Analogical reasoning case to case

Analogical reasoning in law means reasoning from one particular decision to another based on factual comparisons. The current judge studies the facts and outcome of a past case, compares the facts to those of a case now pending before her, then reaches a decision in the pending case based on similarities and differences between the two cases.

Judges use, or claim to use, case-to-case analogies in various ways. First, a judge may find the current case to be factually similar to a prec­edent case and proceed to replicate the outcome of the precedent case (Levi 1948). Second, the judge may find that the outcome of a prec­edent case dictates the outcome of the current case a fortiori, because the new facts present at least as strong a case for the same result (Horty 2004). Thus, in one way or another, the decision in the precedent case constrains the later decision, requiring the later court to reach a paral­lel result even if it believes that the result of the precedent case was wrong, all things considered. In effect, the precedent court exercises authority over later judges by describing a set of facts and determining an outcome that controls later outcomes by virtue of similarity.

The third way in which courts purport to reason by analogy is by ‘distinguishing' rules established by precedent courts (see Raz 1979). If the current case falls within the predicate terms of an existing rule, but its facts differ from the facts of the precedent case in a way the current court finds significant, the current court treats the rule as inapplica­ble. Distinguishing is the flip side of a fortiori decision-making: factual disanalogy allows the current court to escape what appears to be an authoritative rule and return to natural reasoning, without explicitly overruling the precedent rule.

More recently, some have proposed a ‘reasons' analysis of the effect of precedent decisions, which combines several of the elements just described (Lamond 2005; Horty 2011; Lamond forthcoming). On this approach, the current judge studies a prior case that appears similar to the current case and determines the ‘reasons' for the prior decision based on some combination of facts and outcome. Then the judge either applies these reasons to decide the current case as a matter of precedential constraint, or identifies distinguishing facts in the cur­rent case that call for a different result, or, if it has power to do so, decides to overrule the precedent. We consider each of these supposed instances of analogical reasoning in more detail below.

6.1.1 Constraint by similarity

The simplest way in which courts use analogies is by finding that the case before them is similar to a precedent case and then citing this similarity as a reason to reach a parallel result. For example, a judge has before her a nuisance case brought by neighbors against a homeowner who keeps a pet ocelot in his home. The plaintiffs point to a prior case in which a court ordered a homeowner not to keep a pet bear in his residence. The prior court did not establish a rule governing accept­able pets; instead, the plaintiffs argue that an ocelot is like a bear and that this likeness gives the court reason to banish the ocelot.

We have already explained why, for us, the maxim that like cases should be treated alike has no moral weight in the context of sequential legal decisions. If the prior court had good reasons for deciding as it did, and the same reasons apply to the current case, then those reasons support a similar decision now, quite apart from equal treatment. But suppose for the moment that the equal treatment principle is sound. A more significant difficulty with this type of analogical thinking is that without more information, and more thought, it is impossible to say that two cases are alike in any meaningful way.

In fact there are an infinite number of similarities and differences between an ocelot and a bear. Both are predatory animals that could harm a small child; both are difficult to domesticate; and both are furry mammals. On the other hand, the ocelot is very likely smaller than the bear; it is a type of feline indigenous to Belize; and it has spots. The outcome of the precedent case does not pick out which of these similarities and differences matter for purposes of a decision about nuisance. The point is even clearer if another precedent case held that a Dalmatian dog was not a nuisance in a residential neighborhood. The Dalmatian was likely about the same size as the ocelot, and, like the ocelot, had spots. Where are we now in the process of determining likeness?

The problem is that nothing in the story so far shows that the judge has reasoned that the outcome of her ocelot case should be the same as the outcome of the prior case involving a bear. Possibly the two cases should be decided alike, but this is not a reasoned conclusion unless it rests on some general proposition about why certain common charac­teristics of bears and ocelots matter in a nuisance case. Without this additional explanation of the importance of the similarity, which can only be formulated through reasoning, the facts of the prior case have nothing to say about the current case. It may not take much reasoning to link an ocelot to a bear, but facts alone do not supply the link.

