I. A Strange Idea
Courts in common law systems decide cases as they decided similar cases in the past—even if they believe they decided those past cases wrongly. This, roughly, is the practice of precedent. In Whole Woman's Health v. Hellerstedt,Footnote 1 for example, the Supreme Court of the United States struck down as unconstitutional a law that limited access to abortion. Four years later, in June Medical Services v. Russo,Footnote 2 the Court heard a challenge to a nearly identical law in a different state. The Court followed precedent and struck down that law, even though most of its members believed that the Court had made a mistake in Whole Woman's Health. That is, the Court held as unconstitutional a law that it believed should have been held constitutional.
Legal philosophers often regard the practice of precedent as a ”strange idea.”Footnote 3 Courts are in charge of making decisions on legal issues because—one hopes—they are good at it. Their judgment on legal issues is generally sound; at the least, it is sounder than the judgment of alternative decision-makers. Why, then, would courts “do something other than make decisions according to their own best legal judgment?”Footnote 4
There is an easy answer with respect to vertical precedent, that is, the practice of lower courts following the precedents of higher courts. Lower courts are generally less expert in the law than higher courts. If they disagree, probably the higher court is right, and the lower court is wrong. So, lower courts will make correct decisions more often by following the decisions of higher courts than they would by relying on their own judgment.Footnote 5 There is nothing strange here, just the unremarkable fact of differential expertise.Footnote 6
Things are trickier with respect to horizontal precedent, that is, the practice of higher courts (such as the Supreme Court) following their own precedents. Courts do not consistently decline in expertise over time. There is no reason to think that the Court in June, for example, is worse at working out what the law requires than the Court in Whole Woman's Health. As a result, there's no reason to think that a court will make correct decisions more often by following its own precedents.Footnote 7 Some of the strangeness therefore remains.
Here I aim to dispel that remaining strangeness. My question, roughly, is: Should a court follow its own precedents, even when it believes they were wrongly decided? If so, why? I shall provide a more precise formulation of my question at the end of Section II. An argument for a positive answer I shall call an “argument for the practice of precedent” or simply “an argument for precedent.”
Here is how I proceed. Section II describes the practice in more detail. Section III distinguishes two strategies for answering our question and clarifies which one I shall pursue. Sections IV and V are the analytical heart of the article. I defend two main claims. The first is that fairness requires that like cases are treated alike if that means treating them correctly. The second is that a court is, in general, as likely to make a mistake in a current case as it was in a previous case. Together, these claims support the conclusion that it is often better for courts to stand by their previous decisions, even when they believe those decisions are mistaken, than to rely on their current judgment. Sections VI and VII respond to objections. Section VIII explains how overruling fits within my account, while Section IX charts a path from the justification for following precedent to a justification for a doctrine of precedent. Section X concludes.
I should be clear that, while my examples are drawn from US law, that is for simplicity's sake only. Nothing in my argument hinges on the details of Supreme Court practice or federal courts doctrine, and my conclusions are applicable to any common law system.Footnote 8
II. Following and Distinguishing
I start with a more precise description of the practice of precedent. There are several competing accounts. For convenience I proceed based on the influential account proposed by Lamond, as developed by Horty.Footnote 9
In any case, there are two possible outcomes: a decision for the plaintiff or a decision for the defendant. A case includes various factors, which are reasons that favor one or the other of these decisions. A court hears the case and reaches a decision on the balance of these reasons. The ratio decidendi of the case is the court's understanding of which reasons sufficed for its decision. We call the case the “precedent” or “precedent case” and the court the “precedent court.”
The same court may later hear a case that falls within the precedent's ratio. Call the later case the “instant case” and the court the “instant court.” If the court can “distinguish” the precedent, then it will decide the instant case differently than it decided the precedent case. To distinguish a precedent is to identify a reason, present in the instant case but absent in the precedent, that tips the balance of reasons in favor of a different decision on the assumption that the precedent was correctly decided. If the court cannot distinguish the precedent, then (subject to one qualification) it will “follow” precedent, that is, decide the instant case as it decided the precedent case.
