Copr. © West 1999 No Claim to Orig. U.S. Govt. Works

87 MILR 940

(Cite as: 87 Mich. L. Rev. 940)

Michigan Law Review

April, 1989



Lauren K. Robel [FNa]

Copyright 1989 by Lauren K. Robel

Since 1976, every federal appellate court has adopted rules that limit the publication of opinions. [FN1] As a result, only a minority of the federal courts of appeals publish even half of their decisions on the merits. [FN2] Although withholding opinions from publication is meant to reduce or eliminate their applicability to litigation, the large body of unpublished decisions creates a variety of incentives for those litigants who have unusual access to these opinions to use them. The policies reflected by the publication plans do not anticipate these sorts of incentives. Although the rules adopted by the courts of appeals attempt to curtail litigant use of unpublished opinions, the controls do not work because these mechanisms restrict only certain uses-- usually of the most overt kind--such as citation. Not only do the appellate rules fail to destroy the usefulness of unpublished opinions, they also exacerbate the advantages that the selective publication plans give frequent litigants.

This article discusses the courts' adoption of the limited publication plans and analyzes the methods used by the courts to discourage the use of unpublished opinions. It also discusses the results of a survey conducted to determine if, and how, government litigants--some *941 of the chief unanticipated beneficiaries of the publication plans--make use of unpublished opinions. Finally, it challenges the assumption that limited publication is essential in an age of caseload crisis.


There are differences among the publication plans adopted by the circuits, but the assumptions underlying the plans are fairly uniform. [FN3] The central assumption is that not all decisions by the courts of appeals warrant publication. [FN4] This assumption is based on the view that appellate opinions serve two primary functions: first, to resolve particular disputes between litigants; second, to advance the state of the law in some manner. [FN5] All of the publication plans are based on the central assumption that opinions that serve no lawmaking function should not be published. [FN6] Clearly, however, appellate opinions serve a host of other purposes: to supervise the lower courts, for instance, or to provide a mechanism for interested or disinterested observers to keep *942 track of how an agency is administering a statute. For the most part, however, the policies that inform the publication plans do not consider the plans' impact on these purposes. [FN7]

Selective publication plans are also premised on the assumption that publication is costly in a number of ways. [FN8] The judges suffer the costs involved in preparing an opinion for publication. [FN9] Presumably, the plans will not eliminate the costs of researching the issues to be decided and formulating a rationale for a decision, since these must be done regardless of publication. Rather, the plans are meant to minimize additional special production costs associated with publication: those that arise because judges do not simply decide a case but also--like other published authors--attempt to express that resolution felicitously, to shore it up with citations to authority at every turn, and to anticipate in writing possible criticisms of the opinion. [FN10] If judges did not publish, then the range of costs associated with felicitous and authoritative expression would be eliminated. Litigants suffer the costs of delays that occur while judges attempt to write. [FN11] Everyone suffers from added costs associated with increasingly large volumes of the Federal Reporter: research becomes more inefficient and time- consuming as the number of published opinions increases, as does maintaining the libraries and citechecking against the possibilities of missed authority. [FN12]

The argument for nonpublication, then, depends upon the claim that judges (with their staffs) [FN13] can efficiently identify cases that add *943 significantly to the costs of litigation and decision but add nothing to the law. Supposedly, once these cases are identified, withholding them from publication reduces costs associated with them.

The option most responsive to the concerns underlying the publication plans would be to eliminate opinions altogether. [FN14] By doing this, the courts would achieve every cost saving that was hypothesized in connection with nonpublication, but would avoid entirely the risk of creating a twilight zone of written-but-unpublished work. This option, however, has proven unattractive to litigants and judges alike: [FN15] summary decision serves none of the legitimating functions of appeal and may leave parties with the feeling that the court never considered their arguments. [FN16] Instead, the publication plans adopted by the courts are a compromise: they eliminate only publication of the decision, rather than the actual writing of the opinion. The resulting 'unpublished opinions' are sometimes extremely obtuse, and they are often short, but just as often they resemble in every way the published opinions of the courts: facts are stated, the parties' legal arguments are addressed, and authority is cited and explained. [FN17] The result of this tension between the need to explain and legitimate results to the parties and the premises of the selective publication plans has been that unpublished opinions are still opinions--providing insights into *944 a court's reasoning and suggesting to advocates the arguments that could win or lose a case.


Lawyers are trained to use opinions: to make arguments based on them, to try to distinguish them, and to consider how they bear on the outcome of their cases. In order to preserve the savings associated with nonpublication, though, lawyers must be persuaded not to use unpublished opinions. The primary disincentive to using unpublished opinions was to be the opinions' inherent worthlessness as precedent. If these opinions really did nothing to add to what was already available in the Federal Reporter, the courts reasoned, then lawyers would soon realize it, and would not attempt to use them. [FN18] In addition to explicitly disavowing their precedential value, [FN19] courts employ two mechanisms to discourage the use of unpublished opinions.

First, all but four of the circuits circumscribe, by rule or by practice, the distribution of unpublished opinions. [FN20] In most circuits, the opinions are routinely distributed to parties to the case and the lower court judge whose decision was reviewed. [FN21] In six circuits, the opinions are also circulated to all appellate judges on the court. [FN22] By limiting access to the opinions to the parties involved, the courts limit the number of attorneys who can use the opinions. Even though theoretically the opinions are available to anyone who wants to dig through the court archives to retrieve them, practically speaking they are unavailable because they are neither indexed nor filed in a manner that would facilitate retrieval. [FN23] By limiting distribution of the opinions, then, the courts make it unlikely that many litigants will find them useful (or find them at all).

*945 Second, all but the Third and the D.C. Circuits limit citation of the opinions to the court. [FN24] Eight of the circuits forbid citation except in related cases or to support a claim of res judicata. Four of the circuits allow citation when there is no better precedent available. [FN25]

By forbidding citation, the courts hope to conserve the presumed savings of nonpublication. First, if cases could be cited and therefore would be used by a wide audience seeking authority, judges might feel compelled to do a better job writing them, and so the assumed savings in judicial time would be lost. [FN26] Second, savings in consumption costs would be lost because litigants would feel the need to research these opinions, if they could be cited, and publishers would publish them. [FN27] Finally, because the courts' distribution rules assure that access to these opinions will not be uniform, the no-citation rules supposedly insure that those litigants who have unusually large access to unpublished opinions will have no incentive to make use of that access in unfair ways. [FN28]

To achieve the presumed savings of selective publication, then, the courts had to assure in some way that their unpublished opinions were truly disposable. The mechanisms they chose, however--reduced access and no- citation--are among the most controversial aspects of selective publication plans. Before the plans were actually implemented, the Commission on Revision of the Federal Appellate System (chaired by Senator Roman Hruska) held hearings to allow the legal community the opportunity to comment on the proposals. [FN29] While testimony before the Hruska Commission generally supported the view that not all cases warranted published opinions, some witnesses *946 expressed concern about the proposed ban on citation. [FN30]

Most of the concern over the no-citation rules centered on the issue of access to the opinions. Witnesses worried that litigants who appeared frequently before the courts would be able to use the opinions despite the ban on citation, because as parties they had unusual access to the opinions. [FN31] Witnesses also worried that judges would consult the opinions in an effort to avoid intracircuit conflicts and to keep abreast of the work of their courts, and would in doing so rely on opinions not generally available to attorneys appearing before these judges. [FN32]

Thus, the concern arose early on that the publication plans would result in a secret body of applicable and pertinent law available only to certain advantaged litigants and the courts before which they routinely appeared. Further examination bears out the legitimacy of this concern.