There are many who believe in the independent power of analogy, without any need for supporting generalizations. Lloyd Weinreb, for example, cites a case in which the New York Court of Appeals held that the owner of a steamboat was strictly liable for money stolen from a passenger's stateroom.^[58] The court cited two lines of precedent: a series of cases holding that innkeepers are strictly liable for property stolen from guest rooms, and another series of cases holding that railroads are not strictly liable for thefts from sleeping carriages. The Court of Appeals held for the steamboat passenger on the ground that steam­boats are more like inns than like railroads. In Weinreb's view, this decision shows that courts can and do make decisions on the basis of similarity to precedent facts, without the aid of either a rule that ties one set of facts to another or a broader proposition about the moral significance of the factual similarity (Weinreb 2005).

We are unconvinced by this example. The fact that the court did not cite or announce a rule linking steamboats to inns does not establish that it had no such rule in mind. In the days of steamboats, courts were reticent about rule-making through judicial opinions. In any event, we do not mean to suggest that analogical decisions must rest on rules of similarity: our point is simply that to conclude that two sets of facts are importantly similar, the court must refer to some general proposi­tion that links them and justifies the same legal result for each. For example, the Court of Appeals may have judged that businesses that provide lodging to a fixed number of guests for a period of time are in a better position to protect or insure against theft than businesses that provide transportation to a variety of customers who come and go at different times.

Suppose, however, instead that, as Weinreb suggests, the Court of Appeals simply formed an intuition of similarity between steamboats and inns, without relying on a supporting generalization about risks or expectations.

Returning to the nuisance example, there are several ways in which a judge might make reference to a prior case (the case of the bear) in the process of reasoning to a decision in a new case involving an ocelot. She might formulate a tentative moral principle that explains the case before her and then test the principle against the outcome of one or more prior decisions. For example, she might begin with the principle that the liberty of owners to use their property as they wish is subject to a duty not to inflict an unreasonable risk of harm on others. Then, upon finding the case of the bear, she might conclude that her tenta­tive principle provides a morally satisfying answer to the bear case as well. This process of comparison lends support to her principle and also to the answer it provides to the problem of the ocelot (see Brewer 1996; cf. Stevens 2018). The judge might instead announce a rule that provides an answer to both the prior bear case and her current ocelot case: dangerous wild animals should not be kept in residential neigh­borhoods, and ocelots, like bears, are dangerous wild animals. This rule finds support from the broader principle the judge has formu­lated about property and harm, and it has the advantage of settling the matter of wild pets in residential neighborhoods for the future. Either way, analogy plays no direct role in her decision. The result may be that similar pets are treated alike, but the conclusion of similarity rests on reason, not intuition.

A related point is that when a judge uses a prior decision to test a tentative principle, or formulates a rule that fits both a prior case and an attractive moral principle, the outcome of the prior case does not determine the outcome of the case before her.

Our argument so far has been that the process of reasoning that leads to the conclusion that two similar cases should be decided alike is not a special analogical form of reasoning. Instead, it is ordinary reasoning to determine whether and why certain similarities are important. Prior decisions stand as points of comparison in a later court's reflections about how best to resolve a current case.

Another possibility, suggested by Weinreb's discussion of liability for losses from theft, is that courts make decisions on the basis of unmedi­ated perceptions of similarity: ocelots are like bears when placed in residential neighborhoods, so if bears are nuisances so are ocelots. Our first observation is that we think this may be a faulty characterization of what courts do. Rather than simply perceiving relevant similarity, the judge may, over the course of a lifetime, have internalized certain general principles that allow her to make a reason-based connection between bears and ocelots without articulating, or even consciously considering, the principles that guide her. If so, her decision still rests on reason rather than intuition.