The abortion cases illustrate these points. The Texas law in Whole Woman's Health required doctors to have admitting privileges at a hospital within thirty miles of where they performed abortions. The issue was whether this restriction imposed an undue burden on access to abortion. Simplifying greatly, suppose that in Whole Woman's Health the Court thinks that these are the relevant factors:
r 1 = the law led to the closure of half of the state's abortion clinics
r 2 = the law had few if any health benefits
r 3 = the legislature thought that the law was overall beneficial
Whereas r 1 and r 2 favor a decision for the plaintiff, r 3 favors a decision for the defendant state. The Court judges—let us suppose—that r 1 and r 2 outweigh r 3. As a result, it decides for the plaintiff. The ratio of the case is that: if r 1 and r 2, then the plaintiff prevails.
In June, the issue was the constitutionality of a Louisiana law that, like the Texas law in Whole Woman's Health, required doctors to hold admitting privileges at a hospital near to where they performed abortions. There were, let's stipulate, the same set of factors as in Whole Woman's Health: r 1, r 2, and r 3. Since r 1 and r 2 are present, the case fell within the ratio of Whole Woman's Health. Since there were no factors present in June but absent in Whole Woman's Health, the cases were indistinguishable. Accordingly, the Court followed precedent and decided for the plaintiff.
Suppose that a third case now arises. In addition to r 1, r 2, and r 3, there is:
r 4 = the law protects patients by credentialing doctors
This factor favors the defendant. It was absent in Whole Woman's Health (and June). The Court judges that r 4 would tip the balance of reasons in favor of the defendant, even on the assumption that r 1 and r 2 outweigh r 3. The Court therefore distinguishes Whole Woman's Health and decides for the defendant.
I said that there was a qualification; here it is. Instead of following or distinguishing a precedent, a court may “overrule” a precedent, where an overruled case does not constrain any court's decision-making. Courts are not, however, prepared to overrule one of their own precedents just because they believe it was wrongly decided (that is, decided contrary to the balance of reasons). The Court in June, for example, did not overrule Whole Woman's Health, even though it believed that case was wrongly decided. More than mere perceived error is needed for overrule. Exactly what more is needed I leave for Section IX.
With these clarifications in place, I can state my question more precisely: Should a court follow its own decisions in previous and indistinguishable cases, even if it believes those decisions were incorrect? If so, why?
III. Doctrine and Act
We can always ask two kinds of questions in respect of a legal rule or doctrine. We can ask: Should there be a rule requiring x? And should you x? With respect to precedent, there's this question: Should there be a legal rule requiring courts to follow their own precedents? And then there's our question, which is about whether courts should follow their own precedents. The first question concerns what the law should be, the second what courts should do. The first is about doctrine, the second about acts falling under a doctrine.
To answer our question, we might try to first answer the doctrine question and then to work out its implications for our question. This is an indirect strategy for answering our question. For example, we might reason that it's better for judicial decisions to be more predictable than less, other things being equal. Judicial decisions will be more predictable if the law requires courts to follow precedent than if not (assuming judges obey the law). There is therefore a reason for the law to require courts to follow precedent. That is an answer to the doctrine question. On the assumption that judges have a genuine reason to obey the law, it's then a short step to the conclusion that courts have a reason to follow precedent, and hence a short step to an answer to our question.Footnote 10
My strategy is different. I shall argue in favor of following precedent without first arguing in favor of a doctrine of precedent. I shall pursue a direct strategy, in other words. I pursue this strategy partly because I am skeptical that judges have a genuine reason to obey the law.Footnote 11 I am skeptical, as a result, that the “short step” from a reason for a doctrine to a reason for acts under a doctrine is a legitimate step. To be clear, the two strategies are not rivals; the practice of precedent could be justified both directly and indirectly. Even if one of these strategies is successful, it is still interesting and important to know whether the other succeeds as well. I therefore do not presuppose the incorrectness of any indirect argument, including the argument from predictability in the last paragraph.
Finally, I should add that it may also be possible to reason in the other direction, from a reason to follow precedent to a reason to require courts to follow precedent. I explore this possibility in Section IX.
IV. The Fairness Principle
My argument for precedent starts from the principle that like cases should be treated alike. This principle is variously attributed to justice, equality, and fairness. For simplicity I'll call it the fairness principle. I am, of course, far from the first person to think that fairness can help us to justify precedent.Footnote 12 Indeed, fairness is probably the simplest and most intuitive basis for precedent. Ask a layperson why a current case should be decided as a past case was decided, and fairness is likely to figure in the answer. Nowadays, though, a number of philosophers doubt that fairness can justify precedent. One of my aims is to explain why. My other aim is to show that there is, in fact, a good argument from fairness for precedent.Footnote 13
A. Strong
There are different ways to understand the fairness principle. Here's one possibility:
Strong. For any two like cases c 1 and c 2, if you treat c 1 in some way, then it is fair for you to treat c 2 alike and not fair for you to treat c 2 differently.
