Gageler, Stephen; Lim, Brendan --- "Collective Irrationality and the Doctrine of Precedent" [2014] MelbULawRw 21; (2014) 38(2) Melbourne University Law Review 525

COLLECTIVE IRRATIONALITY AND THE DOCTRINE OF PRECEDENT

THE HON JUSTICE STEPHEN GAGELER^[*] AND BRENDAN LIM^[†]

Appellate adjudication is committed to decision-making by groups and it is committed to decision-making by adherence to precedent. Those commitments are in tension. Group decision-making is inherently susceptible to collective irrationality in the form of internal inconsistency. This presents particular problems for achieving consistency through precedent-based decision-making. Exploring insights from the emerging interdisciplinary literature on judgment aggregation, we sketch a defence of how the common law has traditionally managed these thorny issues.

CONTENTS

[W]here the Court, although coming to a firm conclusion in its curial order, does so upon different grounds ... it imparts into the legal principles enunciated under a system of precedents, an uncertainty which is to be deplored.^[1]

Inconsistency is inevitable ... [D]emands for perfect consistency can not be fulfilled, and it is inappropriate to condemn the Court’s performance as an institution simply by pointing out that it sometimes, even frequently, contradicts itself.^[2]

I INTRODUCTION

Decision-making by groups, because it depends upon aggregating decisions by individuals, is inherently susceptible to internal inconsistency. Decision-making by adherence to precedent, because it strives to treat like cases alike, is inherently committed to achieving consistency over time. The common law system of decision-making utilises both groups (multi-member appellate courts) and adherence to precedent (stare decisis). The common law system of decision-making is therefore inherently susceptible to inconsistency within particular decisions, yet inherently committed to consistency between those decisions. How does the common law reconcile that susceptibility with that commitment?

The beginnings of an answer to the question can be found in the rational function performed by some of the formal rules of the doctrine of precedent. By exploring and applying the insights that are emerging from a growing interdisciplinary literature on judgment aggregation, we sketch a defence of the common law’s traditional requirements that, when ascertaining of an appellate decision whether it has a ratio decidendi that binds subsequent courts and if so what that ratio decidendi is, one should ordinarily exclude from consideration reasoning that was unnecessary to the decision and dissenting opinions.

In a previous article, one of us examined the practice of group decision-making in appellate courts from the perspective offered by Condorcet’s jury theorem — the mathematical result that a group deciding by majority rule is, under certain reasonable conditions, more likely to arrive at a correct decision than any individual member of the group deciding alone.^[3] That article considered the conditions under which the theorem holds, principally the decisional competence of the individual members of the group and the decisional independence of the individual members of the group, and identified some of the practical issues that confront multi-member appellate courts in that regard. The article did not set out to address in detail the complications presented by composite propositions or the related issue of the inherent susceptibility to internal inconsistency in group decision-making. Nor did it set out to consider the problems that can arise when groups seek to make multiple decisions systematically over time. Those matters are now taken up in this article.

In Part II, we identify the problem of collective irrationality: how, in group decision-making contexts, the need to aggregate individual decisions is attended by an inherent susceptibility to inconsistency. We explain the practical solutions that the common law deploys in individual decisions, but in doing so highlight the hard choices that must be made. In Part III, we identify the related problem of applying collectively irrational decisions as precedents. These decisions, on their face, appear to support multiple inconsistent propositions. A system of precedent cannot tolerate giving binding force to all of those propositions. It must have a workable method for identifying which propositions are binding and which are not. We explain how the traditional rules of the common law doctrine of precedent have solved this problem.

II COLLECTIVE IRRATIONALITY

A Meaning of ‘Collective Irrationality’

Reasoning is sometimes described as ‘irrational’ merely as an expression of emphatic disagreement.^[4] We do not use the adjective in that sense. We use ‘irrational’ in a true sense to denote the simultaneous acceptance of propositions that are logically inconsistent with each other. Standards of rationality ‘rule out failures of consistency’: they rule out ‘tak[ing] propositions to be true that are not co-realizable’.^[5] For example, it would be truly irrational simultaneously to accept both that ‘Socrates is a man’ and that ‘Socrates is not a man’. It would, similarly, be truly irrational to accept the logical proposition that ‘All men are mortal’, and, while also accepting that ‘Socrates is a man’, accept that ‘Socrates is not mortal’. Failures of consistency such as these may not be the only possible failures of rationality,^[6] but for our present purposes they are the salient failures.

We should also not be misunderstood to accuse any individuals of inconsistency or irrationality. We are concerned with a form of collective irrationality that can arise in a group decision-making context even if each individual member of the group reasons perfectly rationally. Collective irrationality is a function of aggregating individual judgments. It is an incident of the decision-making procedure and not an incident of any faulty reasoning by individual decision-makers.

B Illustration of Collective Irrationality

Almost 70 years ago, the eminent Australian scholars George Paton and Geoffrey Sawer described the essence of the problem of collective irrationality in appellate decisions: ‘where a party relies on two or more points, it is possible for him to obtain a majority judgment in his favour, although a majority of judges rules against each point taken separately’.^[7]

At the time, there was ‘[l]ittle discussion of this problem ... in the authorities, whether judicial opinions or learned writings’.^[8] Today, there is a burgeoning literature on the problem, largely traceable to an article by Lewis Kornhauser and Lawrence Sager, published in the Yale Law Journal in 1986, which explained in detail the phenomenon of collective irrationality.^[9] Kornhauser and Sager extended their analysis in 1993 and named the phenomenon ‘the doctrinal paradox’,^[10] although later work has preferred to call it a ‘discursive dilemma’, because it is not confined to doctrinal situations^[11] and because it presents a ‘hard choice’ between alternatives rather than a true paradox.^[12]

Kornhauser and Sager illustrated their ‘paradox’ with a simple example based on a stylised contract case.^[13] In this stylised case there are two intermediate issues, the resolution of which can be called ‘premises’: (1) whether the parties formed a contract; and (2) whether the defendant breached the contract, assuming it properly to have been formed. The ultimate issue, or ‘conclusion’, regarding liability should depend upon the premises: if both issues are resolved affirmatively, then it should be decided in the conclusion that the defendant is liable. In a three-member court, the following pattern of judgments on the premises and the conclusion might emerge:

Each of Justice A, Justice B and Justice C has, individually, reasoned rationally. Their respective conclusions follow logically from their respective positions on the two premises (and from their acceptance of the legal rule that the defendant is liable if, relevantly, the contract was formed and the contract was breached). But now consider the aggregation of their individual judgments according to majority rule:

The aggregation of the judgments is irrational. It exhibits a failure of consistency in that it accepts a set of jointly inconsistent propositions. It accepts that the parties formed a contract; it accepts that the defendant breached the contract; but it also accepts that the defendant is not liable (and it accepts the legal rule that the defendant is liable if, relevantly, the contract was formed and the contract was breached). If the court, taken to be speaking through its aggregate judgment, were held to the standards of rationality expected of any individual judge, it must be seen to have committed an embarrassingly basic error.