Suppose, however, that it is psychologically possible for the judge simply to perceive an important likeness between ocelots and bears. This is not reasoning to a conclusion, any more than a coin toss is reasoning to a conclusion. A proponent of analogical decision-making in law, however, might respond that the process may not be reasoned, but it is what judges do. They manage to decide cases this way and to maintain some continuity in the development of law. Edward Levi, for example, wrote that

If this is really reasoning, then by common standards... it is imperfect unless some overall rule has announced that this common and ascertain­able similarity is to be decisive. But no such fixed rule exists. It could be suggested that reasoning is not involved at all, that no new insight is arrived at through a comparison of cases. But reasoning appears to be involved; the conclusion was arrived at through a process and was not immediately apparent. It seems better to say that there is reasoning, but it is imperfect. (Levi 1948, 3)

We find this conclusion disheartening. Judicial reasoning is under­standably imperfect in many of its outcomes, but it should not be imperfect in the sense that it does not follow a rational process.

This is not to say that intuition cannot play a role in judicial decision­making, by helping to get the process of reasoning under way. To illustrate, suppose again that a judge is presiding over a nuisance case involving an ocelot. She studies a precedent case involving a bear and has an intuition that there is an important similarity between the two animals. Next, she develops a general proposition that supports her intuition: dangerous wild animals should not be kept in residential neighborhoods because the risk of harm outweighs the owner's liberty and the value of animal companionship. She then confirms that bears and ocelots are comparably dangerous and holds that the ocelot must go. This decision follows a rational process, but the process is not strictly analogical because the general proposition developed by the judge, rather than the judge's initial intuition of similarity, ultimately controls the outcome.

Alternatively, suppose the judge reasons to a tentative conclusion in the current case that the owner of the ocelot should be allowed to keep his pet, based on an appropriate balance of liberty, animal companion­ship, and risk. Then she studies the prior case holding that a bear is a nuisance in a residential neighborhood, and has an intuition of impor­tant similarity between this case and the ocelot case she must decide. Next, she develops a general principle that supports her intuition of similarity: dangerous wild animals present too much risk in residential neighborhoods. She tests this revised principle against several more actual or imagined cases and adjusts her initial conclusion about the ocelot case, deciding instead that the balance of liberty, companion­ship, and risk favors the complaining neighbors. This too describes a process that involves intuition but is not strictly analogical, because in the end the judge relies on her own reasoning rather than directly on the outcome of the prior case involving a bear. The role of the bear case is simply to trigger a reexamination of her initial position.^[59]

We have no difficulty with the methods of decision-making just described. The method we reject is decision-making based on a bare intuition of similarity, unmediated by reasoning. Suppose the judge deciding how to resolve the case of the ocelot studies the case of the bear, notices a similarity between the animals involved (they are wild animals), and intuits that this similarity is important. Assume that this type of intuition is psychologically possible, without reference to a supporting generalization about wild animals that might affect the question of nuisance. Without more, the judge decides that the ocelot should be removed, just as the bear was removed. In this case, then, unlike the cases just described, the primary decision-maker is the prec­edent judge, who reasoned to a conclusion about what should happen to the bear. The current judge's only contribution is the intuitive con­nection she makes between the bear and the ocelot.

We have already noted that we doubt that judges often make decisions in this purely intuitive way; we believe the alternatives sketched above are more plausible. But assuming judges can and do reach decisions in this manner, our principal argument is a normative argument, that decisions of this kind should not count as authoritative settlements of legal disputes. Settlement, as we used the term in the early chapters of this book, does not simply mean ending disputes and enabling actors to coordinate their conduct. In the case of legislation, it implies some combination of reason and political compromise. In the case of judi­cial decision-making, it implies reasoned settlement of current, and to some extent future, controversy. In both settings, members of a community have chosen an authority to translate values they recognize as reasons for action into particular decisions or rules. The authority's conclusion may not always be correct, but we expect that the process of decision-making will be capable, at least in principle, of articulation and justification. Judicial decision-making, as an exercise of authority, ought to meet this minimal requirement; therefore, it should entail more than untested and untestable intuition.

6.1.2 A fortiori constraint

We reject the idea that factual similarity between past and present cases can constrain judicial decision-making. Because similarities and differences between cases are infinite, the current judge must rely on some rule or principle, arrived at by ordinary reasoning, to determine when and to what extent common features point to a common result. Once the court has identified a decisional principle, this principle, rather than the prior decision, governs the result of the current case.