Were the strong fairness principle sound, it would be easy to defend the practice of precedent. Indistinguishable cases are, of course, alike.Footnote 14 The strong principle therefore demands that a court follows its decision in an indistinguishable case. And since the strong principle draws no distinction between correct and incorrect decisions, that principle demands that a court follow even a decision that it believes was incorrect. Fairness would require the Supreme Court in June to follow Whole Woman's Health, for example. But fairness would not require the Court to follow Whole Woman's Health in our imagined third case, because that case can be distinguished from Whole Woman's Health based on r 4. So, this principle could explain the sense in which precedents constrain courts, and the sense in which they leave them free.
But the strong fairness principle is not sound. Suppose that in c 1 an innocent defendant is executed. If the strong fairness principle is correct, then it is fair to execute the equally innocent defendant in c 2 and not fair to do otherwise. But this is absurd: plainly, executing an innocent defendant is not fair. In general, deciding a case incorrectly does not, in itself, make it fair to also decide a like case incorrectly. As Lamond says, “that a mistake was made in the earlier case is not—in itself—an argument for repeating the mistake in the later case.”Footnote 15 This point has been made many times elsewhere in the literature.Footnote 16
If you are attracted to the strong principle, you might reply this way. True, that a court decides a case in some way does not mean that it should all things considered decide a like case alike. But that is perfectly compatible with the strong principle. All that principle says is that it's fair to decide like cases alike. Fairness is a pro tanto reason. Pro tanto reasons can be outweighed. Moreover, it's plausible that fairness is outweighed, when it favors following an incorrect decision, by whatever reasons establish that the decision is incorrect. So, you would conclude, the strong fairness principle can accommodate our intuition that like cases should not be decided alike if that means deciding them incorrectly.
There are two difficulties with this response. First, it is not plausible that deciding a case incorrectly provides even a pro tanto reason to treat a like case incorrectly. The execution of an innocent defendant in one case does not provide even a pro tanto reason to execute another innocent defendant, similarly situated. Or, to take an even more “dramatic example” from Larry Alexander, “past genocide does not generate an equality reason – not even a weak one – for continuing genocide in the present.”Footnote 17 As Alexander says, “[t]he two millionth wrong [is not] somehow less wrong than the first,” no matter how additional wrongs “equalize situations.”Footnote 18
Second, there could be cases in which the reasons for the plaintiff and defendant are closely, but not exactly, balanced. Imagine—again, purely for the sake of argument—that r 1 and r 2 are slightly outweighed by r 3, meaning that the Court decided Whole Woman's Health wrongly. If the difference is sufficiently slight, then fairness would tip the balance in favor of upholding the statute in June. It would tip the balance in favor of repeating what is, we're assuming, the incorrect decision. But we said that fairness does not justify repeating an incorrect decision. So, the original objection stands even if fairness is merely a pro tanto reason.
Another way to try to defend the strong fairness principle is to narrow its scope. Compare two scenarios. In the first, a person was treated worse than they should have been treated (relative to non-fairness-related reasons), and the question now is how to treat a similarly situated person. In other cases, a person was treated better than they should have been treated (relative to non-fairness-related reasons), and the question now is how to treat a similarly situated person. The counterexamples to the strong principle discussed above all involve the first type of scenario. If we narrow the principle's scope to the second type of scenario, we can sidestep these concerns. The narrowed principle would require, roughly, that a person in c 2 be treated like a person in c 1, if the person in c 1 was treated better than they should have been. To see the difference this makes, recall that the original strong principle says that an innocent defendant should be sentenced to death if a like defendant received a like sentence. A truncated principle avoids that absurd result because obviously a death sentence is worse treatment than the innocent defendant deserved.Footnote 19
Whatever its other virtues, a truncated strong principle will not justify a practice of precedent. That is because many legal cases, and most civil cases, are zero-sum. “Better treatment” for one party is “worse treatment” for the other party. Consider a simple personal injury case. The court sides with the defendant. On the plausible assumption that there was a uniquely right answer in the case, one party was treated better than they deserved, while the other was treated worse. What does the truncated strong principle say should be done in a like case? I think the answer is that it does not say anything at all, since this is as much a “worse treatment” case as it is a “better treatment” case. The principle does not apply. The principle has nothing to say about precedent-following in ordinary civil cases. If, on the other hand, the narrowed principle does apply in these cases, then it is subject to counterexamples. Suppose that the injuries suffered by the plaintiffs were horrific, and that they were caused by the defendants’ gross negligence. It is not plausible that the defendant in the next case should avoid liability, merely because of the mistake in the earlier case.