The control mechanisms described above were designed to create disposable opinions--opinions that would disappear from the landscape, leaving no precedential trace behind. If the assumption that a disposable opinion can be created is wrong, or at least wrong very often, then, I argue, the selective publication plans, at least in their present form, cannot be supported as a fair or just way to manage the workload of the courts. Differential access to the opinions favors certain litigants. Moreover, such differential access, with its attendant favoritism, would occur under almost any selective publication plan that also achieved the cost savings sought by the courts.

In this section, I examine critically the assumptions underlying publication and the mechanisms the courts employ to discourage use of these opinions. [FN33] I then discuss the results of a survey of government*947 attorneys who regularly litigate cases before the federal appeals courts, to discover whether the practices of these attorneys pertaining to unpublished opinions bear out the assumptions supporting the courts' publication plans.

The selective publication plans fail in their effort to create a class of disposable opinions, largely because these plans depend on mistaken assumptions. The mistaken assumptions are of three kinds. First, the central assumption, that only lawmaking and not dispute-resolving opinions give litigants useful information, is wrong because it underestimates the varieties of information that lawyers retrieve from opinions. Even if the courts select for publication only those opinions that 'make law,' unpublished opinions contain useful information because opinions tell lawyers more than simply 'what the law is.' Second, the publication plans assume a narrow, rules-oriented concept of precedent, so that if the plans work as expected, they will systematically leave unpublished much of what lawyers would routinely use in their work. Third, the plans do not operate neutrally with regard to the subject matter of the opinions, so that most of the work of the courts in several subject areas appears only in unpublished form. Even were it true, then, that the courts had succeeded in some general sense in separating the routine from the important cases, attorneys who work in these areas cannot develop a sense of what the courts consider 'routine,' or how the courts apply the 'important' cases without looking at the courts' unpublished work.

What kinds of information could litigants get from unpublished opinions that are not foreseen by the publication plans? One important piece of information is the shape of the universe of decisions by a court in a particular area of law. The information might be as simple as knowing the odds. [FN34] For instance, an attorney in California in 1987 might assess differently an appeal of a decision of the Board of Immigration Appeals (BIA) if he knew that the Ninth Circuit published only 27% of its immigration opinions in that year, and that over half of the reversals of BIA decisions occurred in unpublished decisions. [FN35] *948 Similarly, an attorney considering an appeal of an agency denial of disability benefits might evaluate an appeal differently if she knew that the same court published only 25% of its disability review cases, and only seven of twenty-three reversals. [FN36]

The attorney with the immigration case might also benefit from being able to discern trends in the court's review of immigration decisions. If he looked in the Federal Reporter for opinions in 1987 having to do with immigration, he would find twenty-two. But the court decided eighty-eight immigration cases on the merits that year. [FN37]

If the court's behavior in its unpublished decisions simply mirrors what it announces in its published decisions, one might conclude that the publication plans are working well: the attorney has been saved from looking at sixty-six cases. However, as I hope I can demonstrate, it is likely that there will be a disparity between the two, even assuming that judges never consciously misuse publication plans. [FN38] For example, the published immigration opinions of the Ninth Circuit for the eleven months beginning October 1, 1986, included one concurrence [FN39] and one dissent. [FN40] By contrast, there were seven concurrences and seven dissents in the unpublished opinions. [FN41] For the most part, *949 the separate opinions critize the agency for failing to differentiate properly among review standards or object to other continuing problems with the agency's implementation of the statute. [FN42] Some of the opinions, on the other hand, criticize the court for requiring unduly detailed findings from the agency. [FN43] Taken together, the large number of reversals and separate expressions in unpublished opinions reveals a good bit of dissatisfaction on the part of the court with a variety of agency practices, as well as a fair amount of disagreement among members of the court about what it ought to require of the agency. It would be difficult to discern this pattern, however, if one were limited to the court's published expressions. [FN44]

What accounts for the different picture of the work of the court one gets when one compares the unpublished and published opinions? There are at least two explanations. The first concerns the concept of precedent used to establish which cases should be decided through published opinions and which through unpublished opinions. The second concerns the wide variations in publication rates depending on the subject matter of the appeal.

First, and obviously, the rules are designed to reflect a substantive difference between what is published and what is not. One would expect the unpublished opinions to differ from the published ones because they have been systematically sorted by the limited publication rules into different 'stacks.' The plans intend to sort out only routine applications of rules. In theory, one ought to be able to predict, on the basis of the published 'lawmaking' opinions, the outcome of the unpublished opinions with a high degree of accuracy. (This is not to suggest that unpublished opinions are meant to be simply a vehicle for disposing of frivolous appeals. [FN45]) The limited publication plans intend that unpublished opinions are to be used for applications of pre-existing *950 rules, rather than for formulation and articulation of the new rules. Granted, the plans hope to distinguish between routine and novel applications. [FN46] But if the plans worked perfectly, the universe of published opinions would resemble a treatise without footnotes: It would be limited to opinions that create rules, [FN47] opinions that contain significant scholarship (historical treatments of legal rules, or previously unreported legislative history), [FN48] or opinions that are 'newsworthy.' [FN49] This world might be a judge's dream come true, [FN50] but I suspect it would be a practitioner's nightmare: While it might open up the possibility for endlessly creative arguments, it would seriously impede a lawyer's ability to make rational decisions about what to argue and how to argue it. Lawyers need to know how stable published precedents are, how they apply in different fact situations, and whether there are unanticipated problems with the precedent in application. In short, they need to know what courts do with their 'lawmaking' opinions. This information is sometimes available in unpublished form, and if the plans worked perfectly, that is the only place it would be available.

Luckily for lawyers, for the most part the plans do not work perfectly, with one rather large exception, which I will discuss. But they sometimes do, and the results can be illustrated with an example, again from the Ninth Circuit, involving the Racketeer Influenced and Corrupt Organizations Act (RICO). [FN51] A relatively recent statute, RICO contains provisions for civil liability that have proven enormously attractive to plaintiffs (largely due to a treble damages provision) and enormously difficult for courts. [FN52] Many of the most difficult *951 questions have revolved around applying the statute: For example, when are ordinary businesses deemed RICO 'enterprises'? In a recent one-year period, [FN53] the Ninth Circuit decided thirteen appeals involving civil RICO: eight opinions were published, and five (including one reversal) were left unpublished. [FN54] It would be fair to say that these unpublished opinions were not groundbreaking. Most of them, in fact, discuss pleading requirements. In Chang v. California Canadian Bank, [FN55] the plaintiffs had alleged that several banks were involved in an elaborate 'Ponzi' scheme [FN56] involving automobile resales. The appeal challenged the dismissal of the complaint for failure to state a claim. The complaint had premised the RICO claim on alleged mail and wire fraud violations. The eleven- page opinion contains a detailed discussion of the requirements of pleading participation in a RICO enterprise. While this case could easily be characterized as routine, the citations in the opinion on this issue were to district court opinions or opinions from other circuits, rather than opinions of the appeals court, indicating that local circuit authority on the issues in the opinion was sparse.

Similarly, Ormes v. I. A. Fialkoff Administrators, Inc. [FN57] discusses pleading requirements. Ormes, a corporation, alleged that its provider of medical insurance had misrepresented the provisions of the policy and then refused to pay claims filed by Ormes' employees. The complaint is typical of many filed under RICO for business misrepresentations. Ormes' complaint did not specify the relationship between the 'person' engaging in the predicate criminal acts and the RICO 'enterprise,' nor did it allege a 'pattern of racketeering activity.' While the court had recently discussed these requirements in published opinions, [*952 FN58] their application to pleading remains confusing to practitioners and judges alike, [FN59] and publication of opinions like Ormes could be important to illustrating how the pleading requirements operate.

