Appellant’s opening brief cites Rice v. Chater, No. 95-35604, 1996 WL 583605 (9th Cir. Oct.9, 1996). Rice is an unpub lished disposition, not reported in the Federal Reporter except as a one-line entry in a long table of cases. See Decisions Without Published Opinions, 98 F.3d 1345 , 1346 tbl. (9th Cir.1996). The full text of the disposition can be obtained from our clerk’s office, and is available on Westlaw® and LEXIS®. However, it is marked with the following notice: “This disposition is not appropriate for publication and may not be cited to or by the courts of this circuit except as provided by 9th Cir.R. 36-3.” Our local rules are to the same effect: “Unpublished dispositions and orders of this Court are not binding precedent ... [and generally] may not be cited to or by the courts of this circuit .... ” 9th Cir. R. 36-3.

We ordered counsel to show cause as to why he should not be disciplined for violating Ninth Circuit Rule 36-3. Counsel responds by arguing that Rule 36-3 may be unconstitutional. He relies on the Eighth Circuit’s opinion in Anastasoff v. United States, 223 F.3d 898 , vacated as moot on reh’g en banc, 235 F.3d 1054 (8th Cir.2000). Anastasoff, while vacated, continues to have persuasive force. See, e.g., Williams v. Dallas Area Rapid Transit, 256 F.3d 260 (5th Cir.2001) (Smith, J., dissenting from denial of reh’g en banc). It may seduce members of our bar into violating our Rule 36-3 under the mistaken impression that it is unconstitutional. We write to lay these speculations to rest.

A. Anastasoff held that Eighth Circuit Rule 28A(i), which provides that unpublished dispositions are not precedential— and hence not binding on future panels of that court — violates Article III of the Constitution. See 223 F.3d at 899 . According to Anastasoff, exercise of the “judicial Power” precludes federal courts from making rulings that are not binding in future eases. Or, to put it differently, federal judges are not merely required to follow the law, they are also required to make law in every case. To do otherwise, Anastasoff argues, would invite judicial tyranny by freeing courts from the doctrine of precedent: “ ‘A more alarming doctrine could not be promulgated by any American court, than that it was at liberty to disregard all former rules and decisions, and to decide for itself, without reference to the settled course of antecedent principles.’ ” Id. at 904 (quoting Joseph Story, Commentaries on the Constitution of the United States § 377 (1833)).

We believe that Anastasoff overstates the case. Rules that empower courts of appeals to issue nonprecedential decisions do not cut those courts free from all legal rules and precedents; if they did, we might find cause for alarm. But such rules have a much more limited effect: They allow panels of the courts of appeals to determine whether future panels, as well as judges of the inferior courts of the circuit, will be bound by particular rulings. This is hardly the same as turning our back on all precedents, or on the concept of precedent altogether. Rather, it is an effort to deal with precedent in the context of a modern legal system, which has evolved considerably since the early days of common law, and even since the time the Constitution was adopted.

The only constitutional provision on which Anastasoff relies is that portion of Article III that vests the “judicial Power” of the United States in the federal courts. U.S. Const, art. Ill, § 1, cl. 1. Anastasoff may be the first case in the history of the Republic to hold that the phrase “judicial Power” encompasses a specific command that limits the power of the federal courts. There are, of course, other provisions of Article III that have received judicial enforcement, such as the requirement that the courts rule only in “Cases” or “Controversies,” see, e.g., Lujan v. Defenders of Wildlife, 504 U.S. 555, 559 , 112 S.Ct. 2130 , 119 L.Ed.2d 351 (1992), and that the pay of federal judges not be diminished during their good behavior. See, e.g., United States v. Hatter, 532 U.S. 557 , -, 121 S.Ct. 1782, 1790-91 , 149 L.Ed.2d 820 (2001). The judicial power clause, by contrast, has never before been thought to encompass a constitutional limitation on how courts conduct their business.

There are many practices that are common or even universal in the federal courts. Some are set by statute, such as the courts’ basic organization. See, e.g., 28 U.S.C. § 43 (creating a court of appeals for each circuit); 28 U.S.C. § 127 (dividing Virginia into two judicial districts); 28 U.S.C. § 2101 (setting time for direct appeals to the Supreme Court and for applications to the Supreme Court for writs of certiorari). See generally David McGowan, Judicial Writing and the Ethics of the Judicial Office, 14 Geo. J. Legal Ethics 509 , 509-10 (2001). Others are the result of tradition, some dating from the days of the common law, others of more recent origin. Among them are the practices of issuing written opinions that speak for the court rather than for individual judges, adherence to the adversarial (rather than inquisitorial) model of developing cases, limits on the exercise of equitable relief, hearing appeals with panels of three or more judges and countless others that are so much a part of the way we do business that few would think to question them.' While well established, it is unclear that any of these practices have a constitutional foundation; indeed, Hart (no relation so far as we know), in his famous Dialogue, concluded that Congress could abolish the inferior federal courts altogether. See Henry M. Hart, Jr., The Power of Congress to Limit the Jurisdiction of Federal Courts: An Exercise in Dialectic, 66 Harv. L.Rev. 1362, 1363-64 (1953). While the greater power does not always include the lesser, the Dialogue does suggest that much .of what the federal courts do could be modified or eliminated without offending the Constitution.

Anastasoff focused on one aspect of the way federal courts do business — the way they issue opinions — and held that they are subject to a constitutional limitation derived from the Framers’ conception of what it means to exercise the judicial power. Given that no other aspect of the way courts exercise their power has ever been held subject to this limitation, we question whether the “judicial Power” clause contains any limitation at all, separate from the specific limitations of Article III and other parts of the Constitution. The more plausible view is that when the federal courts rule on cases or controversies assigned to them by Congress, comply with due process, accord trial by jury where commanded by the Seventh Amendment and generally comply with the specific constitutional commands applicable to judicial proceedings, they have ipso facto exercised the judicial power of the United States. In other words, the term “judicial Power” in Article III is more likely descriptive than prescriptive.

If we nevertheless were to accept Anas-tasoff’s premise that the phrase “judicial Power” contains limitations separate from those contained elsewhere in the Constitution, we should exercise considerable caution in recognizing those limitations, lest we freeze the law into the mold cast in the eighteenth century. The law has changed in many respects since the time of the Framing, some superficial, others quite fundamental. For example, as Professor William Nelson has convincingly demonstrated, colonial juries “usually possessed the power to find both law and fact in the cases in which they sat,” and were not bound to follow the instructions given to them by judges. See William E. Nelson, Marbury v. Madison: The Origins and Legacy of Judicial Review 16-17 (2000). Today, of course, we would consider it unfair — probably unconstitutional — to allow juries to make up the law as they go along.

Another example: At the time ■ of the Framing, and for some time thereafter, the practice that prevailed both in the United States and England was for judges of appellate courts to express separate opinions, rather than speak with a single (or at least majority) voice. The practice changed around the turn of the nineteenth century, under the leadership of Chief Justice Marshall. See George L. Haskins & Herbert A. Johnson, Foundations of Power: John Marshall, 1801-15, in 2 The Oliver Wendell Holmes Devise: History of the Supreme Court of the United States 382-89 (Paul A. Freund ed., 1981).

And yet another example: At the time of the Framing, and for some time thereafter, it was considered entirely appropriate for a judge to participate in the appeal of his own decision; indeed, before the creation of the Circuit Courts of Appeals, appeals from district court decisions were often taken to a panel consisting of a Supreme Court Justice riding circuit, and the district judge from whom the decision was taken. Act of March 2, 1793, ch. 22, § 1, 1 Stat. 333 ; see also Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure § 3504 (2d ed.1984). Today, of course, it is widely recognized that a judge may not hear the appeal from his own decision. There are doubtless many more such examples.

One danger of giving constitutional status to practices that existed at common law, but have changed over time, is that it tends to freeze certain aspects of the law into place, even as other aspects change significantly. See note 6 supra. This is a particularly dangerous practice when the constitutional rule in question is not explicitly written into the Constitution, but rather is discovered for the first time in a vague, two-centuries-old provision. The risk that this will allow judges to pick and choose those ancient practices they find salutary as a matter of policy, and give them constitutional status, is manifest. Compare Richard S. Arnold, Unpublished Opinions: A Comment, 1 J.App. Prac. & Process 219 (1999) (suggesting that all opinions be published and given prece-dential value), with Anastasoff, 223 F.3d 898 (holding that the Eighth Circuit’s rule barring citation to unpublished opinions violates Article III). Thus, in order to follow the path forged by Anastasoff, we would have to be convinced that the practice in question was one the Framers considered so integral and well-understood that they did not have to bother stating it, even though they spelled out many other limitations in considerable detail. Specifically, to adopt Anastasoff’s position, we would have to be satisfied that the Framers had a very rigid conception of precedent, namely that all judicial decisions necessarily served as binding authority on later courts.