Some have suggested a variation on analogical reasoning in which judges compare the strength of facts present in past and present cases (Horty 2004; for criticism, see Alexander 1989). If the precedent case reached a certain result on certain facts, and the facts of the current case provide more support for the outcome of the precedent case than the facts of the precedent case itself, then it follows ‘a fortiori’ that the current judge should reach a corresponding result. Suppose, for example, that the current judge has before her a case in which a home­owner in a residential neighborhood is keeping a lion, a python, and a bear as pets. A prior case in the jurisdiction held that a bear kept in a residential neighborhood was a nuisance, but the opinion in the prior did not purport to establish a rule (‘bears in residential neighborhoods are nuisances per se'). Despite the absence of a precedent rule, the judge should order the owner in the current case to remove his pets from the neighborhood because, a fortiori, if a lone bear was a nuisance in the first case, then a private menagerie containing a bear must be a nuisance in the current case. On a strict version of this view, the same result should follow even if the current judge believes the current menagerie is carefully supervised and maintained and does not create a nuisance.

This method appears more promising as a method of case-to-case reasoning than simple analogy from particular facts of a prior case to similar facts of a later case. Here, the current judge compares her case to past cases and draws what seems to be a necessary conclusion. We believe, however, both that there are serious flaws in the supposed reasoning associated with this method of decision-making and that the method does not adequately account for the possibility of error in prior decision-making.

One difficulty is that facts may be missing from the report of the prec­edent case. The facts mentioned in the prior opinion will have been selected by the prior judge from what was already a subset of facts deemed important by the parties. In some cases, this will result in too broad an a fortiori effect, if the precedent case is phrased in very broad terms. If, for example, the precedent opinion failed to mention that the bear under discussion in that case had a history of violence against humans, then a fortiori the menagerie in the second case would be deemed to be a nuisance, even if all three animals are elderly and gentle. Overly simple descriptions may also blur the distinction between ana­logical decision-making and rule-governed decision-making.

A different problem is that the a fortiori effect is easy to circumvent: a judge who prefers to avoid the outcome of a prior case usually can locate some feature of the current case that will serve to alter the bal­ance in favor of a different result. If the precedent opinion states only that the defendant was keeping a ‘bear', with no details given, and the later judge is sympathetic to the defendant now before her, she could invoke the gentleness of the current defendant's animals as a fact that distinguishes her case from the precedent case. No two cases are per­fectly identical in their facts, so there normally will be some basis on which the second judge can draw a distinction and avoid the a fortiori effect of the precedent case.

Another difficulty is how to weigh facts in order to reach an a fortiori conclusion. The idea behind a fortiori decision-making is that, given a precedent case with a certain outcome, the facts of the current case provide more support for the same outcome than the precedent facts. In the simplest comparison, the current case is the same as the prec­edent case but presents an additional fact that supports the result of the precedent case, or lacks a fact that supported a different result in the precedent case. For example, both cases were nuisance cases involving bears, but the precedent case involved a declawed bear while the bear in the current case has claws. Typically, however, there will be more factual variation between the cases, including at least some facts that point in both directions, for and against the precedent outcome. In cases of this type, the court must first decide what outcome the variant facts tend to favor, and then somehow assign a weight to that tendency.

Looking first at the problem of tendency: the tendency of a particular fact to support one result or another may seem obvious. A large bear favors a finding of nuisance while a small bear cuts the other way. Notice, though, that the fact of bear size does not speak for itself: bear size must be linked to the question of nuisance by a process of reason­ing involving moral principles and empirical connections: it is wrong to engage in an activity that poses an unreasonable risk to the safety of neighbors, and large bears pose a greater safety risk than small bears. Without this explanatory connection, the tendency of bear size toward a certain result is simply a matter of unfounded intuition.

