row moves us to more familiar terrain but also draws out an underappreciated distinction. In both Option 3 and Option 4, precedent’s mandatory aspect is strong, yielding a general obligation to follow precedent. But what happens when the stringent obligations of stare decisis are met? In Option 3, precedent simply ceases to matter since there is no significant permissive aspect. By contrast, Option 4 includes a strong permissive aspect, so even when the requirements of stare decisis are met, precedent remains a lawful option.

The schema above hardly exhausts the range of options,[26] but it does supply a way of mapping them. The next subpart supplies a more detailed example.

B. Justice Thomas’s View in Gamble

In Gamble, Justice Thomas wrote a separate opinion that outlined a distinctive—and perhaps shocking—approach to precedent.[27] Justice Thomas appeared to suggest that precedent affords only a conditional permission: the Court has the option of following precedent until the wrongfulness of a given precedent is adequately brought to the Court’s attention.[28] Returning to Figure 1, Justice Thomas’s apparent view can be located somewhere between Option 1, which treats precedent as a weak permissive factor, and Option 2, which treats precedent as a (categorical) rule of permission.

That said, Justice Thomas’s views on the permission model are not entirely clear as to either content or justification. Part of the difficulty is that Justice Thomas seemingly advanced two quite distinct ways in which precedent can operate as a permission.

First, Justice Thomas described precedent as a procedural assumption about when a court may take a precedent for granted. This idea is set out in a footnote:

I am not suggesting that the Court must independently assure itself that each precedent relied on in every opinion is correct as a matter of original understanding. We may, consistent with our constitutional duty and the Judiciary’s historical practice, proceed on the understanding that our predecessors properly discharged their constitutional role until we have reason to think otherwise—as, for example, when a party raises the issue or a previous opinion persuasively critiques the disputed precedent.[29]

Permitting judges to assume precedent is correct “until” there is “reason to think otherwise” may not seem like much. In fact, this view casts precedent as even weaker than conventional principles of forfeiture,[30] since not just “a party” who properly raises an issue but also a judge’s “previous opinion” can overcome whatever permission precedent affords. Still, the principle has some effect, as evidenced by the fact that Justice Thomas has relied on it in choosing to reserve judgment on suspect precedents that no party had properly challenged.[31]

Justice Thomas gave no explicit justification for the foregoing principle, but it seemingly rests on an intuitive, pragmatic conception of how judges should decide cases. It simply isn’t feasible for every judge to question everything all the time.[32] So for reasons of efficiency, the judge is permitted to assume that precedent is correct until given proper notice. And once that notice is given, the permission to presume precedent’s correctness expires. This essentially procedural view of precedential permission is more about the when than the whether of overruling. And it also ascribes a significant practical power to litigators and judges, who have the ability to raise—or bury—objections to precedent.[33]

Second, Justice Thomas described precedent as a permission that operates within zones of substantive indeterminacy.[34] Here is the key explanatory passage, with emphasis added:

It is within [the] range of permissible interpretations that precedent is relevant. If, for example, the meaning of a statute has been “liquidated” in a way that is not demonstrably erroneous (i.e., not an impermissible interpretation of the text), the judicial policy of stare decisis permits courts to constitutionally adhere to that interpretation, even if a later court might have ruled another way as a matter of first impression. Of course, a subsequent court may nonetheless conclude that an incorrect precedent should be abandoned, even if the precedent might fall within the range of permissible interpretations. But nothing in the Constitution requires courts to take that step.[35]

The above account draws on scholarship by Professor Caleb Nelson[36] (whom Justice Thomas cites repeatedly) and relates to work by Professors Randy Barnett and Larry Solum.[37] The basic premise is that formal sources of law eventually “run out,” either because the law is expressed in indeterminate language or because relevant legal materials (such as historical evidence bearing on original public meaning) cannot rule out plausible options.[38] When precedent provides an answer within this kind of legal “gap,” sometimes called “the construction zone,”[39] courts have permission to follow the precedent. One might say that precedent always gives way to the correct view of underlying law, but precedent is a permissible option when the underlying law is silent or obscure.

If precedent can operate in zones of indeterminacy or construction, as Justice Thomas suggests, the obvious question is: how large are those zones? In other words, how many wrongheaded constitutional precedents fall short of being “demonstrably erroneous,” such that Justices have permission to retain them? One might imagine that almost all seriously disputed cases involve construction. By contrast, Justice Thomas seems to think that the law is usually determinate, such that few important precedents are safe from challenge. Yet Justice Thomas does not spell out why that would be the case, and different interpreters would presumably take different positions on this issue.[40]

Further, Justice Thomas’s proposal raises questions of implementation. First, from what standpoint is demonstrable error to be assessed? Justice Thomas apparently believes that a judge should regard an error as “demonstrable” whenever she is certain about the error. But instead, “demonstrable” error could require some sort of demonstration capable of yielding expert consensus.[41] Second, how much or what kind of “erroneousness” is necessary? A precedent could be viewed as, say, possibly, probably, or certainly erroneous—and varying that confidence threshold would yield very different implications.

There is also the critical question of justification. Even if precedent occupies a zone of legal indeterminacy, why treat precedent as a permissible option, as opposed to a mandatory duty?[42] Perhaps Justice Thomas thinks there is simply no law at all within zones of legal indeterminacy, but that view is in tension with Justice Thomas’s claim that precedent within such a zone can still be “incorrect” and so overruled. Further, saying that there is an absence of law suggests that following precedent could be either permissible or not in various contexts. Giving permission, after all, is very different from pleading ignorance. To say that precedent “permits” an action, as Justice Thomas does, is not to be agnostic on the propriety of the action, but rather to take a meaningful, specific stance on how precedent relates to proper conduct. Why does that stance make sense when others—like a requirement to follow precedent—are also available?

Further complicating things, Justice Thomas’s account may actually have left room for mandatory precedent. Less than a year after Gamble, Justice Thomas used the phrase “binding precedent” and appeared to view case law as constraining.[43] That phrase could have been a simple slip, or it could reflect a change of heart. Another possibility is that at least some of Justice Thomas’s discussion of permissions was always meant to address a different level of judicial decision-making than case resolution. That is, Justice Thomas’s Gamble opinion may have meant to say that the Court was at one time permitted to adopt a conditionally binding rule of stare decisis, but having created such a rule, the Court must now apply it.[44]

Perhaps Justice Thomas plans to work out the basis and implications of his permissive approach or has simply left that task to the academy. Either way, we have good reason to flesh out and test the idea that precedent is a permission.

II. Assessing the Permission Model

To draw precedent’s permissive aspect into stark relief, this Part contrasts a model of precedent wherein case law operates primarily or exclusively as a permission (the permission model) with a more conventional view of precedent as highly constraining (the binding model). Of course, both of these models are quite simple and even contrived. As we will see, however, comparing the models yields conclusions that are informative even for those who would reject or qualify the permission model in favor of other, perhaps milder ways of viewing precedent as permission.

A. The Permission and Binding Models

Consider two stylized and starkly contrasting models of precedent. Under what I will call the permission model, it is always permissible to follow precedent but never mandatory. More conventional views of precedent, by contrast, approximate what I will call the binding model: following precedent is generally not just permissible but mandatory. These models focus on the perspective of deciding courts as well as the various parties who operate in the shadow of adjudication.

The contrast between the permission and binding models requires some elaboration. The binding model is presumably more familiar given the priority formally afforded to precedent as a matter of stare decisis. Where precedent points, courts should generally follow.[45] Of course, there are various recognized exceptions to, or ways for courts to get around, stare decisis.[46] The more capacious and numerous these exceptions, the less “binding” the model will become. For our purposes, however, we can adopt the simplifying assumption that the binding model generally treats adherence to precedent as mandatory, unless stringent stare decisis requirements are met.

By contrast, the permission model holds that precedent operates not as a factor in favor of some actions or against others but instead (and exclusively) as a rule of permissibility.[47] This rule renders precedent a kind of authority: successfully invoking precedent establishes the lawfulness of a particular course of action without requiring any additional inquiry.[48] Once a judge deems precedent applicable, that is, the judge would know that one lawful option is to follow the precedent.[49] But that is all that would have been established.[50] It would remain fully open to the judge to deviate from the precedent, provided that doing so is otherwise lawful.[51]

That proviso is important: the permission model only guarantees the permissibility of following precedent; it does not speak to whether any particular deviation from precedent is permissible. Thus, judges who issue rulings contrary to precedent, or in ignorance of it, would still have to assure themselves that their decisions accord with the law.[52] Under this approach, judges have reason to consult precedent to the extent they want to assure themselves (and any prospective critics) that they have acted lawfully. But judges could decline to consider precedent at all and simply apply other sources of law without having to worry that precedent could render their decisions unlawful.

But if precedent renders certain rulings lawful, as the permission model maintains, don’t courts still have a responsibility to explain why they choose to follow precedent instead of adopting a different course of action? The short answer is “no.”[53] The point of the permission model is that the reasons in favor of following precedent are of such strength and nature that, once a judge shows that she is following precedent, no further explanation is required.[54] Answering instead with a “yes” would generate a moderate or hybrid version of the model by adopting only a conditional rule of permission.[55]

We can imagine other relatively moderate or sophisticated versions of the permission model. Precedent could do more in that, even after establishing permissibility, it might also counsel in favor of a course of decision. Past decisions might thus operate as both permissive authority and persuasive authority.[56] Or precedent could do less in that it might operate as a factor in favor of permissibility without conclusively establishing permissibility or contributing to any sense of obligation.[57] Additionally, the permission model could be given a limited scope so that it applies only to a certain set of courts, issues, or cases.[58] Those sorts of variations might offer ways of reconciling the permission and binding models of precedent—a topic to be revisited below.

We can also imagine different ways for the permission model to accommodate the concept of “overruling.” On the simplest approach, the permission model would not only leave judges free to deviate from any given precedent but would also allow judges to take the further step of withdrawing the precedent and its attending permission.[59] In other words, the permission model might come with a standing permission to withdraw precedential permissions. But again, more modest options are also available. For instance, the conventional stare decisis factors could be reimagined as criteria for repudiating old permissions. A precedent would thus persist (as a permission) unless and until “overruled” pursuant to a stare decisis analysis, at which point a new permission would take its place.

For now, however, I will bracket those sorts of variations and hybrids so as to focus attention on a simple, pure version of the permission model, which holds that precedent is a permission and nothing else. Before turning to opportunities for compromise, we should carefully explore the model in its undiluted form so as to understand its key features and appeal. The next subpart takes up that task.

B. Permission’s Appeal

Precedential permissions represent a legal practice that is justified, if at all,[60] by its pragmatic consequences.[61] But what are those consequences? To assess the permission model’s appeal, let us roughly distinguish between two types of cases: run-of-the-mill cases and high-salience cases. In run-of-the-mill cases, judges approach issues without strong ideological preconceptions or a significant risk of scrutiny. By contrast, high-salience cases engage judges’ ideological views and may also create higher odds of review, either by actual courts or by the court of public opinion.[62]

Section II(B)(1) focuses on run-of-the-mill cases, arguing that precedent’s epistemic value creates an efficiency-based reason for judges to follow precedent even when doing so is optional. This is precedent operating as a permissible shortcut. Section II(B)(2) then turns to high-salience cases to argue that precedent can desirably offer cover to judges who want to do the “right” thing but may be susceptible to various pressures. Here, precedent operates as a shield, allowing the judge to deflect unwanted criticism. These sections show that precedent can exert not only a binding force but also an attractive force.

1. Convergence from Convenience.

The permission model offers a plausible alternative to the binding model, at least if we focus on the pragmatic goals typically associated with precedent—namely, promoting uniformity, predictability, and consequently reliance—in run-of-the-mill cases.[63] This basis for treating precedent as a permission overlaps with some of the standard arguments for precedent, particularly points bearing on epistemic humility and decisional efficiency. But it turns out that these familiar arguments support an unfamiliar result.

Start with precedent’s epistemic value.[64] The intuition is simple: “If judges presumptively as good as me thought X about this issue, why should I presume that I can do any better?”[65] And: “Even if I personally can do better, would my multimember court do better?”[66] These rhetorical questions suggest that, in general, following appellate precedent is about as likely to yield workable, legally correct outcomes as anything else the deciding judge could realistically do.[67] And insofar as following precedent is probably as good as anything else,[68] it would seem appropriate to view that approach as a permissible option on par with alternatives. This modest conclusion is more natural than the one demanded by the binding model—namely, that the epistemic equivalence of prior adjudicators supplies a reason to favor what those earlier courts have done, as compared with what appears correct to the deciding court today.[69]

That simple epistemic argument leads directly to an efficiency-based case in favor of the permission model.[70] Judges generally want to decide each case in a legally permissible way while maximizing their ability to dispense prompt justice across cases overall.[71] The permission model facilitates that goal by giving judges with diverse views a relatively easy, accurate, and shared means of reaching lawful outcomes in the mine-run of cases—namely, by reading and applying precedent, something that judges are well-trained to do.[72] And in the unusual cases when a judge (or panel) can easily determine that a better outcome is achievable by deviating from precedent, or that it is better to sidestep a tangle of confusing case law,[73] the permission model would not block that desirable outcome. In still other unusual cases, judges may decide to expend the extra effort required to arrive at an unprecedented improvement in legal doctrine, and the permission model would not block that special investment of decisional resources either. Notably, this efficiency depends on a strong rule of permissibility: if precedent were only weakly or too conditionally presumed to be permissible, then the typical case would become mired in debate about whether following precedent is actually allowed.