This is, in fact, a much more rigid view of precedent than we hold today. As we explain below, most decisions of the federal courts are not viewed as binding precedent. No trial court decisions are; almost four-fifths of the merits decisions of courts of appeals-are not. See p. 1177 infra. To be sure, Anastasoff challenges the latter practice. We find it significant, however, that the practice has been in place for a long time, yet no case prior to Anastasoff has challenged its constitutional legitimacy. The overwhelming consensus in the legal community has been that having appellate courts issue nonprecedential decisions is not inconsistent with the exercise of the judicial power.

To accept Anastasoff’s argument, we would have to conclude that the generation of the Framers had a much stronger view of precedent than we do. In fact, as we explain below, our concept of precedent today is far stricter than that which prevailed at the time of the Framing. The Constitution does not contain an express prohibition against issuing nonprecedential opinions because the Framers would have seen nothing wrong with the practice.

B. Modern federal courts are the successors of the English courts that developed the common law, but they are in many ways quite different, including how they understand the concept of precedent. Common law judges did not make law as we understand that concept; rather, they “found” the law with the help of earlier cases that had considered similar matters. An opinion was evidence of what the law is, but it was not an independent source of law. See Theodore F.T. Plucknett, A Concise History of the Common Law 343-44 (5th ed.1956). The law was seen as something that had an existence independent of what judges said: “a miraculous something made by nobody ... and merely declared from time to time by the judges.” 2 John Austin, Lectures on Jurisprudence or The Philosophy of Positive Law 655 (4th ed. 1873) (emphasis omitted). Opinions were merely judges’ efforts to ascertain the law, much like scientific experiments were efforts to ascertain natural laws. If an eighteenth-century judge believed that a prior case was wrongly decided, he could say that the prior judge had erred in his attempt to discern the law. See Bole v. Horton, 124 Eng. Rep. 1113, 1124 (C.P.1673). Neither judges nor lawyers understood precedent to be binding in Anasta-soff’s strict sense.

One impediment to establishing a system of strict binding precedent was the absence at common law of a distinct hierarchy of courts. See Plucknett, supra, at 350. Only towards the end of the nineteenth century, after England had reorganized its courts, was the position of the House of Lords at the head of its judicial hierarchy confirmed. Before that, there was no single high court that could definitively say what the law was. Thus, as late as the middle of the nineteenth century, an English judge might ignore decisions of the House of Lords, and the Exchequer and Queen’s Bench held different views on the same point as late as 1842. See id. at 350. Common law judges looked to earlier cases only as examples of policy or practice, and a single case was generally not binding authority. Eighteenth-century judges did not feel bound to follow most decisions that might lead to inconvenient results, and judges would even blame reporters for cases they disliked. See Pluck-nett, supra, at 349.

The idea that judges declared rather than made the law remained firmly entrenched in English jurisprudence until the early nineteenth century. David M. Walker, The Oxford Companion to Law 977 (1980). Blackstone, who wrote his Commentaries only two decades before the Constitutional Convention and was greatly respected and followed by the generation of the Framers, noted that “the ‘law,’ and the ‘opinion of the judge’ are not ... one and the same thing; since it sometimes may happen that the judge may mistake the law”; in such cases, the precedent simply “was not law.” 1 William Blackstone, Commentaries *70-71 (1765).

For centuries, the most important sources of law were not judicial opinions themselves, but treatises that restated the law, such as the commentaries of Coke and Blackstone. Because published opinions were relatively few, lawyers and judges relied on commentators’ synthesis of decisions rather than the verbatim text of opinions.

Case reporters were entrepreneurs who scribbled down jury charges as they were delivered by judges, then printed and sold them. Or, reporters might cobble together case reports from secondhand sources and notes found in estates, sometimes years after the cases were decided. See Robert C. Berring, Legal Research and Legal Concepts: Where Form Molds Substance, 75 Cal. L.Rev. 15, 18-19 (1987). For example, Heydon’s Case was decided in 1584, but Lord Coke did not publish his account of it until 1602. See Allen Dillard Boyer, “Understanding, Authority, and Will”: Sir Edward Coke and the Elizabethan Origins of Judicial Review, 39 B.C. L.Rev. 43, 79 (1997). Not surprisingly, case reports often contradicted each other in describing the reasoning, and even the names, of particular cases. See Berring, supra, at 18. The value of case reports turned not on the accuracy of the report but on the acuity of their authors. See id. at 18-19.

Coke’s intellectual reputation made him the most valued, and the most famous; of the private reporters. His reports were not verbatim transcriptions of what the judges actually said, but vehicles for Coke’s own jurisprudential and political agenda. See Boyer, supra, at 80 (“In the name of judicial reason, Coke was willing to rewrite the law.... In 1602, his chief way of shaping the law was in the way he reported it.”). Like other reporters, Coke often distorted the language and meaning of prior decisions that were inconsistent with what he considered the correct legal principle. See Harold J. Berman & Charles J. Reid, Jr., The Transformation of English Legal Science: From Hale to Blackstone, 45 Emory L.J. 437 , 447 (1996). “There was no clear boundary in his mind between what a case said and what he thought it ought to say, between the reasons which actually prompted the decision, and the elaborate commentary which he could easily weave around any question.” Plueknett, supra, at 281. Contrary to Anastasoff’s view, it was emphatically not the case that all decisions of common law courts were treated as precedent binding on future, courts unless distinguished or rejected. Rather, case reporters routinely suppressed or altered cases they considered wrongly decided. Indeed, sorting out the decisions that deserved reporting from those that did not became one of their primary functions.

A survey of the legal landscape as it might have been viewed by the generation of the Framers casts serious doubt on the proposition — so readily accepted by Anas- tosojff — that the Framers viewed precedent in the rigid form that we view it today. Indeed, it is unclear that the Framers would have considered our view of precedent desirable. The common law, at its core, was a reflection of custom, and custom had a built-in flexibility that allowed it to change with circumstance. Thus, “when Lord Mansfield incorporated the custom of merchants into the common law, it was a living flexible custom, responding to the growth and change of mercantile habits.” Plueknett, supra, at 350. Embodying that custom into a binding decision raised the danger of ossifying the custom: “[I]f perchance a court has given a decision on a point of that custom, it loses for ever its flexibility and is fixed by the rule of precedent at the point where the court touched it.” Id. It is entirely possible that lawyers of the eighteenth century, had they been confronted with the regime of rigid precedent that is in common use today, would have reacted with alarm.

The modern concept of binding precedent — where a single opinion sets the course on a particular point of law and must be followed by courts at the same level and lower within a pyramidal judicial hierarchy — came about only gradually over the nineteenth and early twentieth centuries. Lawyers began to believe that judges made, not found, the law. This coincided with monumental improvements in the collection and reporting of case authorities. As the concept of law changed and a more comprehensive reporting system began to take hold, it became possible for judicial decisions to serve as binding authority.

Early American reporters resembled their English ancestors — disorganized and meager — but the character of the report ing process began to change, after the Constitution was adopted, with the emergence of official reporters in the late eighteenth century and the early nineteenth century. See Berring, supra, at 20-21. And, later in the nineteenth century, the West Company began to publish standardized case reporters, which were both accurate and comprehensive, making “it possible to publish in written form all of the decisions of courts.” Id. at 21. Case reports grew thicker, and the weight of precedent began to increase — weight, that is, in terms of volume.

The more cases were reported, the harder became the task of searching for relevant decisions. At common law, circuit-riding judges often decided cases without referring to any reporters at all, see Fentum v. Pocock, 5 Taunt. 192, 195, 128 Eng. Rep. 660, 662 (C.P.1813) (Mansfield, C.J.) (“It [was] utterly impossible for any Judge, whatever his learning and abilities may be, to decide at once rightly upon every point which [came] before him at Nisi Prius .... ”), and reporters simply left out decisions they considered wrong or those that merely repeated what had come before. Sir Francis Bacon recommended that cases “merely of iteration and repetition” be omitted from the ease reports altogether, and Coke warned judges against reporting all of their decisions for fear of weighing down the law. See Kirt Shuld-berg, Digital Influence: Technology and Unpublished Opinions in the Federal Courts of Appeals, 85 Cal. L.Rev. 541, 545 & n. 8 (1997). Indeed, the English opinion-reporting system has never published, and does not today publish, every opinion of English appellate courts, even though the total number of opinions issued each year in both the English Court of Appeal and House of Lords combined is little more than 1000 — less than a quarter of the number of dispositions issued annually by the Ninth Circuit in recent years, see note 37 infra. Robert J. Martineau, Appellate Justice in England and the United States: A Comparative Analysis 107, 150 (1990); Robert J. Martineau, Restrictions on Publication and Citation of Judicial Opinions: A Reassessment, 28 U. Mich. J.L. Ref. 119,136 (1995).