Finally, in Wavelength, Inc. v. Edwards, [FN60] the court reversed a dismissal of plaintiff Wavelength's RICO, antitrust, and securities claims. The eighteen-page opinion contains a lengthy discussion of the complicated facts underlying the complaint and an explanation of why the facts pleaded were sufficient to withstand a motion to dismiss the RICO claim. Such discussions are invaluable in a relatively uncharted area like this one. [FN61]

Although the unpublished RICO cases discussed above contain information useful to practitioners, they do a fairly good job of achieving one of the efficiencies sought by the publication plans: the unpublished cases apply rules from published decisions. As I noted earlier, however, the publication plans ordinarily do not work this efficiently to weed out applications of rules. The major exception is in subject-matter areas where unpublished opinions predominate, [FN62] and the reason that the plans work efficiently here is, I suspect, that these are areas where judges believe almost all applications are routine. [FN63] Which areas are these? They include much litigation in which the *953 government is a party, such as review of agency determinations in immigration and social security cases, Federal Tort Claims Act cases, criminal and habeas appeals, civil rights actions, and employment discrimination complaints against the federal government. [FN64] On average, these are low- status, low-resources types of appeals, and they are the kinds of cases judges find tedious, [FN65] with the result that they are often--perhaps usually-- relegated to staff handling from start to finish. [FN66]

In fact, there is some reason to doubt whether judges have much to do with the publication decision in these areas. First, judges themselves do not usually do the initial screening that designates a case as a likely candidate for disposition without argument. That initial decision is made by staff, usually staff attorneys or a circuit executive. A high correlation obtains between cases not argued before the court and cases disposed of without a published opinion. [FN67] Some of the screening procedures used by the circuits identify entire categories of cases by subject matter as likely candidates for the expedited review that results in nonpublication. [FN68] This suggests another reason why nonpublication may not be a good indication of the information value of an opinion: Decisions that result in nonpublication have been made in gross rather than individually, at least on the initial level, and judges *954 have few incentives to examine these initial decisions closely. [FN69] Existing data reveal that judges rarely disagree with the initial decision to decide an appeal on the briefs alone. [FN70] This means that staff determinations about the relative merits of the cases almost always prevail, and as noted above, staff determinations may be guided largely by the subject matter of the opinion.

In effect, then, the determination not to publish occurs very early in the appellate process, and necessarily so, for to delay the publication decision until after an opinion is written would be to lose the time savings the plans hope to achieve by having judges write with nonpublication in mind. Yet it seems unlikely that it will usually be possible to predict the information value of an opinion before it is written. In fact, many judges have noted how frequently a case's complexities are revealed through the process of writing an opinion. [FN71] It seems likely, therefore, that an opinion's ultimate information value would be hard to predict at the time when the publication decisions are usually made.

All of the circuits have mechanisms by which members of the panel of judges that decides the case can force a decision's publication. [FN72] Available information suggests, though, that panel members rarely do this in cases that have been initially identified for nonpublication by staff, perhaps because in many circuits these opinions are drafted by central staff rather than the judges themselves or their own clerks. [FN73] If the opinions are drafted by staff, judges will have little investment in the final product. Even if a staff attorney, then, believes that the decision warrants publication, that attorney will have to do additional work to justify a publication decision, and the judges will have to do additional work to gain confidence in the staff attorney's product (or worse, draft an entirely new opinion themselves). In the face of these extra efforts, institutional pressures operate to preserve the initial decision identifying a case for nonpublication.

A two-tier system emerges. In some subject-matter areas, the courts' screening processes and the assumptions of the publication plans about the nature of precedent combine to assure that large numbers *955 of opinions-- many more than half [FN74]--remain unpublished and inaccessible.


These aspects of the publication plans would be unimportant, perhaps, if people could be prevented entirely from using the unpublished opinions; if, in practice, the opinions could be made truly disposable. But my research on institutional litigants indicates that the methods the courts employ to discourage use of unpublished opinions--limited distribution and no-citation-- do not work. In fact, these methods aggravate and enhance any inherent unfairness the selective publication plans might have.

Because the plans underestimate the kinds of information attorneys derive from opinions and overestimate judges' abilities to implement the plans' central assumptions, those people who have unusual access to unpublished opinions will gain an advantage over those who do not. Moreover, the limited distribution plans currently operating assure that the people with unusual access to these opinions will be the same litigants who enjoy a variety of other institutional advantages in litigation: the frequent litigants. [FN75] As parties to the cases, frequent litigants receive the opinions. Additionally, because the opinions are most often distributed only to parties and judges, the frequent litigants will have unique access to a useful source of information known only to them and the judges before whom they appear. The advantages of this access are exaggerated because unpublished opinions tend to cluster in subject-matter areas that pit frequent litigants against those litigants Marc Galanter has dubbed 'one-shotters.' These areas include criminal appeals, social security cases, and immigration cases, and in these sorts of disputes it is unlikely that the same ability to monitor unpublished opinions exists on both sides. [FN76]

*956 The no-citation rules also present only a minor disincentive against using these opinions, especially for those litigants who have unusual access to them. This is so in part because the 'guts' of the opinion--its reasoning, citations to authority, and such--can still be effectively employed through incorporation in briefs and arguments, and because, as explained above, attorneys gain useful information from these opinions other than the kinds contemplated by the no-citation rules. [FN77]

To test my hypothesis that frequent litigants are not discouraged from using unpublished opinions by the various mechanisms used by the courts, I sent a questionnaire to representatives of that most ubiquitous of frequent litigants, the federal government. [FN78] I chose government attorneys for a number of reasons. First, the government litigates in many subject areas disproportionately treated through unpublished opinions. [FN79] Second, government litigants occupy a unique position under the distribution rules followed by the courts: because they are always parties in cases involving the subject matter for which they are responsible, they always receive copies of the unpublished opinions issued in the cases. (The government is always a party to a criminal appeal, for example, and the National Labor Relations Board is always a party to an enforcement action.) This position enables government *957 litigants to compile and use substantial files that represent the views of the courts of appeals in which they practice on their subjectmatter area. [FN80]

The survey was designed to determine two things: first, whether government offices make these opinions accessible to the attorneys, and second, whether the attorneys use them. The survey went to the heads of six offices in Washington that are responsible for appellate litigation in diverse subject- matter areas. Five of the offices were within the Department of Justice: Immigration Litigation, Consumer Litigation, Tax Division, Civil Rights Division, and the Appellate Staff of the Civil Division. I also sent a survey to the National Labor Relations Board. All of the offices responded. Of the six offices, only the Office of Consumer Litigation, which handles relatively few appeals that result in unpublished opinions, responded that the opinions were neither made accessible nor used in any way. [FN81]

I first asked whether unpublished opinions are circulated to attorneys within the office and whether the opinions are filed. All but the Civil Division Appellate staff responded that they do circulate copies of unpublished opinions. Circulation of opinions increases the likelihood that attorneys will remember the opinions should they be involved in similar litigation. All of the offices responded that the opinions are filed, most often with other materials related to the cases in which they are issued. Three offices also filed them separately. [FN82] Finally, the Office of Immigration Litigation stated that it maintains an index of unpublished opinions.

I asked the attorneys to tell me how they used unpublished opinions they received and suggested a number of ways in which the opinions might be used. All of the offices replied that they used unpublished opinions in making litigation and settlement decisions and in writing briefs. All but the NLRB also stated that they use the opinions in making determinations about whether to appeal or contest appeals in other cases. [FN83]

*958 Finally, I asked whether the offices had ever moved for publication of unpublished opinions. All the courts of appeals allow litigants to move to change the status of opinions from unpublished to published. Research by Galanter and others has suggested that frequent litigants have more incentive to 'play for rules' than do other litigants: That is, they are more likely than other litigants to base litigation decisions on the possibility of creating favorable precedent or avoiding unfavorable precedent, rather than simply obtaining a favorable outcome in the immediate case. [FN84] It seemed plausible, therefore, that frequent litigants would also have more incentive than other litigants to seek publication of previously unpublished opinions when the opinions were favorable to them. All of the offices surveyed stated that they had, in fact, moved for publication of previously unpublished opinions.