Permissive precedent also facilitates compromise. Appellate judges routinely confront cases in which they cannot pursue their own first-principles views of the law and so have reason to seek out common ground, even if the result amounts to a second-best legal rule or outcome. In those situations, the mere permissibility of precedent, independent of whether it is binding, can make it desirable to rely on case law. The key to compromise, after all, is that all judges have the option to converge on a point of tolerable agreement. And that sort of convergence is precisely what the permission model fosters. True, underlying disagreement on first principles is hard to suppress. As a result, the kind of settlement that results from permissive precedent would often be half-hearted, grudging, or temporary—but it would exist nonetheless.[74]

Given these epistemic and efficiency-based points, the permission model can credibly claim to foster consistency across cases. As we have just seen, judges with limited resources will have good reasons, both self- and public-interested, to resolve most cases according to precedent. Because that dynamic is generalizable, courts will converge on the path of least resistance. This picture comports with the idea that judges “follow” precedent, as though it were a path or trail that is made more visible, accessible, and desirable each time it is traversed.[75] The result is a mode of consistent decision-making that does not rely on an internalized norm of restraint. Consistency would instead result from the combination of a permissive norm, the institutional context in which judges decide cases, and an awareness that epistemic and efficiency considerations generally counsel in favor of adhering to precedent. Call it “consistency without constraint.”

2. Precedent-Rhetoric as a Shield.

We have just seen that the permission model’s efficiency allows it to foster uniformity and consistency in outcomes, at least in run-of-the-mill cases. But what about high-salience cases—that is, cases in which judges are not so worried about efficiency and, in addition, may face a special likelihood of scrutiny? The permission model can play a consequential role in those sorts of cases as well, though the best argument for that result takes a different form. In short, precedent’s permissive aspect can help insulate from political, social, and even internal pressures those Justices who are already otherwise drawn toward preserving precedent.

This line of argument embraces perhaps the most fundamental critique normally leveled against precedent—namely, the idea that precedent is mere rhetoric, as opposed to an accurate and complete account of the reasoning that actually underlies judicial decision-making in any given case.[76] Accepting that point is a devastating blow to the binding model since it concedes precedent’s inability to constrain. But it suggests another, more counterintuitive role for precedent to play. For even if precedent is merely a legalistic distraction from the real basis of decision, the law may benefit when judges are able to distract from their actual motivations.

To illustrate this dynamic, let us imagine a stark scenario involving a Justice who, for legal, moral, or institutionalist reasons, is not too sure that it makes sense to topple a landmark precedent—perhaps because the precedent seems correct or perhaps based on the disruption or controversy that would ensue. Yet the Justice also knows that her ideological kin loathe the precedent. A sizable body of empirical work suggests what should be intuitive: the Justice will care about how her friends think of her, including like-minded individuals both on and off the Court.[77] Even if the Justice’s desire to avoid antagonizing like-minded people operated at an unconscious level, it could still be consequential.

So when her friends rage that a precedent is wrong, wrong, wrong, the Justice might want, and benefit from, a way to tell her own frustrated allies: “I cannot be impugned for what I have decided.” Precedent’s permissive aspect addresses that felt need. That is, the permissibility of following precedent allows the Justice to point to a prior decision as proof that she has a legitimate option to rule as she did, notwithstanding the force of her allies’ criticisms on the merits. Moreover, the rhetoric of precedent provides a basis for decision that allows the Justice to avoid disputing whether the precedent is (as alleged) horribly misguided and wrong. Of course, this rhetorical move, like all such moves, is imperfect. The losers will still be upset at having lost. But precedent-rhetoric at least offers a softer, less confrontational, and potentially more legitimate-seeming way to let those allies down. Chief Justice Roberts’s recent vote in June Medical can be viewed as an example: precedent allowed the Chief to moderate and protect the Court while shielding himself from criticism for an about-face.[78]

Ironically, the binding model risks placing our imagined Justice in a more precarious position. Unless it is supplemented with a separate permissive principle,[79] the binding model suggests that the existence of a “special justification” extinguishes the force of precedent altogether.[80] And as critics both on and off the bench point out, several leading stare decisis factors, such as whether a precedent is poorly reasoned or has become a doctrinal outlier, overlap with considerations on the merits.[81] The more these kinds of considerations defeat precedent, the less the Justices can use precedent as a way to change the topic from a debate on the merits. By contrast, strong precedential permissions eliminate any need for consideration of stare decisis factors and so allow the besieged jurist to avoid addressing troublesome merits issues altogether.

So far, I have focused on a scenario in which a Justice reaches for precedent-rhetoric in order to insulate herself from ideological criticism. But there are other good uses for such rhetoric. Consider, for example, a Justice who has previously expressed or agreed with sharp criticisms of a particular precedent that has no particular ideological valence. This Justice would feel a nonideological pressure to overrule—namely, a self-interested desire to appear, not just to others but to oneself, as consistent and principled over time, as opposed to opportunistic and wishy-washy.[82] Here too, precedent-rhetoric offers a way out, particularly for jurists who have themselves espoused the importance of stare decisis, such as in a confirmation hearing. The judge might say: “Yes, I still think x, y, and z about this precedent. But precedent it is, and so it commands special respect.”

The idea of precedent as shield plausibly explains some of the most important instances in which stare decisis pulled out surprise victories in the Court. In these cases, there was ample reason to believe that at least five Justices were supportive of an ongoing effort to overrule a particular doctrine. Think of the challenges to Roe v. Wade before Planned Parenthood v. Casey, to Miranda v. Arizona before Dickerson v. United States, and to Auer v. Robbins before Kisor v. Wilkie.[83] Yet when the opportunity to overrule arrived, one or more Justices managed to wiggle out of doing what they were expected to do.[84] And those jurists all invoked a pliable stare decisis analysis rather than announce a new defense of the underlying precedents or assert that they were mechanically “bound.”[85] Dissenters sometimes call out this dynamic. For example, Justice Scalia’s Dickerson dissent argued that the three “surprise” votes for stare decisis sought to avoid contradicting their own past statements.[86] That argument can be viewed as an effort to pierce the rhetorical shield of stare decisis.

Most importantly, the idea of precedent operating as a shield comports with the core empirical findings that supposedly indict the overall efficacy of stare decisis. Justices would not care to use precedent as either a shortcut or a shield when they strongly desire to vote a particular way on the merits, as in a typical high-salience case. So for a proponent of the permission model, it is unsurprising that votes to overrule overwhelmingly accord with each judge’s ideology.[87] Yet the Justices might still care about precedent when their ideological views are complex or disengaged. Empirical studies often gloss over those votes as “consistent,” or not inconsistent, with the judge’s ideology.[88] But those votes matter. And the idea of precedent as shield suggests that precedential permissions can, for that limited but key set of votes, facilitate adherence to past decisions.

* * *

If permissive precedent can operate as a shortcut or a shield, then it yields benefits capable of justifying its existence. Moreover, the benefits arise even where precedent is neither formally nor practically binding. These claims supply a rejoinder to stare decisis skeptics, including both formalists who would abolish precedent as illegitimate and realists who view case law as an irrelevance or subterfuge.[89] The permission model’s consequences, in other words, yield at least some reason to keep stare decisis around rather than abandon it altogether.

C. Objections and Adjustments

This section critically discusses various objections to the permission model, with an eye to theoretical adjustments that might alleviate any concerns.

1. Too Much Discretion?

There is an obvious objection rooted in worries about judicial discretion.[90] This concern comes in both innocent and cynical varieties. The innocent version is concerned with negligent or reckless deviations from precedent. Once freed from the leash of binding precedent,[91] overconfident judges would too eagerly conclude not only that they can do better than their predecessors but also that the costs of trying are worthwhile.[92] The Court would then reshape the law whenever its membership made old rulings fall out of favor,[93] even if only by a single vote, and the lower courts would rapidly arrive at legal interpretations at odds with one another.[94] The result: greater litigation costs and less predictability.

The cynical version is that judges can opportunistically use precedent-rhetoric to advance a covert and illicit agenda. And the permission model would seem to exacerbate this kind of worry by avowedly expanding the lawful scope of judicial discretion. We might imagine a conservative Justice who leaps at every chance to overrule liberal precedents but who breezily waves away meritorious challenges to conservative case law.[95] Even worse, a Justice might honor precedent only when doing so yields desirable case outcomes. Perhaps Justice Thomas himself has misused his discretion in this way. For instance, Justice Thomas’s recent opinion upholding a religious exemption in Little Sisters[96] ignored what was (at least for him) a glaring nondelegation problem because it had not been raised, yet just days earlier in June Medical, Justice Thomas had argued for overruling all abortion-rights cases, something requested by no party.[97]

These counterarguments are formidable but not decisive. To move beyond critique and establish its claim to superiority, the binding model would have to offer stability without too often stymieing worthwhile doctrinal improvements. Those two demands—constraint and flexibility—are of course in tension.[98] And it is doubtful that any stringent principle of stare decisis could strike the needed balance.[99] Because the demands of change are inherently unknowable in advance, strict adherence to precedent would come at a steep cost. Almost any constitutional ruling whose correctness is nonobvious (and so susceptible to overruling) will rest on contingent premises, such as historical research, political consensus, social norms, technological abilities, or economic practices.[100] And when those premises change, judges have good reason to react, thereby undermining precedent’s bindingness. The choice between the permission and binding models thus poses a set of tradeoffs. For the binding model to win out, the risk of judicial overconfidence and cynicism in the use of precedent would have to outweigh the cost of doctrinal stasis in a dynamic world.

More fundamentally, a proponent of the binding model must show that precedent is realistically capable of imposing a genuine constraint. But serious efforts to instantiate the binding model may be futile at the Court given the lack of an effective disciplinary mechanism to keep the Justices in line.[101] Of course, the precedent-setting Court is not around to enforce its dictates against a later Court. And even if the Justices cared about the legal force of binding precedent, they would also care—at least as much and often much more—about establishing correct principles of law. The result is pressure to play fast and loose with whatever discretion the binding model inevitably allows. Perhaps the lower courts have internalized a stringent norm of vertical precedential constraint, consistent with the binding model. But even if so, that norm’s vitality could depend on the threat of appellate review. That is, adherence to binding norms of precedent may depend on external enforcement.[102] And at least under current conditions, there is no such enforcement when it comes to the Justices themselves.

We should also be careful about giving the binding model too much credit for generating whatever constraint currently exists. Return to the view that precedent is significantly more constraining in the lower courts.[103] That pattern may be explained, at least in large part, by structural factors consistent with the permission model, such as the greater workload pressures that direct judges toward precedent.[104] Even more important is the chastening prospect of appellate review, independent of any binding norm.[105] For a precedent to exist, it must not have been overruled. Thus, a federal judge will perceive both Court and circuit case law as having a presumptive safety about it: by following precedent, the judge takes a path that previously evaded reversal and so reduces the odds that her own decision might trigger unwanted Court review or an en banc proceeding.[106] By contrast, higher-court rulings that portend reversal necessarily have a prohibitory quality—somewhat like a warning shot—even if they lack any legally binding force.[107] These structural dynamics help explain why federal courts of appeals openly claim to give weight to Court “dicta,” which by definition seem to lack formal precedential force.[108] And to the extent that the foregoing structural factors and incentives would still operate even if precedent were permissive, the binding model would not offer any unique constraint.[109]

The binding model may also get undue credit for the effect of widespread moral intuitions that the permission model alone fully protects. Imagine, for example, that a judge believes both (i) that a precedent is wrong and (ii) that deviating from the precedent would be unfair to people who reasonably relied on it.[110] In this familiar situation, precedent affords legal significance to reliance interests and thereby gives the judge a lawful reason to stick with an erroneous decision.[111] But precedent’s permissive aspect is actually doing all the relevant work. The permission, after all, is what provides legal space for the judge’s moral intuitions to influence her decision. True, precedent’s mandatory aspect would go further and command attention to reliance—but that legal command would simply reproduce the judge’s separately existing moral intuitions.[112] So long as the judge harbors moral intuitions regarding the importance of reliance, those intuitions will constrain—even if, consistent with the permission model, the doctrine of precedent itself poses no constraint whatsoever.

But perhaps the discussion so far has been too consequentialist. For some readers, discretionary precedent may be problematic not because it threatens bad outcomes but rather because it is incompatible with the judicial role. For instance, Professor Caleb Nelson has drawn a distinction between (i) principled discretion in the application of a standard and (ii) “arbitrary” discretion subject to no standard at all.[113] In a similar vein, Professor William Baude has suggested that judicial review and judicially imposed punishments can be legitimate only if judges act out of legal duty.[114] On these views, the permission model may be objectionable simply because it creates standardless discretion—namely, discretion to follow precedent—without having to adduce any further justification.[115] So even if the permission model generally yielded good case outcomes, its open embrace of discretion could erode the judiciary’s actual or perceived legitimacy.