The stylised contract case is merely illustrative of a more general problem. One could, of course, change the doctrinal context to any area of law. But it is worth highlighting that one can find doctrinal contexts in which the premises connect to the conclusion disjunctively, instead of conjunctively. In fact, many if not most appeals take this form, when the appellant relies upon independent errors in the decision appealed from, any one of which will be sufficient to have the decision set aside:

On another variation, the connecting rule need not be fixed, in that the applicable doctrine could itself be seen within the decisional matrix as a ‘premise’ upon which individual judges should come to a view, and upon which different judges may reasonably come to different views. One could also increase the number of relevant premises or the number of judges. However one calibrates these variables, the possibility of irrational aggregate decisions remains.

The possibility of such irrational aggregate decisions should not be thought to be remote or fanciful. There are numerous instances in the decided cases. Several were collected by Paton and Sawer.^[14] A recent example is Hepples v Federal Commissioner of Taxation (‘Hepples’),^[15] which will be discussed in a later section.^[16] Another is Commonwealth v Verwayen (‘Verwayen’), which raised issues as to whether the Commonwealth was barred, by estoppel or waiver, from pleading a limitation defence to a claim in negligence brought in the Supreme Court of Victoria.^[17] Putting to one side some of the difficulties explored in the decision about treating estoppel and waiver as distinct issues, the seven separate judgments clearly illustrate the problem of collective irrationality. Deane J and Dawson J held that the Commonwealth was estopped from relying upon the defence.^[18] Toohey J and Gaudron J held the defence to have been waived.^[19] Mason CJ, Brennan J and McHugh J held that there was neither an estoppel justifying the relief sought nor a waiver.^[20] In total, four judges to two decided against the allegation of estoppel; a different five judges to two decided against the allegation of waiver; but four judges to three decided that the Commonwealth could not, for one reason or another, plead the limitation defence. So the respondent prevailed despite having a majority against him on each issue. In tabular form:

In none of these illustrations, stylised or real, can any individual judge be criticised for irrationality. The problems of aggregation could not have been avoided simply by ‘more rationality’ on the part of any individual decision-maker. The collective irrationality arises rather as an incident of the decision-making procedure. Indeed, cases like these often motivate criticisms of the practice of delivering separate reasons for judgment. The article by Paton and Sawer, which advocated joint opinions as the solution to the problems of inconsistency, is just one example of many.^[21] Sir Rupert Cross and Professor Harris also deplored the ‘failure of judicial technique’ that is said to occur when the principle on which the court acted is ‘unascertainable’.^[22] The value in joint opinions cannot, however, come at the price of independent judgment. As one of us previously explained, the rationale for multi-member appellate courts presupposes that individual members of those courts decide independently of each other — not necessarily in isolation or without deliberation, but without compromising their ultimate responsibility to give effect to their own true view of the case.^[23] Decisional independence is not only a common law tradition, but a mathematical condition upon the tendency of multi-member courts to achieve a quality of decision-making that will exceed that of an individual judge.^[24] It is, therefore, somewhat ironic that the institutional structures which secure that condition also increase the opportunity for the inherent susceptibility to collective irrationality to materialise.

C Avoiding Collective Irrationality

Over the last decade there has emerged a rather technical literature on this general problem of judgment aggregation. It is eminently possible that this literature could come to have a broad influence on legal scholarship — not only in relation to the study of courts, but also legislatures and other group decision-making bodies. Christian List and Philip Pettit, in a recent book, present many of the field’s important results in an accessible format.^[25] A central result — proved as a mathematical theorem — is that the possibility of collective irrationality is an incident not only of majoritarian decision-making, but of any ‘aggregation function’ that satisfies three conditions:

1 the function admits all possible, rational individual judgments (‘universal domain’);

2 the function gives each individual judgment equal weight (‘anonymity’); and

3 the function operates so that the group’s decision on any given proposition is dependent only upon the individual decisions on that same proposition, and dependent in the same way for all propositions (‘systematicity’).^[26]

Put another way, there is no possible aggregation function that satisfies these three conditions and that also guarantees collective rationality.^[27]

To preclude the possibility of collective irrationality in the decision-making procedure, at least one of those three conditions would need to be relaxed. For example, it would be theoretically possible to design a decision-making procedure for multi-member appellate courts that gave determinative weight to the opinion of, say, the Chief Justice (this would be to relax ‘anonymity’, the requirement of equal weight). Under this hypothetical aggregation function, provided that the Chief Justice decides the case rationally, the group necessarily decides the case rationally because the group’s decision is, by definition, taken to be the Chief Justice’s decision. Of course, this decision-making procedure would secure collective rationality at a price, in that it would tend to defeat the larger purposes of having a multi-member court in the first place.^[28]

List and Pettit canvass different ‘escape routes’^[29] and conclude that the ‘most promising escape route’,^[30] offering the ‘best prospect’^[31] for achieving collective rationality, is to relax not ‘anonymity’, but ‘systematicity’ — that is, to allow the group to ‘prioritiz[e] some propositions over others and let[] the group attitudes on the first set of propositions determine its attitudes on the second’.^[32] There are many different ways to do this,^[33] but we will confine our consideration to two of the more obvious possibilities in the curial context. In terms of the stylised contract case, the court could decide by ‘prioritizing’ the aggregate decision on the ultimate conclusion (no liability), or it could decide by ‘prioritizing’ the aggregate decisions on the intermediate premises (contract formed and contract breached), and then letting the applicable doctrine accepted by all members of the court dictate the conclusion (liability). The first approach is a ‘conclusion-based procedure’; the second approach is a ‘premise-based procedure’.^[34] Kornhauser and Sager call the alternative approaches ‘case-by-case’ and ‘issue-by-issue’ adjudication:

It could be the ... practice of a court that each judge casts a vote in favor of the outcome of the case as she would have independently decided it, and that the court then acts on the simple majority tally that results (case-by-case adjudication). ... Alternatively, it could be the ... practice of a court that each judge casts a vote in favor of the correct resolution of each issue in the case, that each issue is then decided by a simple majority tally of these votes, and that the case itself is decided by assembling the decided issues in the manner dictated by accepted doctrine (issue-by-issue adjudication).^[35]

Each of these procedures can be conceptualised as a methodical exclusion of certain information from the decision-making process. The conclusion-based procedure excludes information about individual judgments on the premises. The premise-based procedure excludes information about individual judgments on the conclusion. We emphasise the concept of ‘excluding’ information to foreshadow some different methods of excluding information to be considered in the next Part. The exclusion of information can be represented in tabular form. For example, the stylised contract case under the conclusion-based procedure can be shown as follows:

Under the premise-based procedure, it looks like this:

The exclusion of the selected information yields a judgment profile that is not collectively irrational.