We should therefore reject the strong principle as a basis for the practice of precedent. In its original formulation, the principle is unsound. In its revised formulation, it may be sound, but it does not justify precedent. That does not mean we need to abandon the idea that fairness can help to justify precedent, however. There is another, better way to interpret the fairness principle.
B. Weak
Suppose that you and I commit the same offense, the uniquely correct sentence for which is four months. I receive the correct sentence. You receive an unduly lenient three-month sentence. These sentences are not fair.Footnote 20 It is not fair that I receive the correct sentence and you receive a lighter sentence when there is nothing to distinguish us. We need a principle that can vindicate this kind of thought.
Consider, then:
Weak. For any two like cases c 1 and c 2, if you treat c 1 correctly, then it is fair for you to treat c 2 alike and not fair for you to treat c 2 differently.
Because the antecedent is not satisfied unless you decide c 1 correctly, the weak principle does not favor the repetition of mistakes. That avoids the problem with the strong principle. Because the antecedent is satisfied when you decide c 1 correctly, the weak principle explains why it is not fair to decide one case correctly and a like case incorrectly. It explains why, for example, fairness demands that you receive a four-month sentence if I do. Likewise, if a defendant was sentenced to death, then on the assumption that the sentence was correct, fairness would demand that a similarly situated defendant receive the same sentence.
I think that the weak principle is true. I would, of course, be delighted if you agreed. But if you weren't already inclined toward that principle, I suspect you may not be entirely convinced of its merits. Examples are all well and good, you might say, but they are not enough of an argument on their own. That is fine for my purposes; you can still accept the rest of my argument. All I need is some positive probability, however small, that the weak principle is true. I return to this point in Section VII.
You might also worry that the weak fairness principle is superfluous or redundant. How does it do any work, you might ask, if it only favors making a decision that is otherwise correct? Doesn't the principle simply tell courts to do what they should do anyway? It's a fair question. I do have an answer, which is roughly that the principle is not redundant given sufficient uncertainty as to which decision is correct. For the exact answer I'll have to ask you to wait until Section VI.
C. Boxed In?
It may seem that I have boxed myself into a corner by endorsing the weak fairness principle. The strong principle would have justified the practice of precedent—were it sound, which it isn't. The weak principle might be sound, but—so it seems—it doesn't justify the practice. After all, the weak principle only counts in favor of following a previous correctly decided case. But isn't the practice to follow even incorrectly decided cases? If so, the weak principle falls short of justifying the practice.
There is, however, a subtle slip.Footnote 21 One question we could ask is: Should a court follow its decision in an earlier and indistinguishable case, even if that decision was in fact wrong? However, this is not our question. Our question is about what a court should do when it believes that it decided a past case wrongly. The first question concerns actual incorrectness. The second question, our question, concerns apparent incorrectness. The two notions align if a court's beliefs are true. But they can diverge because a court's beliefs—like any agent's beliefs—can be false.
The weak fairness principle does not apply if a precedent was in fact incorrectly decided. As a result, that principle does not yield a positive answer to the first question, about actual incorrectness. There is no room for argument on that point. But that is not the point that matters because—to repeat—our question is not framed objectively.
What matters is whether the weak fairness principle applies if a court believes that a precedent was wrongly decided. Maybe it does apply. Then again, maybe it doesn't. We need more information to say one way or the other. Specifically, we need to know whether the court's belief is accurate, and the precedent really is wrongly decided. If the court's belief is accurate, then that principle does not apply; but if the court's belief is inaccurate, and the precedent was correctly decided, then it does apply.
The lesson is that the weak fairness principle is a poor answer to a question we did not ask. Whether it is a good answer to the question we did ask is, at this point in the discussion, an open matter. I imagine you may now wonder whether we asked the wrong question, back in Section I. Might the important thing be whether courts should follow precedents that are truly, not just in the court's eyes, incorrectly decided?