Assuming that there are factual differences between the precedent case and the current case that tend in different directions, the court must then assign weight to each variant fact in order to arrive at an a fortiori comparison. Suppose, for example, that the current defendant is hous­ing a more dangerous set of animals than the losing defendant in a prior nuisance case; however, the current defendant is also engaged in breeding efforts designed to protect endangered species. Comparing human safety and preservation of species is difficult; on some views these two values may be incommensurable, so that neither has greater lexical or absolute value in comparison to the other (see Symposium 1998). Unless the current judge has some way to rank these values, it is hard to see how she can determine which case is the stronger case for an injunction.

Some moral systems, such as utilitarianism, provide a common metric that allows, at least in principle, for a fortiori comparison of the facts of two cases. Because only utility is in play, no ranking of values is needed. Yet, even when a single metric distinguishes good from bad, a comparison between prior facts and current facts is likely to fail. One problem is the conceptual difficulty of defining and calculating utility. Another problem is that some of the precedents under comparison will have been wrongly decided: even if all judges agree that utility is the right metric, they will make mistakes in applying it, leaving a set of decisions that cannot be fully reconciled.

Suppose, for example, that homeowners are seeking an injunction against the construction of a gas station in their residential neigh­borhood. All judges agree that nuisance cases should be decided by reference to a utilitarian criterion that equates utility with the sum of human happiness. The current judge determines that the proposed gas station will result in decreases in happiness due to traffic, fumes, and aesthetic offense, although these losses will be counterbalanced by increases in happiness due to financial profit, employment, and convenience to local motorists. She calculates that, altogether, the gas station will cause a net loss of 3 utils, and she is inclined to grant an injunction. The defendant, however, points to a precedent case in which the owner of a pet crocodile was allowed to keep her crocodile in a residential neighborhood, and argues, persuasively, that the crocodile imposed a net loss of 6 utils on neighbors. The defendant is now in a position to make an a fortiori argument: given the prior decision in favor of a crocodile owner, any use calculated to cause a loss of fewer than 6 utils should be allowed. Of course, something is wrong with this argument but it is hard to say exactly what it is if an a fortiori comparison of current and past cases is assumed to be a valid basis for legal decision-making.

The problem of comparison is compounded if some precedent deci­sions appear to be correct while others appear incorrect. Suppose the parties to a current nuisance case point to two precedents. In one case, involving a crocodile, the judge denied an injunction, resulting in a loss of 6 utils. In another case, involving a bear, the judge granted an injunction, resulting in a gain of 1 util. If both cases stand as prece­dents, then in any new case in which the challenged activity is expected to cause a loss between 1 and 6 utils, there are a fortiori precedents for both denying and granting an injunction. At this point, some more straightforward form of judicial reasoning clearly is needed.

A fortiori decision-making has one significant virtue: it taps the abil­ity of reasoners to make comparative judgments when they may be unable to make absolute judgments. Comparing the degree to which a certain property is present in two different objects (a large bear is more dangerous than a small bear) appears to be an easier task for human minds to manage than deciding upon an absolute value (how danger­ous is a large bear?) (Dember and Warm 1979; Kahneman and Tversky 1979). As a result, judges can make a fortiori judgments about past and present cases more confidently than they can assess a present case in isolation. Yet, this does not mean that judges can reason reliably from one case to another without a principle that ties their assessment together. To assess large and small bears on a scale of dangerousness, the judge first needs to have in mind a rule or principle such as ‘enjoin pets that pose too high a risk of physical harm to neighbors'.

6.1.3 ‘Reasons?'

Some writers, notably Grant Lamond and John Horty, have supported a ‘reasons' account of judicial decision-making (Lamond 2005; Horty 2011; Lamond forthcoming). This approach is a blend of a fortiori techniques, distinguishing techniques of the type discussed in the next section, and some aspects of analogical decision-making. The aim is to mount an analytic defense of what many lawyers suppose to be the fundamental nature of judge-made law. We admire this work as a description and precisification of widely held understandings of the common law. Nevertheless, we argue that it is a fantasy rather than a logical possibility.