D A Dilemma

Each of these procedures dissolves the collective irrationality of the group decision. But, obviously enough, they lead to different outcomes. They lead to different outcomes precisely because of the collective irrationality that attends the aggregate judgments when the premises and conclusion are taken together. Because the outcome depends on the procedure, the court’s choice of procedure is pivotal.

This dependence was recognised in the course of oral argument before the Privy Council in Commonwealth v Bank of New South Wales (‘Bank Nationalisation Case’).^[36] The respondents had successfully challenged the Banking Act 1947 (Cth) in the High Court and contended that no appeal lay to the Privy Council, consistent with s 74 of the Constitution, without an inter se certificate. The Commonwealth argued that the High Court’s decision against it was based on s 92 of the Constitution so that its appeal was not upon an inter se question requiring a certificate. The respondents countered to the effect that the Privy Council could not properly determine the Commonwealth’s appeal without deciding, in addition to the s 92 point, inter se questions such as the scope of the enumerated heads of Commonwealth legislative power.^[37] The Commonwealth argued that s 74 of the Constitution contemplated appeals to the Privy Council not against the curial order of the High Court but on a ‘question’, or a decision of principle on which the curial order was based.^[38] Ultimately the Privy Council preferred the respondent’s submissions,^[39] but we are more interested for present purposes in some of the comments made in the course of argument.

A contemporaneous note in the Australian Law Journal records Lord Simonds as having remarked to the effect that ‘nobody with a long experience at the Bar would have failed to meet with a case where he succeeded or failed, although on every particular point the majority of the Court was against him or for him as the case may be’.^[40] Lord Simonds had in mind the scenario in which an appellant before a Bench of five judges submits that the decision below was wrong for any one of five reasons and succeeds in persuading a different one of the five judges to each of those five reasons.^[41] That appellant would, according to Lord Simonds, win the appeal unanimously, even though on each of the individual grounds of appeal there was a 4:1 majority against him. Lord Simonds apparently presupposed that a conclusion-based procedure would be used to resolve the appeal. The note continues to record, however, that Lord Normand, a Scottish member of the Privy Council, observed that under Scots law the court would not decide the matter that way, but would instead decide upon each plea in law separately — that is to say, issue-by-issue — so that the proper result in the kind of paradoxical case described by Lord Simonds can be seen to ‘depend[] solely on the procedure of the Court’.^[42] American law, like English law, appears ordinarily to follow the conclusion-based procedure.^[43] In Australia, the conclusion-based procedure has some statutory backing: a division of opinion in the Federal Court of Australia is to be resolved according to the majority opinion ‘as to the judgment to be pronounced’.^[44] In the High Court and the Supreme Court of New South Wales, the applicable statutes provide for majority opinions to prevail in the case of divided opinion on ‘any question’^[45] and ‘the decision’^[46] respectively.

The divergence between the Anglo-American procedure and the Scots procedure nicely illustrates the dilemma. But the fact that these sophisticated legal systems could choose, or at least tacitly adopt,^[47] different solutions to the problem underscores an important point: it is not obvious that either procedure is superior. This is also reflected in American scholarship on the topic: some writers argue in favour of adopting the premise-based procedure;^[48] others argue in favour of adopting the conclusion-based procedure;^[49] others still argue that the court should determine which procedure to follow by casting a ‘metavote’ on a case-by-case basis,^[50] or by adopting some more complicated ‘voting protocol’.^[51]

Current thinking in the interdisciplinary literature suggests that neither the conclusion-based nor the premise-based procedure is optimal for all circumstances. Significant work in this regard has been done on the relative performance of the procedures in correctly forming collective judgments upon interconnected factual propositions. Although this work relies on sophisticated mathematical techniques, its basic results can be reported simply enough. The question is whether one procedure is better than the other at ‘truth-tracking’ in the sense of judging a proposition to be true when it is true, and judging a proposition to be false when it is false.^[52] It turns out that the answer to the question depends not only on the size of the group, and the competence of the members of the group, but also on the nature of the premises and the conclusion, and the logical relationship between them: ‘the precise choice of an aggregation procedure has to be calibrated to the group size and problem specifics’.^[53] In some contexts, the premises may be amenable to quite ‘direct’ perception, while the conclusion will depend upon complex inferences. Then, a premise-based procedure will typically track the truth better than a conclusion-based procedure would. In other contexts, the opposite is true.

The point can be put in terms of Condorcet’s jury theorem, which holds that a group deciding by majority will be more likely than any individual member of the group to judge correctly the truth of a proposition.^[54] The theorem, as we have mentioned, requires of the individual members of the group both decisional independence and decisional competence. Decisional competence requires each member of the group to be more likely than not to judge correctly. Depending on the structure of the composite proposition being adjudicated, it might be the case that individual decision-makers satisfy the requirement of decisional competence in respect of the premises, but not the conclusion, or vice versa. The ‘key to choosing a decision procedure that tracks the truth’ is identifying the propositions on which the condition of decisional competence is met.^[55] Thus, the condition of decisional competence can be seen to be more demanding when the conclusion in question depends upon conjunctive intermediate issues. If there are two conjunctive premises, an individual member of the group will be more likely than not to reach a correct conclusion only if he or she is more than 70 per cent likely to reach a correct conclusion on each of the premises. If there are three premises, he or she must be just under 80 per cent likely. If these conditions do not hold for all members of the group, then a premise-based procedure is likely to be preferable. The condition is less demanding when the premises are disjunctive and may therefore favour a conclusion-based procedure. The only general conclusion is that ‘there may not be a “one size fits all” organizational design that is best for all group agents and all epistemic tasks’.^[56]

The difficulty choosing between the procedures is amplified in the legal context because the measure of a court’s decision-making procedure cannot be purely epistemic. This is the case in relation to fact-finding by trial courts, which do not, of course, pursue the truth at all costs.^[57] And it is even more so the case in relation to appellate courts, whose predominant function may better be conceived as a ‘volitional’^[58] task of declaring the parties’ rights by the application of legal principle. As one of us previously pointed out, however, the application of Condorcet’s jury theorem to legal decision-making ‘does not depend on the adoption of any naïve or absolute notion of what it means for a judgment to be “correct”: relative or evaluative terms such as “preferable” or “better” can be substituted’.^[59]

Nevertheless, there are other criteria, in addition to the likely ‘correctness’ of the outcomes, by which we should also compare and evaluate the decisional procedures followed by a court of law. Those other criteria reflect other important values such as: that the procedure should be fair and transparent; that the procedure should serve to quell the controversy between the parties; that it should provide reasons for decision; and that it should provide guidance on the principles to be applied in future cases.