No. For one thing, we are interested in the justification for an existing legal practice. That practice is typically described by its participants in terms of apparent incorrectness. In Planned Parenthood of Southeastern Pennsylvania v. Casey,Footnote 22 for example, the Supreme Court wrote that a decision to depart from precedent should rest “on some special reason over and above the belief that a prior case was wrongly decided.”Footnote 23 For another, this understanding is clearly assumed in most of the literature. To see that, think back to why lower courts should follow the precedents of higher courts. The reason we gave was that lower courts are generally less expert in the law than higher courts. This is an excellent reason for lower courts to follow precedents that they believe to be incorrectly decided. But it is no reason at all for lower courts to follow precedents that are in fact incorrectly decided. Once we accept the standard account of why lower courts should follow higher courts’ precedents, we're committed to working with apparent incorrectness.
So, the weak fairness principle is a poor answer to a question we did not ask—and were right not to ask. It remains to be seen whether it is a good answer to the question we were right to ask.
V. Fairness and Probability
I now want to answer our question. I'll do that by showing how we can proceed from the weak fairness principle to a justification for courts following precedent.
A. Asymmetry
Assume that there are two indistinguishable cases. Assume that the courts in these cases disagree as to the correct decision. A court is “correct” if the decision it favors is correct, and “incorrect” otherwise.
If the weak principle is correct, then there is an asymmetry between:
• the instant court departing from precedent when it (the instant court) is correct, and
• the instant court following precedent when the precedent court is correct.
If the instant court departs from precedent, when it is correct, then of course it decides the instant case correctly. It does not, however, decide the case fairly.Footnote 24 By contrast, if a court follows precedent, when the precedent court is correct, then it decides the instant case correctly and fairly. It realizes not one reason but two—the reason to decide a case correctly, and the reason to decide a case fairly.
If you have a reason to do something, e, the fact that something else, m, is a means of e-ing may be a reason to m. Principles that specify this relation are principles of instrumental reason. Here I follow Kolodny and assume this principle: if you have a reason to e, and there is a positive probability, conditional on m, that your m-ing helps to bring it about that you e, then that is a reason to m.Footnote 25 (The principle that Kolodny ultimately endorses is more elaborate, but the elaborations aren't important here.) The strength of the reason to m is equal to the importance of e-ing multiplied by the probability.
Let Pr i be the probability that the instant court is correct. Let Pr p be the probability that the precedent court is correct. (Pr i and Pr p need not sum to 1 since there could be multiple correct or multiple incorrect decisions in the two cases.) Let C be the importance of a correct decision and F be the importance of a fair decision.
If Pr i > 0, then departing from precedent is a means of making a correct decision, that is, it probabilizes making the correct decision. The reason to depart from precedent has strength equal to Pr i(C). If Pr p > 0, then following precedent is a means of making a correct and a fair decision. The reason to follow precedent therefore has strength equal to Pr p(C + F). An agent should do what there is most reason to do. So, assuming correctness and fairness are the only relevant reasons, courts should follow precedent, and should not depart from precedent, if and only if:
When is this inequality satisfied? A full answer would require me to work out the relative importance of fairness and correctness. I don't know how to do that. But I can provide a partial answer.
B. Peerhood
Let us say that agents who are equally likely to be correct on some matter are epistemic peers with respect to that matter.Footnote 26 In general, agents are epistemic peers if they have
• the same ability to evaluate evidence on a matter, and
• the same evidence on the matter.
Agents who have the same ability to evaluate evidence on a matter are evaluative peers. Agents who have the same evidence on the matter are evidential peers.Footnote 27 In rare cases, agents are epistemic peers on some matter despite divergent abilities or evidence. Nonetheless, being evaluative and evidential peers is a good predictor of being epistemic peers.
Is a precedent court the evaluative peer of an instant court? A court's capacity to evaluate evidence does not, in general, quickly improve. The number of judges, their skill, their education levels, the time they have to deliberate, the quality of the lawyers appearing before them—all stay about the same, year after year. Matters may be different over decades; over centuries, they almost certainly improve. But over relatively short periods, it is plausible that precedent and instant courts are evaluative peers.
The Supreme Court in the abortion cases may be an example. The panel in Whole Woman's Health had eight judges. The panel in June had nine. There were about the same opportunities to share experience and insight, as a result. Of the eight judges who heard Whole Woman's Health, seven also heard June. The judges who joined the court in between the two cases had similar educational and professional backgrounds as the judges who left. The Court in Whole Woman's Health deliberated for seventeen weeks—just as the Court did in June. It is reasonable to think that the two courts are evaluative peers.