Yet the permission model embraces the judicial duty to decide cases according to law. The only question is whether that law may contain a limited set of permissive rules. It would therefore be a mistake to view the permission model as opposed to traditional, formal notions of judicial duty. In creating a bounded space where the law admits of different outcomes, permissions resemble discretion-conferring standards—which all agree are legitimate.[116] Of course, there can be too many permissions. Yet for any given standard, an equivalent amount of discretion or indeterminacy may be conferred through a permission. Which of these two flavors of discretion is preferable, and in what mixture, does not admit of a categorical answer. Perhaps the invocation of a permission is more visible than the application of a loose standard and so offers a riper target for critics of discretionary judging. But even if so, the permission’s relative transparency is plausibly counted as a virtue: it makes plain what might otherwise pass without notice, accommodation, or rebuke.

Further, judicial discretion, even standardless discretion, is in fact commonplace.[117] For example, the Court’s own rules posit that “[r]eview on a writ of certiorari is not a matter of right, but of judicial discretion.”[118] In effect, this rule grants the Justices permission to deny certiorari.[119] An order to deny certiorari thus operates as a reason-stopper, obviating the need for any further analysis or explanation.[120] Further, judges appear to have virtually complete, if implicit, discretion when it comes to including alternative holdings or choosing to write more or less broad opinions. No law purported to control, for example, whether Brown v. Board of Education[121] should have rested on the distinctive features of public secondary education, as opposed to a broader and more deeply reasoned theory of race and social equality. The Court instead determined Brown’s scope, as well as the scope of other consequential rulings, based on a complex set of unstated, nonlegal considerations that can be grouped under the heading of judicial statesmanship.[122] And that reality is widely accepted.[123] Here too, longstanding practice seems open to characterization in terms of permission. The permission model thus fits neatly among other forms of standardless judicial discretion.

Perhaps worries about the judicial role are best viewed not as decisive or categorical objections to the permission model but rather as reasons to consider tempering the model while also adopting a loose standard to guide—even if not constrain—judicial decision-making. In particular, precedent could be only formally mandatory: given an adequately merits-sensitive standard of stare decisis, that is, all might understand that merits views are generally decisive. We will revisit that way of blending the permission and binding models and see that it roughly captures current practice.[124]

2. Inadequate Rhetoric?

Another, more sympathetic objection would ask whether the permission model can really function as an effective rhetorical shield, particularly as compared with the binding model. True, the permission model allows a Justice to say, “I am following precedent,” when she would otherwise have to address the underlying merits. But the permission model does not allow the Justice to say, “I have to follow precedent.” That legal inability creates a rhetorical vulnerability: the Justice’s critics can plausibly respond, “Well, why did you choose to follow precedent when we all agree that you had a choice not to?” Under the permission model, there is no legal need to answer that question. But if there is nonetheless a rhetorical need to do so, perhaps because important exercises of discretion generally call for case-specific justification,[125] then the permission model could intensify pressure on the Justice rather than supply cover. By contrast, the binding model can supply the desired rhetorical cover by allowing the Justice to strike a more stable rhetorical pose, observing—even lamenting—her own legal inability to dislodge a wrongheaded precedent.

Yet the permission model does not require that judges forgo a rich rhetoric of precedential fidelity. Yes, precedent would be a legally sufficient reason for decision, and the Justices will often feel comfortable saying no more than, “We are following precedent” (much as they now say, “We are voting to deny certiorari”). At other times, the Justices may feel pressed to say more—in which case they need only access the judiciary’s storehouse of stare decisis aphorisms and clichés. The Justices could note, for example, the social importance of reliance, the value of consistency, and the wisdom of their venerable predecessors. These sorts of remarks are of course characteristic of virtually every major ruling on stare decisis—whether or not precedent is overruled.[126] Moreover, these remarks are both true and vacuous in that they reveal laudable principles but do not dictate what to do in any particular case. By helping themselves to this sort of language and reasoning, the Justices can give rhetorical heft to permissive precedent.

Further, a practice of reciting precedent-rhetoric fosters an expectation that judges state their merits-based reasoning when they deviate from precedent. Again, the permission model makes precedent an allowable option simply by virtue of its being precedent. But no such permission would apply to rulings that break from precedent.[127] Without the benefit of precedential justification, a judge who deviates from case law must resort to the merits to justify her ruling.[128] This essentially procedural need to give reasons would make it harder—not impossible, but harder—for cynical deviations from precedent to take place. Judges who have to engage the merits are also more likely to mark their own prejudices.[129] And a judge who has cynical motives for breaking from case law would at least sometimes have to invoke unpersuasive or uncharacteristic arguments on the merits.[130]

If the foregoing points fail to persuade, a proponent of the permission model still has one option, albeit one that pushes up against the binding model: precedent could be viewed as only formally mandatory. That is, precedent could be associated with an explicitly mandatory rule whose content is so malleable that, in practice, it exerts little or no constraint. The result would be a superior rhetorical pose for Justices seeking cover without any significant reduction in decisional flexibility. Precedent’s bindingness would thus be valuable not because it actually binds but rather because it helps precedent operate as a permission. As we will see, current doctrine and practice roughly fit that description. In other words, the Justices may help themselves to a rhetoric of bindingness while engaging in a practice that is best described in terms of permissions.

One might respond that merely “formal” mandatoriness has a whiff of deception or bad faith. In extolling the benefits of binding precedent, the Court might mislead practitioners and the public alike into believing that precedent is a real constraint. This charge is serious but misplaced. Already, the Court sometimes acknowledges that it is engaged in a substantially discretionary enterprise when it applies stare decisis. To wit, the Court has bluntly stated that stare decisis does not constrain when a past error is clear.[131] So the Court is already taking advantage of precedent-rhetoric in a way that is justificatory, self-protecting, and reasonably candid.[132] Deception or bad faith would arise only if the Justices denied the ample discretion that they in fact possess. And the binding model poses a much greater risk of prompting that sort of false assurance.

* * *

The permission model offers a surprisingly attractive alternative to the binding model, at least when it comes to horizontal precedent at the Court. As a descriptive matter, the permission model offers a more plausible account of how precedent does—and does not—influence the Justices. Even in the lower courts, the permission model has significant (albeit reduced) explanatory power. Moreover, there are plausible normative reasons to favor the permission model.

That said, both models have distinctive sources of appeal, and neither can claim complete superiority over the other. More importantly, these models are stylized simplifications, and the law is free to blend both approaches so as to benefit from each model’s insights. The next Part explores that possibility.

III. Lessons for Stare Decisis

In light of the foregoing case in favor of the permission model, what should the content of stare decisis be? This Part outlines several broad answers, each of which cuts against the grain of scholarly commentary but reflects underappreciated features of actual judicial practice, particularly in the Court. Two conclusions bear emphasis. First, the ease with which the Court evades horizontal precedent is no accident or calamity but rather an unavoidable and indispensable feature of stare decisis. Second and counterintuitively, the best way to strengthen precedent may be to make it more of a permission.

A. The Need for Merits-Sensitive Stare Decisis

At present, case law often casts stare decisis in mandatory terms.[133] Does that practice defy the permission model by placing stock in precedent as a constraint? Perhaps not, so long as the formally mandatory aspects of stare decisis are adequately merits-sensitive. In other words, even formally mandatory principles of stare decisis do not constrain to the extent there is overlap between (i) the merits of a legal issue and (ii) the factors that purportedly dictate whether to follow precedent on that issue.

The permission model accordingly provides reason to celebrate merits-sensitivity in the law of stare decisis. As we have seen, the permission model aims to give judges the flexibility to follow precedent as a good enough shortcut, hide behind it as a shield, or abandon it as harmfully erroneous. To create that wide range of legal and rhetorical options, and more besides,[134] a formally binding doctrine of stare decisis has to include some considerations that are merits-sensitive as well as factors that are relatively merits-insensitive. Further, the proper way to balance those factors must be significantly indeterminate so that the Justices can vary their emphasis on the merits to whatever degree the situation demands. And indeed, that facially confounding description fits well with current doctrine in its evidently fallen state. There are lots of vague stare decisis considerations, affording determined jurists many ways of getting around almost any wrongheaded precedent. But the number and interconnectedness of the various considerations also prompt explanation while creating opportunities for dissenters to embarrass cynical or slapdash jurists.[135] The result is a stare decisis of speed bumps, as opposed to roadblocks.

The idea that stare decisis should be merits-sensitive runs headlong against some rather aspirational views of precedent, including views espoused by the Court. For example, the Justices (especially Justice Elena Kagan) often posit that only merits-neutral “special justifications” can justify overruling precedent.[136] And Professor Randy Kozel has lately championed a similar view.[137] How, after all, could precedent possibly constrain if it can be circumvented whenever it seems incorrect?[138] But as we have seen, that oft-repeated question is inapt: precedent can usefully foster adherence to past decisions through its permissive aspect alone, without exerting any constraint whatsoever.

The permission model thus frees us to recognize that stare decisis cannot work in a truly merits-neutral fashion.[139] The law of precedent is largely concerned with how and when to correct legal error. And it is impossible to arrive at an intelligent view of error-correction without assessing the nature of the error in question. If a past decision reflects a misguided view of legal arcana, perhaps it is better to have settlement than correctness. But if the precedent instead strikes at the heart of a cherished legal value, that fact should have a powerful effect on a court’s willingness to overrule.[140] So in heeding Justice Robert Jackson’s call for “a sober appraisal of the disadvantages of the innovation as well as those of the questioned case,” the Justices are not escaping the merits so much as recasting them.[141] A Justice’s views of the merits, if suitably strong and properly expressed, always remain available as a basis to overrule, even—or especially—when the Court engages in a “weighing of practical effects.”[142]

We have already seen that case law on stare decisis exhibits substantial merits-sensitivity insofar as it highlights factors such as the quality of a past decision’s reasoning.[143] But even seemingly merits-neutral factors are also, on reflection, blended with the merits. What counts as cognizable reliance, for instance, is inextricably linked to what interests are legally recognized and condemned. Take segregationists’ reliance on the “separate but equal” principle of Plessy v. Ferguson.[144] From one standpoint, there was vast reliance on Plessy, which underwrote much of southern society. But it is impossible, or repulsive, to evaluate that reliance without reference to an underlying theory of legal rights. Segregationists’ reliance on Plessy, no matter how vast, cannot possibly “count”—perhaps not at all, but certainly not in a way that might override the interests of persons legally entitled to equality in basic aspects of life.[145]

Reliance interests are similarly merits-sensitive in many other circumstances.[146] Citizens United v. FEC[147] offers a more recent example: whether to “count” the government’s reliance on campaign finance case law largely depends on one’s view of the First Amendment.[148] Much as with Plessy, deliberate reliance can easily be recast as ill-gotten gains. So again, even relatively merits-insensitive factors do and must allow for some consideration of both merits and non-merits considerations, rather than stringently prioritizing the latter. A truly merits-insensitive principle of mandatory precedent would be inflexible and stifling. The Justices neither would nor should abide it.

Instead of trying to root out merits-sensitive overrulings, the Court might try to foster greater deliberation and transparency. Take the doctrinal spectrum of reliance interests, which ascribes special force to reliance on contractual and property rights while discounting reliance in cases involving constitutional rights or powers.[149] Recognizing that spectrum does not block the Court from overruling any particular kind of decision, but it does help draw attention to discrete considerations in certain cases. The spectrum also acknowledges that constitutional cases are especially likely to implicate merits-sensitive forms of reliance. The reliance spectrum thus offers a set of merits-based categories that can guide judicial reasoning without seriously attempting to constrain the Court. As a modest reform, the Court might elaborate and refine these categories.

Merits-sensitivity is also consistent with other efforts to protect reliance interests.[150] Consider “the doctrine of one last chance,” which requires the Court to give notice of disruptive rulings, thereby reducing or mitigating their harmful effects.[151] That mandatory, but essentially procedural, precept does not attempt to “constrain” in the sense of preventing the Court from reaching a particular legal result. A determined Court could simply deny that its decision is sufficiently disruptive—who’s going to stop it?—or else wait until a later case to issue a decisive ruling. Yet this precept would still mitigate reliance costs. Further, procedural principles that slow the overall pace of overruling can prevent the Court from simply doing whatever its current membership desires, even if no particular precedent is deemed binding.[152] Rather than ping-ponging between competing views as rival ideologies hold sway, the Court would instead chart a more gradual course—and so seem more like a judicial body in the common law tradition.

Embracing stare decisis’s merits-sensitivity, or otherwise moving toward the permission model, could have broader benefits for legal culture. When nominees to the Court extol stare decisis, sophisticates sometimes worry that members of the public, even the Senate, are being misled.[153] Who, after all, can really count on precedent to constrain the Justices? But once precedent is viewed primarily as a permission, both the value and limits of precedent-rhetoric come into focus. Nominees’ now-routine practice of genuflecting before stare decisis helps maintain a culture of respect for precedent, thereby facilitating case law’s genuine ability to affect outcomes by operating as a shortcut or shield. At the same time, permissive precedent offers greater transparency about what precedent cannot accomplish—namely, a serious check on the Justices’ pursuit of their clear, first-principles views of the law. The result, on balance, would be less mystification and an improved public understanding of the Court’s role.

B. Combining Permissive and Mandatory Rules

The permission model also points toward new ways of structuring the stare decisis inquiry. The key takeaway is this: precedent is most likely to foster adherence to past decisions not when it is maximally mandatory but rather when it exhibits a mix of both permissions and mandates.