Having regard to those broader concerns of the common law system of decision-making, there are competing intuitions for and against each of the conclusion-based and premise-based procedures. On the one hand, the intuitive appeal of the conclusion-based procedure attaches to the basic premise that judges are entrusted with the responsibility to quell controversies and to determine with finality the rights of parties. They do this by deciding cases, not by deciding issues. The result of an appeal should therefore depend upon the individual judges’ views on the case, and not necessarily on the issues. Further support for the approach may be found in the difficulties that could arise if there is disagreement about what the salient issues are or at what level of generality the issues should be stated for determination. Indeed, on one influential view, the ability for the court to reach stable conclusions, even though they may be ‘incompletely theorized’, is a virtue of the common law’s orientation towards decisional minimalism.^[60]

On the other hand, commitments to providing reasoned decisions and guidance for future cases underpin the competing intuition favouring the premise-based procedure. The idea is that, in appellate courts and especially ultimate appellate courts, reasons matter more than outcomes matter. This can be seen to be so if one views the court as an institution ‘continu[ing] in time’,^[61] independently of changes in its individual composition. Subjects of the law ought to be able to rely, both now and over time, upon the reasoning of the institution itself, and ought to be able to contest, both now and over time, the reasoning of the institution itself, independently of its changeable individual members. On this view of a court, it may be more important for the discipline of reason to apply collectively than individually.^[62]

The High Court considered these issues to some extent in Hepples,^[63] which presented a serious doctrinal paradox. The case concerned the construction of ‘extraordinarily complex’^[64] capital gains tax provisions in the Income Tax Assessment Act 1936 (Cth). Put simply, the issue was whether a payment, made to the taxpayer by his former employer in consideration for certain covenants of secrecy and non-competition, was assessable as a capital gain. The issue turned upon whether the payment came within either of s 160M(6) or s 160M(7) of the Income Tax Assessment Act 1936 (Cth). It is not necessary to understand the detail of those provisions in order to understand the doctrinal paradox. All that is relevant for present purposes is that the payment would be assessable if and only if it came within one or the other of those two subsections. Producing seven separate reasons for judgment, the members of the High Court delivered a collectively irrational decision:^[65]

The question thus confronting the High Court was, in its own words: ‘What order should [it] make when a majority would dismiss the appeal but for discrepant reasons and each of those reasons is rejected by a majority differently constituted?’^[66] Unlike in Verwayen, the Court invited the parties to make submissions on the question. The taxpayer urged a premise-based procedure to resolve the appeal in his favour.^[67] The Commissioner urged a conclusion-based procedure, arguing that ‘[t]he Court does not exist to declare legal principle but to determine disputes between parties’.^[68]

The Court unanimously decided, contrary to the Commissioner’s submission, that the taxpayer’s appeal should be allowed. It followed a premise-based procedure, in effect ‘declar[ing] legal principle’ rather than concluding the parties’ rights. It did so, however, because of the peculiar nature of the underlying proceeding, which called for a declaration of principle rather than a conclusion of the parties’ rights. The taxpayer had sought review of the Commissioner’s assessment in the Administrative Appeals Tribunal, which then stated a question of law for determination by a Full Court of the Federal Court.^[69] The nature of the appeal to the High Court from the Full Court’s answer to that question was said to be materially different from the ordinary kind of appeal:

The question arises in an appeal from a judgment which is intended to determine an issue of law arising in proceedings pending in the Administrative Appeals Tribunal; it does not arise in an appeal from a final judgment which concludes the rights of the parties or in an appeal which, if successful, would conclude the rights of the parties.^[70]

The Court held that the question that had in fact been stated was properly construed to have asked two different questions of law — one about s 160M(6) and one about s 160M(7). It held that it should answer each of those questions of law individually:

when an issue of law is determined for the purposes of proceedings pending in a court or tribunal, an order on appeal must declare the majority opinion as to the issue of law, irrespective of any conclusion as to the ultimate rights of the parties to which the reasons of the respective Justices would lead.^[71]

The order in Hepples was not only unanimous, but also led to a conclusion that was contrary to the conclusion that a majority of the judges would have reached had they decided the case individually.

The unanimous decision did, however, leave open a number of questions. First, the High Court demonstrated that it would not necessarily regard itself as bound by the formulation of the question of law acted upon in the court below. It was, therefore, fortunate that the judges agreed on the proper formulation of the questions. It is not difficult to imagine scenarios in which such agreement may prove to be more elusive. Furthermore, to say that the Court should adopt issue-by-issue adjudication ‘when an issue of law is determined for the purposes of proceedings pending in a court or tribunal’^[72] assumes that the relevant issue of law does not itself decompose into further premises upon which there may be collectively irrational judgment. The proper ‘decomposition’ of legal propositions may be contestable.

Secondly, although it distinguished appeals on stated questions of law from appeals concluding the rights of parties, the High Court did not unequivocally commit to following a conclusion-based procedure in the latter kind of appeal. It said only that such an appeal has ‘traditionally been determined’ that way — that is, ‘according to the opinion of a majority as to the order which gives effect to the legal rights of the parties irrespective of the steps by which each of the Justices in the majority reaches the conclusion’.^[73] There is much to be said for adhering to the ‘traditional’ course, which in a practical sense has worked well enough for many years. On the other hand, there is also much to be said for giving conscious attention to the choice in the cases in which it arises and, as in Hepples, making the choice on the basis of argument and published reasons so that, over time, rules and principles ‘emerge’ to guide the choice.^[74]

Thirdly, and relatedly, one of the reasons that the High Court gave for following the premise-based procedure in appeals on stated questions of law might sometimes be applicable to appeals concluding the rights of the parties. The High Court said that it would ‘work a miscarriage of justice’ to dispose of the taxpayer’s appeal according to a conclusion-based procedure because it would bind the Administrative Appeals Tribunal to hold that the payment was assessable in the instant case, whereas ‘[i]n future cases, though the facts be indistinguishable, the Tribunal would be constrained’, at least in a practical sense if not by force of the formal doctrine of precedent, to hold that the payment was not assessable.^[75] That reasoning could be applied to any collectively irrational decision, regardless of the nature of the underlying procedure. For example, the High Court’s decision in Verwayen bound the Supreme Court of Victoria not to permit the Commonwealth to plead the limitation defence. And yet, to paraphrase Hepples, though the facts be indistinguishable in future cases, the Supreme Court could be constrained (in a practical if not formal sense) to hold that those indistinguishable facts did not give rise either to an estoppel or a waiver. This consideration implicates the question of the approach that should be taken to applying collectively irrational judgments as precedents. It is to that topic we now turn.