Federal courts today do follow some common law traditions. When ruling on a novel issue of law, they will generally consider how other courts have ruled on the same issue. This consideration will not be limited to courts at the same or higher level, or even to courts within the same system of sovereignty. Federal courts of appeals will cite decisions of district courts, even those in other circuits; the Supreme Court may cite the decisions of the inferior courts, see, e.g., City of Richmond v. J.A. Croson Co., 488 U.S. 469, 491 , 109 S.Ct. 706 , 102 L.Ed.2d 854 (1989) (citing Associated Gen. Contractors of Cal. v. City & County of San Francisco, 813 F.2d 922 , 929 (9th Cir.1987)), or those of the state courts, see, e.g., Lujan v. G & G Fire Sprinklers, Inc., 532 U.S. 189 , 121 S.Ct. 1446, 1452 , 149 L.Ed.2d 391 (2001) (citing J & K Painting Co. v. Bradshaw, 45 Cal.App.4th 1394, 1402 , 53 Cal.Rptr.2d 496 (Cal.Ct.App.1996)). It is not unusual to cite the decision of courts in foreign jurisdictions, so long as they speak to a matter relevant to the issue before us. See, e.g., Mozes v. Mozes, 239 F.3d 1067, 1071 (9th Cir.2001). The process even extends to non-case authorities, such as treatises and law review articles. See id. at 1071 & n. 7.

Citing a precedent is, of course, not the same as following it; “respectfully disagree” within five words of “learned colleagues” is almost a cliche. After carefully considering and digesting the views of other courts and commentators — often giving conflicting guidance on a novel legal issue — courts will then proceed to follow one line of authority or another, or sometimes strike out in a completely different direction. While we would consider it bad form to ignore contrary authority by failing even to acknowledge its existence, it is well understood that — in the absence of binding precedent — courts may forge a different path than suggested by prior authorities that have considered the issue. So long as the earlier authority is acknowledged and considered, courts are deemed to have complied with their common law responsibilities.

But precedent also serves a very different function in the federal courts today, one related to the horizontal and vertical organization of those courts. See John Harrison, The Power of Congress Over The Rules of Precedent, 50 Duke L.J. 503 (2000). A district judge may not respectfully (or disrespectfully) disagree with his learned colleagues on his own court of appeals who have ruled on a controlling legal issue, or with Supreme Court Justices writing for a majority of the Court. Binding authority within this regime cannot be considered and cast aside; it is not merely evidence of what the law is. Rather, caselaw on point is the law. If a court must decide an issue governed by a prior opinion that constitutes binding authority, the later court is bound to reach the same result, even if it considers the rule unwise or incorrect. Binding authority must be followed unless and until overruled by a body competent to do so.

In determining whether it is bound by an earlier decision, a court considers not merely the “reason and spirit of cases” but also “the letter of particular precedents.” Fisher v. Prince, 97 Eng. Rep. 876, 876 (K.B.1762). This includes not only the rule announced, but also the facts giving rise to the dispute, other rules considered and rejected and the views expressed in response to any dissent or concurrence. Thus, when crafting binding authority, the precise language employed is often crucial to the contours and scope of the rule announced.

Obviously, binding authority is very powerful medicine. A decision of the Supreme Court will control that corner of the law. unless and until the Supreme Court itself overrules or modifies it. Judges of the inferior courts may voice their criticisms, but follow it they must. See, e.g., Ortega v. United States, 861 F.2d 600 , 603 & n. 4 (9th Cir.1988) (“This case is squarely controlled by the Supreme Court’s recent decision.... [We] agree[ ] with the dissent that [appellant] deserves better treatment from our Government. Unfortunately, legal precedent deprives us of discretion to do equity.”). The same is true as to circuit authority, although it usually covers a much smaller geographic area. Circuit law, a concept wholly unknown at the time of the Framing, see Danny J. Boggs & Brian P. Brooks, Unpublished Opinions & the Nature of Precedent, 4 Green Bag 2d 17 , 22 (2000), binds all courts within a particular circuit, including the court of appeals itself. Thus, the first panel to consider an issue sets the law not only for all the inferior courts in the circuit, but also future panels of the court of appeals.

Once a panel resolves an issue in a precedential opinion, the matter is deemed resolved, unless overruled by the court itself sitting en banc, or by the Supreme Court. As Anastasojf itself states, a later three-judge panel considering a case that is controlled by the rule announced in an earlier panel’s opinion has no choice but to apply the earlier-adopted rule; it may not any more disregard the earlier panel’s opinion than it may disregard a ruling of the Supreme Court. Anastasoff, 223 F.3d at 904 ; see also Santamaria v. Horsley, 110 F.3d 1352, 1355 (9th Cir.1997) (“It is settled law that one three-judge panel of this court cannot ordinarily reconsider or overrule the decision of a prior panel.”), rev’d, 133 F.3d 1242 (9th Cir.) (en banc), amended by 138 F.3d 1280 (9th Cir.), cert. denied, 525 U.S. 823 -24, 119 S.Ct. 68 , 142 L.Ed.2d 53 (1998); Montesano v. Seafirst Commercial Corp., 818 F.2d 423, 425-26 (5th Cir.1987) (A “purpose of institutional orderliness [is served] by our insistence that, in the absence of intervening Supreme Court precedent, one panel cannot overturn another panel, regardless of how wrong the earlier panel decision may seem to be.”)• Designating an opinion as binding circuit authority is a weighty decision that cannot be taken lightly, because its effects are not easily reversed. Whether done by the Supreme Court or the court of appeals through its “unwieldy” and time-consuming en banc procedures, Richard A. Posner, The Federal Courts: Crisis and Reform 101 (1985), overruling such authority requires a substantial amount of courts’ time and attention — two commodities already in very short supply.

Controlling authority has much in common with persuasive authority. Using the techniques developed at common law, a court confronted with apparently controlling authority must parse the precedent in light of the facts presented and the rule announced. Insofar as there may be factual differences between the current case and the earlier one, the court must determine whether those differences are material to the application of the rule or allow the precedent to be distinguished on a principled basis. Courts occasionally must reconcile seemingly inconsistent precedents and determine whether the current case is closer to one or the other of the earlier opinions. See, e.g., Mont. Chamber of Commerce v. Argeribright, 226 F.3d 1049 , 1057 (9th Cir.2000).

But there are also very important differences between controlling and persuasive authority. As noted, one of these is that, if a controlling precedent is determined to be on point, it must be followed. Another important distinction concerns the scope of controlling authority. Thus, an opinion of our court is binding within our circuit, not elsewhere in the country. The courts of appeals, and even the lower courts of other circuits, may decline to follow the rule we announce — and often do. This ability to develop different interpretations of the law among the circuits is considered a strength of our system. It allows experimentation with different approaches to the same legal problem, so that when the Supreme Court eventually reviews the issue it has the benefit of “percolation” within the lower courts. See Samuel Estreicher & John E. Sexton, A Managerial Theory of the Supreme Court’s Responsibilities: An Empirical Study, 59 N.Y.U. L.Rev. 681, 716 (1984). Indeed, the Supreme Court sometimes chooses not to grant certiorari on an issue, even though it might deserve definitive resolution, so it will have the benefit of a variety of views from the inferior courts before it chooses an approach to a legal problem. See McCray v. New York, 461 U.S. 961, 963 , 103 S.Ct. 2438 , 77 L.Ed.2d 1322 (1983) (Stevens, J., respecting denial of petitions for writs of certiorari) (“[I]t is a sound exercise of discretion for the Court to allow [other courts] to serve as laboratories in which the issue receives further study before it is addressed by this Court.”).

The various rules pertaining to the development and application of binding authority do not reflect the developments of the English common law. They reflect, rather, the organization and structure of the federal courts and certain policy judgments about the effective administration of justice. See Payne v. Tennessee, 501 U.S. 808, 828 , 111 S.Ct. 2597 , 115 L.Ed.2d 720 (1991) (stare decisis is a “principle of policy,” and “not an inexorable command”); see, e.g., Textile Mills Secs. Corp. v. Comm’r, 314 U.S. 326, 334-35 , 62 S.Ct. 272 , 86 L.Ed. 249 (1941) (en banc rehearing “makes for more effective judicial administration”). Circuit boundaries are set by statute and can be changed by statute. When that happens, and a new circuit is created, it starts without any circuit law and must make an affirmative decision whether to create its circuit law from scratch or to adopt the law of another circuit — generally the circuit from which it was carved — as its own. Compare Bonner v. City of Prichard, 661 F.2d 1206, 1209 (11th Cir.1981) (en banc) (adopting as binding precedent all decisions issued by the former Fifth Circuit before its split into the Fifth and Eleventh Circuits), and South Corp. v. United States, 690 F.2d 1368, 1370-71 (Fed.Cir.1982) (en banc) (adopting as binding precedent all decisions of the Federal Circuit’s predecessor courts, the Court of Claims and the Court of Customs and Patent Appeals), with Estate of McMorris v. Comm’r, 243 F.3d 1254, 1258 (10th Cir.2001) (“[W]e have never held that the decisions of our predecessor circuit [the former Eighth Circuit] are controlling in this court.”). The decision whether to adopt wholesale the circuit law of another court is a matter of judicial policy, not a constitutional command.