Is a precedent court the evidential peer of an instant court? An instant and precedent court will almost never have the same evidence. A lot of the evidence in a case is about the specific parties and dispute in that case. The evidence in Whole Woman's Health was partly about a Texas law, whereas the evidence in June was partly about a Louisiana law. The question that matters for peerhood, though, is whether there is additional evidence in the instant case that suggests that a different decision would be correct than in the precedent case. The answer is that it depends on the instant case. Sometimes, the two courts will be evidential peers. In June, for instance, not only was the law at issue nearly identical to the one in Whole Woman's Health; the facts “mirror[ed] those . . . in Whole Woman's Health in every relevant respect.”Footnote 28 Other times, the instant court will be the evidential superior of the precedent court. I doubt that we can say anything more general than that.
In summary: an instant court and a precedent court are ordinarily epistemic peers—over the short or medium term, and absent new evidence. I return to these qualifications in Section VIII when I discuss overruling. For the time being, I assume that an instant and precedent court are peers.
C. The Tiebreaker Argument
If the precedent and instant courts are epistemic peers with respect to the correct decision in indistinguishable cases, then they are equally likely to be correct when they disagree. That is: Pr i = Pr p. Equation 1 simplifies to:
There is most reason to follow precedent if and only if F is of positive value, which of course it is.
The question was: Should a court follow its decision in an earlier and indistinguishable case, even if it believes that decision was wrong? Answer: yes, if the courts in the two cases are peers. The explanation is that peerhood takes correctness out of the equation, leaving only fairness. In other words, fairness breaks a tie between two decisions that are correct with equal probability. Call this the tiebreaker argument for precedent.Footnote 29
I said that it is reasonable to think that the Court in Whole Woman's Health is the evaluative and evidential, and therefore epistemic, peer of the Court in June. If that's right, then the Court is equally likely to make the correct decision in the two cases. Since it is more likely to make the fair decision by following its previous decision than by acting on its current judgment, it should follow its previous decision. It should therefore strike down the Louisiana abortion restrictions—even though it thinks that the decision to strike down the identical Texas restrictions was a mistake.Footnote 30 In essence, the Court is faced with a tie, which fairness breaks in favor of following precedent.Footnote 31
I want to stress the generality of the tiebreaker argument. A no-doubt common scenario is one where the instant and precedent courts differ, one is right, and the other is wrong. In this scenario, both Pr i = (1 − Pr p) and Pr p = (1 − Pr i). I suspect—though I do not wish to put it any more strongly—that this scenario describes Whole Woman's Health and June. The abortion restrictions were either constitutional or unconstitutional, but not both. The precedent court thought they were unconstitutional, while the instant court thought they were constitutional. One was right, the other wrong.
If this is right, and if the courts in the two cases are peers, then Table 1 illustrates the choice in June:
One correct decision
The two courts were equally likely to be correct and incorrect. Given a choice of row, we're equally likely to end up in either column. That makes following precedent the best choice for the court in June.
There are, however, other scenarios. In one, there are multiple correct decisions, such that while the instant and precedent courts disagree, they could both be correct. Indeed, it might be certain that they are both correct. In that event both Pr i = 1 and Pr p = 1. Table 2 illustrates:
Multiple correct decisions
In this scenario, whether the court follows or departs from precedent, it decides correctly. But, if it follows precedent, it also decides fairly. Again, fairness breaks the tie in favor of following precedent.
I mention this second scenario partly because other scholars have noted that the weak fairness principle, or something like it, can break ties when the precedent and instant court favor different correct decisions.Footnote 32 This is true, as far as it goes. But it misses the larger picture, which is that fairness breaks ties whenever the precedent and instant courts are equally likely to favor a correct decision. It follows that fairness breaks a tie when it is certain that the decisions favored by the precedent and instant courts are each correct, but that is just a special case of the much more general claim I have argued for.
I also need to emphasize the limits of the tiebreaker argument. I have identified a sufficient condition for following precedent despite disagreement. I do not present this as a necessary condition. And, indeed, my analysis suggests that it is not necessary.
If the instant court is more likely to reach the correct decision than the precedent court, then there is no tie to break, and the tiebreaker argument does not apply. Nonetheless, it may be better to follow precedent than to depart from it. If there is some chance the precedent court was correct, there is some chance that by following precedent the instant court will act both correctly and fairly. Even a slim chance of realizing two goods can be more attractive than a substantial chance of realizing only one of those goods. To illustrate, suppose that the instant court is somewhat more likely to be correct than the precedent court, such that Pr i = 0.6 and Pr p = 0.4. It will still be better to follow precedent, provided that deciding a case fairly is more than half as good as deciding it correctly.Footnote 33 Whether or not this assumption holds true, the fact remains: it may make sense to follow precedent, even though the instant court is more likely correct than the precedent court.