III DOCTRINE OF PRECEDENT

The premise-based and conclusion-based procedures allow multi-member courts to manage their inherent susceptibility to collective irrationality within decisions. But what about consistency between decisions?

The common law’s commitment to consistency over time is worked out in its rules for determining the ratio decidendi of a decision, which is then binding on courts making subsequent decisions. The defining characteristic of a collectively irrational decision, as we have seen, is a ‘divergence of reason and outcome’^[76] exhibited by different majorities supporting the ultimate disposition (conclusion) and the resolution of intermediate issues (premises). How, if at all, is the ratio decidendi of a collectively irrational decision to be identified? Which of the inconsistent propositions, all of them attracting the support of a majority, bind courts making subsequent decisions? Gummow and Hayne JJ described these questions as raising ‘a matter of debate’ without finding it necessary to resolve the debate.^[77]

It is a basic precept of justice and a motivating norm of precedent-based decision-making that like cases should be treated alike.^[78] That precept provides a starting point for considering the precedential status of decisions that exhibit collective irrationality. The starting point is that the actual outcome in the case of a doctrinal paradox, whether reached by the conclusion-based procedure or by the premise-based procedure, must bind in subsequent cases. To treat like cases ‘alike’ is to treat them alike in respect of their actual outcomes. As McHugh J explained on a number of occasions, a decision not overruled remains authority at least ‘for what it decided’,^[79] in the sense that the outcome in a subsequent case cannot be different when the circumstances of that subsequent case ‘are not reasonably distinguishable from those which gave rise to the decision’.^[80]

It follows that it would be wrong to treat as binding any principle of law which, although seemingly extracted from the case, is inconsistent with the outcome of the case.^[81] If the outcome has been reached by a premise-based procedure, then the majority views on each premise will necessarily be consistent with the outcome. But if the outcome has been reached by the traditional conclusion-based procedure, then the difficulties in according precedential weight to the reasoning on the intermediate issues are more acute.

Take as an example, once again, the stylised contract case under a conclusion-based procedure. The outcome is that the defendant is not liable. But there was majority support for the proposition that a contract had been formed, and majority support for the proposition that the contract had been breached. Strictly speaking, neither one of these propositions, taken alone, is inconsistent with the outcome of no liability. But the consistency with the outcome of either one of the propositions would require the other proposition to be false, contrary to the majority holding on that other proposition. It would be wrong to treat the stylised contract case as a binding precedent either on the issue of contract formation or on the issue of breach.

The formal rules of precedent, unsurprisingly, produce that result. Two of those rules are particularly instructive on the point. The first is the rule that reasoning not necessary to an individual judge’s decision is obiter dicta. That is to say, in discerning the ratio decidendi, one is to ignore unnecessary reasoning. The second is the rule that dissenting opinions are to be ignored. Both of these rules can be seen to be justified by reference to the problem of collective irrationality. They share a functional similarity with the conclusion-based procedure and premise-based procedure in that they methodically exclude certain information from the decision-making process in the interests of collective rationality. But while the conclusion-based procedure and premise-based procedure are directed towards reaching a decision in an instant case, ignoring unnecessary reasoning and ignoring dissenting opinions are directed towards applying a decision in future cases.

A Ignoring Unnecessary Reasoning

Everyone agrees that not everything that is said in a judgment is part of its ratio decidendi: ‘The propositions which bind are often much narrower than the totality of what was said’.^[82] The ratio decidendi is, literally, only the ‘reason for deciding’.^[83] Reasons that have been expressed, but were not ‘for’ deciding the case, are unnecessary and therefore merely obiter dicta. How to distinguish ratio from obiter is, however, a topic on which commentators express diverse views and to which judges themselves have ‘devoted little explicit attention’.^[84]

Different views have been taken as to when actual reasoning can be designated as having been ‘unnecessary’ for a decision. One view is that reasoning is ratio decidendi if the judge in fact adopted it as an actual step in reasoning to the ultimate conclusion — that is to say, if it is ‘treated by the judge as a necessary step in reaching his conclusion, having regard to the line of reasoning adopted by him’.^[85]

Another view is that reasoning is unnecessary and therefore obiter dicta if it was objectively non-dispositive. Reasoning on an issue can be said to be objectively ‘non-dispositive’ if the outcome of the case would have been the same even if that particular issue had been decided the opposite way. This is not to be confused with saying that an issue might have been decided the same way for reasons narrower than those in fact relied upon.^[86] Nor is it to say that if a case is disposed of for two or more independent reasons either one of those reasons may be treated as non-dispositive:

it is impossible to treat a proposition which the Court declares to be a distinct and sufficient ground for its decision as a mere dictum, simply because there is also another ground stated upon which, standing alone, the case might have been determined.^[87]

In such a case, each of the reasons is a separate ratio.^[88]

Finkelstein J, as a member of a Full Court of the Federal Court of Australia, carefully illustrated the two competing approaches by reference to seemingly divergent English case law on the point.^[89] He preferred the first approach, concluding that the ratio of a case ‘should at least include every ruling on a point of law that is treated by the judge as a necessary step in reaching his ultimate conclusion’ and perhaps also ‘any ruling on a point of law that is put in issue by the parties ... and which the judge decides that he should resolve’.^[90] The other members of the Full Court, Black CJ and Lehane J, did not find it necessary to deal with the question.

Despite this unresolved controversy about precisely how to identify ‘unnecessary’ reasoning, the rule that unnecessary reasoning, however identified, is to be ignored in extracting the ratio decidendi of a case may have some utility when dealing with precedents that exhibit collective irrationality. Sometimes, ignoring the unnecessary reasoning will dissolve the collective irrationality for the purposes of applying the precedent to future cases.

Take the stylised contract case again as an illustration. Justice A ruled for the defendant on the ground that there had been no breach of contract. Therefore, Justice A’s reasoning on the issue of whether a contract had been formed could not have affected his or her ultimate disposition of the matter, and his or her conclusion that a contract had been formed was truly non-dispositive. On one view, the reasoning on the issue of formation was, as a prelude to the reasoning on breach, an actual step towards the conclusion, and therefore part of the ratio decidendi. On the other view, the non-dispositive quality of the reasoning on that issue renders it unnecessary to the decision, and therefore obiter dicta. Justice B’s decision turned on the issue of formation (not breach), so that his or her decision on breach was obiter dicta. Justice C’s decision depended on his or her determination of both issues, thus having two rationes decidendi.