How binding authority is overruled is another question that was resolved by trial and error with due regard to principles of sound judicial administration. Early in the last century, when the courts of appeals first grew beyond three judges, the question arose whether the courts could sit en banc to rehear cases already decided by a three-judge panel. The lower courts disagreed, but in Textile Mills Securities Corporation v. Commissioner, the Supreme Court sustained the authority of the courts of appeals to sit en banc. Textile Mills Secs. Corp. v. Comm’r, 314 U.S. 326, 335 , 62 S.Ct. 272 , 86 L.Ed. 249 (1943) (“Conflicts within a circuit will be avoided. Finality of decision in the circuit courts of appeal will be promoted. Those considerations are especially important in view of the fact that in our federal judicial system these courts are the courts of last resort in the run of ordinary cases.”)• En banc rehearing would give all active judges an opportunity to hear a case “[wjhere ... there is a difference in view among the judges upon a question of fundamental importance, and especially in a ease where two of the three judges sitting in a case may have a view contrary to that of the other ... judges of the court.” Comm’r v. Textile Mills Secs. Corp., 117 F.2d 62, 70 (3d Cir.1940), aff'd, 314 U.S. 326 , 62 S.Ct. 272 , 86 L.Ed. 249 (1943). Congress codified the Textile Mills decision just five years later in 28 U.S.C. § 46 (c), leaving the courts of appeals “free to devise [their] own administrative machinery to provide the means whereby a majority may order such a hearing.” W. Pac. R.R. v. W. Pac. R.R., 345 U.S. 247, 250 , 73 S.Ct. 656 , 97 L.Ed. 986 (1953).

That the binding authority principle applies only to appellate decisions, and not to trial court decisions, is yet another policy choice. There is nothing inevitable about this; the rule could just as easily operate so that the first district judge to decide an issue within a district, or even within a circuit, would bind all similarly situated district judges, but it does not. The very existence of the binding authority principle is not inevitable. The federal courts could operate, though much less efficiently, if judges of inferior courts had discretion to consider the opinions of higher courts, but “respectfully disagree” with them for good and sufficient reasons.

Ill

While we agree with Anastasoff that the principle of precedent was well established in the common law courts by the time Article III of the Constitution was written, we do not agree that it was known and applied in the strict sense in which we apply binding authority today. It may be true, as Anastasoff notes, that “judges and lawyers of the day recognized the authority of unpublished decisions even when they were established only by memory or by a lawyer’s unpublished memorandum,” 223 F.3d at 903 , but precedents brought to the attention of the court in that fashion obviously could not serve as the kind of rigid constraint that binding authority provides today. Unlike our practice today, a single case was not sufficient to establish a particular rule of law, and case reporters often filtered out cases that they considered wrong, or inconsistent with their view of how the law should develop. See pp. 1166-67 supra. The concept of binding case precedent, though it was known at common law, see note 13 supra, was used exceedingly sparingly. For the most part, common law courts felt free to depart from precedent where they considered the earlier-adopted rule to be no longer workable or appropriate.

Case precedent at common law thus resembled much more what we call persuasive authority than the binding authority which is the backbone of much of the federal judicial system today. The concept of binding precedent could only develop once two conditions were met: The development of a hierarchical system of appellate courts with clear lines of authority, and a case reporting system that enabled later courts to know precisely what was said in earlier opinions. See note 21 supra. As we have seen, these developments did not come about — either here or in England — until the nineteenth century, long after Article III of the Constitution was written.

While many consider the principle of binding authority indispensable — perhaps even inevitable — it is important to note that it is not an unalloyed good. While bringing to the law important values such as predictability and consistency, it also (for the very same reason) deprives the law of flexibility and adaptability. See Planned Parenthood v. Casey, 505 U.S. 833, 868 , 112 S.Ct. 2791 , 120 L.Ed.2d 674 (1992) (“The promise of constancy, once given, binds its maker for as long as the power to stand by the decision survives and the understanding of the issue has not changed so fundamentally as to render the commitment obsolete.”). A district court bound by circuit authority, for example, has no choice but to follow it, even if convinced that such authority was wrongly decided. Appellate courts often tolerate errors in their caselaw because the rigors of the en banc process make it impossible to correct all errors. See note 29 supra.

A system of strict binding precedent also suffers from the defect that it gives undue weight to the first case to raise a particular issue. This is especially true in the circuit courts, where the first panel to consider an issue and publish a prece-dential opinion occupies the field, whether or not the lawyers have done an adequate job of developing and arguing the issue.

The question raised by Anastasoff is whether one particular aspect of the binding authority principle — the decision of which rulings of an appellate court are binding — is a matter of judicial policy or constitutional imperative. We believe Anastasoff erred in holding that, as a constitutional matter, courts of appeals may not decide which of their opinions will be deemed binding on themselves and the courts below them. For the reasons explained, the principle of strict binding authority is itself not constitutional, but rather a matter of judicial policy. Were it otherwise, it would cast doubt on the federal court practice of limiting the binding effect of appellate decisions to the courts of a particular circuit. Circuit bound aries — and the very system of circuit courts — are a matter of judicial administration, not constitutional law. If, as Anastasoff suggests, the Constitution dictates that every “declaration of law ... must be applied in subsequent cases to similarly situated parties,” 223 F.3d at 900 , then the Second Circuit would have no authority to disagree with a ruling of the Eighth Circuit that is directly on point, and the first circuit to rule on a legal issue would then bind not only itself and the courts within its own circuit, but all inferi- or federal courts.

Another consequence of Anastasoff’s reasoning would be to cast doubt on the authority of courts of appeals to adopt a body of circuit law on a wholesale basis, as did the Eleventh Circuit in Bonner, and the Federal Circuit in South Corp. See p. 1173 supra. Circuits could, of course, adopt individual cases from other circuits as binding in a case raising a particular legal issue. See, e.g., Charles v. Lundgren & Assocs., P.C., 119 F.3d 739 , 742 (9th Cir.) (“Because we have the benefit of the Seventh Circuit’s cogent analysis, we will not replow plowed ground. Instead, we adopt the reasoning of the Seventh Circuit ....” ) cert. denied, 522 U.S. 1028 , 118 S.Ct. 627 , 139 L.Ed.2d 607 (1997). But adopting a whole body of law, encompassing countless rules on matters wholly unrelated to the issues raised in a particular case, is a very different matter. If binding authority were a constitutional imperative, it could only be created through individual case adjudication, not by a decision unconstrained by the facts before the court or its prior caselaw.

Nor is it clear, under the reasoning of Anastasoff, how courts could limit the binding effect of their rulings to appellate decisions. Under Anastasoff’s reasoning, district court opinions should bind district courts, at least in the same district, or even nationwide. After all, the Constitution vests the same “judicial Power” in all federal courts, so Anastasoff’s conclusion that judicial decisions must have prece-dential effect would apply equally to the thousands of unpublished decisions of the district courts.

No doubt the most serious implication of Anastasoff’s constitutional rule is that it would preclude appellate courts from developing a coherent and internally consistent body of caselaw to serve as binding authority for themselves and the courts below them. Writing an opinion is not simply a matter of laying out the facts and announcing a rule of decision. Prece-dential opinions are meant to govern not merely the cases for which they are written, but future cases as well.

In writing an opinion, the court must be careful to recite all facts that are relevant to its ruling, while omitting facts that it considers irrelevant. Omitting relevant facts will make the ruling unintelligible to those not already familiar with the case; including inconsequential facts can provide a spurious basis for distinguishing the case in the future. The rule of decision cannot simply be announced, it must be selected after due consideration of the relevant legal and policy considerations. Where more than one rule could be followed — which is often the case — the court must explain why it is selecting one and rejecting the others. Moreover, the rule must be phrased with precision and with due regard to how it will be applied in future cases. A judge drafting a precedential opinion must not only consider the facts of the immediate case, but must also envision the countless permutations of facts that might arise in the universe of future cases. Modern opinions generally call for the most precise drafting and re-drafting to ensure that the rule announced sweeps neither too broadly nor too narrowly, and that it does not collide with other binding precedent that bears on the issue. See Fred A. Bernstein, How to Write it Right, Cal. Lawyer, at 42 (June 2000). Writing a precedential opinion, thus, involves much more than deciding who wins and who loses in a particular case. It is a solemn judicial act that sets the course of the law for hundreds or thousands of litigants and potential litigants. When properly done, it is an exacting and extremely time-consuming task.

It goes without saying that few, if any, appellate courts have the resources to write precedential opinions in every case that comes before them. The Supreme Court certainly does not. Rather, it uses its discretionary review authority to limit its merits docket to a handful of opinions per justice, from the approximately 9000 cases that seek review every Term. While federal courts of appeals generally lack discretionary review authority, they use their authority to decide cases by unpublished — and nonprecedential — dispositions to achieve the same end: They select a manageable number of cases in which to publish precedential opinions, and leave the rest to be decided by unpublished dispositions or judgment orders. In our circuit, published dispositions make up approximately 16 percent of decided cases; in other circuits, the percentage ranges from 10 to 44, the national average being 20 percent. Administrative Office of the United States Courts, Judicial Business of the United States Courts 44 tbl. S-3 (2000).