The reasons account proposes that a current judge begins by studying a prior case that is factually similar to the dispute to be resolved. The judge extracts ‘reasons' for the prior decision based on a combination of facts and outcome. If a prior opinion held for the plaintiff and cited facts a, b, and c in support, this establishes a reason for decision (ratio decidendi): ‘a, b, and c n'. At this point, assuming the prior judge and the current judge occupy equivalent positions in the judicial hierarchy, the current judge compares the facts of her current case. If facts a, b, and c are present, the judge must either (1) decide for the plaintiff as a matter of precedential constraint; (2) locate some distinguishing fact of the current case that was not present in the prior case; or (3) perhaps, make a decision to override the precedent decision. Prior judgments, on this view, enter into the fabric of the common law and bind later judges, not as rules, but as determinations of legal questions in context. In this way the common law is elaborated and extended over time, but modifications normally are incremental and closely tied to particular facts.

An important assumption here is that, as the process of decision­making occurs, the prior judge is making law and the current judge is following it, with a certain amount of leeway to alter the path of the law as it develops. The first flaw in this design lies in how exactly the reason associated with the precedent case comes into being. If the current judge derives a legal reason that includes all facts mentioned in the prior opinion, that reason will rarely apply to the new case because no two cases are exactly alike. If the current judge selects among facts mentioned in the prior opinion, the reason is generated at least in part by the current judge's selection rather than by the prior judge's deci­sion. Its source is not the common law, but the interpreting judge.

A related problem arises from the nature of the facts that generate the reason associated with a precedent case. Suppose the precedent case was a nuisance case involving a pet bear; the current case is a nuisance case involving a pet ocelot. The prior opinion may state that the bear was large, young, untrained, and allowed to roam free in the owner's yard, and thus was a nuisance when located in a residential neighbor­hood. This description generates a precedent reason that might guide a future judge in deciding a future case involving a closely similar bear. If, on the other hand, the prior opinion states that the bear was unrea­sonably dangerous and unsatisfactorily restrained, a future judge must interpret these facts to arrive at a reason for a decision in a new case. In this situation, the operative reason comes not from the pre-existing common law but from the new judge.

Another difficulty with the reasons account of precedent is that effec­tive governance sometimes requires rules that are preemptive of moral reasoning. As discussed in Chapter 5, an authoritative rule that speci­fies a set of outcomes in determinate terms and remains in effect until definitively overruled may, in some circumstances, produce better results over the long run than case-by-case judicial reasoning. The reasons account of the common law, however, prevents judges from announcing authoritative rules. On the reasons account, the rule asso­ciated with a past case depends on the decision the precedent judge made in response to particular facts rather than statements of law the precedent judge may have made. Moreover, later judges can avoid the implication of both announced rules and precedent reasons by making factual distinctions that the precedent judge might not have made. It follows that determinate rules are not a tool of governance available to judges under this approach.

Finally, it is not clear to us why we would want current judges to extract legal reasons from prior cases rather than deciding the cases before them based on their own best judgment. In the absence of definitive precedent rules, it is always open to current courts to study past cases and derive insight from the relations between the facts and outcomes of those cases. Responsible decision-making also requires current courts to consider the extent to which both parties and non­parties have relied on the reasons for decision expressed in prior judi­cial opinions. But in the absence of authoritative rules, it seems to us that judges should look to the past for guidance rather than direction.

6.1.4 Distinguishing precedent decisions

Another way in which courts rely, or purport to rely, on factual com­parisons between past and present cases is the practice of distinguish­ing rules set out in prior decisions. Typically, a prior court has decided a case raising similar issues and, in the course of its decision, has announced what appears to be an authoritative rule. A new case arises that falls within the terms of the prior court's rule but does not per­fectly match the prior court's description of the precedent facts - some facts present in the prior case may be missing in the new case, or some new facts may be present. In these circumstances, the court in the new case may ‘distinguish' the prior case and in this way avoid the effects of the rule.

Distinguishing precedent rules can be viewed as the reverse of using factual similarities between cases to expand the implications of prec­edent decisions. Factual dissimilarities allow the court to avoid the consequences of a rule. Distinguishing may also function as a form of reverse a fortiori reasoning. If the facts of the new case provide weaker support for the outcome reached in the prior case, then the current court can use this technique to avoid the implications of the prior holding.