I won't explore this possibility further since it would mean comparing the importance of fairness and correctness. It seems plausible to me that fairness is not vastly less important than correctness. So, it seems plausible to me that, in the example in the last paragraph, the instant court should follow precedent. But I admit to not having a proper argument for that view.
VI. Redundancy?
When I introduced the weak fairness principle in Section IV, I said that many people worry that the principle is redundant or superfluous because it only provides a reason to do what there is already a decisive reason to do. I promised I would respond to that worry; now I can.
The weak fairness principle is indeed redundant if:
• uniqueness: either the precedent or instant court is correct but not both; and
• certainty: it is certain which court is correct.
Were it certain that the Court was correct in Whole Woman's Health and incorrect in June, then the Court in June should follow precedent. We don't need fairness to reach that conclusion. Conversely, were it certain that the Court was incorrect in Whole Woman's Health and correct in June, then the Court in June should depart from precedent. Fairness isn't at issue. Either way, fairness is irrelevant.
Drop either of these conditions, though, and fairness can make a difference. That's obvious if we drop the uniqueness condition: if both courts are right, then a court does the right thing and the fair thing by following precedent, but only the right thing by departing from it. Fairness is decisive. (This is just the scenario in Table 2.)
Dropping the certainty condition also makes room for fairness to play a role. Fairness doesn't make a difference to which decision has greater value; the correct decision always has greater value. But fairness can make a difference to which decision has greater expected value. It does so by increasing the payoff from making a certain decision and that decision turning out correct.
Here's an analogy. There are two closed cardboard boxes, side by side on the table. One contains $100, the other contains nothing. You can choose either box and keep what's inside. Now I say: if you choose the right-hand box, and it contains $100, I'll give you a bonus of $50. Which box should you choose? If you can open the boxes and look inside before you choose, then of course you should pick whichever box has the $100. The bonus doesn't change that. But if we stipulate that you can't open the boxes beforehand, that the boxes are opaque, that you weren't watching when the money was added, etc., then the answer is: choose the right-hand box. The reason is simple: choosing the right-hand box maximizes the expected amount received (0.5 × ($ 100 + $ 50) > 0.5 × $ 100). The bonus is decisive, even though you only receive it if you choose the box that, in some sense, you should have chosen anyway.
Likewise, if a court is equally likely to make the correct decision if it follows precedent or departs from precedent, but it obtains the “fairness bonus” only if it makes the correct decision and follows precedent, then it should follow precedent. Fairness makes a difference, even though it only favors a decision that the court, in some sense, should have made anyway.
VII. Dominance
The tiebreaker argument relies on the weak fairness principle. In Section IV I motivated this principle with an example, in which I claimed that the weak fairness principle yielded the intuitive result. I also acknowledged that you might have reservations about that principle. If so, you'll doubt whether fairness has anything to say about the treatment of like cases. (I assume that no one would, on reflection, be attracted to the strong principle. And I assume that the strong and weak interpretations are the only sensible ways to develop the fairness principle.)
I can't allay those doubts. I can do something almost as good, though, which is to show that those doubts don't stand in the way of you accepting the tiebreaker argument.
Suppose that you are very skeptical of the weak fairness principle. You think that the probability that it is true is just 0.001, or 1 in 1000. The probability that fairness has nothing to say about how to treat like cases is 0.999, or 999 in 1000. That might seem to threaten my conclusions. In fact, it does not. Here's why. Assume that the weak principle is false. Between following and departing from precedent, fairness is not in issue. The expected value of following and departing from precedent is the same (given peerhood). Assume, conversely, that the weak principle is true. Following precedent maximizes expected value (given peerhood). Following precedent is therefore the dominant option: it is worse in no state of affairs and better in one state of affairs (that is, the state in which the weak principle is true).Footnote 34
So, provided that the probability that the weak fairness principle is true is not zero, the tiebreaker argument holds.
VIII. Overruling
I said in Section V that a court ought to follow indistinguishable precedents—if it is the peer of the precedent court. I also conceded that an instant and precedent court will not always be peers. So, my argument has a limited scope. Given its limited scope, my argument falls short of justifying the practice—unless, that is, the scope of the practice is similarly limited. Is it?