That a case is decided without a prece-dential opinion does not mean it is not fully considered, or that the disposition does not reflect a reasoned analysis of the issues presented. What it does mean is that the disposition is not written in a way that will be fully intelligible to those unfamiliar with the case, and the rule of law is not announced in a way that makes it suitable for governing future cases. As the Federal Judicial Center recognized, “the judicial time and effort essential for the development of an opinion to be published for posterity and widely distributed is necessarily greater than that sufficient to enable the judge to provide a statement so that the parties can understand the reasons for the decision.” Federal Judicial Center, Standards for Publication of Judicial Opinions 3 (1973). An unpublished disposition is, more or less, a letter from the court to parties familiar with the facts, announcing the result and the essential rationale of the court’s decision. Deciding a large portion of our cases in this fashion frees us to spend the requisite time drafting precedential opinions in the remaining, cases.

Should courts allow parties to cite to these dispositions, however, much of the time gained would likely vanish. Without comprehensive factual accounts and precisely crafted holdings to guide them, zealous counsel would be tempted to seize upon superficial similarities between their clients’ cases and unpublished dispositions. Faced with the prospect of parties citing these dispositions as precedent, conscientious judges would have to pay much closer attention to the way they word their unpublished rulings. Language adequate to inform the parties how their case has been decided might well be inadequate if applied to future cases arising from different facts. And, although three judges might agree on the outcome of the case before them, they might not agree on the precise reasoning or the rule to be applied to future cases. Unpublished concurrences and dissents would become much more common, as individual judges would feel obligated to clarify their differences with the majority, even when those differences had no bearing on the case before them. In short, judges would have to start treating unpublished dispositions— those they write, those written by other judges on their panels, and those written by judges on other panels — as mini-opinions. This new responsibility would cut severely into the time judges need to fulfill their paramount duties: producing well-reasoned published opinions and keeping the law of the circuit consistent through the en banc process. The quality of published opinions would sink as judges were forced to devote less and less time to each opinion.

Increasing the number of opinions by a factor of five, as Anastasoff suggests, doesn’t seem to us a sensible idea, even if we had the resources to do so. Adding endlessly to the body of precedent — especially binding precedent — can lead to confusion and unnecessary conflict. Judges have a responsibility to keep the body of law “cohesive and understandable, and not muddy[ ] the water with a needless torrent of published opinions.” Martin, note 36 supra, at 192. Cases decided by nonpre-cedential disposition generally involve facts that are materially indistinguishable from those of prior published opinions. Writing a second, third or tenth opinion in the same area of the law, based on materially indistinguishable facts will, at best, clutter up the law books and databases with redundant and thus unhelpful authority. Yet once they are designated as precedent, they will have to be read and analyzed by lawyers researching the issue, materially increasing the costs to the client for absolutely no legitimate reason. Worse still, publishing redundant opinions will multiply significantly the number of inadvertent and unnecessary conflicts, because different opinion writers may use slightly differ ent language to express the same idea. As lawyers well know, even small differences in language can have significantly different implications when read in light of future fact patterns, so differences in phrasing that seem trivial when written can later take on a substantive significance.

The risk that this may happen vastly increases if judges are required to write many more precedential opinions than they do now, leaving much less time to devote to each. Because conflicts — even inadvertent ones — can only be resolved by the exceedingly time-consuming and inefficient process of en banc review, see Atonio v. Wards Cove Packing Co., 810 F.2d 1477, 1478-79 (9th Cir.1987) (en banc) (conflict in panel opinions must be resolved by en banc court), cert. denied, 485 U.S. 989 , 108 S.Ct. 1293 , 99 L.Ed.2d 503 (1988), an increase in intracircuit conflicts would leave much less time for us to devote to normal panel opinions. Maintaining a coherent, consistent and intelligible body of caselaw is not served by writing more opinions; it is served by taking the time to make the precedential opinions we do write as lucid and consistent as humanly possible.

Unlike the Anastasoff court, we are unable to find within Article III of the Constitution a requirement that all case dispositions and orders issued by appellate courts be binding authority. On the contrary, we believe that an inherent aspect of our function as Article III judges is managing precedent to develop a coherent body of circuit law to govern litigation in our court and the other courts of this circuit. We agree with Anastasoff that we — and all courts — must follow the law. But we do not think that this means we must also make binding law every time we issue a merits decision. The common law has long recognized that certain types of eases do not deserve to be authorities, and that one important aspect of the judicial function is separating the cases that should be precedent from those that should not. Without clearer guidance than that offered in Anastasoff, we see no constitutional basis for abdicating this important aspect of our judicial responsibility.

Contrary to counsel’s contention, then, we conclude that Rule 36-3 is constitutional. We also find that counsel violated the rule. Nevertheless, we are aware that Anastasoff may have cast doubt on our rule’s constitutional validity. Our rules are obviously not meant to punish attorneys who, in good faith, seek to test a rule’s constitutionality. We’ therefore conclude that the violation was not willful and exercise our discretion not to impose sanctions.

The order to show cause is DIS' CHARGED.