Consider the underappreciated question whether overruling precedent should ever be mandated. Some Justices embrace mandatory overruling,[154] but the permission model counsels just the opposite. To the extent precedent-following is impermissible, judges cannot rely on precedent as a shortcut or hide behind it as a shield.[155] Still, there are reasons to make overruling at least formally mandatory. For example, mandatory rules could help prompt greater explanation as well as focused consideration of discrete topics, even if they cannot meaningfully constrain the ultimate outcome.[156] And we have also seen that formally binding principles of stare decisis, even if unconstraining, can help precedent operate as a shield.[157] These points suggest a basis for adopting a hybrid decisional structure with both mandates and permissions.

Justice Thomas’s Gamble opinion offers an example of such a hybrid. As we have seen, Justice Thomas suggested that overruling is conditionally mandatory—that is, mandatory when certain procedural and substantive conditions are met.[158] The fact that Justice Thomas’s rule is mandatory helps prompt judicial explanation and so guards against cynical exploitation of precedent’s permissive aspect. But the conditional nature of Justice Thomas’s rule is also important: the rule establishes a presumption in favor of precedent’s permissibility even when traditional stare decisis factors point toward overruling. As a result, Justice Thomas can usually rely on precedent’s permissive aspect to brush aside troublesome requests to overrule. Otherwise, he would have to test every precedent against the stare decisis factors, undermining efficiency. Justice Thomas thus aims at a fruitful compromise by blending permissions and mandates.

Still, Justice Thomas’s attempt to strike that compromise is subject to criticism. First, his proposal—essentially, that precedent is permissible except that it must be overruled if demonstrably erroneous—omits all non-merits factors. Rather than directing judges to consider reliance or other relevant interests, Justice Thomas focuses exclusively on the merits. That simplicity is elegant but also passes up an opportunity to encourage more fulsome judicial deliberation and explanation, with all their attending benefits.[159] Second, Justice Thomas’s proposal places too large a condition on the permission that precedent affords. Under his approach, overruling is mandatory whenever there is a sufficiently strong argument on the merits. That limitation undermines precedent’s ability to operate as a shield since every attempt to invoke precedent invites a decisive merits-based objection. Surprisingly, Justice Thomas’s version of the permission model isn’t permissive enough.

We can improve on Justice Thomas’s bare-bones approach by revisiting the familiar stare decisis factors recognized in current doctrine. In short, the stare decisis factors might best operate not by adding weights to either side of a scale but rather by altering whether it is mandatory to follow, mandatory to overrule, or permissible to do either.

As a first step, consider that some reasons to overrule seem much weaker, or less constraining, than others. Concluding that a precedent was “poorly reasoned,” for instance, is relevant to many Justices but cannot be as compelling as a finding that a precedential rule is “unworkable.”[160] A poorly reasoned case could still be right, whereas an unworkable one seems necessarily incorrect. And intuitively, many poorly reasoned precedents should be, and are in fact, retained. By contrast, it is hard to see why any truly unworkable rulings should—or even could—survive. These disparities can be fleshed out using our now-familiar distinction between permissions and mandates. Perhaps the quality of a decision’s reasoning can render overruling permissible, whereas a decision’s workability sheds light on whether overruling is mandatory.

A similar analysis could apply to factors that trigger heightened stare decisis protections. For example, Justice Kagan has recently argued that reliance interests do not create stare decisis obligations but instead render them “superpowered.”[161] And the fact that a precedent interpreted a statute is often thought to have a similar effect, yielding “enhanced” stare decisis.[162] Perhaps these considerations negate the option to overrule, making it mandatory to follow.

Building on these points, we can outline the resulting decision procedure, or an example of one, as follows:

Figure 2: Schema of Permissive and Mandatory Rules

Rule 1 By default, it is mandatory to follow precedent.
Rule 2 Notwithstanding the above, it is permissible (not mandatory) to follow a precedent that is poorly reasoned or a doctrinal outlier.
Rule 3 Notwithstanding the above, it is mandatory to follow precedents that interpret a statute or protect large reliance interests.
Rule 4 Notwithstanding the above, it is mandatory to overrule precedents that are unworkable.


This kind of schema[163] structures judicial decision-making, rendering stare decisis somewhat more comprehensible and rule-like, as compared with simply throwing qualitatively different factors into a big doctrinal soup.[164] Many stare decisis reformers, including opponents of the permission model, could find this basic approach useful.

Yet the foregoing schema still largely accords with the permission model. At the outset, for example, judges can move from Rule 1 to Rule 2 whenever precedent is “poorly reasoned,” thereby undermining Rule 1’s ability to constrain. Further, Rule 2 itself establishes a permission and so leaves room for overruling while also enabling precedent to operate as a shortcut or a shield (or both). True, Rule 2’s permission is contingent on later rules, but those principles are merits-sensitive: at least in constitutional cases, the merits will play a significant role in ascertaining legitimate reliance under Rule 3 and even workability under Rule 4.[165]

Moreover, there is good reason to think that introducing permissions would actually make precedent more effective. A maximally binding rule of stare decisis would run headlong into the need for merits-sensitivity,[166] prompting Justices to defect from the rule. And a purely mandatory principle that is merits-sensitive would undermine precedent’s ability to shield Justices from criticism and embarrassment.[167] By contrast, strong permissions can both serve as a shield and leave room for flexibility based on merits considerations. In the above schema, for instance, Rule 2 empowers a Justice to say: “In light of the respect owed to venerable case law, I can properly choose to follow precedent even if it may be poorly reasoned or a doctrinal outlier.” The appeal of being able to make that claim would do more to prompt judicial adherence to precedent than any formally mandatory—but actually toothless—rule.

Of course, some readers may desire greater constraint or a different mix of constraint and permission. But if so, a path forward is available: by imposing different or more stringent, merits-neutral triggering conditions, the foregoing schema can move closer to the binding model, even as it preserves a useful zone of permissive discretion. The permission model thus yields insights that can be deployed by proponents of conventional stare decisis or the binding model.

C. The Legitimacy of Personalized Precedent

Precedent is sometimes held out as being capable of overcoming ideological and methodological divides among diverse jurists.[168] The permission model both undermines and fulfills that aspirational picture.

Permissions, by their nature, invite personal variation. Whether a judge views any given precedent as “good enough” for efficient decision-making, as opposed to clearly wrong, will depend on that judge’s first-principles views of legal correctness.[169] Likewise, whether a Justice uses precedent-rhetoric as a shield will depend on that particular Justice’s view of right decision-making as well as on her susceptibility to various decisional pressures. For these reasons, different jurists would invoke precedent in different situations and to different degrees. And that is what actually happens.[170] Justice Kagan, for instance, is a precedent stalwart, whereas Justice Thomas sees little or no compulsion to follow precedent. This state of affairs must be objectionable for an adherent of the binding model since it demonstrates the ease of molding stare decisis to suit one’s own view of judicial decision-making.

By contrast, the permission model both recasts and justifies “personal precedent,” or the idea that each Justice should regard her own opinions as precedential.[171] Consider perhaps the most controversial form of personal precedent—namely, the “perpetual dissent,” whereby a Justice who dissented from an admittedly precedential decision feels free to deviate from the decision while calling for its overruling.[172] This practice follows from the permission model. A Justice who has stated a firm view can more easily see that contrary precedent is wrong by her lights, undercutting efficiency-based reasons for following precedent. And of course, the permission model views precedent as nonbinding. Thus, the persistent dissenter can point out her own prior dissent as justification enough for ongoing defiance.

At the same time, the permission afforded by Court precedent would still serve a desirable function: it supplies a face-saving way for a perpetual dissenter to stop dissenting. Even the most fervent objector might eventually come to see that her past views are wrong or just not worth adhering to. And with varying degrees of fanfare, the Justices take that opportunity, acquiescing to rulings that they had previously resisted.[173] This phenomenon is, of course, another example of precedent operating as a shield. So the permission model not only tolerates perpetual dissents but approves them. And a good thing too, since stare decisis seems unable to prevent the Justices from behaving any differently.

Finally, the permission model’s personal adaptability contributes to its appeal and practical viability. Because a permission is itself subject to interpretation, a jurist’s general attitude toward interpretation can affect the total range of permissions available to her. A judge who is inclined to find texts to be determinate, for instance, might think that the permissions afforded by case law (a kind of text) are relatively narrow in scope and few in number. The permission model would thus afford that jurist a relatively limited increase in lawful discretion. By contrast, a judge who generally finds legal materials to be indeterminate might perceive a wider range of precedential permissions—allowing her to reach a wider range of legal outcomes. One might say the permission model adapts to its user in that it tends to offer more permissions to more permissive interpreters.

That interpretive adaptability suggests that the permission model should prove attractive to diverse jurists—even ones who usually abhor discretion. Whenever a judge strongly believes that underlying law commands a specific outcome, she will appreciate that the permission model allows her to reach the right result.[174] And judges who believe that there is already extensive discretion built into underlying law will still be at home in the added freedom the permission model affords. The permission model thus has something to offer every Justice—a feature that may help explain why Justice Thomas, a noted skeptic of judicial discretion, has marked himself as the Court’s foremost proponent of permissive precedent, at least in some form.[175] By contrast, binding approaches to precedent must often frustrate not only Justices who perceive clear answers in underlying law but also Justices who desire decisional flexibility.

There are of course worries attending the personalization of precedent. Particularly those who view underlying law as highly indeterminate, or who otherwise foresee rampant divergence among judges, might seek fixity in case law. And only mandatory precedent, it seems, can offer the determinacy and stability that the rule of law demands.[176] Strict stare decisis, in other words, might compensate for indeterminacy in underlying law.

But that critique overlooks both the frailty of precedential mandates and the attractive force of precedential permissions. When correct legal answers are hard to discern, sharply disputed, or open to endless second-guessing, determined Justices will have little difficulty finding their way around whatever pesky stare decisis factors stand in their way. Yet those same jurists would still find it desirable to seek out the security of precedential permissions. By offering an assurance of lawfulness, precedent lets beleaguered Justices sidestep difficult questions, find common ground, and escape criticism.[177]

So yes, there is a need for judges to fashion some order amidst chaos. But permissions, apart from mandates or prohibitions, can go a long way toward meeting that need. And an exclusive focus on precedential mandates will give false comfort while distracting from auspicious reforms. A more nuanced approach would combine permissions and mandates, placing greater emphasis on one or the other as the situation demands.


The operation may have been a success. But did the patient die? Perhaps so. Viewing precedent as largely or entirely permissive does mean expecting less. In particular, it means giving up on the idea that Court precedent is, or might one day become, a strong constraint on the Justices. Wherever a precedential permission abides, so too does a limited discretion—namely, the discretion either to follow case law or, instead, to pursue the underlying merits. So much the worse for the rule of law, some might say.

Yet attention to permissions also teaches us to expect more from precedent. It shows us how precedent can help the Justices accomplish their goals without tying their hands, cluttering their work, or succumbing to internal and external pressures. And it also points toward new ways of fostering adherence to precedent, including ways that would retain at least some of precedent’s traditional binding quality. It turns out that precedent has always been doing more than just rendering particular outcomes formally mandatory. And by attending to those overlooked functions, stare decisis reformers can find more effective ways of achieving their goals.

In short, focusing on precedent’s permissive aspect entails a shift in expectations, demanding both more and less and so, on balance, something different. The upshot is a new way of understanding, and revising, stare decisis.