For example: the current case is a nuisance claim against the owner of a large dog. The judge in a prior case ordered the owner of a retired circus bear to remove it from a residential neighborhood, stating that ‘large animals are nuisances per se when kept in residential neigh­borhoods'. The rule appears to cover the current case, requiring the current judge either to follow it or to overrule it. The practice of dis­tinguishing precedent, however, allows the current judge to permit the dog to stay on the ground that the dog is a domestic animal, and the rule does not apply by its terms to large domestic animals. Without saying so explicitly, the current judge has modified the rule to say that ‘large wild animals are nuisances per se in residential neighborhoods'.

This process of distinguishing may appear to support a modified, more nuanced, application of the rule model of precedent: the rule is pre­served but the court makes an exception for a case that differs from the case that first generated the rule. In fact, however, the rule plays no role in the current judge's decision (see Lamond 2005; Stevens 2018). Something like the original rule will remain on the books, but as long as creative judges (or parties) can identify factual distinctions among cases, what was intended to be an authoritative rule will have no effect on the outcome of subsequent cases. Rules, by nature, must be more general than the facts of any particular case they cover. Consequently, there will always be some difference between one covered case and another, which can serve as a point of distinction. The rule established in the case of the circus bear held that ‘large animals are nuisances per se'. The next case involved a dog, so the rule became ‘large wild ani­mals are nuisances per se'. If the next case involves a crocodile, and the judge is sympathetic to the crocodile's owner, the judge can note that the rule first applied to a bear, and does not mention animals without fur, and therefore can fairly be understood to mean that ‘large furry wild animals are nuisances per se'. It follows that the rule itself, even if intended to be authoritative in the range of cases it described, has no constraining effect.

Joseph Raz has argued that the practice of distinguishing precedents, properly understood, does constrain the manner in which judges can modify precedent rules (Raz 1979)· Raz suggests that the ability of judges to alter precedent rules by adding distinctions is subject to two conditions: the modified rule that emerges from the process must be the same as the precedent rule, with further conditions added; and the modified rule must continue to support the outcome of the precedent case in which the rule was announced· Thus (using Raz's example): assume a precedent case involving facts, a, b, c, d, and e. The outcome of the case was X, and the precedent court announced a rule ‘If A, B, and C, then X. In a new case, the facts are a, b, c, d, andf, but not e (fact e is now missing and a new fact f is present). In these circumstances, the new court would like to reach the result not-X. According to Raz, it can do this, even though A, B, and C are present. Specifically, it can dis­tinguish the current case from the precedent case by pointing out that fact e is not now present, or fact f is now present; then it can modify the precedent rule to provide ‘If A, B, C, and E, then X or ‘If A, B, C, and not F, then X, and proceed to hold not-X. Either of these modified rules supports both the outcome of the precedent case and the desired outcome in the new case. But the court cannot modify the precedent rule to provide ‘If A, B, C, and not D, then X, because this rule does not support the outcome of the precedent case.

In our view, the constraint Raz describes is illusory. Consider the example above, in which a precedent rule states that ‘wild animals in residential neighborhoods are nuisances per se'. The original case involved a bear, and the judge in the later case involving a crocodile avoids the rule by pointing out that the bear was a furry wild animal and the crocodile lacks fur. This decision meets Raz's conditions, but if this is an acceptable pattern of decision-making, then as long as a later judge can identify any factual difference between cases, the precedent rule exerts no influence over later decisions. Moreover, reason plays no role in the pattern of decision-making over time. As a result, the evolution of the common law depends not on articulated changes in values but on the creativity of judges in avoiding precedents they find unacceptable on unarticulated grounds.