. See also Coleen M. Barger, Anastasoff, Unpublished Opinions, and ''No-Citation” Rules, 3 J.App. Prac. & Process 169, 169-70 (2001). Barger notes that "[t]he chief judge of the District of Massachusetts seems determined to force the issue in the First Circuit,” citing 1st Cir. R. 36(b)(2)(F) ("Unpublished opinions may be cited only in related cases .... ”), "as he has begun to routinely insert the following footnote in his opinions whenever he cites unpublished opinions to support his reasoning”: For the propriety of citing unpublished decisions, see Anastasoff v. United States, 223 F.3d 898, 899-905 (8th Cir.) (R. Arnold, J.) (holding that unpublished opinions have precedential effect), vacated as moot, No. 99-3917, 2000 WL 1863092 (8th Cir. Dec. 18, 2000); Giese v. Pierce Chem. Co., 43 F.Supp.2d 98, 103 (D.Mass.1999) (relying on unpublished opinions' persuasive authority), and Richard S. Arnold, Unpublished Opinions: A Comment, 1 J.App. Prac. & Process 219 (1999). See, e.g., Suboh v. City of Revere, 141 F.Supp.2d 124 , 144 n. 18 (D.Mass.2001) (Young, C.J.). . Our rule operates somewhat differently from that of the Eighth Circuit, though it is in essential respects the same. While Eighth Circuit Rule 28A(i) says that "[u]npublished decisions are not precedent,” we say that unpublished dispositions are "not binding precedent.” Our rule, unlike that of the Eighth Circuit, prohibits citation of an unpublished disposition to any of the courts of our circuit. The Eighth Circuit's rule allows citation in some circumstances, but provides that the authority is persuasive rather than binding. See 8th Cir. R. 28A(i) ("Parties may ... cite an unpublished opinion of this court if the opinion has persuasive value on a material issue and no published opinion of this or another court would serve as well.”). The difference is not material to the rationale of Anastasoff because both rules free later panels of the court, as well as lower courts within the circuit, to disregard earlier rulings that are designated as nonprecedential. For a comprehensive table of nonpublication and noncitation rules across all circuits and states, see Melissa M. Serfass & Jessie L. Cranford, Federal and State Court Rules Governing Publication and Citation of Opinions, 3 J.App. Prac. & Process 251, 253-85 tbl. 1 (2001). . In the passage cited by Anastasoff, Justice Story argued only that the judicial decisions of the Supreme Court were “conclusive and binding,” and that inferior courts were not free to disregard the “decisions of the highest tribunal.” He said nothing to suggest that the principle of binding authority constrained the “judicial Power,” as Anastasoff does; rather, he recognized that the decisions of the Supreme Court were binding upon the states because they were the "supreme law of the land.” Story, supra, §§ 376-78. . To be sure, exercise of the judicial power is subject to a number of explicit constraints, such as the requirements of due process, trial by jury, the availability of counsel in criminal cases, the ex post facto clause and the prohibition against bills of attainder — to name just a few. . Because the matter arises so seldom, there is little authority on this point, but the authority that does exist supports the view that the text of the judicial power clause is merely descriptive. For example, United States v. Ferreira, 54 U.S. (13 How.) 40 , 14 L.Ed. 40 (1851), considered whether decisions of district courts as to whether certain Spanish citizens were entitled to compensation pursuant to a treaty between Spain and the United States were an exercise of the judicial power. If the district judges found the claimants entitled to compensation, they were to recommend that the Secretary of the Treasury make such payments, and the latter could (but was not required to) pay the claim. In concluding that such recommendations did not constitute an exercise of the judicial power (and hence were not reviewable by the Supreme Court), the opinion noted the ways in which the procedures for establishing these claims differed from "the ordinary forms of a court of justice”: For there is to be no suit; no parties in the legal acceptance of the term, are to be made- — no process to issue; and no one is authorized to appear on behalf of the United States, or to summon witnesses in the case. The proceeding is altogether ex parte; and all that the judge is required to do, is to receive the claim when the party presents it, and to adjust it upon such evidence as he may have before him, or be able himself to obtain. But neither the evidence, nor his award, are to be filed in the court in which he presides, nor recorded there; but he is required to transmit, both the decision and the evidence upon which he decided, to the Secretary of the Treasury; and the claim is to be paid if the Secretary thinks it just and equitable, but not otherwise. It is to be a debt from the United States upon the decision of the Secretary, but not upon that of the judge. See also Missouri v. Jenkins, 515 U.S. 70, 130-33 , 115 S.Ct. 2038 , 132 L.Ed.2d 63 (1995) (Thomas, J., concurring) (listing various functional limitations on the exercise of the judicial power, including federalism, separation of powers and the prohibition against decid ing political questions); Young v. United States ex rel. Vuitton et Fils S.A., 481 U.S. 787, 815-18 , 107 S.Ct. 2124 , 95 L.Ed.2d 740 (1987) (Scalia, J., concurring) (discussing the functional limitation of separation of powers on the exercise of the judicial power). . The three examples we have given, though apparently disparate, actually bear on the question of what weight was given to precedent at the time of the Framing. In a regime where juries have power to decide the law, the concept of ''binding” precedent has a very different, and much more diluted, meaning than in the current regime where jury verdicts are routinely reversed if they are not supported by the evidence in light of the applicable law. Similarly, binding precedent means something different altogether when a court speaks with seven or nine voices than with a single voice. Nine judges speaking separately may well agree on the outcome of a case, but they cannot give the kind of specific guidance as to the conduct of future cases that can be found in a single opinion speaking for the court. Finally, during the time when appeals were conducted by two-judge panels consisting of the circuit justice flanked by the district judge whose ruling was being appealed produced remarkably few — if any — written rulings. The precedential value of rulings from such panels was, for obvious reasons, not particularly valuable guidance in future cases. Anastasojf's view that the judicial process underwent such fundamental changes, yet the process of producing prece-dential opinions remained essentially unchanged, strikes us as inherently doubtful. Anastasoff’s historical analysis has been called into question even by academics who generally agree with the result. See, e.g., Polly J. Price, Precedent and Judicial Power After the Founding, 42 B.C. L.Rev. 81, 84, 90-93 (2000); Salem M. Katsh & Alex V. Chachkes, Constitutionality of "No-Citation” Rules, 3 J.App. Prac. & Process 287, 288 & n. 5 (2001). . Rules limiting the precedential effect of unpublished decisions exist in every federal circuit and all but four states (Connecticut, Delaware, New York and North Dakota). See Serfass & Cranford, note 2 supra, at 260-61 tbl. 1, 273-74 tbl. 1. But see Eaton v. Chahal, 146 Misc.2d 977 , 553 N.Y.S.2d 642, 646 (N.Y.Sup.Cl.1990) ("[U]nreported decisions issued by judges of coordinate jurisdiction ... are not binding precedent upon this court....”) The near-universal adoption of the practice illustrates not only that the practice is consistent with the prevailing conception of the judicial power, but also that it reflects sound judicial policy. . As Hale described it, judicial decisions "do not make a Law properly so-called,” but "they have a great Weight and Authority in Expounding, Declaring, and Publishing what the Law of this Kingdom is, [and] are a greater Evidence [of a law] than the Opinion of any private Persons, as such, whatsoever.” Sir Matthew Hale, The History of the Common Law of England 68 (London, Nutt & Gosling 1739). In Lord Mansfield’s view, "[t]he reason and spirit of cases make law; not the letter of particular precedents.” Fisher v. Prince, 97 Eng. Rep. 876, 876 (K.B.1762). . As Holdsworth put it: The general rule is clear. Decided cases which lay down a rule of law are authoritative and must be followed. But in very many of the statements of this general rule there are reservations of different kinds.... The fundamental principle, upon which all these reservations ultimately rest, is the principle stated by Coke, Hale and Blackstone, that these cases do not make law, but are only the best evidence of what the law is. They are not, as Hale said, "law properly so called,” but only very strong evidence of the law. They are evidence, as Coke said, of the existence of those usages which go to make up the common law; and, conversely, the fact that no case can be produced to prove the existence of an alleged usage is evidence that there is no such usage. This principle is the natural, though undesigned, result of the unofficial character of the reports; and it is clear that its adoption gives the courts power to mould as they please the conditions in which they will accept a decided case or a series of decided cases as authoritative. If the cases are only evidence of what the law is the courts must decide what weight is to be attached to this evidence in different sets of circumstances. The manner in which they have decided this question has left them many means of escape from the necessity of literal obedience to the general rule that decided cases must always be followed. They have allowed many exceptions to, and modifications of, this rule if, in their opinion, a literal obedience to it would produce either technical departures from established principles, or substantial inconveniences which would be contrary to public policy. Sir William Holdsworth, 12 A History of English Law 150-51 (1938) (footnotes omitted) (emphasis added). .As one commentator has noted: [T]wo conditions had to be satisfied before the doctrine of stare decisis could be established. (1) There had to exist reliable reports of cases. It is obvious that if cases are to be binding, there should be precise records of what they lay down. (2) There had also to be a settled judicial hierarchy. Equally obvious is it that until this was settled it could not be known which decisions were binding. Not until roughly the middle of the last century were these conditions fulfilled, and it is from about then that the modern doctrine [of stare decisis] emerges. R.W.M. Dias, Jurisprudence 30-31 (2d ed.1964). . One reason that House of Lords decisions commanded little respect was that as late as 1844, judicial deliberations could be conducted by lay peers, who brought far less training and experience to bear on legal issues than did the judges of the Exchequer Chamber. Dias, note 10 supra, at 32-33. . The three common law courts of first instance — the King's (or Queen’s) Bench, Common Pleas and Exchequer — had overlapping jurisdiction in many common classes of cases. See Plucknett, supra, at 210. . The absence of an appellate hierarchy that could definitively settle legal issues was a continuing problem until the nineteenth century. The need for such definitive resolution nevertheless existed and the common law judges invented a substitute: the Exchequer Chamber. When a particularly vexing legal issue arose that was common to two or more of the courts, all the judges would meet, sometimes including the Lord Chancellor, the barons of the Exchequer, the members of the Council and the serjeants. See Plucknett, supra, at 151 (the Council consisted of the King’s closest advisers); id. at 224 (serjeants were, essentially, lawyers known for wearing the coif, "a close-fitting cap of white silk or linen fastened under the chin; hence the