  1. .See, e.g., P.S. Atiyah, Form and Substance in Legal Reasoning: The Case of Contract, in The Legal Mind: Essays for Tony Honoré 19, 27 (Neil MacCormick & Peter Birks eds., 1986) (“The concept of a system of precedent is that it constrains judges in some cases to follow decisions they do not agree with.”); see infra note 2.
  2. .The Supreme Court routinely presents stare decisis as a requirement, albeit one subject to conditions. See, e.g., Kisor v. Wilkie, 139 S. Ct. 2400, 2423 (2019) (“[W]e would need a particularly ‘special justification’ to now reverse Auer.”); Halliburton Co. v. Erica P. John Fund, Inc., 573 U.S. 258, 266 (2014) (“Before overturning a long-settled precedent . . . we require ‘special justification’ . . . .”). Even when describing it as the “preferred course” or “wise policy,” the Court seems to view stare decisis as a potentially decisive factor. E.g., Payne v. Tennessee, 501 U.S. 808, 827 (1991) (quoting Burnet v. Coronado Oil & Gas Co., 285 U.S. 393, 406 (1932) (Brandeis, J., dissenting)); see also infra notes 85, 131.
  3. .Notably, the Justice who has lately sounded the alarm over abortion precedent, Stephen Breyer, is also the Court’s foremost advocate of overruling extensive case law allowing for capital punishment. Compare Franchise Tax Bd. of Cal. v. Hyatt, 139 S. Ct. 1485, 1504–05 (2019) (Breyer, J., dissenting) (invoking stare decisis while conspicuously citing Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 854–55 (1992)), with Glossip v. Gross, 135 S. Ct. 2726, 2762 (2015) (Breyer, J., dissenting) (casting doubt on the Court’s Eighth Amendment precedents). As stare decisis skeptics observe, this sort of tension is hardly unique to Breyer.
  4. .The most recent examples include Janus, Hyatt, and Knick, which all overruled precedent, as well as Gamble, Gundy, and Kisor, which did not. These cases are cited in other notes.
  5. .See, e.g., Knick v. Twp. of Scott, 139 S. Ct. 2162, 2190 (2019) (Kagan, J., dissenting) (“If that is the way the majority means to proceed—relying on one subversion of stare decisis to support another—we may as well not have principles about precedents at all.”).
  6. .Ramos v. Louisiana, 140 S. Ct. 1390, 1414 (2020) (Kavanaugh, J., concurring in part). On Kavanaugh’s proposed way forward, see text accompanying infra note 141.
  7. .Frederick Schauer, Stare Decisis—Rhetoric and Reality in the Supreme Court, 2018 Sup. Ct. Rev. 121, 143 (2019). Schauer notes that “[t]here are, of course, exceptions” to the generalization that stare decisis poses no serious constraint. Id. at 133.
  8. .Online Shop of Strict Scrutiny Podcast, https://‌strict-scrutiny-podcast-shop‌.myshopify‌.com‌/ [https://‌perma‌.cc‌/JY8T-ZNLU].
  9. .139 S. Ct. 1960, 1980 (2019) (Thomas, J., concurring).
  10. .See id. at 1986 (describing stare decisis as a source of “permissible interpretations”).
  11. .Id. For more on Justice Thomas’s views, see infra subpart I(B).
  12. .See infra subpart I(B).
  13. .I assume, consistent with practice, that the Supreme Court (and possibly also Congress) may lawfully create and revise rules of stare decisis on pragmatic grounds. See infra notes 61, 136.
  14. .Permissive precedent is both an accommodation and a rejoinder to some strands of legal realism: even if precedent is purely rhetorical and not at all constraining, as realists are often thought to claim, precedent can still be consequential—even a consequential authority. See infra note 76 and accompanying text.
  15. .On whether vertical stare decisis comports with viewing precedent as permission, see infra text accompanying note 107.
  16. .June Med. Servs. L.L.C. v. Russo, 140 S. Ct. 2103, 2133 (2020) (Roberts, C.J., concurring in the judgment).
  17. .Id. at 2133–34 (applying Whole Woman’s Health v. Hellerstedt, 136 S. Ct. 2292 (2016)); see infra text accompanying note 78. To a great extent, the question in June Medical was whether to narrow Hellerstedt. See Richard M. Re, Narrowing Precedent in the Supreme Court, 114 Colum. L. Rev. 1861, 1889 (2014) (distinguishing between narrowing and overruling). And in fact, Roberts himself purported to revise or narrow Hellerstedt, even as he relied on it. See June Medical, 140 S. Ct. at 2138–39 (Roberts, C.J., concurring in the judgment).
  18. .See Robert Barnes, With Abortion Ruling, Roberts Reasserts His Role and Supreme Court’s Independence, Wash. Post (June 29, 2020, 7:18 PM), https://‌www‌.washingtonpost‌.com‌/politics‌/courts‌_law‌/john-roberts-supreme-court-abortion-ruling‌/2020‌/06‌/29‌/64dd30a6-ba3b-11ea-80b9-40ece9a701dc‌_story.html [https://‌‌/CL87-6A33] (speculating that Roberts’s vote in June Medical was “intended to keep the court from moving too quickly to the right”).
  19. .See infra notes 81, 154 and accompanying text.
  20. .See infra subpart III(A).
  21. .The scholarly view closest to the permission model is probably McConnell’s. Michael W. McConnell, Time, Institutions, and Interpretation, 95 B.U. L. Rev. 1745, 1769 (2015). McConnell’s view is summarized in three propositions: (1) “The existence of precedent always counts as a good reason in support of the conclusion that flows from that precedent.” Id. at 1769; (2) “If a particular outcome of the case follows from precedent, judges are obliged either to follow the precedent or to explain why not—and if the latter, to establish a new, contrary precedent.” Id.; and (3) “The weight to be given questionable precedents is a matter of comity within the Court.” Id. at 1770. For more on McConnell’s insightful view, see infra notes 57, 64, 68, 106.
  22. .For a recent critique, see Michael Gentithes, Janus-Faced Judging: How the Supreme Court Is Radically Weakening Stare Decisis, 62 Wm. & Mary L. Rev. 83 (2020).
  23. .On this view, a court has a legal power to permit its successors to rule as it has done. See also infra note 47 and accompanying text (discussing how precedential permissions can be authoritative).
  24. .One interesting variation is that judges might be permitted (not mandated) to invoke certain kinds of authorities as reasons for taking a particular action. To wit, Frederick Schauer argues that “persuasive” authorities, such as extra-jurisdictional precedents, are better labeled “optional” or “permissive” on the ground that judges have “discretion” as to whether to invoke these authorities at all. Frederick Schauer, Authority and Authorities, 94 Va. L. Rev. 1931, 1946–47, 1946 n.53 (2008). Insofar as an “optional” authority can render a particular outcome lawful, it would indirectly give rise to a permission as I use that term here.
  25. .See Larry Alexander, Constrained by Precedent, 63 S. Cal. L. Rev. 1, 5–7 (1989); Lawrence B. Solum, How NFIB v. Sebelius Affects the Constitutional Gestalt, 91 Wash. U. L. Rev. 1, 40 (2013) (explaining that “persuasive authority . . . can be overcome by the balance of reasons” despite its “epistemic, predictive, and legitimating authority”); see infra text accompanying note 56.
  26. .For example, the permission model could be supplemented with a rule as to when a court must withdraw a permission, thereby yielding a form of stare decisis with an additional mandatory aspect apart from the mandate described in the Table’s Option 3. See infra note 59.
  27. .See Gamble v. United States, 139 S. Ct. 1960, 1986 (2019) (Thomas, J., concurring).
  28. .Id. at 1986 & n.6.
  29. .Id. at 1986 n.6.
  30. .See United States v. Olano, 507 U.S. 725, 733 (1993).
  31. .See, e.g., Ziglar v. Abbasi, 137 S. Ct. 1843, 1870 (2017) (Thomas, J., concurring in part and concurring in the judgment) (“The Court correctly applies our precedents, which no party has asked us to reconsider.”); Shepard v. United States, 544 U.S. 13, 28 (2005) (Thomas, J., concurring in part and concurring in the judgment) (“The parties do not request it here, but in an appropriate case, this Court should consider Almendarez-Torres’ continuing viability.”).
  32. .See Benjamin N. Cardozo, The Nature of the Judicial Process 149 (1921) (“[T]he labor of judges would be increased almost to the breaking point if every past decision could be reopened in every case . . . .”); Jonathan R. Macey, The Internal and External Costs and Benefits of Stare Decisis, 65 Chi.-Kent L. Rev. 93, 96 (1989).
  33. .See Amy Coney Barrett, Originalism and Stare Decisis, 92 Notre Dame L. Rev. 1921, 1930–31 (2017) (noting that certain settled questions, like the lawfulness of the Social Security Act, are not raised or, via certiorari, entertained, regardless of their susceptibility to legal challenge, particularly on originalist grounds).
  34. .See Gamble, 139 S. Ct. at 1984 (Thomas, J., concurring) (“Federal courts may (but need not) adhere to an incorrect decision as precedent, but only when traditional tools of legal interpretation show that the earlier decision adopted a textually permissible interpretation of the law.”) (emphasis added).
  35. .Id. at 1986 (emphasis added and original italicization altered).
  36. .See generally Caleb Nelson, Stare Decisis and Demonstrably Erroneous Precedents, 87 Va. L. Rev. 1 (2001). Justice Thomas also drew directly from Madison. See Letter from James Madison to N.P. Trist (Dec. 1831), in 9 The Writings of James Madison 471, 477 (Gaillard Hunt ed., 1910). For related historical work, see John O. McGinnis & Michael B. Rappaport, Originalism and the Good Constitution 154–74 (2013); Randy J. Kozel, Original Meaning and the Precedent Fallback, 68 Vand. L. Rev. 105, 150 (2015) (“[T]here is a basis for concluding that, at very least, reliance on precedent was a permissible response to linguistic indeterminacy at the time of the Founding.”). But see infra note 60 (citing Lawson).
  37. .See Randy E. Barnett, The Gravitational Force of Originalism, 82 Fordham L. Rev. 411, 418 (2013); Lawrence B. Solum, Originalist Theory and Precedent: A Public Meaning Approach, 33 Const. Comment. 451 (2018).
  38. .See H.L.A. Hart, The Concept of Law 138–43 (1961) (discussing judicial discretion in hard cases); Ken Kress, A Preface to Epistemological Indeterminacy, 85 Nw. U. L. Rev. 134, 138–39 (1990) (explaining that “[m]etaphysical indeterminacy speaks to whether there is law”).
  39. .See generally Solum, supra note 37.
  40. .Cf. Frederick Liu, Book Note, The Supreme Court Appointments Process and the Real Divide Between Liberals and Conservatives, 117 Yale L.J. 1947, 1956 (2008) (reviewing Christopher L. Eisgruber, The Next Justice: Repairing the Supreme Court Appointments Process (2007)) (“The real divide between judicial liberals and judicial conservatives lies in their views of the relative number of hard cases the Supreme Court hears.”).
  41. .Current doctrine implicitly rests on the view that the Justices, by virtue of their legitimacy and expertise, should privilege their own first-person perspectives on legal questions over more widely shared or predictable alternatives. That perspectival choice, while sensible, creates a foreseeable risk—namely, that the gap between mere error and “demonstrable” error will shrink or collapse. See generally Richard M. Re, Clarity Doctrines, 86 U. Chi. L. Rev. 1497 (2019). The upshot is that such a test might not actually constrain at all.
  42. .Nelson himself envisioned a mandatory component to precedent, particularly within zones of underlying legal indeterminacy, based in part on Madison’s theory of “liquidation”:For Madison, then, when the early interpreters of a statute or constitutional provision that was obscure or ‘controverted’ gave it a permissible construction, they helped to ‘settle its meaning’; subsequent interpreters could be bound to follow that construction even if they would have adopted a different one as an original matter.Nelson, supra note 36, at 13–14; see also The Federalist No. 37 (James Madison); William Baude, Constitutional Liquidation, 71 Stan. L. Rev. 1, 16, 36–42 (2019). Nelson suggested that, within “the space left by the indeterminacy of the underlying rules of decision,” “it is perfectly sensible for courts to apply a rebuttable presumption against overruling precedents.” Nelson, supra note 36, at 84. Further, a “special reason” might call for preserving even demonstrably erroneous precedent. Id. at 5.
  43. .See Allen v. Cooper, 140 S. Ct. 994, 1007 (2020) (Thomas, J., concurring in part and concurring in the judgment) (“[A]lthough I agree that Florida Prepaid Postsecondary Ed. Expense Bd. v. College Savings Bank is binding precedent, I cannot join the Court’s discussion of stare decisis.”) (internal citations omitted).
  44. .See Akhil Reed Amar, The Supreme Court, 1999 Term—Foreword: The Document and the Doctrine, 114 Harv. L. Rev. 26, 81 (2000) (“[I]t seems permissible for the Court to give its past decisions a rebuttable presumption of correctness.”). Here too, we must distinguish (i) a permission to follow precedent and (ii) a permission to create a conditional mandate in favor of precedent.
  45. .See supra note 1.
  46. .See Payne v. Tennessee, 501 U.S. 808, 828 (1991).
  47. .In Raz’s framework, the permission I have in mind would be both “explicit” and “strong” because it flows from an authorizing norm rather than resulting simply from the absence of relevant prohibitory norms. See Joseph Raz, The Authority of Law: Essays on Law and Morality 67, 117 n.4 (2d ed. 2009). But can a permission be truly authoritative? Raz himself posits that “a law is authoritative if its existence is a reason for conforming action and for excluding conflicting considerations.” Id. at 29. Precedential permissions would exclude legal reasons for criticizing judges who follow precedent. And perhaps the result—legal permissibility—can itself be viewed as a “reason for conforming action.” See also infra note 49.