Allowing motions for publication creates a serious flaw in the publication plans because, depending upon the frequency with which they are granted, they allow frequent litigants to stack the precedential deck. None of the courts of appeals keep any records on the frequency of motions to publish or the parties involved, so it is difficult to know much about this practice. [FN85] I was able to identify, though, all of the Seventh Circuit opinions that have changed publication status in the last five years. [FN86] What I discovered tends to confirm the suspicion that frequent litigants take advantage of this mechanism more often than others. In 1982, for example, thirty opinions changed status from unpublished to published. Twenty-two of them, or 73%, involved a government litigant, and fifteen of these were federal litigants. In all but three of the cases involving the government, the result of the opinion favored the government, and among cases involving federal litigants, the result was always favorable.

Because the United States Attorneys are responsible for so much appellate litigation, I attempted to check my initial findings through surveys directed at all of the United States Attorney offices, both main *959 offices and branch offices, asking these attorneys the same questions that I asked those in the Washington offices. Unfortunately, I received only a twenty percent response. [FN87]

While I hesitate to draw any conclusions because of this low response rate, those attorneys responding generally confirmed the earlier findings from the Washington offices. Ninety percent of those responding used unpublished opinions in writing briefs, and 70% said they used the opinions in making litigation or settlement decisions and in determining whether to appeal or oppose an appeal. Only 60% of the offices stated that they had moved for publication of previously unpublished opinions, but several of the responses were from branch offices, and some respondents believed that these motions had been made by the main office in their district.


I conclude from these surveys that the publication plans are seriously flawed. The mechanisms that supposedly ensure the disposability of unpublished opinions do not work and in fact increase the likelihood that the plans are substantively unfair. Motions for publication, selective distribution of unpublished opinions, and rules against citation do little more than create possibilities for frequent litigants to manipulate precedent through manipulating publication, while they provide insignificant disincentives for these litigants to use the opinions.

Despite the problems with the limited publication plans, Judge Posner has argued that they are a necessary evil. While he accepts in theory many of the points aruged here, [FN88] he argues that unpublished opinions are not prepared with the same care as those that are published and that this fact significantly decreases their information value. If the quality of the opinions cannot be increased, [FN89] he argues, then the opinions ought not to be used because 'their information value would be slight--maybe even negative.' [FN90]

*960 In what sense could this be true? The information value of these opinions might be 'negative' if the results in the opinions were whimsical; if, for instance, they had no predictive value whatsover. Or the opinions' value might be negative if it were generally true that those decisions reached in unpublished opinions were in some sense wrong. I don't expect that Judge Posner meant either of these things. Rather, I believe that his essential objection is that these opinions are not judge-work but staff-work, and that they might not accurately reflect what judges would do in a perfect world with unlimited time--or even with a published opinion. But I think this objection misses a central point of Posner's own book: that judges haven't more time, and so the system we have--at least as it concerns judges--is what we are likely to have in the future. The relevant question, then, is not whether judges ought to be expected to increase their involvement in producing unpublished opinions, but whether practitioners ought to be denied access to the work product of those people who do produce the opinions--the staff. It is very little answer to a practitioner who works in one of the subject-matter areas in which unpublished opinions predominate that he should not have that access because these opinions are ghost-written. In reality, this practitioner needs precisely to know the contents of unpublished opinions prepared by central staff and law clerks.

The nature of central staff and law clerk work argues against Posner's earlier conclusion that the information value of these opinions is slight. If, as he argues elsewhere in his book, [FN91] clerks and staff attorneys are by nature cautious, and if they deal with appeals involving the same subject- matter areas repeatedly, it is likely that they will take advantage of their earlier work (or even the work of their predecessors) by filing it and referring to it. Further, if Posner is correct that staff attorneys are less constrained by the values of the judges than law clerks are (since they do not often work with one judge for long), [FN92] then it seems even more plausible that the notion of a second tier--with its own set of 'precedents' and culture of decision--exists in those areas that are routinely relegated to staff handling. If the numbers of appeals that are routed through this second tier were trivial, perhaps the loss to the courts in credibility or authoritativeness that would come from publicly accepting the current state of affairs would outweigh the fact that some practitioners are able to exploit the present system. But the numbers are not trivial. [FN93]

*961 The biggest objection most judges have to changing the present system is that any change will increase the demands on judges. Judges generally agree that they cannot be expected to increase their output significantly, nor can they be expected to take on the task of authoring decisions in the large number of unpublished cases. While the lack of judge involvement in unpublished opinions poses a serious critique of the publication plans generally, [FN94] this critique does not attack the inequalities in access to the opinions and the information they contain, which is an objection of a different order. There is no reason why judges would have to change their behavior to accommodate universal publication. Those opinions that now remain unpublished could instead be published with a legend indicating that they are not to be cited except in those limited instances [FN95] that unpublished opinions can now be cited. By continuing to limit citation, judges would be encouraged not to change their present behavior, since they would not have to take account of these opinions in any more instances than they now do. We could then allow attorneys to make whatever use of the opinions is rational, and I suspect that attorneys can be trusted to behave rationally in this regard. [FN96]

Finally, universal publication would have other advantages. My survey of government attorneys and my analysis of existing data about government involvement in unpublished opinions raise a number of other issues. Significant government involvement in a body of cases that is relatively inaccessible and unexamined is inherently troubling. In many of the subject- matter areas routinely consigned to nonpublication, the standard of review employed by the court is already extremely deferential. [FN97] For example, in social security or immigration *962 cases, the relevant agency usually needs only to meet standards asking whether the agency's determination is supported and is not arbitrary or capricious. What is the consequence of consigning large numbers of these cases to routine handling by staff for ultimate disposition through unpublished opinion? One possible consequence might be the inability of the traditional critics--the bar and the scholarly community--as well as the bench to discern trends in agency decisionmaking. Not only do trends in the executive's administration of a statute become more difficult to determine, but difficulties in the administration of a statute or in the statutes themselves become more difficult to detect. In essence, the practice of selective publication obstructs effective oversight of government litigation and decisionmaking by both interested and disinterested observers. A system of universal publication with limited citation would eliminate this obstruction.


The courts adopted selective publication plans over a decade ago with the idea of alleviating severe pressures on judges' time. Little thought has been given since to the question of how litigants have adapted their practices to accommodate the plans. [FN98] Yet answers to what litigants actually do with the opinions are central to evaluating whether selective publication is an equitable way to decrease the workload of the courts.

As my survey shows, some litigants can and do use these opinions in ways that call into question the fairness of the current publication policies. This problem need not be accepted as an inevitable byproduct of rising caseloads. It can, and should, be remedied.

FNa Associate Professor of Law, Indiana University--Bloomington. B.A. 1978, Auburn University; J.D. 1983, Indiana University.--Ed. The author is grateful to Terry Bethel, Thomas Davies, Joseph Hoffman, John Oakley, Richard Posner, and Jeff Stake for helpful comments on an earlier draft of this article, and to Marcella DePeters for research assistance. I am especially grateful to Donna Stienstra for help with advice, information, and wisdom at all stages of this project. I would like to thank the Indiana University School of Law for research support.

FN1. See D.C. CIR. R. 14; 1ST CIR. R. 36.1-36.2; 2D CIR. R. 0.23; 3D CIR. INTERNAL OPERATING PROCEDURES [hereinafter IOP] 5(F); 4TH CIR. IOP 36.3-36.4; 5TH CIR. R. 47.5; 6TH CIR. R. 24; 7TH CIR. R. 53; 8TH CIR. IOP VI(B) & appx.; 9TH CIR. R. 21; 10TH CIR. R. 36; 11TH CIR. IOP 36; FED. CIR. R. 18.