term 'order of the coif.' ”) The Exchequer Chamber debated particular legal issues and came up with a definitive ruling, which was then announced in the court where the case raising the issue originated. Id. at 162-63. The Exchequer Chamber was not a separate court; it was referred to by that name because these meetings were held in the court of the Exchequer, which "had ample office accommodation” to allow all the judges to meet in one place. Pluck-nett, supra, at 162 n. 7. The Exchequer Chamber might best be viewed as a super-en banc court including all of England’s judicial officers. Unlike other decisions at common law, decisions reached by the Exchequer Chamber were considered binding precedent and, according to Plucknett, this is the first time we find “the principle that a single case may be precedent.” Id. at 348. The Exchequer Chamber is significant for our analysis because it clearly suggests common law judges knew the distinction between binding and persuasive precedent. The vast majority of precedents at common law were considered more or less persuasive; only the few decisions agreed-to by all English judges sitting together were afforded the status that the Anastasoff court would now afford to every decision of a three-judge court of appeals as a matter of constitutional imperative. . In the first century of American jurisprudence, Blackstone's "Commentaries were not merely an approach to the study of law; for most lawyers they constituted all there was of the law.” Daniel J. Boorstin, The Mysterious Science of the Law 3 (1941). . For example, "Clerk v. Day was reported in four different books, and in not one of them correctly — not even as to name.... Arbitrary spelling of the names of cases is a bibliographical irritation, and sometimes a difficulty. Fetter v. Beal ... is a pretty good disguise for Fitter v. Veal Percy H. Winfield, The Chief Sources of English Legal History 185 n. 3 (1925) (citations omitted). . As Holdsworth wrote: [I]n the eighteenth century, because the reports were made by private reporters, the reports of decided cases possessed, as we have seen, very different degrees of authority. It was always possible for a judge who was trying a case to decry the authority of a report which laid down a rule with which he disagreed. We have seen that Lord Mansfield, when he was pressed by a case which laid down a rule with which he did not like, was rather too apt to take this line. It is no doubt a line which it became less possible to take as the reports improved in quality, and as reporting became more standardized and more stereotyped. But within limits this censorship of reports is both legitimate and necessary.... Thus in the case of Chillingworth v. Esche [1924] 1 Ch. at pp. 112-113 Warrington L.J. said, "there are one or two points raised by Mr. Micklem with which I think I ought to deal. He relies on Moeserv. Wisher ((1871) L.R. 6 C.P. 120). In my opinion that is a case which never ought to have been reported. It was an ex parte application. The judges seized on a single fact, and decided on that fact. The purchaser in that case had no opportunity of stating his view.” Holdsworth, note 9 supra, at 154 & 154 n. 3 (footnotes omitted). . Coke was not alone in this practice: [B]arristers have sometimes exercised some kind of censorship over the cases which they have reported.... For instance ... Atlay, The Victorian Chancellors ii 138, says, "Campbell was no mere stenographer; he exercised an absolute discretion as to what decisions he reported and what he suppressed, and sternly rejected any which appeared to him inconsistent with former rulings or recognised principles. He jocularly took credit for helping to establish the Chief Justice's reputation as a' lawyer, and he used to boast that he had, in one of his drawers, material for an additional volume in the shape of 'bad Ellenborough law'.” Holdsworth, note 9 supra, at 158 & 158 n. 1. . As one commentator has noted: It would appear also that from about 1785 judges were beginning to favour particular reporters chosen for each court and to prefer citation from them and no other.... The question what cases should be reported bristles with problems. The decision rests ultimately with the individual reporter. Dias, note 10 supra, at 33. . As another commentator has noted: The Framers were familiar with the idea of precedent. But ... [t]he whole idea of just what precedent entailed was unclear. The relative uncertainty over precedent in 1789 also reflects the fact that "many state courts were manned by laymen, and state law and procedure were frequently in unsettled condition. The colonial and state courts did not enjoy high prestige, and their opinions were not even deemed worthy of publication.” Henry Paul Monaghan, Stare Decisis and Constitutional Adjudication, 88 Colum. L.Rev. 723, 770 n. 267 (1988) (citations omitted). See also Melissa H. Weresh, The Unpublished, Non-Precedential Decision: An Uncomfortable Legality?, 3 J.App. Prac. & Process 175, 186 (2001) ("Stare decisis and the American common law system have never required the publication of all decisions.”) .Far from being the strict and uncontro-verted doctrine that Anastasoff attempts to portray, the concept of precedent at the time of the Framers was the subject-of lively debate. Adherence to the common law was not "inevitable and unopposed.” Robert H. Jackson, The Supreme Court in the American System of Government 29 (1955), "[T]he parameters of judicial power were highly contested in the late colonial and early Republic periods. ... [N]o one knew the exact role that judges would have in the new experiment in government that formed the United States.” R. Ben Brown, Judging in the Days of the Early Republic: A Critique of Judge Richard Arnold’s Use of History in Anastasoff v. United States, 3 J.App. Prac. & Process 355, 375, 383 (2001). Therefore, "lawyers, judges and legal commentators contested the question of just what body of law judges should use to decide cases in the early Republic.” Id. at 358. On one side of the debate was Blackstone himself. "Far from providing support for Judge Arnold’s claim that the colonial judiciary was bound by common law precedent, Blaclcstone’s thesis was just the opposite”: that American courts were not bound by English precedent. Id. at 357 (footnotes omitted). St. George Tucker, a prominent nineteenth-century American scholar, disagreed. Id. at 358. Amidst this disagreement, American judges not only routinely picked and chose which English precedents to follow, but also felt free to ignore their own decisions. Id. at 359, 360-63 (discussing Fitch v. Brainerd, 2 Day 163 (Conn.1805) (available at 1805 WL 203 ), in which the Connecticut Supreme Court declared, without explanation, that its prior decision adopting an English precedent authored by Lord Mansfield, "was not law.”) Such cavalier treatment of precedent — the Fitch court did not acknowledge the precedent as binding and distinguish or reject it, but simply declared it “was not law” — illustrates that precedent at the time of the Framers was a far more fluid concept than it is today, and certainly more so than the strict form advocated by Anastasoff. . As Plucknett notes, "[tjhe nineteenth century produced the changes which were necessary for the establishment of the rigid and symmetrical theory [of case precedent] as it exists today.” Plucknett, supra, at 350. Among the changes he points to was the establishment of a strict appellate hierarchy and . the standardization of case law reporting. Id. . The first volumes of the United States Reports reveal the idiosyncratic and sometimes unreliable character of the early reporters. The first volume contains not a single decision of the United States Supreme Court. See Craig Joyce, The Rise of the Supreme Court Reporter: An Institutional Perspective on Marshall Court Ascendancy, 83 Mich. L.Rev. 1291, 1296 (1985). The reporter, Alexander James Dallas, began his career by publishing decisions of the Pennsylvania and Delaware courts, but not until 1806 were Pennsylvania judges required to reduce their opinions to writing (and then only at the parties’ request). Dallas’s first volume therefore contains only brief descriptions of the earliest decisions, based on notes preserved by judges and lawyers. See id. at 1295-98. And, while his second volume does contain decisions of the United States Supreme Court, Dallas could not always rely on a written opinion as the basis of his report because the Court did not invariably reduce its opinions to writing: Not a single formal manuscript opinion is known to have survived from the Court’s first decade; and few, if any, may ever have existed for Dallas to draw upon. Nor may it be confidently assumed that in all instances Dallas was present in court to take down what the Justices said, or that he was able afterwards to consult any notes they may have kept of the opinions they announced .... Delay, expense, omission and inaccuracy: these were among the hallmarks of Dallas’ work. Id. at 1305 (footnotes omitted). At that time, the Supreme Court had no official reporter and cases were never printed. United States v. Yale Todd, decided by the Supreme Court in 1784, is a typical example. Because "[t]here was no official reporter at that time, [the] case has not been printed.” United States v. Ferreira, 54 U.S. (13 How.) 40, 52 , 14 L.Ed. 40 (1851). So said Chief Justice Taney in a note added following Fer-reira, describing Yale Todd. "[A]s the subject is one of much interest, and concerns the nature and extent of judicial power, the sub stance of the decision in Yale Todd’s case is inserted here, in order that it may not be overlooked, if similar questions should hereafter arise.” Id. . In 1986, only 39% of the 884 opinions of the English Court of Appeal were reported. Martineau, Appellate Justice, supra, at 107, 150. “Although technically a judgment need not be reported to be cited as precedent [in England] ... the reality is that unless a judgment is reported it is not likely to be used as precedent.” Id. at 104. Nevertheless, "[t]here does not appear to be among the judges and the bar any current dissatisfaction with the system except that some believe too many, not too few, judgments are reported.” Id. at 107. . The same practice is followed in the state courts as well. See, e.g., Auto Equity Sales, Inc. v. Superior Court of Santa Clara County, 57 Cal.2d 450 , 20 Cal.Rptr. 321 , 369 P.2d 937, 940 (Cal.1962) (“Courts exercising inferi- or jurisdiction must accept the law declared by courts of superior jurisdiction. It is not their function to attempt to overrule decisions of a higher court.”). . For example, in Dun & Bradstreet, Inc. v. Greenmoss Builders, Inc., 472 U.S. 749 , 105 S.Ct. 2939 , 86 L.Ed.2d 593 (1985), a majority held that the rule announced in Gertz v. Robert Welch, Inc., 418 U.S. 323 , 94 S.Ct. 2997 , 41 L.Ed.2d 789 (1974) (plaintiff must show “actual malice” to obtain punitive damages for false and defamatory statements), applies only to statements involving matters of public concern. Relying on the language and context of Gertz , the Court rejected the dissenters' claim that the Gertz rule applied to all defamatory statements, and instead concluded that Gertz left it an open question whether the rule applied to statements not of public concern. Compare Dun & Bradstreet, 472 U.S. at 757 n. 4, 105 S.Ct. 2939 (“The dissent states that '[a]t several points the Court in Gertz makes perfectly clear [that] the restrictions of presumed and punitive damages were to apply in all cases.' Given the context of Gertz , however, the Court could have made ‘perfectly clear' only that these restrictions applied in cases involving public speech." (citations omitted)), with id. at 785 n. 11, 105 S.Ct. 2939 ("Distrust of placing in the courts the power to decide what speech was of public concern was precisely the rationale Gertz offered for rejecting [an alternative] approach. It would have been incongruous for the Court to go on to circumscribe the protection against presumed and punitive damages by reference