  48. .A permission differs from a standard: rather than surfacing deliberation as to factors, the permission (where it applies) can substitute for consideration of deeper legal purposes. So a permission can stand in for any reason, including political reasons that, even if legitimate, sit uncomfortably within the law. Cf. Tara Leigh Grove, The Supreme Court’s Legitimacy Dilemma, 132 Harv. L. Rev. 2240, 2240 (2019) (reviewing Richard H. Fallon, Jr., Law and Legitimacy in the Supreme Court (2018)) (discussing possible tensions between legal and sociological legitimacy); see supra note 17 and accompanying text (discussing June Medical). In this way, precedential permissions can create legal space for judges to act on moral and other nonlegal considerations. See also text accompanying infra note 111.
  49. .The precedent is authoritative in the sense of being content-independent. See generally H.L.A. Hart, Commands and Authoritative Legal Reasons, in Essays on Bentham: Studies in Jurisprudence and Political Theory 243 (1982). Hart critically discusses Bentham’s notion that whatever is not proscribed is permitted, id. at 250, whereas I posit that, when precedent neither proscribes nor permits, a judge may fall back on her first-principles approach to legal correctness.
  50. .The permission can be viewed as the rule-based component of a safe harbor given that a safe harbor not only “provides by rule that particular facts comply with the law” but also “leaves other facts to be judged by a standard.” Susan C. Morse, Safe Harbors, Sure Shipwrecks, 49 U.C. Davis L. Rev. 1385, 1387 (2016).
  51. .Consider case reasoning that the Court deviates from, thereby creating a new precedent without overruling. See, e.g., Michael C. Dorf, Dicta and Article III, 142 U. Pa. L. Rev. 1997, 2019–22 (1994). The older precedents seem not to be mandatory, but perhaps they abide as permissions. Examples arise in substantive due process, nondelegation, and removal case law.
  52. .The permission is thus asymmetric: it is not a permission not to follow case law. By making it easier to follow precedent, the permission model makes it relatively difficult to deviate from it. And it does so without requiring a “special justification” for overruling.
  53. .For a longer answer, see infra section II(C)(2).
  54. .The permission operates as a reason-stopper and so checks the tendency for accumulated reason-giving to constrain in the manner associated with the common law. Absent the permission, that is, judges’ discussions of precedent would gradually establish when adherence to precedent both is and is not justifiable. Cf. Frederick Schauer, Giving Reasons, 47 Stan. L. Rev. 633, 649–58 (1995).
  55. .A related form of moderation would answer the main text’s question with a presumption: “No, unless there is a showing that the judge acted on an impermissible reason.”
  56. .See supra note 25.
  57. .Advancing a similar but more constraint-oriented view, McConnell argues that “at least sometimes it is legitimate to treat precedent as the decisive factor” and suggests that “it also seems to follow that . . . judges always act legitimately when they decide cases on the basis of precedent (except perhaps in the rare instance when other considerations, such as text and original meaning, overwhelmingly support the other side).” McConnell, supra note 21, at 1769.
  58. .For example, the model might apply only to the Supreme Court, to constitutional issues, or to fragmented rulings.
  59. .We might say that Lawrence v. Texas did not just deviate from, but also withdrew, the permission previously afforded in Bowers v. Hardwick. See Lawrence v. Texas, 539 U.S. 558, 577–78 (2003) (overruling Bowers v. Hardwick, 478 U.S. 186 (1986)).
  60. .I assume the conventional view that the Constitution itself permits some legal preference for precedent over constitutional first principles, perhaps because stare decisis is encompassed in the “judicial Power” of Article III and supported by longstanding practice. See supra note 36 (collecting sources). But see Gary Lawson, The Constitutional Case Against Precedent, 17 Harv. J.L. & Pub. Pol’y 23, 31 (1994) (arguing that horizontal precedent cannot lawfully overcome the Constitution’s “objectively ascertainable meaning”).
  61. .Permissions afforded by judicial decisions thus resemble other normative powers operative in everyday life. Cf. Joseph Raz, Practical Reason and Norms 140 (1975) (concerning the obligation of people to keep their promises).
  62. .For my purposes, a case can be salient whenever it engages the judge’s first-principles views of the law, what might be called the jurist’s “judicial ideology,” whether or not the issue is of public notice. See generally Lee Epstein & Jeffrey A. Segal, Measuring Issue Salience, 44 Am. J. Pol. Sci. 66, 72–77 (2000) (discussing metrics for salience). Cf. Joshua B. Fischman & David S. Law, What Is Judicial Ideology, and How Should We Measure It?, 29 Wash. U. J.L. & Pol’y 133, 136–37 (2009) (exploring the ambiguity surrounding “judicial ideology”). Others assess case salience specifically based on public awareness. See Forrest Maltzman, James F. Spriggs II & Paul J. Wahlbeck, Crafting Law on the Supreme Court: The Collegial Game 45–46 (2000) (distinguishing legal and political salience); Frederick Schauer, The Supreme Court, 2005 Term—Foreword: The Court’s Agenda—and the Nation’s, 120 Harv. L. Rev. 4, 11–12, 49 (2006).
  63. .Cf. Adam J. Hirsch, Cognitive Jurisprudence, 76 S. Cal. L. Rev. 599, 613 (2003) (suggesting that “judges will have greater recourse to convenient, effort-saving devices,” such as formal precedent, “when deciding cases that they deem less important (or simply duller)”).
  64. .See generally Deborah Hellman, An Epistemic Defense of Precedent, in Precedent in the United States Supreme Court 63, 66 (Christopher J. Peters ed., 2013) (arguing that epistemic reasons can “provide a justification for the rule” of precedent); see also Amar, supra note 44, at 81 (“a Justice may rightly give a precedent epistemic weight”).
  65. .This line of thought presupposes equal authority or aptitude among judges across time. See Geoffrey R. Stone, Precedent, the Amendment Process, and Evolution in Constitutional Doctrine, 11 Harv. J.L. & Pub. Pol’y 67, 71 (1988) (arguing that today’s judges “have no greater interpretive authority than their predecessors”).
  66. .The main text’s limitation to multimember appellate courts reflects a premise of the epistemic argument—namely, a supposition that the prior ruling adequately considered the relevant question, such that the mere fact of a second look does not give the later court an edge. Unsurprisingly, then, trial courts generally do not regard one another’s rulings—or even their own—as precedential. See Camreta v. Greene, 563 U.S. 692, 709 n.7 (2011).
  67. .Equivalent efficacy among judges might be accepted as a descriptive fact or as a normative presupposition. See Frederick Schauer, Precedent, 39 Stan. L. Rev. 571, 599 (1987) (noting that, for the epistemic argument to work, “we must either perceive some actual similarity between the decisionmakers or believe that there is some reason to impose a presumed similarity where none in fact exists”). Both sorts of considerations counsel against the Supreme Court’s treating, say, circuit precedent as a permission.
  68. .Precedent’s forcefulness might be linked to whether it is longstanding and widespread. See McConnell, supra note 21, at 1768 (“If we wanted a genuinely evolutionary system of stare decisis, we would allow judges to make up their own minds about the wisdom of decisions, until enough time had passed and enough judges had concurred to establish a consensus.”); see also Peter M. Tiersma, The Textualization of Precedent, 82 Notre Dame L. Rev. 1187, 1225–27 (2007) (discussing common law precedent in contrast with stare decisis). More abstractly, perhaps precedent’s force could properly vary depending on how many judges have approved the precedent, over what span of time, and in what array of deliberative contexts. When a “consensus” exists, perhaps precedent should operate not just as a permission but also (as McConnell proposes) as a reason in favor of a particular ruling. See supra note 57. Yet even that argument for eventually treating precedent as binding would require that case law initially be treated as anything but: a pattern of decisions is instructive only if earlier courts could have rejected precedent.
  69. .See Thomas R. Lee, Stare Decisis in Historical Perspective: From the Founding Era to the Rehnquist Court, 52 Vand. L. Rev. 647, 654 (1999) (“The difficult issue for any system of precedent is not whether a court may rely on precedent in the run-of-the-mill case, but whether it must do so when it perceives an error in the ways of the past.”).
  70. .Schauer substantially explores this point within the context of binding or mandatory precedent. See Schauer, supra note 67.
  71. .See Marin K. Levy, Judicial Attention as a Scarce Resource: A Preliminary Defense of How Judges Allocate Time Across Cases in the Federal Courts of Appeals, 81 Geo. Wash. L. Rev. 401, 407–09, 413–14 (2013); Richard A. Posner, What Do Judges and Justices Maximize? (The Same Thing Everybody Else Does), 3 Sup. Ct. Econ. Rev. 1, 14 (1993).
  72. .The Court thus errs in suggesting that precedent matters only when precedent is deemed incorrect. Compare Charles L. Barzun, Impeaching Precedent, 80 U. Chi. L. Rev. 1625, 1650 n.107 (2013) (“[U]nder the efficiency-of-decision-making view, if the rule of a previous decision is clearly applicable to the set of facts with which a court is presented, then in theory the court does not ever reach an independent conclusion about what the right outcome would be but for the binding precedent.”) (emphasis omitted), with, e.g., Kimble v. Marvel Entm’t, LLC, 576 U.S. 446, 455 (2015) (“[S]tare decisis has consequence only to the extent it sustains incorrect decisions; correct judgments have no need for that principle to prop them up.”).
  73. .See Randy J. Kozel, Settled Versus Right: A Theory of Precedent 37 (2017) (noting the risk of case-law analysis being inefficient in some circumstances).
  74. .See, e.g., Gundy v. United States, 139 S. Ct. 2116, 2130–31 (2019) (Alito, J., concurring in the judgment) (adhering to nondelegation precedents despite opposition to them on the merits, largely because a majority was not at that time prepared to undertake a deep revision of that case law).
  75. .A tweaked metaphor would capture the idea that precedent can develop deliberately, in one fell swoop. Imagine a builder who inherits an open plot and constructs a walkway on it. Many people might then come to walk together along the path—not because the fields are walled off but because the road is easy. For more on the “path” metaphor, see Michael Stokes Paulsen, Steven G. Calabresi, Michael W. McConnell & Samuel L. Bray, The Constitution of the United States 1605 (3rd ed.) (2016) (quoting “The Calf-Path” by Sam Walter Foss in discussing precedent); see also Max Radin, The Trail of the Calf, 32 Cornell L.Q. 137, 137, 139 (1946) (critically discussing the influential poem).
  76. .This perspective is associated (perhaps not entirely correctly) with the legal realists. Compare, e.g., Martha Chamallas & Linda K. Kerber, Women, Mothers, and the Law of Fright: A History, 88 Mich. L. Rev. 814, 842 (1990) (“Realists suspected that the rhetoric of precedent and tradition had been a vehicle for mystifying convenient political choices.”), with Karl N. Llewellyn, The Common Law Tradition: Deciding Appeals 179 (1960) (arguing that legal rules “guide” but do not “control”), and Brian Leiter, Rethinking Legal Realism: Toward a Naturalized Jurisprudence, 76 Texas L. Rev. 267, 268 (1997). Those who view precedent as constraining may nonetheless believe it is also (not merely) rhetorical. E.g., Earl M. Maltz, Abortion, Precedent, and the Constitution: A Comment on Planned Parenthood of Southeast Pennsylvania v. Casey, 68 Notre Dame L. Rev. 11, 11 (1992).
  77. .A recent work argues that the Justices have an acute, sometimes subconscious interest in fostering both the Court’s and their own prestige, particularly among the Justices’ elite social groups. See Neal Devins & Lawrence Baum, The Company They Keep: How Partisan Divisions Came to the Supreme Court 8–11 (2019).
  78. .See supra notes 16–18 and accompanying text.
  79. .See infra subpart III(B) (discussing this possibility).
  80. .This characterization of the binding model corresponds to Option 3 in Figure 1. In other words, the main text assumes that a special circumstance renders overruling not just available but obligatory whenever a precedent is erroneous.
  81. .See, e.g., Knick v. Twp. of Scott, 139 S. Ct. 2162, 2190 (2019) (Kagan, J., dissenting) (“The majority repurposes all its merits arguments . . . to justify its overruling.”); Franchise Tax Bd. of Cal. v. Hyatt, 139 S. Ct. 1485, 1499 (2019) (beginning a list of stare decisis factors with “the quality of the decision’s reasoning” and “its consistency with related decisions”); Janus v. AFSCME, 138 S. Ct. 2448, 2478–79 (2018) (naming “quality of [decision’s] reasoning” and “consistency with other related decisions” as stare decisis factors). See generally Gentithes, supra note 22 (arguing that the stare decisis factors recognized in Janus weaken the force of precedent).
  82. .The appeal of being perceived to be consistent is widely recognized—and exploited. See generally Robert B. Cialdini, Influence: The Psychology of Persuasion (1984) (examining different tools of persuasion, including the desire to be and be perceived as consistent, which is characterized as a “highly potent weapon”).
  83. .Roe v. Wade, 410 U.S. 113 (1973); Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833 (1992); Miranda v. Arizona, 384 U.S. 436 (1966); Dickerson v. United States, 530 U.S. 428 (2000); Auer v. Robbins, 519 U.S. 452 (1997); Kisor v. Wilkie, 139 S. Ct. 2400 (2019).
  84. .In Casey, the surprises were Kennedy and Souter. In Dickerson, Rehnquist, Kennedy, and O’Connor. And in Wilkie, Roberts.