FN2. D. STIENSTRA, UNPUBLISHED DISPOSITIONS: PROBLEMS OF ACCESS AND USE IN THE COURTS OF APPEALS 40 (Table 2) (1985). As of 1984, only four circuits published a majority of their opinions. The Third Circuit published a mere 22.9% of its opinions that year. Even more dramatic difference within circuits are possible when one looks at publication rates within subject-matter areas. See infra notes 62-70 and accompanying text.

FN3. The central difference is in the amount of direction the plans provide the judges in making publication decisions. Some of the publication plans are quite detailed. See 4TH CIR. IOP 36.3 (Opinions will be published only 'if the author or a majority of the joining judges believe the opinion satisfies one or more of the standards for publication. These standards are: i. It establishes, alters, modifies, clarifies, or explains a rule of law within this Circuit; or ii. It involves a legal issue of continuing public interest; or iii. It criticizes existing law; or iv. It contains a[ ] historical review of a legal rule that is not duplicative; or v. It resolves a conflict between panels of this Court, or creates a conflict with a decision in another circuit.'); 7TH CIR. R. 53(c)(1) ('A published opinion will be filed when the decision (i) establishes a new, or changes an existing rule of law; (ii) involves an issue of continuing public interest; (iii) criticizes or questions law; (iv) constitutes a significant and nonduplicative contribution to the legal literature (a) by a historical review of law, (b) by describing legislative history, or (c) by resolving or creating a conflict in the law; (v) reverses a judgment or denies enforcement of an order when the lower court or agency has published an opinion supporting the judgment or order; or (vi) is pursuant to an order of remand from the Supreme Court and is not rendered merely in ministerial obedience to specific directions of that Court.'). Other plans are less comprehensive. See 3D CIR. IOP (5)(F) ('The criterion normally applied [to determine publication of signed opinions] is whether or not the opinion has precedential or institutional value.'). All of the plans state, in one form or another, that opinions which establish new rules or law should be published.

FN4. This assumption underlies even those limited publication plans that have a presumption in favor of publication. See, e.g., 1ST CIR. R. 36.2(a) ('In general, the court thinks it desirable that opinions be published. . . . This policy may be overcome in some situations where an opinion does not articulate a new rule of law, modify an established rule, apply an established rule to novel facts, or serve otherwise as a significant guide to future litigants.').

FN5. Reynolds & Richman, An Evaluation of Limited Publication in the United States Courts of Appeals: The Price of Reform, 48 U. CHI. L. REV. 573, 579 (1981) [hereinafter Reynolds & Richman I]; Reynolds & Richman, The Non- Precedential Precedent--Limited Publication and No-Citation Rules in the United States Courts of Appeals, 78 COLUM. L. REV. 1167, 1182 (1978) [hereinafter Reynolds & Richman II]; Leflar, Some Observations Concerning Judicial Opinions, 61 COLUM. L. REV. 810 (1961); ADVISORY COUNCIL ON APP. JUSTICE, STANDARDS FOR PUBLICATION OF JUDICIAL OPINIONS 2-3 (1973) [hereinafter ADVISORY COUNCIL ON APP. JUSTICE]. For a comprehensive discussion of arguments for and against limited publication, see Reynolds & Richman II, supra.

FN6. Most of the publication plans also allow publication when the case is particularly newsworthy. See, e.g., D.C. CIR. R. 14(b)(7).

FN7. See infra text accompanying notes 18-32.

FN8. See R. POSNER, THE FEDERAL COURTS 124-26 (1985); Reynolds & Richman I, supra note 5, at 580.

FN9. By some accounts, opinion writing is the most time-consuming part of a judge's job. U.S. COMM. ON REVISION OF THE FED. APP. CT. SYS., STRUCTURE AND INTERNAL PROCEDURES: RECOMMENDATIONS FOR CHANGE 49 (1975) [hereinafter COMMISSION] (results of Third Circuit Time Study revealed that in 1971-72 judges spent almost half of their time writing opinions); ADVISORY COUNCIL ON APP. JUSTICE, supra note 5, at 1 (state court judges reported that opinion writing took more of their time than any of their other tasks).

FN10. The always provocative Judge Posner argues that while the desire for felicitous prose should be encouraged, the tendency toward excessive citation is a disease (of which law clerks are the carriers) that should be eradicated. R. POSNER, supra note 8, at 108-09.

FN11. Interestingly, one study of limited publication plans indicated that, while cases decided through unpublished disposition are typically decided more quickly than those that are not, there is no empirical support for the hypothesis that limited publication enhances overall productivity. Reynolds & Richman, I, supra note 5, at 594-97. Daniel Hoffman has also been unable to conclude that limited publication really saves time for judge. Hoffman, Nonpublication of Federal Appellate Court Opinions, 6 JUST. SYS. J. 405, 419-20 (1981).

FN12. Some of the limited publication rules make specific mention of these costs. See, e.g., 1ST CIR. R. 36.1 (limited publication adopted 'in the interests both of expedition in the particular case, and of saving time and effort in research on the part of future litigants').

FN13. See R. POSNER, supra note 8, at 24; J. CECIL & D. STIENSTRA, DECIDING CASES WITHOUT ARGUMENT: AN EXAMINATION OF FOUR COURTS OF APPEALS (1987) (describing staff function in screening cases for expedited treatment); cf. Davies, Gresham's Law Revisited: Expedited Processing Techniques and the Allocation of Appellate Resources, 6 JUST. SYS. J. 372, 375 (1981) (discussing suggested reforms directed toward expedited processing of civil claims, in order to dismiss frivolous claims quickly).

FN14. This could be done by summarily affirming or reversing the decision below. See, e.g., 5TH CIR. R. 47.6 ('Affirmance Without Opinion').

FN15. The Fifth Circuit attempted for a time to eliminate opinions altogether in a substantial number of cases. See Shuchman & Gelfand, The Use of Local Rule 21 in the Fifth Circuit: Can Judges Select Cases of 'No Precedential Value'?, 29 EMORY L.J. 195 (1980). Several other circuits also have rules which would allow summary disposition without opinion. Reynolds & Richman II, supra note 5, at 1174-75 & n.37. Reynolds and Richman note that summary disposition without opinion has been almost universally condemned. Id. at 1174. See also COMMISSION, supra note 9, at 49 (attorneys opposed to doing away with opinions in any number of cases).

FN16. See P. CARRINGTON, D. MEADOR & M. ROSENBERG, JUSTICE ON APPEAL 10 (1976) ('The obligation to give reasons is vital to [the functions of appellate review] . . .. An unreasoned decision has very little claim to acceptance by the defeated party, and is difficult or impossible to accept as an act reflecting systematic application of legal principles.').

FN17. There is considerable stylistic variation in unpublished opinions even within circuits. In the Ninth Circuit, for example, it is not at all unusual to see lengthy unpublished opinions even in social security disability cases, in which the record is conscientiously and painstakingly reviewed, and the applicable regulations discussed and applied. See, e.g., Rose v. Heckler, No. 84-4246 (9th Cir. Feb. 19, 1986) (affirming denial of benefits after three- page discussion of facts and three-page discussion of standard of review, relevant regulations, and how regulations properly applied). On the other hand, even unpublished reversals in the Ninth Circuit can be relatively brief. See, e.g., Allsopp v. Schweiker, No. 84-6536 (9th Cir. Feb. 14, 1986) (two-page opinion reversing denial of disability benefits and discussing several errors in Social Security Appeals Council's consideration of evidence and procedures).

FN18. COMMISSION, supra note 9, at 52.

FN19. Most of the circuits disavow the precedential value of unpublished opinions through forbidding their citation as authority before the court. See infra text accompanying note 24.