The approach to identifying the ratio decidendi, which would exclude the ‘non-dispositive’ aspects of each judge’s reasoning would produce the following judgment profile:

This modified set of judgments does not exhibit collective irrationality — or at least it does not exhibit true irrationality in the sense we have defined it. That qualification is necessary because, there being no majority view either way on the issue of formation or the issue of breach, one might argue that there is some irrationality in accepting the conclusion that the defendant is not liable without forming any collective judgment on the premises that support that conclusion. But this is certainly a weaker form of irrationality than forming judgments on the premises that contradict the conclusion. In addition to dissolving the collective irrationality in this case, the approach also indicates why neither premise can be taken as a ratio decidendi of the decision. The case would have no ratio, other than that it stands for the conclusion that the defendant was not liable.

The approach to identifying the ratio decidendi which would not exclude reasoning in fact adopted by a judge as a step in his or her reasoning would have to include at least Justice A’s reasoning on formation, and perhaps also Justice B’s reasoning on breach, depending on precisely how the reasoning was expressed. It may not solve the problem of collective irrationality in this case. There is, however, another rule of the doctrine of precedent, which properly works in tandem with the rule excluding unnecessary reasoning. That rule is that dissenting opinions are to be ignored.

B Ignoring Dissenting Opinions

Cross and Harris, writing in 1991, appeared to regard as an open question in English law whether dissenting opinions were to be disregarded in ascertaining the ratio decidendi of a case. The learned authors wrote of a ‘requirement, if there is one, that dissenting judgments should be disregarded for the purpose of ascertaining the authoritative effect of the decisions of appellate courts’.^[91] In Australia, the requirement is supported by high authority. Mason CJ, Wilson, Dawson and Toohey JJ said that ‘it would not be proper to seek to extract a binding authority from an opinion expressed in a dissenting judgment’.^[92]

The problem of collective irrationality offers a justification for this rule. Consider again the stylised contract case, assuming it to have been decided according to the traditional conclusion-based procedure. Justice C dissented in the conclusion, but his reasoning on the intermediate issues gave the appearance of majority support for the premises that a contract was formed, and that the contract was breached. If it were permissible to take Justice C’s reasons into account, it would be permissible to embrace the irrationality of the aggregate decision by discerning binding principles in favour of both formation and breach. On the other hand, if the dissenting opinion of Justice C is excluded, then the following profile of judgments remains:

By excluding the reasoning of the dissenting judge from consideration, we arrive at an aggregation of judgments which does not exhibit collective irrationality, or at least true collective irrationality in the sense previously discussed. The exclusion of the information does not affect the majority decision on the conclusion, and it simultaneously indicates in relation to the premises that the case is one with no ratio, other than that it stands for the conclusion that the defendant was not liable. Significantly, this result obtains here even if the non-dispositive reasoning of Justice A and Justice B is not ignored.

Similar results follow in the other examples previously examined. In the disjunctive example offered by Verwayen, excluding the dissenting opinions of Mason CJ, Brennan J and McHugh J sufficiently dissolves the collective irrationality:

As in the contract case, the rule that dissenting judgments are to be ignored leads to the conclusion that Verwayen has no binding ratio on these points.^[93]

A decision with no dissenting judgments to ignore can still be collectively irrational. That was the scenario postulated by Lord Simonds in the Bank Nationalisation Case.^[94] And even if a case does have one or more dissenting opinions, ignoring those dissenting opinions does not necessarily dissolve any collective irrationality. This can be illustrated by adapting Lord Simonds’ example, which postulated five judges upholding five independent grounds of appeal, so that the fifth judge instead dissents as to the outcome, by rejecting all of the grounds of appeal. In that adapted example, a majority of 4:1 allows the appeal and, even ignoring the dissenting opinion, there is a majority of at least three judges against each ground of appeal:

Although the rule for ignoring dissenting opinions is of no assistance in such a case, ignoring unnecessary reasoning, subject to some of the subtleties discussed previously, may be an alternative solution:

Both of these formal rules of precedent can be seen to operate together to minimise the impact of collective irrationality upon the consistency over time to which a precedent-based system of decision-making is committed.

C No Ratio Decidendi

The notion of a binding ratio is crucial to the application of the doctrine of precedent. The doctrine is not directed to following a decision ‘because it is the right decision, because it is logical, because it is just, because it accords with the weight of authority, because it has been generally accepted and acted on, [or] because it secures a beneficial result to the community’.^[95] It is directed rather to when a decision ‘must be followed because it is a previous decision and for no other reason’.^[96] We intend no irony in relying upon this passage, which Heydon J recently quoted in a dissenting opinion.^[97]

The doctrine would be problematic were it to require the application of inconsistent propositions emerging from collectively irrational decisions. The typical consequence of applying its formal rules, as the examples amply illustrate, is that the collectively irrational decision in question is seen not to have a discernible ratio decidendi. The binding influence that the collectively irrational decision may exert on future cases is in that way limited to the outcome of the decision without limiting the persuasive influence that the different strands of reasoning in the decision may exert. That is because formally to exclude unnecessary reasoning and dissenting opinions for the purposes of ascertaining a binding ratio decidendi (or, as is more likely in the case of a collectively irrational decision, to ascertain that a binding ratio decidendi does not exist) is not to detract from whatever persuasive force that the unnecessary reasoning or dissenting opinions may otherwise have. The point is that even if the problem of collective irrationality properly deprives certain lines of reasoning of binding force, those lines of reasoning undoubtedly retain their intrinsic persuasive force.

The doctrine of precedent in this way accommodates the inherent susceptibility of group decision-making to collective irrationality, and clears the way for inconsistencies to be resolved over time by a subsequent working-out of principle. The value of the doctrine of precedent, as Duxbury has observed, thus ‘rests ... in its capacity simultaneously to create constraint and allow a degree of discretion’.^[98]

IV CONCLUSION

At the beginning of this article, we extracted two somewhat contradictory quotations. The first was taken from a 1949 note in the Australian Law Journal on the very point explored in this article and is representative of a familiar sort of criticism of appellate judgments that, because of their multiplicity, do not provide clear rationes decidendi.^[99] The second was taken from a 1982 article by Frank Easterbrook, now an appellate judge in the United States, which applied the insights of the then-nascent field of public choice to conclude that ‘inconsistency is inevitable’.^[100] In 2014, the now-nascent field of judgment aggregation bears out Easterbrook’s conclusion. Collective irrationality in group decision-making cannot be avoided, and can be managed only by carefully designed decision-making procedures. The common law has developed some of the procedures for managing the problem, both in individual cases and systematically over time. These procedures work by methodically excluding selected information, in different ways for different purposes, from the aggregation of individual decisions. For the purpose of deciding an instant case: conclusion-based procedures exclude information about individual judgments on the premises, while premise-based procedures exclude information about individual judgments on the conclusion. For the purpose of subsequently applying a decision as a precedent: ignoring unnecessary reasoning and ignoring dissenting opinions both exclude certain information in order to advance the basic tenet of precedential systems that like cases should be treated alike and in order to allow the rational development of legal principle over time.