Raz suggests a third condition on the practice of distinguishing rules that might, depending on its application, give more weight to precedent rules in judicial reasoning: courts should modify prior announced rules only in the way that ‘will best improve' the original rules. The sugges­tion is that precedent rules should be preserved in some sense that is illuminated by the reasons that originally motivated them. Thus, if the original nuisance rule forbidding large pets were based on concerns about danger to neighbors, a distinction that allows large domestic animals could be seen as an improvement to a rule that is overbroad in relation to its purpose. Yet, from the standpoint of a judge who believes that the precedent rule was poorly conceived from the outset, the way to improve the rule is to cut it back in whatever way works to limit its effects in future cases. If the current judge believes that owners should be allowed maximum liberty to choose their pets as long as the animal has no history of harmful behavior, a distinction permitting wild animals without fur results in a better rule.

Another difficulty is that in practice, judges may distinguish precedent rules in response to unsatisfactory results in particular cases rather than inadequacies in the rule. The precedent rule may be a perfectly good rule that yields a better sum of outcomes than particularized decision-making would yield, and yet, because it is phrased in determi­nate terms, it will sometimes dictate a bad result. If the current judge is more interested in avoiding a single bad outcome than fine-tuning the rule, distinguishing may result in the substitution of an awkward new formula for a normally workable rule. In the example above, the modified rule ‘large, wild, furry animals are nuisances per se' is not a rule consciously designed to produce the best sum of results in cases that fall within its terms; it is simply what happens to be left of the original rule after an exception is carved out to accommodate a par­ticular friendly crocodile. We accept that, subject to rules that establish judicial hierarchies, judges have the power to alter rules; but we would prefer that they exercise this power in a forthright manner and not in the guise of a minor tweak.

In our view, therefore, the process of distinguishing precedents is most charitably understood as a disguise for ordinary reasoning that refers to moral principles to determine what outcome currently is best, all things considered. If this view of distinguishing precedents is correct, then, in effect, courts are employing ordinary moral and empirical skills to reach a conclusion, then using their overruling and rule-making powers to substitute new rules for precedent rules. The main drawback of the distinguishing mechanism is that the pretense of retaining modified versions of old rules may produce a somewhat strained and awkward version of the common law.

6.1.5 Summary: why purely analogical decision-making does not exist

Analogical reasoning is often thought to act as a constraint on judicial decision-making. Findings of factual similarity or difference are said to require parallel results in similar cases, or determine results a fortiori, or determine the conditions under which judges can avoid precedent rules, or reveal ‘reasons' embedded in the facts and outcomes of past cases. We hope we have shown that it does none of these things.

Analogical decision-making based on factual similarity between cases is either intuitive or deductive. If the process of identifying important similarities is intuitive, then precedent cases do not constrain the out­comes of current cases in any predictable, or even detectable, way. If the process is deductive, then the rules or principles the judge applies to determine similarity, rather than the outcomes of precedent cases, determine the results of later cases.

A fortiori decision-making based on factual comparisons between past and present cases imposes no real constraint. Precedent cases cannot determine the results of new cases a fortiori because a later judge can always cite some fact about her new case that favors a different result, and in this way dispel the a fortiori effect of the precedent case. A further problem with a fortiori decision-making is that the notion of weighing facts is problematic. To weigh two different sets of facts, a judge must identify a common metric for comparison. No such metric may be available, and if a common metric exists, its application to a body of precedent that includes incorrect decisions can lead to legal chaos.

Decision-making based on the reasons for decision, or ratios, expressed in prior decisions is the classic method associated with the common law. This method, however, is illusory because prior reasons typically are either too narrow to control current decisions or too open ended to be applied without significant reinterpretation by current judges. Reasons of this type also supersede authoritative judicial rules and undermine the benefits associated with determinate guidance.

Distinguishing precedent rules is often characterized as a way in which later courts can preserve a precedent rule but adjust it in minor ways to avoid applying to cases in which the rule produces an unwar­ranted result. In our view, distinguishing is not a modest way to avoid unintended consequences while preserving the precedent rule. It amounts instead to overruling the precedent rule and leaving in its place a new rule that may not adequately serve the objectives of the original rule. The precedent rule, and the decision in which it was announced, impose no constraint on the process of distinguishing, the outcome of the current case, or the content of the replacement rule.

6.2