FN20. D. STIENSTRA, supra note 2, at 15-22. Stienstra discovered that practices vary from circuit to circuit. Three circuits (the Fourth, Sixth, and Ninth) allow subscriptions to unpublished opinions despite the fact that the rules in these circuits appear to limit distribution of the opinions. Id. at 19.

FN21. Id. at 21.

FN22. Id. at 20.

FN23. The Tenth Circuit indexed the opinions for a while but has ceased this practice. Id. at 19-20 n.39; most courts simply file them chronologically in the clerk's office and list the outcomes of the cases on tables in the Federal Reporter. However, only three of the circuits that list the results in this manner include all unpublished decisions in the list sent to West Publishing Company for inclusion in the Reporter. Id. at 21. With the growth of legal databases such as Westlaw and Lexis, unpublished opinions may become more accessible. Both of these services have begun to include many unpublished opinions from federal appellate courts. However, the availability of unpublished dispositions through these databases is not uniform across the circuits.

FN24. 1ST CIR. R. 36.2(6); 2D CIR. R. 0.23; 4TH CIR. IOP 36.5; 5TH CIR. R. 47.5.3; 6TH CIR. R. 24(b); 7TH CIR. R. 53(b)(2)(iv); 8TH CIR. R. 8(i); 9TH CIR. R. 36.3; 10TH CIR. R. 36.3. The Eleventh Circuit has no formal rule, but in practice limits citation. D. STIENSTRA, supra note 2, at 22-23. Interestingly, the prohibition on citation of opinions extends only to citation within the circuit. Because citation rules are promulgated by each circuit independently, there is no prohibition on citing an unpublished case to a court in another circuit, and some attorneys who responded to the survey stated that they had done so. See infra text accompanying notes 78-84.

FN25. For a complete summary of the rules and practices of the circuits on citation, see D. STIENSTRA, supra note 2, at 51-52. She reports, based on interviews with court staff, that prohibitions on citation of unpublished materials have been effective in preventing attorneys from including the materials in briefs, although she notes that the courts have no way of determining whether the opinions are being used without citation. Id. at 24.

FN26. See supra text accompanying note 10.

FN27. See 1 Hearings Before the Commn. on Revision of the Federal Court Appellate System, Second Phase, 93d Cong., 1st Sess. 476 (1974) [hereinafter 1 Hearings] (letter of Dwight Opperman, President, West Publishing Co.).

FN28. See, e.g., 2 Hearings Before the Commn. on Revision of the Federal Court Appellate System, 94th Cong., 2d Sess. 1072 (1975) [hereinafter 2 Hearings] (testimony of Robert Stern); 1 Hearings, supra note 27, at 556 (testimony of Willard Lassers).

FN29. See 1 Hearings, supra note 27; 2 Hearings, supra note 28; COMMISSION, supra note 9.

FN30. See, e.g., 1 Hearings, supra note 27, at 557 (testimony of Willard Lassers).

FN31. Cf. 1 Hearings, supra note 27, at 452 (statement of the Bar Association of the Seventh Federal Circuit) ('If [unpublished opinions] can be cited, they become a source of secret law available to regular litigants before the court but not to the remainder of the bar, thereby giving unfair advantage to such regular litigants as the U.S. Attorney.').

FN32. Judge Sprecher noted this possibility in testimony before the Hruska Commission, 1 Hearings, supra note 27, at 536 (court might have to resort to 'some kind of an intracourt index of unpublished opinions . . . available for the court, even though they cannot be cited by the court or to the court'). In fact, many of the circuits do circulate all these opinions to judges within the circuit. D. STIENSTRA, supra note 2, at 18. Court staff attorneys often use the opinions as well, creating banks of opinions for future reference.

FN33. My discussion is based on research conducted in the Court of Appeals for the Ninth Circuit. Robel, Unpublished Opinions in the Court of Appeals for the Ninth Circuit (1988) (unpublished report to the United States Court of Appeals for the Ninth Circuit on file with author). This report was prepared for use by the court in assessing its limited publication plan. The author studied all opinions from that court, published and unpublished, during the period from October, 1986 through September, 1987.

FN34. One might argue that at least in those circuits which publish results of appeals in tables in the Federal Reporter, an attorney could in fact know the 'odds.' This is hardly possible, though, without an understanding of the subject matter of the litigation and the result in the district court.

FN35. Robel, supra note 33, at 19. Rates of publication in immigration may have been even lower in earlier years. Statistics compiled by the Administrative Office of the United States Courts reveal that the court published only 14% of the cases in which the INS was a party in 1983, and only 10% in 1984. Federal Judicial Center, Federal Court Cases, 1970-1984: Integrated Database (Inter-university Consortium for Political and Social Research, Appellate Terminations Database #8429) [hereinafter Database] (information about how figures were retrieved from database on file with author).

FN36. Robel, supra note 33, at 21. Of the unpublished reversals, all but one involved the appellate court's determination that benefits had been erroneously denied.

FN37. Id.

FN38. There is a good possibility that judges do sometimes use the publication plans for reasons not contemplated by the rules. I have found examples, for instance, of 'private' reprimands to district court judges and Assistant United States Attorneys. Consider the history of Kling v. County of Los Angeles, 633 F.2d 876 (9th Cir. 1980) (Kling I). Kling sued the City, alleging that she had been denied admission to nursing school because of a disability, in violation of federal law. After a hearing, the district court denied a preliminary injunction. The court of appeals reversed and remanded with instructions to enter the injunction. 633 F.2d at 876. After a trial, the same district court judge found that Kling was not handicapped within the meaning of federal law and that she had not suffered discrimination. On June 13, 1985, the Ninth Circuit reversed again in an unpublished opinion. No 83- 6193 (9th Cir. June 13, 1985). The court noted that the trial had not produced substantially different evidence from that considered in Kling I, that the district judge's conclusions were inconsistent and clearly erroneous, and that the judge's findings were directly contrary to the appeals court's decision in the first appeal. The court ordered the case remanded, to a different judge, for entry of damages. No. 83-6193. On July 16, 1985, the court amended its earlier opinion to take out language about the trial judge and (after certiorari was granted in the case) ordered the opinion published. Kling v. County of Los Angeles, 769 F.2d 532 (9th Cir. 1985) (Kling II). Finally, the Supreme Court summarily reversed the Ninth Circuit. County of Los Angeles v. Kling, 474 U.S. 936 (1985). In a dissent to the summary reversal, Justice Stevens argues that the use of unpublished opinions and summary reversals encourages 'decisionmaking without the discipline and accountability that the preparation of opinions requires.' 474 U.S. at 940.

FN39. Mada-Luna v. INS, 813 F.2d 1006, 1018 (9th Cir. 1987) (Hill, J., concurring).

FN40. Shoo Hwan Kim v. Meese, 810 F.2d 1494, 1498 (9th Cir. 1987) (Norris, J., dissenting).

FN41. See Robel, supra note 33, at 19-20. Twelve different members of the court wrote separately in unpublished opinions during this period. During previous periods, the rates may have been higher. Data for 1983 show thirty- nine concurrences in unpublished immigration opinions, and 1984 shows an astonishing eighty-four concurrences. Database, supra note 35.

FN42. Robel, supra note 33, at 20.

FN43. Id.

FN44. One of the respondents in the survey, discussed infra at notes 78-84 and accompanying text, candidly stated that one of the uses he makes of unpublished opinions is to monitor court trends and to gather intelligence about the judges.

FN45. R. POSNER, supra note 8, at 122 ('These are not frivolous appeals.'). My study of the Ninth Circuit's unpublished work confirms Posner's point. There were very few unpublished cases in which Rule 11 sanctions were imposed, with the exception of the cases involving tax protesters. Even using the term 'frivolous appeal' in a less technical sense, I do not believe that most of the cases decided through unpublished opinions raise no or few issues worthy of consideration, or even that their outcome is necessarily foreordained. Cf. Davies, Affirmed: A Study of Criminal Appeals and Decision-Making Norms in a California Court of Appeal, 1982 AM. B. FOUND. RES. J. 543, 582 (arguing that the concept of a 'frivolous appeal' is unhelpful in explaining high criminal appeal affirmance rate.)