^[*] Justice of the High Court of Australia.

^[†] Barrister, New South Wales Bar; Centre Fellow, Gilbert + Tobin Centre of Public Law, Faculty of Law, The University of New South Wales.

^[1] A M, ‘Ratio Decidendi in Appellate Courts’ (1949) 23 Australian Law Journal 355, 355.

^[2] Frank H Easterbrook, ‘Ways of Criticizing the Court’ (1982) 95 Harvard Law Review 802, 813.

^[3] Justice Stephen Gageler, ‘Why Write Judgments?’ [2014] SydLawRw 9; (2014) 36 Sydney Law Review 189.

^[4] Minister for Immigration and Citizenship v SZMDS (2010) 240 CLR 611, 646 [124] (Crennan and Bell JJ); Re Minister for Immigration and Multicultural Affairs; Ex parte Applicant S20/2002 [2003] HCA 30; (2003) 198 ALR 59, 61 [5] (Gleeson CJ); Minister for Immigration and Multicultural Affairs v Eshetu [1999] HCA 21; (1999) 197 CLR 611, 626 [40] (Gleeson CJ and McHugh J).

^[5] Christian List and Philip Pettit, Group Agency: The Possibility, Design, and Status of Corporate Agents (Oxford University Press, 2011) 24.

^[6] Ibid 24–5.

^[7] G W Paton and G Sawer, ‘Ratio Decidendi and Obiter Dictum in Appellate Courts’ (1947) 63 Law Quarterly Review 461, 461; see also at 465–70. See also A M, above n 1.

^[8] Paton and Sawer, above n 7, 461.

^[9] Lewis A Kornhauser and Lawrence G Sager, ‘Unpacking the Court’ (1986) 96 Yale Law Journal 82. Precursors in the Italian, French, and mathematical literatures are identified in List and Pettit, above n 5, 43. Cognate issues were considered in Easterbrook, above n 2.

^[10] Lewis A Kornhauser and Lawrence G Sager, ‘The One and the Many: Adjudication in Collegial Courts’ (1993) 81 California Law Review 1, 3.

^[11] Philip Pettit, ‘Deliberative Democracy and the Discursive Dilemma’ (2001) 11 Philosophical Issues 268, 272.

^[12] Ibid 277, 280.

^[13] Kornhauser and Sager, ‘The One and the Many’, above n 10, 10–12.

^[14] Paton and Sawer, above n 7, 465–70. Several American examples are examined in Kornhauser and Sager, ‘The One and the Many’, above n 10.

^[15] [1992] HCA 3; (1992) 173 CLR 492.

^[16] See below nn 63–74 and accompanying text.

^[17] (1990) 170 CLR 394.

^[18] Ibid 446 (Deane J), 462 (Dawson J).

^[19] Ibid 475 (Toohey J), 487 (Gaudron J).

^[20] Ibid 417 (Mason CJ), 427–8, 430 (Brennan J), 499, 504 (McHugh J).

^[21] Paton and Sawer, above n 7, 483–5.

^[22] Sir Rupert Cross and J W Harris, Precedent in English Law (Clarendon Press, 4^th ed, 1991) 93.

^[23] Justice Gageler, above n 3, 196.

^[24] Ibid 193–6.

^[25] See List and Pettit, above n 5; see especially at ch 2.

^[26] Ibid 49–50.

^[27] Ibid 50.

^[28] See generally Justice Gageler, above n 3. For a discussion of the different compromises involved in other decision-making procedures that can arise in practice, see Skulander v Willoughby City Council [2007] NSWCA 116; (2007) 73 NSWLR 44, 46–52 [42]–[76] (Mason P).

^[29] List and Pettit, above n 5, 51.

^[30] Ibid 58.

^[31] Ibid 55.

^[32] Ibid 56; see also at 58.

^[33] See ibid 56–7.

^[34] Ibid 56.

^[35] Kornhauser and Sager, ‘The One and the Many’, above n 10, 22.

^[36] [1949] HCA 47; (1949) 79 CLR 497. The report of the oral argument in Commonwealth v Bank of New South Wales [1950] AC 235, 262–3 (Sir Cyril Radcliffe KC) (during argument), 276 (Sir David Maxwell Fyfe KC) (during argument) records some related aspects.

^[37] See Bank Nationalisation Case [1949] HCA 47; (1949) 79 CLR 497, 623–4 (Lord Porter for Lords Porter, Simonds, Normand, Morton and MacDermott).

^[38] Ibid 625.

^[39] Ibid 625–6.

^[40] A M, above n 1, 355. But see Robert Megarry’s note on the topic, doubting whether the phenomenon was quite as common as Lord Simonds suggested: R E M, ‘Note’ (1950) 66 Law Quarterly Review 298, 298.

^[41] A M, above n 1, 355. Cross and Harris identify Boys v Chaplin [1971] AC 356 as an actual example with almost as extreme an outcome (five Law Lords unanimous as to the outcome, but dividing 2:2:1 on three different rationes): Cross and Harris, above n 22, 93.

^[42] A M, above n 1, 355.

^[43] Kornhauser and Sager, ‘The One and the Many’, above n 10, 20.

^[44] Federal Court of Australia Act 1976 (Cth) s 16.

^[45] Judiciary Act 1903 (Cth) s 23(2).

^[46] Supreme Court Act 1970 (NSW) s 45(1).

^[47] See Kornhauser and Sager, ‘The One and the Many’, above n 10, 20.

^[48] David Post and Steven C Salop, ‘Rowing against the Tidewater: A Theory of Voting by Multijudge Panels’ (1992) 80 Georgetown Law Journal 743; David G Post and Steven C Salop, ‘Issues and Outcomes, Guidance, and Indeterminacy: A Reply to Professor John Rogers and Others’ (1996) 49 Vanderbilt Law Review 1069.

^[49] John M Rogers, ‘“Issue Voting” by Multimember Appellate Courts: A Response to Some Radical Proposals’ (1996) 49 Vanderbilt Law Review 997.

^[50] Kornhauser and Sager, ‘The One and the Many’, above n 10, 30–3.

^[51] Jonathan Remy Nash, ‘A Context-Sensitive Voting Protocol Paradigm for Multimember Courts’ (2003) 56 Stanford Law Review 75, 146–57.

^[52] See Robert Nozick, Philosophical Explanations (Harvard University Press, 1981) 178. See also List and Pettit, above n 5, 82.

^[53] Stephan Hartmann, Gabriella Pigozzi and Jan Sprenger, ‘Reliable Methods of Judgement Aggregation’ (2010) 20 Journal of Logic and Computation 603, 613.