to a judicial judgment as to whether the speech at issue involved matters of public concern.” (citation omitted)). .This is consistent with the practice in our court — and all other collegial courts of which we are aware — in which the judges who join an opinion authored by another judge make substantive suggestions, often conditioning their votes on reaching agreement on mutually acceptable language. . The exception is the Federal Circuit, which has a geographic area precisely the same as the Supreme Court, but much narrower subject-matter jurisdiction. See 28 U.S.C. § 1295 (a). . Or, unless Congress changes the law. See, e.g., Van Tran v. Lindsey, 212 F.3d 1143, 1149 (9th Cir.) (earlier caselaw established that mixed questions in habeas petitions were reviewed de novo, but under the Anti-Terrorism and Effective Death Penalty Act of 1996, the standard of review is governed by 28 U.S.C. § 2254 (d)), cert. denied, 531 U.S. 944 , 121 S.Ct. 340 , 148 L.Ed.2d 274 (2000). . An impressive array of judges and academics have noted the rigors of en banc procedures. See Richard S. Arnold, Why Judges Don't Like Petitions for Rehearing, 3 J.App. Prac. & Process 29, 37 (2001) ("[0]n many days, I confess, I find myself wishing that there were no such thing [as en banc rehearing].”); Pamela Ann Rymer, How Big Is Too Big?, 15 J.L. & Pol. 383 , 392 (1999) ("expensive and time consuming”); Joseph T. Sneed, The Judging Cycle: Federal Circuit Court Style, 57 Ohio St. L.J. 939 , 942 (1996) ("time consuming and complex”); James Oakes, Personal Reflections on Learned Hand and the Second Circuit, 47 Stan. L.Rev. 387, 393 (1995) ("enormously time-consuming and expensive”); Deanell Reece Tacha, The "C” Word: On Collegiality, 56 Ohio St. L.J. 585 , 590 (1995) ("time-consuming and expensive”); Irving R. Kaufman, Do the Costs of the En Banc Proceeding Outweigh Its Advantages?, 69 Judicature 7, 7 (1985) ("the most time consuming and inefficient device in the appellate judiciary’s repertoire”); J. Woodford Howard, Jr., Courts of Appeals in the Federal Judicial System: A Study of the Second, Fifth, and District of Columbia Circuits 217 (1981) ("most circuit judges regard en bancs as a 'damned nuisance' ”). Because they are so cumbersome, en banc procedures are seldom used merely to correct the errors of individual panels: "[W]e do not take cases en banc merely because of disagreement with a panel's decision, or rather a piece of a decision.... We take cases en banc to answer questions of general importance likely to recur, or to resolve intracircuit conflicts, or to address issues of transcendent public significance — perhaps even to curb a 'runaway' panel — but not just to review a panel opinion for error, even in cases that particularly agitate judges....” EEOC v. Ind. Bell Tel. Co., 256 F.3d 516 , 86 Fair Empl. Prac. Cas. (BNA) 1 , 2001 WL 717685 , at *11 (7th Cir.2001) (en banc) (Posner, J., concurring). See also Fed. R.App. P. 35(a) ("An en banc hearing or rehearing is not favored and ordinarily will not be ordered unless: (1) en banc consideration is necessary to secure or maintain uniformity of the court’s decisions; or (2) the proceeding involves a question of exceptional importance.”); Arnold, supra, at 36 ("Petitions for rehearing are generally denied unless something of unusual importance — such as a life — is at stake, or a real and significant error was made by the original panel, or there is conflict within the circuit on a point of law.”) It is therefore very important that three-judge panel opinions be decided correctly and that they state their holdings in a way that is easily understood and applied in future cases. . Some state court systems apply the binding authority principle differently than do the federal courts. In California, for example, an opinion by one of the courts of appeal is binding on all trial courts in the state, not merely those in the same district. Judicial Council of California, Report of the Appellate Process Task Force 59 (2000); Jon B. Eisen-berg, Ellis J. Horvitz & Justice Howard B. Wiener, California Practice Guide: Civil Appeals and Writs § 14:193 (2000) ("A court of appeal decision must be followed by all superior and municipal courts, regardless of which appellate district rendered the opinion.”) However, court of appeal panels are not bound by the opinions of other panels, even those within the same district. In re Marriage of Shaban, 88 Cal.App.4th 398 , 105 Cal.Rptr.2d 863, 870-71 (2001) ("[Because there is no 'horizontal stare decisis' within the Court of Appeal, intermediate appellate court precedent that might otherwise be binding on a trial court ... is not absolutely binding on a different panel of the appellate court." (citations omitted)). See also Report of the Appellate Process Task Force, supra, at 60-61; Ei-senberg, Horvitz & Wiener, supra, § 14:193.1 ("In contrast, a decision by one court of appeal is not binding on other courts of appeal.”) California's management of precedent differs from that of the federal courts in another important respect: The California Supreme Court may "depublish” a court of appeal opinion — i.e., strip a published decision of its precedential effect. See Cal. R. Ct. 976(c)(2); Steven B. Katz, California’s Curious Practice of "Pocket Review”, 3 J.App. Prac. & Process 385 (2001). California's depublication practice shows that it is possible to adopt more aggressive methods of managing precedent than those used by the federal courts. . It also forces judges in certain instances to act in ways they may consider to be contrary to the Constitution. Some have argued that the duty of judges to follow the Constitution stands on a higher footing than the rule requiring adherence to precedent, and judges should not follow precedent when they believe that to do so would violate the Constitution. See Gary Lawson, The Constitutional Case Against Precedent, 11 Harv. J.L. & Pub. Pol’y 23, 27-28 (1994). . Opinion, writing is a "reflective art,” an absolute necessity of which is "fully adequate time to contemplate, think, write and rewrite.” Howard T. Markey, On the Present Deterioration of the Federal Appellate Process: Never Another Learned Hand, 33 S.D. L.Rev. 371, 379, 384 (1988). Judge Markey rightly mourns the age when a judge could, as Judge Hand did, talk at length about each case, “with his feet on the desk and hands behind his head,” and "having reached his decision, ... wr[i]te the entire opinion in longhand.” Id. at 380. Today, "[t]here simply isn’t time” to engage in such "reflective personal craftsmanship.” Id. at 379-80. . As Judge Posner has noted: Given the workload of the federal courts of appeals today, the realistic choice is not between limited publication, on the one hand, and, on the other, improving and then publishing all the opinions that are not published today; it is between preparing but not publishing opinions in many cases and preparing no opinions in those cases. It is a choice, in other words, between giving the parties reasons for the decision of their appeal and not giving them reasons even though the appeal is not frivolous. Richard A. Posner, The Federal Courts: Challenge and Reform 168-69 (1996). . The United States Supreme Court decided seventy-seven cases in October Term 1999, which represents less than nine opinions per justice. Statistics for the Supreme Court's October Term 1999, 69 U.S.L.W. 3076 (BNA 2000). By comparison, in 1999, each active judge in our court heard an average of 450 cases and had writing responsibility for an average of twenty opinions and 130 unpublished dispositions. See infra note 37. . Sufficient restrictions on judicial decision-making exist to allay fears of irresponsible and unaccountable practices such as “burying” inconvenient decisions through nonpub-lication. In Unpublished Decisions in the Federal Courts of Appeals: Making the Decision to Publish, 3 J.App. Prac. & Process 325 (2001), Professor Stephen L. Wasby concludes, after "extended observation of the ... Ninth Circuit,” id. at 331 , that formal publication guidelines and judges’ enforcement of them through their interactions with each other, keep judges honest in deciding whether or not to publish. See also Martineau, Restrictions on Publication and Citation of Judicial Opinions: A Reassessment, supra, at 132 ("American appellate systems ... have many built-in protections to prevent against [judicial] irresponsibility without mandatory publication of opinions.") . See Boyce F. Martin, Jr., In Defense of Unpublished Opinions, 60 Ohio St. L.J. 177 , 196 ("[I]t will not save us any time if [unpublished opinions] are being cited back to us. We will have to prepare unpublished opinions as we do published opinions — as if they were creating precedent."). . Recent figures tell a striking story. In 1999, our court decided some 4500 cases on the merits, about 700 by opinion and 3800 by unpublished disposition. Each active judge heard an average of 450 cases as pari of a three-judge panel and had writing responsibility in a third of those cases. That works out to an average of 150 dispositions — 20 opinions and 130 unpublished dispositions— per judge. In addition, each judge had to review, comment on, and eventually join or dissent from 40 opinions and 260 unpublished dispositions circulated by other judges with whom he sat. See Alex Kozinski & Stephen Reinhardt, Please Don’t Cite This! Why We Don’t Allow Citation to Unpublished Dispositions, Cal. Law., June 2000, at 44; see also Report of the Federal Courts Study Committee 109 (Apr. 2, 1990) (noting the federal appellate courts’ "crisis of volume"). . Concerned that judges spend too little time writing (as opposed to editing) precedential opinions, commentators have suggested that judges should do the preliminary drafting of all published opinions. See, e.g., David McGowan, Judicial Writing and the Ethics of the Judicial Office, 14 Geo. J. Legal Ethics 509 , 514, 555-56 (2001). Adoption of such proposals would, however, "produce fewer published opinions [and] more unpublished • dispositions.” Id. at 593 . By preventing judges from determining which of their opinions will be citable as precedent, Anastasoff would have precisely the opposite effect, forcing judges to spread their resources more thinly, resulting in even less judicial involvement in precedential opinions. . Anastasoff suggests that the appointment of more judges would enable courts to write binding opinions in every case. See 223 F.3d at 904 . We take no position as to whether there should be more federal judges, that being a policy question for Congress to decide. We note, however, that Congress would have to increase the number of judges by something like a factor of five to allocate to each judge a manageable number of opinions each year. But adding more judges, and more binding precedents, creates its own set of problems by significantly increasing the possibility of conflict within the same circuit as each judge will have an increased body of binding caselaw to consider and reconcile. That problem, in turn, could be ameliorated by increasing the number of circuits, but that would increase the number of inter-circuit conflicts, moving the problem up the chain of command to the Supreme Court, which likewise does not have the capacity to significantly increase the number of opinions it issues each year. See Wisniewski v. United States, 353 U.S. 901, 901-02 , 77 S.Ct. 633 , 1 L.Ed.2d 658 (1957) (per curiam) (noting the problems of intra-circuit consistency raised by the growing number of circuit judgeships). In the end, we do not believe that more law makes for better law. . This is hardly a novel view: [C]ertain types of cases do not deserve to be authorities. One type, already alluded to, is that in which there is no discoverable ratio decidendi. Others are cases turning purely on fact, those involving the exercise of discretion, and those which judges themselves do not think worthy of being precedents. Dias, note 10 supra, at 55 (footnotes omitted) (citing R. v. Stokesley (Yorkshire) Justices, Ex parte Bartram [1956] 1 All E.R. 563 at 565).