FN46. See, e.g., 5TH CIR. R. 47.5.1 (publication appropriate when opinion 'applies an established rule of law to facts significantly different from those in previous published opinions applying the rule').

FN47. All of the publication plans require publication when an opinion 'establishes . . . a rule of law within [the] Circuit.' 4TH CIR. IOP 36.3.

FN48. See, e.g., 7TH CIR. R. 53 (opinion published if it 'constitutes a significant and non-duplicative [sic] contribution to legal literature (a) by a historical review of law, (b) by describing legislative history, or (c) by resolving or creating a conflict in the law').

FN49. This is the only category routinely included in the publication plans that, presumably, has nothing to do with assessment of precedential value. See, e.g., 7TH CIR. R. 53 (publication when opinion 'involves an issue of continuing public interest').

FN50. There may be, however, some reason to doubt this. Judge Posner has recently written that '[d]espite the vast number of published opinions, most federal circuit judges will confess that a surprising fraction of federal appeals are difficult to decide, not because there are too many precedents but because there are too few on point.' Posner, supra note 8, at 123.

FN51. 18 U.S.C. 1961-1968 (1982 & Supp. IV 1986).

FN52. Judge Burns, a district court judge sitting by designation on one of the panels that decided a RICO appeal during the period studied, noted, 'RICO is for me (and many, if not most, of my district court colleagues) an agonizingly difficult and confusing area of the law.' Sun Savings & Loan Assn. v. Dierdorff, 825 F.2d 187, 196 (9th Cir. 1987) (Burns, J., concurring).

FN53. The period of the Ninth Circuit study, Robel, supra note 33.

FN54. Id. at 16. Most of the published RICO opinions during this period were not particularly noteworthy. They include two per curiam decisions, Sigmond v. Brown, 828 F.2d 8 (9th Cir. 1987) and Televideo Sys., Inc. v. Heidenthal, 826 F.2d 915 (9th Cir. 1987) (a sanctions case more than a RICO case), and an opinion that announced a new rule on statutes of limitations in RICO cases, State Farm Mut. Auto. Ins. Co. v. Ammann, 828 F.2d 4 (9th Cir. 1987). Two published opinions might be characterized as RICO opinions, although RICO was not the focus of the discussion. See Volk v. D. A. Davidson & Co., 816 F.2d 1406 (9th Cir. 1987) (discussing securities fraud statute of limitations) and California Architecutral Bldg. Prods. v. Franciscan Ceramics, 818 F.2d 1466 (9th Cir. 1987) (short discussion of RICO 'pattern' requirement). Three of the published opinions during this period contain substantial discussions of RICO, Wilcox v. First Interstate Bank, 815 F.2d 522 (9th Cir. 1987) (applying recent Supreme Court case); Schreiber Distrib. Co. v. Serv-Well Furniture Co., 806 F.2d 1393 (9th Cir. 1987) (substantial discussion of pleading requirements); Sun Savings & Loan Assn. v. Dierdorff, 825 F.2d 187 (9th Cir. 1987) (applying recent Supreme Court case).

FN55. No. 86-6529 (9th Cir. July 17, 1987) (WESTLAW).

FN56. For a description of a 'Ponzi' scheme and the history of the scheme's namesake, see Cunningham v. Brown, 265 U.S. 1 (1923).

FN57. No. 83-5709 (9th Cir, Aug. 13, 1987) (WESTLAW).

FN58. See Schreiber Distrib. Co. v. Serv-Well Furniture Co., 806 F.2d 1393 (9th Cir. 1987) (discussing pleading requirements).

FN59. See, e.g., Judge Burns' comments, supra note 52.

FN60. No. 85-6468 (9th Cir. Nov. 3, 1986) (WESTLAW).

FN61. It could be argued that the Ninth Circuit's rule on publication would be expected to provide for the publication of opinions like these, since they could be deemed to 'clarify' rules of law. 9TH CIR. R. 36-2(a). The RICO examples illustrate, however, a shortcoming in the criteria for publication: the distinctions among rule development, rule clarification, and rule application are too fine for routine use. So, some cases that clarify rules will be relegated to unpublished opinions on the assumption that they merely call for application.

FN62. If publication rules operated neutrally (i.e., without regard to the subject matter of cases), then one would expect that the rate of publication would remain constant across all categories of cases. It does not. In the Ninth Circuit during the period of the study, supra note 33, the court published an average of 37% of its opinions. But the rates within subject areas varied widely. The court published 70% of the cases involving securities law during this period, for instance, but only 10% of the civil rights cases filed by prisoners. Ignoring higher than average rates of publication, I assumed that lower than average rates of publication were not the result of random fluctuation for the year examined in the study. This assumption was confirmed by examining rates of publication across subject areas for different years. (The data for comparison, compiled by the Administrative Office of the United States Courts, are available through the Institute for Social Research at the University of Michigan.) Because the rates remain fairly constant from year to year, I assumed that the higher than average rates were the result of more systematic factors.

FN63. The exception to this is diversity cases. During the study period, the Ninth Circuit published only 38% of its diversity opinions. That 23% of those cases were reversals suggests that there are difficult issues involved in unpublished diversity opinions. Because federal interpretations of state law rules are by nature provisionslly authoritative and because judges realize that their opportunity to interpret that law rests on the chance of diversity jurisdiction, judges may well suppose that the need to publish these opinions is not great.

FN64. Robel, supra note 33, at Table 1. During the study period, the Ninth Circuit's publication rates in these areas were as follows: immigration 24%; social security 31% (disability review only 26%); Federal Tort Claims Act 37%; criminal appeals 37%; state prisoner, habeas corpus 34%; civil rights 29%; employment discrimination against federal government 25%.

FN65. Judge Rubin notes, for instance, that not many highly qualified people would want to be federal judges if the job involved a lot of social security and personal injury cases and nothing more. Rubin, Bureaucratization of the Federal Courts: The Tension Between Justice and Efficiency, 55 NOTRE DAME LAW. 648, 657 (1980).

FN66. Judge Posner argues that the 'conventional argument against limited publication is flawed' because it fails to recognize that 'the preparation of unpublished opinions is delegated to law clerks and staff attorneys . . ..' R. POSNER, supra note 8, at 124. Posner may not be describing a practice uniform in all the circuits. Professor Hellman, who served as the supervising staff attorney at the Ninth Circuit, writes that staff drafting of opinions 'remained the exception rather than the rule' during his tenure with the court. Hellman, Central Staff in Appellate Courts: The Experience of the Ninth Circuit, 68 CALIF. L. REV. 937, 982 (1980). See J. CECIL & D. STIENSTRA, DECIDING CASES WITHOUT ARGUMENT: A DESCRIPTION OF PROCEDURES IN THE COURTS OF APPEALS 24-25 (1985) [hereinafter DECIDING CASES I] (describing staff attorney functions in providing draft opinions in nonargument cases).

FN67. J. CECIL & D. STIENSTRA, DECIDING CASES WITHOUT ARGUMENT: AN EXAMINATION OF FOUR COURTS OF APPEALS 31 (1987) [hereinafter DECIDING CASES II]. In the Fourth Circuit, for example, only 1% of cases not argued before the court are decided by published opinion. The highest such percentage is 22%, in the Fifth Circuit. Id.