^[54] See generally Justice Gageler, above n 3, 193–6.

^[55] Christian List, ‘The Discursive Dilemma and Public Reason’ (2006) 116 Ethics 362, 386.

^[56] List and Pettit, above n 5, 102; see generally at 92–103.

^[57] See J J Spigelman, ‘Truth and the Law’ in Justice Nye Perram and Justice Rachel Pepper (eds), The Byers Lectures 2000–2012 (Federation Press, 2012) 232.

^[58] See Chief Justice R S French, ‘The Executive Power’ (Speech delivered at the Inaugural George Winterton Lecture, Sydney Law School, The University of Sydney, 18 February 2010).

^[59] Justice Gageler, above n 3, 194.

^[60] See Cass R Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court (Harvard University Press, 1999); Cass R Sunstein, ‘Foreword: Leaving Things Undecided’ (1996) 110 Harvard Law Review 4; Cass R Sunstein, ‘Incompletely Theorized Agreements’ (1995) 108 Harvard Law Review 1733. See also Justice J D Heydon, ‘How Far Can Trial Courts and Intermediate Appellate Courts Develop the Law?’ (2009) 9 Oxford University Commonwealth Law Journal 1, 36–8.

^[61] Pettit, above n 11, 285.

^[62] See generally ibid, considering the issue in the context of republicanism and deliberative democracy.

^[63] [1992] HCA 3; (1992) 173 CLR 492.

^[64] Ibid 497 (Mason CJ).

^[65] See ibid 497–8 (Mason CJ), 509 (Brennan J), 518 (Deane J), 518–19 (Dawson J), 527 (Toohey J), 528 (Gaudron J), 544, 549 (McHugh J).

^[66] Ibid 550 (Mason CJ, Brennan, Deane, Dawson, Toohey, Gaudron and McHugh JJ).

^[67] Ibid 549 (D H Bloom QC) (during argument).

^[68] Ibid 550 (B J Shaw QC) (during argument).

^[69] See Administrative Appeals Tribunal Act 1975 (Cth) s 45.

^[70] Hepples [1992] HCA 3; (1992) 173 CLR 492, 550 (Mason CJ, Brennan, Deane, Dawson, Toohey, Gaudron and McHugh JJ).

^[71] Ibid 551.

^[72] Ibid.

^[73] Ibid.

^[74] Cf Thomas v Mowbray [2007] HCA 33; (2007) 233 CLR 307, 351 [91] (Gummow and Crennan JJ), quoting Leslie Zines, The High Court and the Constitution (Butterworths, 4^th ed, 1997) 195.

^[75] Hepples [1992] HCA 3; (1992) 173 CLR 492, 553 (Mason CJ, Brennan, Deane, Dawson, Toohey, Gaudron and McHugh JJ).

^[76] Kornhauser and Sager, ‘The One and the Many’, above n 10, 33.

^[77] Jones v Bartlett (2000) 205 CLR 166, 224–5 [203]–[207].

^[78] See generally Justice Heydon, above n 60, 9–13.

^[79] Re Tyler; Ex parte Foley [1994] HCA 25; (1994) 181 CLR 18, 37.

^[80] Scruttons Ltd v Midland Silicones Ltd [1961] UKHL 4; [1962] AC 446, 479 (Lord Reid), quoted in Re Tyler; Ex parte Foley [1994] HCA 25; (1994) 181 CLR 18, 37 (McHugh J). See also Shaw v Minister for Immigration and Multicultural Affairs [2003] HCA 72; (2003) 218 CLR 28, 47–8 [50] (McHugh J); D’Orta-Ekenaike v Victoria Legal Aid [2005] HCA 12; (2005) 223 CLR 1, 46–7 [133] (McHugh J).

^[81] See, eg, the application of Hepples [1992] HCA 3; (1992) 173 CLR 492 in Paykel v Commissioner of Taxation (1994) 49 FCR 41, 50–1 (Heerey J) (‘Paykel’). Paykel is discussed in Alastair MacAdam and John Pyke, Judicial Reasoning and the Doctrine of Precedent in Australia (Butterworths, 1998) 212 [10.38].

^[82] Justice Heydon, above n 60, 5.

^[83] G W Paton and David P Derham (eds), A Textbook of Jurisprudence (Oxford University Press, 4^th ed, 1972) 210. See also Cross and Harris, above n 22, 39.

^[84] Justice Heydon, above n 60, 6.

^[85] Cross and Harris, above n 22, 72.

^[86] Cf XYZ v Commonwealth [2006] HCA 25; (2006) 227 CLR 532, 547 [34] (Gummow, Hayne and Crennan JJ), 558 [71] (Kirby J).

^[87] Commissioners of Taxation (NSW) v Palmer [1907] UKLawRpAC 3; [1907] AC 179, 184 (Lord Macnaghten for Lords Macnaghten, Davey, Robertson and Atkinson).

^[88] See, eg, Ex parte King; Re Blackley [1938] NSWStRp 26; (1938) 38 SR (NSW) 483, 490 (Jordan CJ); Paton and Sawer, above n 7, 470.

^[89] Bristol-Myers Squibb Co v F H Faulding & Co Ltd [2000] FCA 316; (2000) 97 FCR 524, 570–2 [148]–[156].

^[90] Ibid 573 [160].

^[91] Cross and Harris, above n 22, 91 (emphasis added).

^[92] Federation Insurance Ltd v Wasson [1987] HCA 34; (1987) 163 CLR 303, 314, citing Dickenson’s Arcade Pty Ltd v Tasmania (1974) 130 CLR 177, 188 (Barwick CJ) and Great Western Railway Co v Owners of SS Mostyn; The Mostyn [1928] AC 57, 73–4 (Viscount Dunedin). Cf Jones v Bartlett (2000) 205 CLR 166, 224–5 [206] (Gummow and Hayne JJ).

^[93] See China Ocean Shipping Co Ltd v P S Chellaram & Co Ltd (1990) 28 NSWLR 354, 379–80 (Kirby P).

^[94] See above nn 40–2 and accompanying text.

^[95] Max Radin, ‘Case Law and Stare Decisis: Concerning Präjudizienrecht in Amerika’ (1933) 33 Columbia Law Review 199, 200.

^[96] Ibid. See also Jones v DPP [1962] AC 635, 711 (Lord Devlin).

^[97] Australian Crime Commission v Stoddart [2011] HCA 47; (2011) 244 CLR 554, 574–5 [57].

^[98] Neil Duxbury, The Nature and Authority of Precedent (Cambridge University Press, 2008) 183.

^[99] See A M, above n 1.

^[100] Easterbrook, above n 2, 813.