Here's what I need to show. I need for the instant court to depart from precedent only if it is the epistemic superior, not peer, of the precedent court. That means departing from precedent only if the instant court is either better at evaluating evidence, or has better evidence, than the precedent court.
As I said in Section II, a court does not follow a precedent, even in a like case, when it overrules the precedent. So, the question is: Do courts overrule precedents only when (they believe) they are the epistemic superior of the precedent court?Footnote 35 Overruling is a big topic, and its details vary between common law jurisdictions. So, I can't fully answer this question here. What I can do is say why a positive answer is plausible.
Sticking with US law for simplicity's sake, here is what the Supreme Court said when it was asked to overrule Roe v. Wade in Planned Parenthood:
[W]e may enquire whether Roe's central rule has been found unworkable; . . . whether the law's growth in the intervening years has left Roe's central rule a doctrinal anachronism discounted by society; and whether Roe's premises of fact have so far changed in the ensuing decades as to render its central holding somehow irrelevant or unjustifiable in dealing with the issue it addressed.Footnote 36
This is not an exhaustive statement of Supreme Court practice. But it is the “most prominent”Footnote 37 account of the factors that favor overruling precedents, and more recent statements list similar factors.Footnote 38
The point to notice is that these inquiries are all about evidence that could be before the instant court but could not have been before the precedent court. Evidence as to the “law's growth” since Roe, for example, is evidence that would only be available to the instant court. Likewise, while an instant court might know whether Roe's “central rule” has turned out to be “unworkable,” or whether its “premises of fact” have radically changed, the court that decided Roe could not. A court with more evidence is the epistemic superior of a court with less evidence, other things being equal. So, the Court is asking the right kind of questions if the aim is to work out whether it is the evidential superior of the precedent court.Footnote 39
The Court's list does not include factors relevant to evaluative superiority, but a comprehensive list would. One factor is time for deliberation. A precedent court may have to decide a complicated issue quickly, whereas the instant court has time to consider it deeply. Another example (not relevant to the US Supreme Court) is size of panel. A court may sit in a panel of five or seven when it first hears an issue, but in a panel of nine when it next encounters that issue, meaning it has greater opportunity to share insight and expertise.Footnote 40 Courts may also improve over time at evaluating certain moral and especially rights-centered issues. Such factors make the instant court the precedent court's evaluative superior, other things being equal. They may, as a result, entitle the instant court to act on its own judgment instead of on the precedent court's.Footnote 41
IX. Doctrine
I distinguished in Section III our question and the question whether there should be a doctrine of—a legal requirement to follow—precedent. I said I would not try to answer our question by drawing out the implications of an answer to the doctrine question. But the implications can also flow the other way, from an argument for following precedent to an argument for requiring courts to do so.
Suppose that the tiebreaker argument is correct. Suppose, too, that an instant court is disposed to depart from a precedent whenever it disagrees with precedent. Left to its own devices, the instant court will depart from precedent too often. A doctrine of precedent could help. It would require the instant court to stick with precedent unless it has a reason—beyond the mere fact of disagreement—to think it knows better than the precedent court. Thereby disciplined to depart from precedent only in certain circumstances, the instant court will better conform to the reasons anyway applicable to it.
The argument in the last paragraph relies on empirical assumptions—the dispositions of instant courts, the likely responses of courts to doctrinal requirements, etc.—that I have no way of verifying, and that may hold in some jurisdictions or at some times but not others. So, I do not claim to have justified a doctrine of precedent. All I mean is that my argument is compatible with, and can potentially provide, that justification.
X. Conclusion
The practice of precedent turns out to be less strange than legal philosophers sometimes suggest. True, courts sometimes let past decisions guide their present decision-making, even when they think the past is a poor guide to what they should now decide. But there is a good reason to do so: fairness. I expect this answer to our question is, in some ways, disappointing. Isn't it just the answer that lawyers and laypeople would have assumed was true anyway? Maybe so. But I think a conclusion is better for being simpler and more intuitive, other things being equal. In any event, while my conclusion is not original, my argument for that conclusion is, as far as I know, novel.
Acknowledgments
I thank Tom Adams, Kate Greasley, Jean Thomas, Jonathan Turner, Sandy Steel, Gregoire Webber, and three anonymous reviewers for their comments
Open access