FN68. In the Fifth Circuit, for instance, staff attorneys responsible for the initial determination whether to recommend argument routinely screen

prisoner cases with and without counsel, section 2255 cases with and without counsel, civil federal question cases, civil cases in which the United States is a party (e.g., federal tort claims act cases, bankruptcy cases, and [federal] agency cases other than tax cases), civil rights cases other than title VII, and Social Security cases.

DECIDING CASES I, supra note 66, at 23.

FN69. The benefits of screening cases for expedited review are lost if judges must reevaluate every initial screening decision.

FN70. See DECIDING CASES II, supra note 67, at 55-56, 82-83, 103-04 (judges reject initial determinations by staff in only 10-20% of cases).

FN71. The statements to this effect by judges are legion. See, e.g., F. COFFIN, THE WAYS OF A JUDGE 57-58 (1980). Thus, the decision not to publish an opinion because the appeal is not complex might be self-fulfilling prophecy. See R. POSNER, supra note 8, at 124.

FN72. D. STIENSTRA, supra note 2, at 32.

FN73. See supra note 13.

FN74. In the most recent year for which figures are available, the average percentage of unpublished opinions for all the federal appeals courts was 61%. Two courts still publish a majority of their opinions, the Seventh Circuit (34.2% unpublished) and the First Circuit (37.7% unpublished). The Fourth Circuit left an astonishing 80.2% of its opinions unpublished, with the runner- up for taciturnity being the Sixth Circuit, which left 77.6% unpublished. Figures are available from Federal Judicial Center Integrated Data Base (includes information about all federal district and appellate court cases active during the period from Statistical Year (SY) 1970-SY 1987). If the substance of the workload does not differ substantially among these courts, one could surmise that more significant information might be available in unpublished form in a low-publication rate circuit than in, say, the Seventh Circuit.

FN75. Marc Galanter has described many of these advantages in Galanter, Why the 'Haves' Come out Ahead, 9 LAW & SOCY. REV. 95 (1974).

FN76. This advantage will remain, I suspect, even with the advent of increased electronic reporting of these opinions through the commercial databases, since access to these databases is still relatively expensive. It is possible that in some areas the 'one-shotters' may be represented by a cohesive bar that could share these opinions or induce a publisher to publish them. There has been very little research regarding information sharing by practitioners. Most of the information I found about this topic centered on mass-tort litigation groups. See, e.g., Ranii, How the Plaintiffs' Bar Shares Its Information, Natl. L.J., July 23, 1984, at 1; McHugh, Specialties Create Need for Exotic Publications, Chi. Daily L. Bull., Mar. 21, 1983, at 1.

FN77. See supra note 31 and accompanying text.

FN78. The text of the survey is as follows:

Questionnaire on Unpublished Decisions of the U.S. Courts of Appeals

1. Are unpublished opinions circulated to attorneys within the office?

Are they circulated to regional offices?

Anyone else?

2. Are unpublished opinions filed?

If yes, are they filed with other material concerning the case in which they are issued?

Are they also filed separately?

Are they indexed?

3. Are unpublished opinions ever reviewed systematically within the office?

4. Are unpublished decisions ever consulted or used in any of the following ways?

a. in making litigation and settlement decision

b. in writing briefs

c. in determining whether to appeal or contest an appeal

d. other

5. Has this office ever moved for publication of unpublished decisions pursuant to court rules that allow for such motions?

If yes, who makes the determination to move for publication? 6. Comments

(Completed questionnaires are on file with author.).

FN79. During the period of this study, Robel, supra note 33, the Ninth Circuit published only 31% of the opinions in which the Department of Health and Human Services was a party, 24% of the opinions in which the Immigration and Naturalization Service was a party, and 39% of the decisions in which the National Labor Relations Board was a party. In non-agency review civil cases in which the U.S. was a party, the court published 26% of its opinions. Id. at Table 1.

FN80. See infra text accompanying notes 81-83.

FN81. The Office of Consumer Litigation may be unique among the offices responding to the survey in that it deals primarily with client offices like the Food and Drug Administration. It also is involved in comparatively few appeals. In FY 1987, for example, the Office opened 21 cases on the appellate level. Telephone interview with Ken Maddox, Office of Consumer Litigation (Aug. 4, 1988).

FN82. Immigration Litigation, NLRB, and the Civil Rights Division file the opinions separately.

FN83. The Attorney General (usually through the Office of the Solicitor General) is responsible for all final determinations concerning appeals by the United States. 28 U.S.C. 516 (1982). However, the Solicitor General has historically depended on recommendations of staff in the appellate sections of the Justice Department, and, in some instances, upon recommendations from agencies, in making appeals determinations. This has especially been the case in appeals to the courts of appeals. See Carrington, United States Appeals in Civil Case: A Field and Statistical Study, 11 HOUS. L. REV. 1101 (1974).

FN84. Galanter, supra note 75, at 100. See also Carrington, supra note 83, at 1102 (United States is a 'cautions and successful' litigant that approaches appeals decisions from perspective of rule development).

FN85. Correspondence with court clerks (on file with author).

FN86. Seventh Circuit opinions that are first released as unpublished opinions and are then published carry the following legend: 'This case was previously decided by unpublished order according to Circuit Rule 35.' Although these opinions can be identified, it is not always possible to tell whether the court itself later published the opinion or whether one of the parties moved for publication. The Fourth and Eleventh Circuits also indicate when an opinion has changed status, and in the District of Columbia Circuit, the authoring judge has the option of indicating a change in status. None of the other circuits routinely indicate when changes occur. Correspondence with court clerks (on file with author).

FN87. The Executive Office of the United States Attorneys requires all surveys of U.S. Attorneys to receive approval from the Executive Office before attorneys respond. When I became aware of the rule, I asked that the Executive Office allow the attorneys to reply to my survey, but I was told that the office had determined that responses would not be in the best interest of the Justice Department. I was also forwarded a copy of an order from the Executive Office to all U.S. Attorneys that requested that they not cooperate with me. Letter from Jason Green, Executive Office of United States Attorneys (Mar. 20, 1987) (on file with author).

FN88. R. POSNER, supra note 8, at 123 ('almost by definition, all opinions have some actual or precedential value').

FN89. And, he argues, given caseload pressures, it cannot be. Id. at 124.

FN90. Id.

FN91. Id. at 108-13.

FN92. Id. at 113.

FN93. See supra note 74.

FN94. This lack of involvement raises a host of issues about the credibility of the courts. For a poetic example of some aspects of the debate, see J. VINING, THE AUTHORITATIVE AND THE AUTHORITARIAN (1986).

FN95. See supra notes 24-32 and accompanying text.

FN96. Since there is no rule that prohibits anyone from publishing these opinions now, one might ask whether the publishing companies have already made a determination that no one wants access to these opinions. In fact, the contrary is true. Unpublished opinions are often 'reported' on the LEXIS and WESTLAW computer databases and in some specialty reporters, such as BNA's Labor Relations Reference Manual. However, the decisions about which opinions to report and whether to report them are made on an ad hoc basis by various companies, with the result that significant amounts of information remain unavailable. Because of the special relationship between the federal courts and West Publishing Company (the official reporter for the federal courts), I believe West would offer the unpublished cases were the courts to tender them for publication.

Of course, one of the goals of limited publication plans was to reduce the bar's cost of research by excluding sterile decisions from the research pool. If the operation of the plans cannot be trusted to have this result, as I argue, then decisions about how heavily to research this new source of information should be left to attorneys.

FN97. Cf. Davies, supra note 45, at 591-619 (discussing role of standard of review in affirmance rates).

FN98. Nor has much thought been given to how judges actually use the plans. For instance, it has been known for some time that publication plans containing presumptions for publication do not produce high publication rates nor do plans that contain presumptions against publication necessarily produce low publication rates. Reynolds & Richman, Limited Publication in the Fourth and Sixth Circuits, 1979 DUKE L.J. 807, 815. More systematic study of the actual factors that affect publication is needed.