  85. .The relevant Justices tended to frame their analysis as reasoned discretion and even altered aspects of the precedents they upheld. See, e.g., Kisor, 139 S. Ct. at 2424 (Roberts, C.J., concurring in part) (“I agree that overruling those precedents is not warranted.”); Dickerson, 530 U.S. at 444 (“Following the rule of stare decisis, we decline to overrule Miranda ourselves.”); Casey, 505 U.S. at 845–46 (“After considering the fundamental constitutional questions resolved by Roe, principles of institutional integrity, and the rule of stare decisis, we are led to conclude this: the essential holding of Roe v. Wade should be retained and once again reaffirmed.”).
  86. .Dickerson, 530 U.S. at 445 (Scalia, J., dissenting).
  87. .See Saul Brenner & Harold J. Spaeth, Stare Indecisis: The Alteration of Precedent on the Supreme Court, 1946–1992, at 106–09 (1995); Jack Knight & Lee Epstein, The Norm of Stare Decisis, 40 Am. J. Pol. Sci. 1018, 1019 (1996); see also Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited 288–310 (2002); Harold J. Spaeth & Jeffrey A. Segal, Majority Rule or Minority Will: Adherence to Precedent on the U.S. Supreme Court 287 (1999); cf. Thomas G. Hansford & James F. Spriggs II, The Politics of Precedent on the U.S. Supreme Court (2006) (discussing a model that casts precedent as both a constraint and an opportunity).
  88. .For example, Casey is often viewed as a case in which stare decisis imposed no constraint because the ideologies of the plurality Justices were consistent with the case outcome. E.g., Segal & Spaeth, supra note 87, at 291 (“[I]t is difficult to categorize [Justice] Kennedy as an opponent of Roe . . . .”); Brenner & Spaeth, supra note 87, at 106 (“Although the major thrust of Planned Parenthood was liberal, the overruled decisions were even more so.”).
  89. .See supra notes 60 (citing Lawson), 76 (legal realists) & 87 (political scientists).
  90. .See The Federalist No. 78 (Alexander Hamilton) (“To avoid an arbitrary discretion in the courts, it is indispensable that they should be bound down by strict rules and precedents . . . .”). Both supporters and skeptics of binding precedent can harbor worries about discretion.
  91. .The fact that a ruling today binds tomorrow is often thought to have a sobering effect that is sometimes glossed as a “constraint.” See, e.g., Richard H. Fallon, Jr., Constitutional Precedent Viewed Through the Lens of Hartian Positivist Jurisprudence, 86 N.C. L. Rev. 1107, 1156 (2008) (“[T]he Supreme Court is constrained by the need to create only such precedents . . . as will have acceptable future consequences.”). But even if deemed a “constraint,” that effect would persist under the permission model given that precedent would still have substantial—even if reduced—implications for future rulings.
  92. .Cf. Daniel Kahneman & Amos Tversky, On the Reality of Cognitive Illusions, 103 Psychol. Rev. 582, 587–89 (1996).
  93. .This concern is often overstated: the Court’s case law will exhibit substantial consistency and predictability, even without any precedent at all, simply because the Court decides all cases en banc and is staffed by a fairly stable group of personnel. But see, e.g., Citizens United v. FEC, 558 U.S. 310, 398 (2010) (Stevens, J., concurring in part and dissenting in part); Rogers v. Bellei, 401 U.S. 815, 837 (1971) (Black., J., dissenting) (arguing that a constitutional principle “should not be blown around by every passing political wind that changes the composition of this Court”). Instability will arise only upon the change of a median justice, and even then, the resulting effect on the law will often be predictable.
  94. .See, e.g., Hutto v. Davis, 454 U.S. 370, 375 (1982) (per curiam) (worrying about “anarchy” absent vertical stare decisis); Thurston Motor Lines, Inc. v. Jordan K. Rand, Ltd., 460 U.S. 533, 535 (1983) (per curiam).
  95. .Cf. Richard H. Fallon Jr., Law and Legitimacy in the Supreme Court 5 (2018) (discussing affirmative action decisions).
  96. .Little Sisters of the Poor Saints Peter and Paul Home v. Pennsylvania, 140 S. Ct. 2367 (2020).
  97. .Compare id. at 2380, 2382 (bemoaning that Congress gave the agency “virtually unbridled discretion” but acknowledging that “[n]o party has pressed a constitutional challenge to the breadth of the delegation involved here”) (citing Gundy v. United States, 139 S. Ct. 2116 (2019)), with June Med. Servs. LLC v. Russo, 140 S. Ct. 2103, 2142 (2020) (Thomas, J., dissenting) (“Our abortion precedents are grievously wrong and should be overruled.”). Thomas had previously written and joined opinions criticizing both non-delegation and abortion precedents.
  98. .See, e.g., Joseph William Singer, Legal Realism Now, 76 Cal. L. Rev. 465, 472 n.19 (1988) (reviewing Laura Kalman, Legal Realism at Yale: 1927–1960 (1986)) (quoting Interview with Archibald Cox: On Law, the Constitution, and the Supreme Court, Law. Monthly 3 (Aug. 1987) (discussing tension in the law of stare decisis as identified by Judge Learned Hand)).
  99. .For less stringent “hybrid” approaches, see Part III below.
  100. .To illustrate, take views suggesting that constitutional doctrine does or should reflect forms of political mobilization. See, e.g., Barry Friedman, The Will of the People: How Public Opinion Has Influenced the Supreme Court and Shaped the Meaning of the Constitution (2009); Jeffrey Rosen, The Most Democratic Branch: How the Courts Serve America (2006); 2 Bruce Ackerman, We the People: Transformations (1998); Robert Post & Reva Siegel, Roe Rage: Democratic Constitutionalism and Backlash, 42 Harv. C.R.-C.L. L. Rev. 373 (2007). When political views inevitably shift, should stare decisis block corresponding changes in constitutional doctrine? The binding model problematically suggests that the answer is yes. To avoid this difficulty, a proponent of the binding model may be tempted to allow overruling specifically when there are changes in popular views—but that kind of exception would of course align with the merits-based reasons to disagree with the precedent in the first place. The point is generalizable: any plausible theory of constitutional law will call for ample merits-based grounds for overruling, thereby undermining the binding model. See also infra subpart III(A).
  101. .Antifederalists raised similar doubts about founding-era claims that precedent would constrain the Supreme Court. Compare The Antifederalist Papers Nos. 78–80 (Brutus) (Morton Borden ed., 1965), with The Federalist No. 78 (Alexander Hamilton).
  102. .See Rebecca Stone, Legal Design for the “Good Man, 102 Va. L. Rev. 1767, 1776 (2016) (discussing varied modes of rule internalization, some of which are enforcement-independent).
  103. .See, e.g., Michael Abramowicz & Maxwell Stearns, Defining Dicta, 57 Stan. L. Rev. 953, 957 (2005) (“Vertical stare decisis is generally considered absolute . . . .”). The absoluteness of vertical stare decisis is often overstated. See Richard M. Re, Narrowing Supreme Court Precedent from Below, 104 Geo. L.J. 921, 923 (2016).
  104. .See Richard A. Posner, How Judges Think 145 (2008) (positing that the federal courts of appeals “are strongly motivated to adhere to precedent, not only because they want to encourage adherence to the precedents they create, but also—and this is more important to most judges—because they want to limit their workloads”). Similarly, permissive precedent might operate as a shortcut in the certiorari process—a high-volume mode of decision-making where precedent is often thought to be influential. See Barrett, supra note 33, at 1929–30.
  105. .It is commonplace that judges do or should strive to avoid being reversed, for both public-spirited reasons (such as to avoid gratuitous litigation) and self-interested ones (such as to enhance one’s prestige). Some theorists—indeed, some active judges—have embraced that insight so firmly as to adopt a “prediction model” of vertical stare decisis, which holds that higher-court precedent imposes obligations only insofar as it aids in the prediction of how cases will turn out on appeal. E.g., Transcript of Oral Argument at 9–10, Hughes v. United States, 138 S. Ct. 1765 (2018) (No. 17-155) (Roberts, C.J.); Spector Motor Serv., Inc. v. Walsh, 139 F.2d 809, 823 (2d Cir. 1944) (Hand, J., dissenting), vacated sub nom. Spector Motor Serv., Inc. v. McLaughlin, 323 U.S. 101 (1944); John M. Rogers, Lower Court Application of the “Overruling Law” of Higher Courts, 1 Legal Theory 179 (1995).
  106. .McConnell similarly notes that “en banc proceedings [are] time-consuming and inconvenient” and suggests that there “is no need for a jurisprudence of stare decisis when there is a practical cost to departing from precedent.” McConnell, supra note 21, at 1770.
  107. .Three-judge district courts offer an example that might somewhat cut against the account in the main text, insofar as those courts are bound by circuit precedent, even though they are not reviewed by circuits. Such courts might then view in-circuit case law more in the way that circuit judges view out-of-circuit case law: as informative but not authoritative. Compare Joshua A. Douglas & Michael E. Solimine, Precedent, Three-Judge District Courts, and the Law of Democracy, 107 Geo. L.J. 413, 431–36 (2019) (taking this view), with Michael T. Morley, Vertical Stare Decisis and Three-Judge District Courts, 108 Geo. L.J. 699, 706 (2020) (disagreeing).
  108. .See David Klein & Neal Devins, Dicta, Schmicta: Theory Versus Practice in Lower Court Decision Making, 54 Wm. & Mary L. Rev. 2021, 2049–50 (2013).
  109. .An additional structural factor relates to the risk that disregarding someone else’s past decisions could increase the odds that other judges will disregard your past decisions. See Michael J. Gerhardt, The Power of Precedent 79 (2008) (discussing the “golden rule of precedent”—essentially, do as to others’ precedents as you would have them do unto your own); see also McConnell, supra note 21, at 1770. That risk is higher in a large court that generally hears cases in smaller panels since majority coalitions are less stable. By comparison, in a court that hears cases en banc, like the Supreme Court, a judge need only worry about coalition breakdown over time.
  110. .See generally Neil MacCormick, Legal Reasoning and Legal Theory 73–99 (1978); Rupert Cross, Precedent in English Law 4–5 (3d ed. 1977); see also Mark Greenberg, The Moral Impact Theory of Law, 123 Yale L.J. 1288, 1317 (2014) (maintaining that “fairness and democracy” may “explain the careful attention given to” past judicial opinions); Christopher J. Peters, Foolish Consistency: On Equality, Integrity, and Justice in Stare Decisis, 105 Yale L.J. 2031, 2056 (1996) (“Courts commonly intone egalitarian mantras like ‘fairness’ and ‘the principle of deciding like cases alike’ in assessing stare decisis . . . .”).
  111. .That is, the judge might believe that her legal views have priority over her moral views or that her legal views have decisive moral force.
  112. .Similar intuitions may underlie the Court’s assertion that stare decisis is “a principle of policy.” Payne v. Tennessee, 501 U.S. 808, 828 (1991) (quoting Helvering v. Hallock, 309 U.S. 106, 119 (1940)). But see Stephen E. Sachs, Precedent and the Semblance of Law, 33 Const. Comment. 417, 423 (2018) (reviewing Kozel, supra note 73) (“Courts sometimes call stare decisis ‘a principle of policy,’ but they don’t really mean it.”) (footnote omitted). Sachs is right that stare decisis must be a “separate legal rule,” id., but perhaps it is a legal rule that works by permitting consideration of certain nonlegal policies.
  113. .See Nelson, supra note 36, at 9–10 (discussing historical sources in the context of this distinction).
  114. .William Baude, Precedent and Discretion, 2019 Sup. Ct. Rev. 313, 330–32 (2020). Baude’s claim might not cut against permissions so much as dictate how to use them. That is, if judges may not take certain actions (such as ordering that someone be incarcerated) unless obligated to do so by law, then perhaps a judge must take advantage of whatever permissions, if any, would avert those actions.
  115. .Standardless discretion can arise for reasons independent of permissions, such as a simple absence of law to apply. Cf. supra note 47 and accompanying text (distinguishing “explicit,” “strong” permissions).
  116. .More broadly, legal positivists acknowledge that the law often “runs out.” See generally Hart, supra note 38.
  117. .On the prevalence and appeal of discretionary judicial reasoning, see generally David L. Shapiro, Jurisdiction and Discretion, 60 N.Y.U. L. Rev. 543 (1985).
  118. .Sup. Ct. R. 10. The Rule then lists considerations that are expressly labeled as not “controlling.” Id.
  119. .See Stephen M. Shapiro, Kenneth S. Geller, Timothy S. Bishop, Edward A. Hartnett & Dan Himmelfarb, Supreme Court Practice 329 (10th ed. 2013) (noting “the highly discretionary nature of the Court’s certiorari jurisdiction, whereby each individual Justice is free to cast a negative vote for whatever reason he or she sees fit”).
  120. .See, e.g., Teague v. Lane, 489 U.S. 288, 296 (1989) (noting that denials of certiorari have no precedential value). On reason-stopping, see supra note 54 and accompanying text.
  121. .347 U.S. 483 (1954).
  122. .See generally Neil S. Siegel, The Virtue of Judicial Statesmanship, 86 Texas L. Rev. 959, 987, 987 n.148 (2008) (discussing judicial statesmanship in connection with Brown).
  123. .See Paul J. Watford, Richard C. Chen & Marco Basile, Crafting Precedent, 131 Harv. L. Rev. 543, 574 (2017) (reviewing Bryan A. Garner, Carlos Bea, Rebecca White Berch, Neil M. Gorsuch, Harris L. Hartz, Nathan L. Hecht, Brett M. Kavanaugh, Alex Kozinski, Sandra L. Lynch, William H. Pryor Jr., Thomas M. Reavley, Jeffrey S. Sutton & Diane P. Wood, The Law of Judicial Precedent (2016)) (“[W]hen identifying the governing law, the court has discretion about how strictly or loosely it chooses to articulate the rule it is creating . . . .”).
  124. .See infra subpart III(A).
  125. .Absent such justification, one might worry not only that the deciding Justices would be embarrassed but also that the Court itself would be delegitimized. See Deborah Hellman, The Importance of Appearing Principled, 37 Ariz. L. Rev. 1107, 1125 (1995).
  126. .See, e.g., Michigan v. Bay Mills Indian Cmty., 572 U.S. 782, 798 (2014) (“[S]tare decisis is a foundation stone of the rule of law . . . .”); Burnet v. Coronado Oil & Gas Co., 285 U.S. 393, 406 (1932) (Brandeis, J., dissenting) (“[I]n most matters it is more important that the applicable rule of law be settled than that it be settled right.”).
  127. .See supra note 52.
  128. .When deviating from precedent under the permission model, the fact of a past ruling might still be relevant—not as an authority, but rather as a background fact when evaluating the merits. For example, a Justice who interprets the Constitution in light of pragmatic considerations would naturally be concerned with protecting reliance on precedent. See Alexander, supra note 25, at 7–8 (discussing the “natural” model of precedent); supra note 52 and accompanying text (noting that permissive precedent is asymmetric in nature).
  129. .This point is a staple of reason-giving requirements, including the judicial duty to issue a reasoned decision on the merits. See, e.g., Robin J. Effron, Reason Giving and Rule Making in Procedural Law, 65 Ala. L. Rev. 683, 713 (2014); Schauer, supra note 54, at 657.
  130. .Imagine a judge who decides whether to follow precedent by flipping a coin. When the coin flip dictates a deviation from precedent, the judge would lack the authoritative shield of permissive precedent and so would have to provide a merits-based rationale for her decision—something that might simply be infeasible. Further, the judge’s rationale would presumably have to be candid and so would entail disclosure of the illegitimate coin flip as the “real” basis for decision. See infra note 132 (discussing duties of candor).
  131. .See, e.g., Smith v. Allwright, 321 U.S. 649, 665 (1944) (“However, when convinced of former error, this Court has never felt constrained to follow precedent.”); Payne v. Tennessee, 501 U.S. 808, 828 (1991) (similar). Perhaps, in this instance, we should take the Justices at their word.
  132. .A rhetoric of “formally” mandatory precedent is sincere and intelligible but imperfect in terms of how much it discloses. That approach seems legitimate even if it falls short of certain judicial ideals. See Richard H. Fallon, Jr., A Theory of Judicial Candor, 117 Colum. L. Rev. 2265, 2308 (2017); Micah Schwartzman, Judicial Sincerity, 94 Va. L. Rev. 987, 1000 (2008).
  133. .See supra note 1; cf. supra note 85 (noting the Court’s frequent acknowledgment of discretionary judgment in connection with stare decisis).
  134. .For instance, precedent (even if binding) often is and should be narrowed with an eye to the merits. See Re, supra note 103, at 971.
  135. .See infra note 156 and accompanying text; cf. Seana Valentine Shiffrin, Inducing Moral Deliberation: On the Occasional Virtues of Fog, 123 Harv. L. Rev. 1214, 1222–25 (2010) (discussing the benefits of “induced moral deliberation” occasioned by ostensibly hazy standards).
  136. .Knick v. Twp. of Scott, 139 S. Ct. 2162, 2190 (2019) (Kagan, J., dissenting); Janus v. AFSCME, 138 S. Ct. 2448, 2497 (2018) (Kagan, J., dissenting). Interestingly, the provenance of the “special justification” test suggests that it might have originally been intended to require only an explanation—not necessarily a justification that is merits-neutral. See Arizona v. Rumsey, 467 U.S. 203, 212 (1984) (“Although adherence to precedent is not rigidly required in constitutional cases, any departure from the doctrine of stare decisis demands special justification.”). The idea of merits-neutral special justifications crystallized in Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 864 (1992). The special-justification rule thus illustrates how the Court treats stare decisis as subject to modification.
  137. .See Kozel, supra note 73, at 107, 118.
  138. .See Hubbard v. United States, 514 U.S. 695, 716 (1995) (Scalia, J., concurring in part and concurring in the judgment) (maintaining that overruling requires “reasons that go beyond mere demonstration that the overruled opinion was wrong (otherwise the doctrine would be no doctrine at all)”); Randy J. Kozel, Precedent and Constitutional Structure, 112 Nw. U. L. Rev. 789, 823 (2018) (“A Justice’s conclusion that she would have decided a case differently is not enough to warrant an overruling. If it were, any constraining effect of precedent would vanish.”).
  139. .Kozel insightfully seeks out a “second-best” approach to stare decisis, given the Justices’ deep disagreements. Kozel, supra note 73, at 107. But Kozel’s proposal still clings to a kind of first-best optimism in assuming that Justices espousing different first principles will converge on a set of merits-neutral stare decisis criteria, such as reliance and unworkability. That optimism is unwarranted. Justices with different views of legal rights and interests both will and should assess considerations like reliance differently from one another.
  140. .In other words, a sufficiently determinate, discoverable, and valued merits principle can make up for a shortage of merits-neutral reasons for overruling. Cf. supra note 134 (discussing the role of merits in narrowing precedent).
  141. .Hon. Robert H. Jackson, Decisional Law and Stare Decisis, 30 A.B.A. J. 334 (1944), cited in Citizens United v. FEC, 558 U.S. 310, 378 (2010) (Roberts, C.J., concurring) and Ramos v. Louisiana, 140 S. Ct. 1390, 1415 (2020) (Kavanaugh, J., concurring in part). Both Chief Justice Roberts in Citizens United and Justice Kavanaugh in Ramos exhibited merits-sensitive overruling by respectively emphasizing the special value of free speech and racial equality.
  142. .Jackson, supra note 141, at 334.
  143. .See supra note 81 and accompanying text.
  144. .163 U.S. 537 (1896).
  145. .Kozel would approve Plessy’s overruling based on an exception for “profoundly immoral” rulings. Kozel, supra note 73, at 122. But that approach seems to ascribe a cognizable reliance interest to segregation. At any rate, the exception Kozel envisions is of course itself merits-sensitive—and more malleable than one might think. See Corinna Barrett Lain, Mostly Settled, But Right for Now, 33 Const. Comment. 355, 367–68 (2018) (reviewing Kozel, supra note 73).
  146. .Justice Gorsuch has recently voiced a similar concern about considering governmental reliance on convictions obtained through constitutional violations. See Transcript of Oral Argument at 73, Gamble v. United States, 139 S. Ct. 1960 (2019) (No. 17-646) (“[Y]ou haven’t suggested, I don’t think, that—that a prosecutor has a right to rely on an unconstitutional rule to put someone in prison. I mean, that wouldn’t be a thing, would it?”).
  147. .558 U.S. 310 (2010).
  148. .Compare id. at 365 (2010) (dismissing governmental reliance on decisions allowing campaign-finance regulation), with id. at 411–12 (Stevens, J., concurring in part and dissenting in part) (not).
  149. .Payne v. Tennessee, 501 U.S. 808, 828 (1991); see Garcia v. San Antonio Metro. Transit Auth., 469 U.S. 528, 557 (1985) (“We have not hesitated [to overrule] when it has become apparent that a prior decision has departed from a proper understanding of congressional power under the Commerce Clause.”); see also FEC v. Wisconsin Right to Life, Inc., 551 U.S. 449, 500 (2007) (Scalia, J., concurring in part and concurring in the judgment) (“This Court has not hesitated to overrule decisions offensive to the First Amendment . . . .”).
  150. .Many procedural and remedial principles, such as preclusion, immunity, and retroactivity doctrines, can likewise be viewed as ways of honoring reliance interests while enabling merits-sensitive overruling. See generally Toby J. Heytens, The Framework(s) of Legal Change, 97 Cornell L. Rev. 595, 599 (2012); John C. Jeffries, Jr., The Right-Remedy Gap in Constitutional Law, 109 Yale L.J. 87, 113 (1999).
  151. .Richard M. Re, The Doctrine of One Last Chance, 17 Green Bag 2d 173, 174 (2014).
  152. .Imagine a rule allowing only a certain number of overrulings over a particular period of time. See generally Fallon, supra note 91.
  153. .See, e.g., Ian Millhiser, Brett Kavanaugh Thinks That Sen. Susan Collins Is Stupid, ThinkProgress (Aug. 21, 2018, 2:15 PM), https://‌‌/brett-kavanaugh-susan-collins-roe-0d34a78c453d‌/ [https://‌‌/998E-RDH5] (criticizing Justice Kavanaugh and Chief Justice Roberts for trumpeting “several vague principles that should guide judges in deciding whether to overrule precedent”).
  154. .See, e.g., Ramos v. Louisiana, 140 S. Ct. 1390, 1408 (2020) (Sotomayor, J., concurring in part) (“[O]verruling precedent here is not only warranted, but compelled.”); supra subpart I(B) (discussing Justice Thomas’s views).
  155. .See supra text accompanying notes 19, 81.
  156. .See Hellman, supra note 64, at 70–73 (arguing that giving precedent weight, or requiring adherence to precedent unless a special reason can be adduced, is one way—“not the only way”—to prompt overconfident judges to consider opposing views and check their own hubris); see also Randy J. Kozel, Stare Decisis as Authority and Aspiration, 96 Notre Dame L. Rev. (forthcoming 2021) (manuscript) (describing precedent as a “double check”). This line of thought helps explain why stare decisis both does and should direct judicial attention toward a number of open-ended factors; the practice of ticking through various factors is a nudge, encouraging judges to address a range of considerations, without dictating whether to overrule.
  157. .See supra section II(C)(2).
  158. .See supra subpart I(B). Justice Thomas’s “demonstrably erroneous” test is, of course, merits-sensitive.
  159. .See supra notes 129, 156, and accompanying text.
  160. .Cf. Audrey Lynn, Note, Let’s (Not) Make This Work! Why Stare Decisis Workability Should Be a Sword but Not a Shield, 31 Regent U. L. Rev. 91, 108 (2018).
  161. .Kimble v. Marvel Entm’t, LLC, 576 U.S. 446, 457–58 (2015), cited in Knick v. Twp. of Scott, 139 S. Ct. 2162, 2190 (2019) (Kagan, J., dissenting).
  162. .E.g., Kimble, 576 U.S. at 456.
  163. .For a similar schematization of the U.K. rules of statutory interpretation, including (in essence) both mandates and permissions, see Rupert Cross, Statutory Interpretation 42–43 (1976).
  164. .For a critical history and assessment of the use of such step-based decision procedures, see Mitchell Chervu Johnston, Stepification, 116 Nw. U. L. Rev. (forthcoming 2021), https://‌ssrn‌.com‌/abstract‌=3542568 [https://‌perma‌.cc‌/DT2P-AY57].
  165. .Unworkability can be merits-sensitive if, for instance, the merits place a premium on notice and predictability. The schema also admits of additional discretion-conferring nuances; for instance, Rule 3 might not apply to judicial interpretations of “common law statutes.” See generally Margaret H. Lemos, Interpretive Methodology and Delegations to Courts: Are “Common Law Statutes” Different?, in Intellectual Property and the Common Law 89 (Shyamkrishna Balganesh ed., 2013) (critically discussing the idea of “common law statutes”).
  166. .See supra subpart III(A).
  167. .As we have seen, precedent can operate as a shield if and only if judges are able to avoid engagement with the merits. See supra text accompanying note 79.
  168. .See, e.g., Kozel, supra note 73, at 107, 122; Janus v. AFSCME, 138 S. Ct. 2448, 2497 (2018) (Kagan, J., dissenting).
  169. .See Re, supra note 41, at 1509–10 (distinguishing first- and third-person clarity assessments).
  170. .For an empirical showing that Justices exhibit “personal stare decisis,” which includes voting to overrule a case from which the Justice dissented, see Brenner & Spaeth, supra note 87, at 77–88; see also Nina Varsava, Precedent and Conflicts of Interpretation 4 (Univ. of Wisconsin Legal Studies Research Paper, No. 1495, 2020), https://‌ssrn‌.com‌/abstract=3238221 [https://‌perma‌.cc‌/L849-K5KD] (“[M]ethods of precedential interpretation . . . can and do differ across jurisdictions and courts.”).
  171. .See, e.g., Craig S. Lerner & Nelson Lund, Judicial Duty and the Supreme Court’s Cult of Celebrity, 78 Geo. Wash. L. Rev. 1255, 1278 (2010) (“Justices write separately in opinion after opinion, striving to preserve consistency with their own stock of personal precedents.”).
  172. .See Allison Orr Larsen, Perpetual Dissents, 15 Geo. Mason L. Rev. 447, 449–50 (2008); see also Frederick Schauer, Has Precedent Ever Really Mattered in the Supreme Court?, 24 Ga. St. U. L. Rev. 381, 395–97 (2007) (citing persistent dissents as evidence that the stare decisis norm hardly exists).
  173. .E.g., Alleyne v. United States, 570 U.S. 99, 122 (2013) (Breyer, J., concurring in part and concurring in the judgment); United States v. Navajo Nation, 556 U.S. 287, 302 (2009) (Souter, J., concurring).
  174. .Judges who often face this situation might be labeled “formalists,” but a functionalist (or a moralist, or most anyone else) will sometimes feel the same way. See, e.g., Richard Primus, Debate, The Cost of the Text, 102 Cornell L. Rev. 1651, 1655–56, 1656 n.20 (2017) (“Given the statute overall, giving legal force to the plain meaning of section 1311 [in King v. Burwell] would have been unreasonable.”). Thus, Justices may take turns perceiving either fixity or fluidity in underlying law.
  175. .See supra subpart I(B) (discussing Justice Thomas’s views).
  176. .See supra text accompanying notes 36, 90.
  177. .See supra subpart II(B) (discussing permission as shortcut or shield).